1

MOTIVATION BELIEFS OF INDO-CANADIAN AND ANGLO-CANADIAN EARLY

ADOLESCENTS: A CROSS-CULTURAL INVESTIGATION OF SELF- AND
COLLECTIVE EFFICACY
by
Robert M. Klassen
M.A., University of British Columbia, 1994
THESIS SUBMITTED IN PARTIAL FULFILLMENT OF
THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
in the Faculty of Education
Robert M. Klassen 2002
SIMON FRASER UNIVERSITY
DECEMBER 2002
All rights reserved. This work may not be
reproduced in whole or in part, by photocopy
or other means, without permission of the author.
2
Abstract
This study examined the self- and collective efficacy beliefs of Indo-Canadian and Anglo-
Canadian early adolescent students. The research participants included 112 Anglo-
Canadian and 158 Indo-Canadian (children of Punjabi Sikh immigrants) grade 7 students.
On a 22-item measure of math performance, the Indo-Canadian students earned a
significantly higher score than the Anglo-Canadian students and also rated their self-
efficacy at a higher level. There were no differences between the groups in terms of
calibration of self-efficacy and performance. In a multiple regression analysis, self-
efficacy was the only motivation variable that predicted math performance for both
groups. For the Anglo-Canadian students previous math grade was the only other
significant predictor of performance; for the Indo-Canadian students, math self-concept
was the other significant predictor. Of the hypothesized four sources of efficacy beliefs,
emotional arousal was the strongest predictor of efficacy for both cultural groups. Past
performance was the next strongest predictor for Anglo-Canadians, while for the Indo-
Canadian students, vicarious experience was the second significant predictor of self-
efficacy. For the sample as a whole, the students who were most accurate in their
calibration performed at a higher level than students who under-estimated or over-
estimated their performance. For the group task, collective efficacy was the best predictor
of group performance for the Anglo-Canadians, whereas previous math grade predicted
group performance more strongly for the Indo-Canadians. Counter to predictions of
higher overall collectivism, the Indo-Canadians displayed higher levels of vertical
individualism and vertical collectivism than the other group. Implications of the influence
of the vertical aspect of I/C on self-appraisal are discussed.
3
Motivation beliefs of Indo-Canadian and Anglo-Canadian early adolescents:
A cross-cultural investigation of self- and collective efficacy
Introduction
Bandura claimed Among the mechanisms of personal agency, none is more
central or pervasive than peoples beliefs about their capabilities to exercise control over
events that affect their lives (1989, p. 1175). Self-efficacy has been defined as beliefs in
ones capabilities to organize and execute the courses of action required to produce given
attainments (Bandura, 1997, p. 2). These perceptions of self-capabilities or self-efficacy
have been identified as key factors affecting thought patterns and performance in a wide
variety of tasks. In fact, Bandura (1997) suggested that motivation, affect, and actions are
more strongly influenced by self-beliefs than by what is objectively true. Self-beliefs of
ones efficacy influence choice of activity, task perseverance, level of effort expended, and
ultimately, degree of success achieved. Inaccurate estimates of self-efficacy may develop
from misunderstandings of tasks or from a lack of self-knowledge (Bandura & Schunk,
1981). Beliefs of inefficacy hamper ones functioning: students who doubt their academic
efficacy, for example, set lower performance goals, and more readily give up when faced
with challenging tasks or difficulties. Conversely, students who have similar abilities but
possess optimistic self-efficacy beliefs set more challenging goals and persist in the face
of obstacles (Pajares, 1996a).
Considerable recent research converges on the idea that many psychological
processes, previously thought to be universally applicable, are, in fact, culturally
4
contingent, with the result that many of the findings currently regarded as basic to
psychology are a function of particular cultural frameworks (Fiske, Kitayama, Markus,
& Nisbett, 1998, p. 916). Triandis (1996) notes that psychological theories are often built
upon unexamined cultural assumptions, with the result that our theories often reflect our
culture (p. 407). Pajares (2000) calls for a culturally attentive educational psychology
that investigates human functioning in a range of social and cultural settings. Considerable
research has been devoted to the study of self-efficacy beliefs in education (e.g., Pajares &
Graham, 1999; Schunk, 1991; Zimmerman, 2000) and other diverse fields, but to this
point, most of the work has involved Western, typically American, participants. Even
though self-efficacy has been shown to be a strong predictor of academic functioning with
this particular population, less is known about how self-efficacy beliefs operate with
non-Western cultural groups. Pajares submits that cross-cultural research can help clarify
how different cultural practices shape efficacy beliefs, and how efficacy beliefs might
operate as a function of culture (2000).
An understanding of cultural conceptions of self and self-beliefs is a critical step in
gaining an understanding of the functioning of diverse populations in a multicultural
society. Formation of self-beliefs, like self-efficacy, is influenced by culturally held
values, assumptions, and beliefs (Markus & Kitayama, 1991). Fiske et al. (1998) claim
that culture and the psyche are mutually constitutive, resulting in dramatic divergence in
psychological functioning between European-Americans and East Asians (p. 916).
Again, with references to East Asian individuals, Markus and Kitayama (1991) propose
that self-phenomena may assume a different form according to the culturally influenced
5
relationship of self with others. Although the emerging findings from initial cross-cultural
research of self-beliefs have suggested that these beliefs operate differently for different
cultural groups, the research up to this point has explored only a limited number of
cultural groups in a restricted range of settings.
This thesis examines self-efficacy from a cross-cultural perspective; more
specifically, the thesis investigates the mathematics efficacy beliefs of early adolescents
from South Asian Indo-Canadian immigrant and Anglo-Canadian non-immigrant
backgrounds. In this chapter, I begin by describing self-efficacy, its formation or sources,
and the calibration of efficacy beliefs and performance. Vertical and horizontal dimensions
of individualism and collectivism are used to frame an examination of cultural influences,
and differences in the concept of self in individualist and collectivist cultures are
discussed, as is the role played by collective efficacy in human functioning. Next, I report
Banduras defense of how self-efficacy remains valid in non-Western collectivist
settings. Following this, I provide an overview of many of the studies published that
explore self-efficacy beliefs using cross-cultural comparison groups, and I highlight
themes and problems across the studies and speculate about gaps in that body of
research. Subsequently, I briefly explore psychological and anthropological research
examining a specific cultural groupIndian Sikhs who have migrated from the state of
Punjab to Canada, the U.S., and England. A brief discussion of cross-cultural research
methodology ensues, after which specific elements and research questions of this thesis
are listed.
6
A central question addressed in this thesis is If conception of self is markedly
different in non-Western cultures, and self-beliefs reflect these differences, are self-efficacy
beliefs relevant in all cultures? Specifically, do efficacy beliefsself- and
collectiveoperate differently in Indo-Canadian as compared to Anglo-Canadian early
adolescents?
Self-Efficacy
Self-efficacy as a motivational factor has been frequently examined subsequent to
Banduras 1977 seminal publication Self-efficacy: Toward a Unifying Theory of
Behavioral Change. Efficacy beliefs are context-specific evaluations of the capability to
successfully complete a task, and are formed through mastery experiences (past
performance), vicarious experiences (observation of others), social/verbal persuasion, and
interpretations of physiological and emotional states (Bandura, 1995). These beliefs
contribute to prediction of academic outcomes beyond the contributions offered by
ability, previous attainments, knowledge, and skill alone. According to Bandura (1993),
students need more than ability and skills in order to perform successfully; they also need
the sense of efficacy to use them well and to regulate their learning.
Self-efficacy beliefs differ from related constructs such as competence beliefs and
self-concept in that they are more task-specific, and are established through normative
criteria rather than through comparison with others (Bong & Clark, 1999; Zimmerman,
1995). For example, scales measuring math self-concept or math competence might ask
Are you good at math (compared to others in your class)?, whereas a scale measuring
math self-efficacy might contain the item How confident are you in solving a 2-digit
7
multiplication question? Conceptually, self-concept consists of both cognitions and
related feelings, whereas self-efficacy tends to be comprised of cognitions or beliefs about
capabilities (Bong & Clark, 1999). Efficacy beliefs play a role in managing motivation in
expectancy-value theory, which asserts that individuals evaluate courses of behavior for
their value or potential to produce certain outcomes. An expectancy-value item might ask
How useful is it to learn fractions? Shell, Murphy, and Bruning (1989) found that
adding a self-efficacy component significantly increased the predictiveness of expectancy-
value constructs. Self-efficacy beliefs, then, consist of the degree to which individuals
believe they can control their level of performance and their environment (Bandura, 1997).
In spite of Banduras (1997, p.6) caution that self-efficacy beliefs may vary
across realms of activity, different levels of task demands within a given activity domain,
and under different situational circumstances much of the self-efficacy research
conducted has been more global in scope (Pajares, 1996b). Although there has been much
research examining the self-efficacy of college undergraduates (e.g., Pajares & Johnson,
1994; Shell, Murphy, & Bruning, 1989; Zimmerman & Bandura, 1994) and some research
exploring the efficacy of younger children (e.g., Pajares & Valiante, 1997; Schunk &
Swartz, 1993), there has been less work charting the development of the self-efficacy of
early adolescents (Anderman & Maehr, 1994) and findings from different situational
circumstances may not hold true for early-adolescent students. In the place of research
investigating global self-beliefs, Pajares (1996a) calls for further studies of self-efficacy
investigating specific settings: More information is also required about how students at
various ages, academic levels, or grades use the diverse sources of efficacy information in
8
developing self-efficacy beliefs (p. 567). Self-efficacy is conceptualized as a context-
specific belief, and research exploring this motivational belief is needed in a variety of
specific contexts.
Sources of Self-Efficacy
Self-efficacy beliefs are formed from four main sources. First, mastery or enactive
experiences (sometimes referred to as past performance) provide direct evidence that one
has been capable of performing a certain task. Bandura (1995, 1997) proposes that
mastery experience is in most cases the strongest source of efficacy beliefs. A person
might learn through vicarious experience, which refers to evidence from similar others that
the task is manageable. Observation of successful others raises the chance of success,
whereas observation of failures undermines motivation. Third, social persuasion can
strengthen efficacy beliefs and persuade people that a task can be successfully completed.
Bandura (1997) notes that social persuasion alone will not boost self-efficacy beliefs for
long. Finally, self-efficacy beliefs are formed through an analysis of ones physiological or
emotional state. People interpret physiological signs of stress, for example, as evidence
that they may have difficulty completing a specific task. Studies that investigate the
sources of self-efficacy increase understanding of how efficacy beliefs are formed and may
lead to interventions designed to improve performance.
Hampton (1998) examined the sources of academic self-efficacy beliefs of upper-
level high school students with and without learning disabilities. She found that neither
social persuasion nor physical/emotional arousal contributed significantly to predictions
of mathematics self-efficacy beliefs. In a study investigating the mathematics self-efficacy
9
of Japanese undergraduate students, Matsui, Matsui and Ohnishi (1990) found
significantly higher self-efficacy scores for men, but no sex differences in the four self-
efficacy sources. Similar to the findings in Hamptons (1998) study, verbal or social
persuasion did not make a unique contribution to self-efficacy. Anderson and Betz (2001)
investigated the factor structure of the sources of self-efficacy and found one cluster
comprised of direct experience factorspast performance, emotional arousal, and social
persuasionand a second cluster reflecting indirect experiencevicarious experience.
Past performance and emotional arousal were the strongest predictors of social self-
efficacy in the undergraduate sample while vicarious experience did not add to prediction
of the criterion. In contrast, Lent, Lopez, and Bieschke (1991) found past performance
was the only source variable that contributed to prediction of undergraduates
mathematics self-efficacy. The researchers hypothesized that the three non-predictive
source variables might become less salient as students become more experienced in a
certain field; that is, social persuasion, vicarious experience, and emotional arousal may
provide secondary or even redundant information to students who have developed an
awareness of their academic skills and abilities through extensive enactive experience.
Although some research has investigated the sources of efficacy beliefs, the
universality of findings is open to question. Most of the studies conducted examine self-
efficacy sources for undergraduate or senior high school students without reference to
cultural or ethnic backgrounds of the participants. Cultural context or cultural dimension
(like individualism and collectivism) may play a key role in influencing how self-efficacy
beliefs are formed in diverse settings. Of particular interest for this thesis is an
10
investigation into the sources of efficacy beliefs for different cultural groupshow
enactive experience, vicarious experience, verbal persuasion, and physiological/emotional
status contribute to the formation of efficacy beliefs for early adolescent students.
Oettingen (1995) proposes that cultural differences on the collectivism/individualism
continuum may affect the appraisal of the information used to determine self-efficacy
beliefs. To illustrate how a specific cultural dimension might influence the sources of
efficacy beliefs, Oettingen offers the example of schoolchildren in a collectivist setting
who are given feedback for their individual performance (as would be the case in
individualist societies) but also feedback about their in-group or collective performance.
Further, she suggests that sources are differentially valued according to cultural
dimension, with individual attainment being more highly valued in individualist settings. A
few studies have investigated the sources of mathematics self-efficacy (Lent et al., 1991;
Lopez & Lent, 1992; Matsui et al., 1990) but none have used a cross-cultural perspective.
Calibration
Calibration addresses the accuracy of ones beliefs about potential functioning. In
the measurement of self-efficacy for academic functioning, students might be asked to rate
how confident they are that they can perform a certain academic task. For example,
Graham and Harris (1989) measured students self-efficacy for writing essays using a 5-
item, 10-point scale that asked about perceived ability to write an essay with a good
beginning, that gave three reasons to support the premise, and had a good ending (p.
206). Students then responded on a scale that ranged from not sure (10) through
maybe (40), pretty sure (70) to real sure (100) in 10-point increments. The
11
calibration of beliefs with performance is assessed by comparing mean efficacy ratings
with task performance. In this example, the students displayed a low level of essay-
writing performance, although measured self-efficacy beliefs showed a mean of close to 70
or pretty sure (i.e., pretty sure that I can write a good essay). These assessments of
self-efficacy can be viewed as a function of metacognitive knowledge, and are derived
from the understandings that students build about themselves (Butler, 1998). In the area
of writing, and to a certain extent, reading, the comparison of self-efficacy ratings with the
performance criteria is indirect; typically, researchers construct self-efficacy scales that
reflect specific elements of the criterial task, and the self-efficacy score is a composite of
the students confidence to complete these individual elements.
In the mathematics domain, self-efficacy assessment can be measured at the
domain, or item, level. Domain level efficacy questions might ask How confident are you
that you can correctly do math? For item level efficacy, students are briefly shown a
math question, and then queried about their confidence to complete the actual, or a
similar, item. Pajares and colleagues (e.g., Pajares & Graham, 1999; Pajares & Miller,
1997) presented students with a number of math problems, and asked students to
estimate their confidence to solve the problems. Calibration was measured through mean
bias (which reflects the direction of judgment errors) and mean accuracy (which reflects
the magnitude of judgment errors). Pajares and Graham (1999) found gifted students
displayed better calibration (less biased towards overconfidence, and more accurate) than
regular education students. In a review of studies investigating the self-efficacy beliefs of
students with learning disabilities, Klassen (2002a) found a higher degree of calibration
12
accuracy in studies of mathematics functioning, whereas studies investigating the domain
of writing reported more frequent mis-calibration (predominantly overconfidence) of
efficacy beliefs.
According to Bandura (e.g., 1995) reasonably optimistic efficacy beliefs are
essential for optimal functioning. Realistic efficacy beliefs, in fact, are claimed to hinder
significant accomplishments:
One rarely finds realists in the ranks of innovators and great
achievers.... The successful, the venturesome, the sociable... and
the innovators take an optimistic view of their personal
capabilities to exercise influence over events that affect their
lives.... such (optimistic) personal beliefs foster positive well-
being and human accomplishments. (1995, p. 13)
In investigations of academic self-efficacy, there are many examples of overconfidence to
complete academic tasks (e.g., Pajares, 1996a). Optimistic estimates of ones efficacy are
hypothesized to increase effort and persistence, and to promote accomplishment in
challenging circumstances (Bandura, 1997). However, in collectivist settings, efficacy
predictions have been found to be typically lower and more accurate (e.g., Oettingen,
1995). It is possible that the optimistic bias remarked on by Pajares (1996) may not
apply to some non-Western cultural groups. Investigation into the accuracy of the
calibration of efficacy beliefs with subsequent performance helps illuminate the
differential role that self-efficacy might play for people from non-Western cultural
backgrounds.
13
Individualism, Collectivism, and other Cultural Dimensions
The cultural dimensions of individualism/collectivism refer to the degree of
separateness and connectedness of individuals and groups (Markus & Kitayama, 1991;
Triandis, 1995). Individualist cultures tend to emphasize I consciousness,
independence, individual initiative and right to privacy (Kim, Triandis, Kagitcibasi, Choi,
& Yoon, 1994). In contrast, collectivist cultures have a tendency to stress we
consciousness, collective identity, group solidarity, and duty (Kim et al., 1994). It is
important to note that these categories, as conceptualized by Hofstede (1980) and others,
were originally viewed as polar ends of the same continuum (Coon & Kemmelmeier,
2001). However, individualism and collectivism have more recently (e.g., Triandis, 1996,
2001a) been understood as discrete dimensions which exist on separate continua, with
considerable intra-group and even intra-individual variation, according to changing settings
(Sinha & Tripathi, 1994). Triandis emphatically states, Individualism and collectivism
are not opposites. They must be conceived as multidimensional constructs (2001a, p.
38). Furthermore, the dimensions of I/C are contextual: people may behave very
differently depending on the setting and whether others are from an in-group or out-
group. The individualism and collectivism concepts applied to individuals are termed
idiocentrism and allocentrism, respectively; this terminology allows for discussion of
idiocentrics in collectivist culture, and allocentrics in individualist cultures (Triandis,
1990).
The results from recent research have suggested that efficacy beliefs operate
differently in different cultures (e.g., Earley, Gibson, & Chen, 1999; Schaubroeck, Lam, &
14
Xie, 2000). An individuals sense of self is, to some extent, culture-specific, with
different cultural groups emphasizing varied aspects of self-functioning (Markus &
Kitayama, 1991). Asian cultures have been described as tending toward collectivism and
Western cultures, especially North American culture, as strongly individualistic. Indian
culture has been described as collectivist (Laungani, 1999; Verma & Triandis, 1999) as
well as mixed (Mishra, 1994; Sinha & Tripathi, 1994), and although no studies have
investigated the efficacy beliefs of immigrant Indian groups (such as Indo-Canadians)
other collectivist immigrant groups have been shown to retain characteristics of their
culture of origin (e.g., Asian Americans in Coon & Kemmelmeier, 2001; Eaton & Dembo,
1997, Whang & Hancock, 1994).
The dimensions of individualism and collectivism have been further refined with
the additional overlay of vertical and horizontal elements (Triandis, 1996), resulting in the
four dimensions of vertical individualism, horizontal individualism, vertical collectivism,
and horizontal collectivism (Singelis, Triandis, Bhawuk, & Gelfand, 1995; Triandis &
Gelfand, 1998). The vertical component of collectivism and individualism, which is
related to Hofstedes (1980) concept of power distance, stresses hierarchy, inequality,
power, status, competition, conformity, and authoritarianism, whereas the horizontal
component stresses equality, self-direction, benevolence, universalism, and sociability
(Singelis et al., 1995; Soh & Leong, 2002; Triandis, 1996). Individualism and collectivism
are considered to be independent (or co-existent, see Sinha & Tripathi, 1994) dimensions;
individuals and cultural groups might score high or low on each dimension, or have
divergent scores on the two dimensions (Triandis, 1996; Triandis & Gelfand, 1998).
15
Previous I/C scales have captured one or other of the vertical and horizontal elements,
resulting in measures of individualism that capture only self-reliance (horizontal
individualism, or HI) but not competition (VI); or interdependence and sociability (HC)
but not family integrity (VC) (Triandis & Gelfand, 1998). Soh and Leong (2002) found
support for the cross-cultural validity of the four constructs, but mixed support for how
the constructs were operationalized in Triandis and Gelfands (1998) 16-item measure.
The addition of the vertical and horizontal elements adds descriptive sophistication and
explanatory power to the well-used dimensions of individualism and collectivism.
There has been recent criticism of the I/C dimensions. Fiske (2002) argues that
individualism is based on how Americans wish to perceive themselves, whereas
collectivism is a simplistic representation of the antithetical other (p. 84). The use of
self-report and surveys to measure cultural dimensions has been questioned (Bond, 2002;
Kitayama, 2002) and characterized as distorted and biased (Fiske, 2002). Oyserman,
Coon, and Kemmelmeier (2002) conducted a meta-analysis and qualitative review of all of
the English-language I/C research, and concluded that there is evidence that these cultural
dimensions do influence psychological processes, but the empirical basis for this
conclusion is not as firm as might be desired (p. 43). Furthermore, they found most of
the research (about 80%) was based on (likely unrepresentative) undergraduate samples,
and that cross-national differences were too readily accepted as evidence of I/C
differences.
In a review of 15 studies comparing Americans and Japanese on I/C measures,
Takano and Osaka (1999) found only one study supported the expected pattern of
16
Japanese collectivism and American individualism, nine studies showed no significant
differences, and five studies revealed differences in the opposite direction. Matsumoto
(1999) reviewed studies exploring I/C between the U.S. and Japan and found 17 of 18
studies did not support the expected stereotypic findings of American individualism
and Japanese collectivism. After reviewing much of the I/C research comparing these two
countries, Matsumoto asserts The evidence available to date, therefore, overwhelmingly
indicates that the Japanese are not more collectivist than Americans; if anything, in some
cases the Japanese are more individualistic than Americans (1999, p. 298). In order to
address the complexity of cross-cultural self-phenomena, Matsumoto recommends the
adoption of qualitative and multimethod approaches. Hong and Chiu (2001) advise that
cross-cultural differences in social behavior may be most valid when researchers
contextualize their work, and focus on context-specific cultural beliefs. In response to
these criticisms, the I/C framework continues to be refined (e.g., including horizontal and
vertical dimensions of individualism and collectivism), and continues to have a large
impact on psychology (Triandis, 2001a).
Individualism and collectivism are not the only possible ways of providing a
framework to the construct of cultureother cultural dimensions have been proposed.
Hofstedes (1980) factor analysis of his massive data set gathered from over 100,000
workers in over 60 countries led to the proposal of four primary cultural dimensions:
Individualism/Collectivism, Power distance, Uncertainty avoidance, and
Masculinity/Femininity. His work continues to be used to provide baseline data for
researchers interested in contextualizing their work (Bond, 1994). Laungani (1999)
17
proposed four core values or cultural dimensions to separate Western from Eastern,
specifically Indian, cultures: Individualism/Communalism; Cognitivism/Emotionalism;
Free will/Determinism; and Materialism/Spiritualism. Laungani cautioned that people
seldom fit into neat theoretically formulated and/or empirically derived categories (p.
193) and posited that dimension shifts along the continuum take place for groups and for
individuals. There have been attempts to link the contrasts of agency and communion (see
Gilligan, 1982) with individualism and collectivism, but with mixed empirical results
(Bond, 1994). Of all these and other possible cultural dimensions, it is the
individualism/collectivism dimension that has most frequentlyand successfullybeen
adopted to provide structure to the challenging concept of culture (Kim et al., 1994).
Collective Efficacy and Sampling Aspects of Self
Bandura makes reference to a group-level efficacy beliefcollective efficacythat
may have implications for differential cross-cultural functioning. Collective efficacy is
defined as a groups shared belief in its conjoint capabilities to organize and execute the
courses of action required to produce given levels of attainments (1997, p. 477).
Collective efficacy is not defined by Bandura as a collectivist substitute for self-efficacy,
but rather as a separate, group-oriented attribute that acts in addition to self-efficacy, and
is rooted in self-efficacy (1997, p. 480). In terms of measurement, some researchers
have made a distinction between group efficacy, which refers to a groups consensus
about ability to perform a task, and collective efficacy, which is a measure of each
group members beliefs about the groups ability to perform a task (e.g., Earley, 1999;
Mischel & Northcraft, 1997). Also, measurement of collective efficacy can vary much in
18
terms of specificity and correspondence, ranging from item-level collective efficacy
(How confident are you that your group can solve this task?), to more global collective
efficacy (The members of my group work well together as suggested by Riggs, Warka,
Babasa, Betancourt, and Hooker, 1994). Variations in collective efficacy measurement
may have a significant impact on the way in which collective efficacy operates in
comparison to other variables predicting performance.
Much of the previous research has investigated collective efficacy in business
settings (Mischel & Northcraft, 1997; Riggs et al., 1994; Schaubroeck et al., 2000).
Recently, research has been conducted investigating the collective efficacy of schools as a
unit (e.g., Goddard, 2001; Goddard, Hoy, & Woolfolk Hoy, 2000); but few, if any studies
have investigated collective efficacy of small groups in a school setting. Other units of
collectiveness that have been investigated include classroom, work unit, sports team,
and cultural group (Bandura, 1997). As will be seen below, a few researchers (Earley,
1994; Earley, Gibson, & Chen, 1999; Schaubroeck et al., 2000) have investigated the role
of self-efficacy and group functioning, and collective efficacy from a cross-cultural
perspective.
There are some parallels that can be drawn between beliefs in collective efficacy
and collective aspects of self. Triandis (1989) proposes that people sample different
aspects of selfprivate, public, and collectivedepending on their cultural background
and the current setting. The private self refers to cognitions that involve traits, states, or
behaviors of the person (e.g., I am introverted) (p. 507). The public self includes
cognitions concerning the generalized others view of the self, such as People think I am
19
introverted (p. 507). The collective self pertains to cognitions concerning a view of the
self that is found in some collective (e.g., family, coworkers, tribe, scientific society); for
instance, My family thinks I am introverted (p. 507). Depending on cultural context,
certain aspects of the self are more highly developed and complex: in a collectivist setting,
the collective self develops greater complexity and contains many more collective-self
units than public or private units, and is more likely to be sampled in most situations. A
person from an individualist cultural background will be more likely to sample from the
highly developed private self. Changes in environment can alter the probability of certain
selves being sampled. Triandis (1989) suggests that when life changes result in increased
choice of in-groups (through migration, for example) conflicting norms lead individuals to
turn inward to decide what to do. Thus, they are more likely to sample the private self
(p. 513). In other words, the self that is sampled may vary with context, and may shift
according to life changes. In the context of the cultural dimensions of individualism and
collectivism, Oyserman et al. (2002) suggest that all people are capable of viewing the
world through either an individualistic or collectivistic lens, but they differ in what is
likely to come to mind (i.e. which self is sampled) in any given situation.
The Self in Collectivist Cultures
Non-Western, often collectivist, cultures are typically considered to stress a more
relational, interdependent self-concept than do Western cultures. One of the primary
attributes imputed to many non-Western cultural conceptions of self is that of connection
or interdependence. Markus and Kitayama (1991) suggest that people in interdependent
cultures view themselves, connected with relevant others, as an integral part of the social
20
context; consequently, they view the self not as a circumscribed whole, but as
something that varies with context. Furthermore, it is the relationship with othersnot
the inner selfthat is the focal point of inward cognition (Markus, Mullally, &
Kitayama, 1997). The construal of self in many non-Western settings is holistic: persons
separated from others and from a social context are seen as incomplete and not fully
understandable. Nonetheless, this interdependent construal of self does not imply that
personal agency does not exist; rather, agency is directed at maintaining harmonious
relationships with others, and control is focused inwardly, in contrast to Western cultures
(and self-efficacy theory) in which control of outward behaviour and environment is often
the goal (Markus and Kitayama, 1991). However, as has been noted, characterizing all
non-Western cultures as wholly and unchangeably collectivistic is problematic in many
cases.
The Self in Self-Efficacy
It is important to consider just how differing self-conceptsparticularly a
concept of self stressing interdependencemight result in variations in self-beliefs, in this
case, beliefs of self-efficacy. For example, if a self, at its core, is conceived as tightly
connected with others, beliefs in ones ability to exercise control over a situation would be
less relevant than beliefs about what is expected in a particular context of the self-in-
relation to important others. Self-efficacy, while still perhaps a factor influencing
behaviour, might not be as influential as it would be in more individualist settings. Self-
efficacy theory is based on the belief that individuals seek to exercise control over their
lives because it provides innumerable personal and social benefits (Bandura, 1997, p.2).
21
In Banduras conceptualization of human functioning, human agency, which reflects the
concept of self, is situated in the context of three interdependent determinantspersonal
factors (including cognitive, affective, biological, and motivational), behavioral factors, and
environmental factors. These three factors influence each other reciprocally, hence
Banduras description of triadic reciprocal causation (1997). The three factors are not
equally influential in all situations, but rather vary in power depending on the
circumstances. Bandura contends that through reciprocal causation, people have the
opportunity to exercise control over life events and their performance, and that behaviors
or actions reflexively influence both personal factors (self-efficacy beliefs, for example)
and the environment.
The self in self-efficacy is portrayed as independent and autonomousit is
responsive to the surrounding environment, and it both influences and is influenced by
behaviorbut it is, in its essence, an individual, independent, separate self: Unless
people believe they can produce desired actions by their actions, they have little incentive
to act.... Beliefs of personal efficacy constitute the key factor of human agency
(Bandura, 1997, p. 3). This conception of self as largely independent and autonomous is
reflective of most Western conceptions of self. Markus and Kitayama (1991) for example,
note In many Western cultures, there is a faith in the inherent separateness of distinct
persons. The normative imperative of this culture is to become independent from others
and to discover and express ones unique attributes (p. 226). The self is influenced by,
but separate from others, and although the inner self is shaped by important others, it
is assumed to be distinct and moving towards the culturally valued goal of independence.
22
A central construction in this construal of self is that of control: individuals
independently strive for control over their settings; indeed, self-efficacy is described as
the exercise of control.
Banduras Defense of Self-Efficacy in Collectivist Settings
Self-efficacy, with its emphasis on personal controlPeople guide their lives by
their beliefs of personal efficacy (Bandura, 1997, p. 3)is firmly seated in a Western,
independent, individualist context. However, Bandura (1997) rejects the notion that self-
efficacy plays a lesser role in collectivist cultures: People live their lives neither entirely
autonomously nor entirely interdependently in any society. Interdependence does not
obliterate a personal self (p. 32). He points out that cultures within the collectivist
dimension vary greatly, and that individuals, too, adjust their behaviour depending on the
context. With in-group members, collectivists display a high level of communalism; with
out-group members, collectivists behave differently. Self-efficacy, Bandura argues, is
equally important for collectivists who carry out collectively oriented tasks, and for
individualists who focus on individually-based tasks. Bandura is hesitant to embrace the
idea that a differently construed self might negate the influence of self-efficacy beliefs. He
argues that personal efficacy is equally valued by collectivists because without a resilient
sense of self, people are easily overwhelmed by adversities in their attempts to improve
their group life through collective effort (p. 32). What Bandura rejects, then, is the idea
that the self might vary in its fundamental composition in a different cultural setting; that
is, by claiming that a resilient sense of (personal, private) self is essential for
23
achievement, he seems to deny the existence of we consciousness (to the exclusion of
an I-referenced self) referred to by Kim et al. (1994).
Review of Cross-Cultural Self-Efficacy Research
In the next section, I review much of the cross-cultural self-efficacy literature that
has been published in the last 30 years. Most of the research in this area has been
conducted very recently, with the majority of articles found (14 of 18) published between
1997 and 2001. The age-range and domains investigated are evenly distributed, with three
studies involving school-aged children, five studies including high school students, five
involving college undergraduates, four involving managers/employees in business settings,
and one study involving unemployed youth. The domains investigated include academic
efficacy (eight studies), efficacy for specific business tasks (four studies), career or job-
seeking efficacy (three studies), general self-efficacy (two studies), and self-efficacy for
social functioning (one study). The cultural or national groups in these studies is wide-
ranging, but with a heavy representation of North Americans (including participants
labeled as Asian-American, Hispanic, non-Hispanic, African American, Canadian, White,
Caucasian) and Asians (from Hong Kong, Taiwan, China, and India). (Ethnic group names
used in the reviewed articles are retained here). European representation comes from
Germany (East and West Germany, as well as unified Germany), Russia, the Czech
Republic, and Poland.
Academic self-efficacy. There is considerable evidence that students efficacy
beliefs influence academic task outcomes in Western contexts (e.g., Zimmerman, 2000),
but the universality of these findings has not yet been established. Eaton and Dembo
24
(1997) examined achievement and a variety of motivational beliefs in Asian American and
non-Asian high school students. Self-efficacy beliefs were rated higher by the non-Asian
students, but achievement was lower. Results from a multiple regression analysis
predicting achievement showed that for the Asian Americans, fear of academic failure was
the strongest predictor of achievement, followed by self-efficacy; for the non-Asians, fear
of failure was not a significant predictor, while self-efficacy beliefs provided a significant
contribution. Most interestingly, the Asian American students rated their self-efficacy
beliefs lower than did the non-Asian students, but displayed higher levels of performance
on the achievement task. Contrary to Banduras predictions (1997) about the role of
optimistic efficacy beliefs, the Asian American students displayed modest levels of
confidence in their abilities, but out-performed their more confident non-Asian
classmates.
In a study investigating the self-efficacy beliefs of Hispanic and non-Hispanic high
school students, Bong (1999) found the Hispanic students rated their academic efficacy
lower than did non-Hispanic students, with the exception of Spanish class, where
Hispanic students expressed significantly higher self-efficacy beliefs. Bong concluded that
personal factors (such as language or cultural background) operate on students
appraisal of their efficacy. Pajares and Johnson (1996) also compared Hispanic and non-
Hispanic students and found Hispanic students displayed lower self-efficacy for writing
than non-Hispanic White students. In this case, however, writing performance was also
lower, suggesting that efficacy beliefs were consistent with subsequent performance.
Similarly, a study comparing West Berlin and East Berlin school children before
25
unification found the East German children displayed lower but better-calibrated self-
efficacy; that is, they performed at a lower level, but displayed a higher level of accuracy
in the calibration of efficacy beliefs and school performance (Oettingen, Little,
Lindenberger, & Baltes, 1994).
Oettingen and Maier (1999) found that children in East Berlin rated their efficacy
beliefs lower than schoolchildren in Moscow; likewise, children in West Berlin rated their
self-efficacy lower than children in Los Angeles. In both of these comparisons, the
children with lower efficacy beliefs also weighted their teachers evaluations more
strongly than did children from the higher efficacy groups. For the latter comparison
group, the authors attributed differences in efficacy beliefs to differences in individualism
(with students from Los Angeles described as more individualistic) and power distance
(with the German school system described as incorporating a large power distance
between students and teachers).
As a further exploration of efficacy beliefs in former East-Bloc countries,
Oettingen and her colleagues (Oettingen, Maier, Kotaskova, Smolenska, & Stetsenko
1997, as cited in Oettingen & Maier, 1999) assessed efficacy beliefs of children in Prague
and Warsaw in 1991 and made comparisons with the data previously collected in East
Berlin and Moscow. In this four-way comparison, children in East Berlin evinced the
lowest levels of efficacy beliefs, a finding the authors attribute to the concurrence of
higher collectivism and political conformity, along with greater power distance. In this
scenario, political context, coupled with cultural values together influenced the formation
of childrens academic self-efficacy beliefs.
26
Randhawa and Gupta (2000) found that math self-efficacy was a stronger
predictor of achievement for a group of Canadian high school students than it was for the
presumably more collectivist Indian students (from private high schools in New Delhi).
The Canadian adolescents displayed higher efficacy beliefs for performing everyday
math tasks and for completing math courses, but lower beliefs for solving specific math
problems. The Indian students in this study performed at a significantly higher level on
the performance measure than did the Canadian students, but still rated themselves lower
in at least two of the measures of efficacy beliefs, a finding the authors attribute to the
higher academic expectations and greater academic competition found in the Indian private
school.
Changes resulting from migration. Salili, Chiu, and Lai (2001) also explored the
efficacy beliefs of Canadian high school students, but in this case comparisons were made
among European Canadian, Canadian Chinese, and Hong Kong Chinese adolescents for
general academic efficacy beliefs. The authors found academic self-efficacy beliefs to vary
according to culture. On a 7-point scale, the Hong Kong group mean was 3.99, the
Canadian Chinese mean 4.47, and the European Canadian mean was 4.76each of the
three means significantly different from the others. However, correlationswhich might
be viewed as a measure of calibration or accuracybetween self-efficacy beliefs and
academic performance (average grade in last term exam) did not follow the same pattern.
For the Hong Kong Chinese, the correlation coefficient was .19, for the Canadian Chinese
the correlation was .46, and for the European Canadians the correlation was .35. In other
words, the Canadian Chinese group displayed the best calibration of efficacy beliefs and
27
performance, whereas the Hong Kong Chinese showed significantly lower calibration
between their self-efficacy and their academic performance. Regarding efficacy beliefs, the
authors conclude that (a) self-efficacy was the single most important predictor of
academic achievement, and (b) the context of learning appeared to have a moderating
effect on cultural influences (p. 241). In other words, in this cross-sectional sample, the
Canadian Chinese group appeared to be in the process of a shift from collectivist to
individualist orientation, at least in the academic setting.
Changes in political systems. In studies of efficacy beliefs of students in
contrasting political settings, such as East and West Berlin, Oettingen and her colleagues
(Oettingen, 1995; Oettingen et al.,1994; Oettingen & Maier, 1999) persuasively argued
that the East German regime promoted a strongly collectivist education system in its 40
or so years of power and that children in school were urged to function as members of the
class collective (Oettingen & Maier, 1999). The West German children, of course,
experienced a considerably more individualist upbringing and cultural milieu. Self-efficacy
mean ratings were the same in grade 2, but were significantly different at every
subsequent grade (i.e., grades 3-6), with the West German children displaying higher
efficacy ratings (Oettingen, 1995). However, the East German children displayed
significantly more accurate calibration of efficacy and performance (course grades) at four
out of five grade levels. It is important to note that these differences evolved over a
relatively short period, and that the two groups were presumably culturally homogeneous
in most regards before the city and country was divided.
28
The authors attributed this greater realism to the collectivist character of former
East German society: students in this educational system regularly participated in peer
and teacher-directed appraisal before the class collective, teaching strategies were
group-oriented, and students were held accountable for the performance of their class
collective. Although the East Berlin and West Berlin political and educational systems
evolved into distinctly different entities, the two halves of the separated city did share a
long history of mutual culture, values, and self-beliefs that was broken for a period of
only about 40 years. In terms of collectivist/individualist orientation, this relatively brief
political fracture apparently resulted in people with differing self-beliefs, at least for
children in the school setting.
Non-academic self-efficacy beliefs of undergraduates. Five of the eighteen studies
investigated the efficacy beliefs of college or university undergraduates. In a study
exploring career decision-making self-efficacy among American and Taiwanese university
students, Mau (2000) found that students from Taiwan scored significantly lower on
decision-making self-efficacy than did the American students. The author concluded the
collective-oriented culture may have influenced Taiwanese students to rely less on
individual abilities than on group efforts (p. 374). Career decision-making self-efficacy
was the subject of investigation by Gloria and Hird (1999) but in an American context.
The racial/ethnic minority undergraduates displayed significantly lower efficacy beliefs
than did the White students, but higher trait anxiety. Nesdale and Pinter (2000) examined
a similar constructjob-seeking self-efficacyand found that efficacy beliefs were not
29
predictive of job-seeking behaviors among the ethnic youth (who were college-aged) in
their sample.
Schwarzer, Bassler, Kwiatek, Schroder, and Zhang (1997) compared general
self-efficacy among large groups of undergraduate students from Germany, Costa Rica,
and Hong Kong and found the Asian students had the lowest measured efficacy scores.
For the German and Hong Kong samples, the score for males was significantly higher than
for females. In the discussion section of the article, a call is made for future comparison
studies investigating corresponding levels of collective self-efficacy. In another study of
general self-efficacy beliefs, Feasel (1999) probed the role of collectivism and
individualism in relation to the efficacy beliefs of a variety of cultural groups in the U.S.
She found that general self-efficacy was a stronger predictor of domain specific self-
efficacy for individualists, whereas for collectivists, social sources of information were
better predictors of domain specific functioning. In her sample, Asian undergraduates
were more collectivist than the other groups (Blacks, Latinos, and Whites) and displayed
lower general self-efficacy. In light of these findings, the author questions the universal
nature of self-efficacy beliefs.
In a study involving college undergraduates, Zane, Sue, Hu, and Kwon (1991)
compared self-efficacy for assertiveness between Asian American and Caucasian
students. Interestingly, Zane et al. found no self-efficacy differences between the two
groups for assertiveness in situations involving acquaintances or intimates; however, the
Asian American university students reported significantly lower efficacy to socially
assert themselves in the presence of strangers.
30
Self- and collective efficacy in vocational settings. Several studies have investigated
cross-cultural issues pertaining to self-efficacy in work settings. Earley (1994)
investigated self-efficacy and work performance differences as a function of group and
individual training for managers from collectivist and individualist backgrounds.
Collectivists (from Hong Kong and the Peoples Republic of China) responded best to
group-focused training information whereas individualists (from the United States)
responded best to individualized training; both groups, however, responded better to
some trainingeven of the wrong orientationthan to no training at all. Earley
acknowledges self-efficacy is influenced by different sources of information that are
more or less persuasive depending on a persons cultural values (p. 114). In an extension
of the previous study, Earley et al. (1999) explored the effects that different types of
feedbackpersonal or group-basedhad on the self-efficacy and performance of workers
from collectivist and individualist cultures. For the individualists, the results were
straightforward: efficacy beliefs were most strongly influenced by personal, rather than
group feedback. For the collectivists, however, either personal or group feedback resulted
in increases in efficacy judgments. The authors concluded a collectivists sense of self is
based on both personal and group-based information (p. 614).
Earley (1999) also investigated the influence of the cultural dimension of power
distance on collective efficacy beliefs in a group setting. In his comparison of groups of
senior managers from four countries, Earley found that the efficacy judgments of higher
status group members in the high power distance countries (France and Thailand) were
more influential than the judgments of higher status group members in the lower power
31
distance countries (England and United States). In other words, the status of individuals
differentially influenced group decision-making depending on the cultural context.
The influences of cultural differences on job control and efficacy perceptions in
large samples of American and Hong Kong Chinese bank tellers were examined by
Schaubroeck et al., (2000). The researchers found that the American workers rated their
job self-efficacy at a significantly higher level than did the Asian workers. Conforming to
expectations, collective or group efficacy was significantly stronger among the Asian
workers. The authors primary concern, however, was the interaction between efficacy
beliefs, job demands, job control, and coping responses. Although the pattern of
interaction between the variables was generally the same between cultures, the nature of
the efficacy belief constructcollective efficacy or self-efficacydiffered between the
groups. That is, the role that efficacy beliefs played in Western and non-Western cultural
groups appeared similar, but the type of efficacy beliefs varied, with collective efficacy
and self-efficacy playing equivalent roles for the Hong Kong and American workers,
respectively.
Findings from Literature Review
There is some evidence from this brief summary of the research that efficacy
beliefs operate differently in non-Western cultures than they do in Western cultures. In
almost all of the studies reviewed, self-efficacy beliefs were higher for participants from
Western, individualist cultures than for the comparison groups from collectivist settings.
On first glance, the results from this review support Eaton and Dembos claim that
traditional measures of self-efficacy may be less valid for this (collectivist) population
32
(1997, p. 438). Other investigators of self-concepts have arrived at similar conclusions:
motivational concepts generated from research conducted in Western countries may only
be useful for understanding people from those countries (Whang & Hancock, 1994, p.
315). A theoretical basis for these differences has been discussed: one would predict that
an interdependent or collectivist sense of self would be reflected in differing beliefs about
self; self-efficacy, as a kind of self-belief reflects these differences. But while the results
from this review support a finding of differential functioning of efficacy beliefs depending
on cultural context, three other findings need to be emphasized.
First, and perhaps surprisingly, there is some evidence that among collectivists,
efficacy beliefs may be more predictive of performance, and that the calibration of
efficacy beliefs and subsequent functioning may be more accurate than among
individualists. Second, concepts of self, like self-efficacy, appear not to be fixed, but are
amenable to change, depending on the context. Finally, the role played by collective
efficacy may be more salient among those with collectivist orientations and may, to some
extent, replace the role played by self-efficacy.
Optimistic calibration of self-efficacy. When self-efficacy comparisons between
individualist and collectivist cultures were made, the results almost invariably showed
that the non-Western participants rated their efficacy lower than Western participants,
regardless of level of performance. When calibration was assessed, in some cases the more
collectivist, less optimistic groups displayed better calibration (Eaton & Dembo, 1997;
Oettingen, 1995; Salili et al., 2001); that is, participants from the collectivist groups
inclined towards greater accuracy or realism in predicting subsequent performance. In
33
the Eaton and Dembo study, the authors make note of the lower efficacy beliefs of Asian-
American students, but they fail to note the stronger correlation between these students
efficacy beliefs and subsequent achievement. Oettingen (1995) comments that in East
Berlin (before unification), optimistic efficacy beliefs incurred social censure and were
detrimental to successful classroom functioning. She concluded that strong self-efficacy
beliefs were relevant and desirable for all cultural groups, because they are indeed
universal they are founded in basic psychological principles and mechanisms common
to human agency in general (p. 169).
Salili et al. (2001) found that self-efficacy beliefs were more strongly correlated
with academic performance among Canadian Chinese than European Canadian students,
but were least strongly correlated for the Hong Kong Chinese students. The authors
suggest that the extremely competitive and stringent grading practices in Hong Kong may
have adversely affected the accuracy of these students efficacy beliefs. Of the two groups
with the same educational context (European Canadian and Canadian Chinese), the latter
group displayed lower but more accurate efficacy beliefs along with higher academic
performance. It seems, then, that the claim (e.g., Bandura, 1995; Pajares, 1996a) that
optimistic self-efficacy promotes the completion of challenging tasks may not hold true for
all groups. The lower, but more realistic self-efficacy beliefs held by many collectivists do
not seem to hamper high levels of performance: for example, East Asian Americans, with
a more collectivist orientation, and more modest efficacy beliefs, often out-perform
European Americans on academic measures (e.g., Fulgini, 1997). Banduras assertion
that an optimistic sense of efficacy fosters psychological well-being and personal
34
accomplishments (1997, p. 75) may apply to individuals from Western cultures, but this
positive mis-calibration of beliefs may not operate the same way for other groups. It was
not possible to assess calibration between efficacy and performance in all studies: some
studies did not include a performance task, and other studies did not provide sufficient
information to judge calibration.
Efficacy beliefs and life changes. Another finding of interest pertains to ratings of
self-efficacy beliefs in relation to life changes. Research investigating self-efficacy beliefs
longitudinally among immigrant populations has not been conducted, but examining the
efficacy beliefs of immigrant cultural groups (Salili et al., 2001) and cultural groups that
have been separated by political forces (Oettingen, 1995) lends preliminary insight into
how groups, and perhaps individuals, might change important self-beliefs over time. Some
aspects of the self, and the make-up of an individuals concept of self, show evidence
of change in parallel with contextual changes. In the research investigating European
Canadian, Chinese Canadian, and Hong Kong Chinese students (Salili et al., 2001), the
immigrant group rated their self-efficacy beliefs between the individualist European
Canadian group and the collectivist Hong Kong Chinese group. From these results, one
might speculate that the immigrant groupthe Chinese Canadian grouphad undergone a
cultural shift in construal of self-beliefs: their self-efficacy, and perhaps their conception
of self had metamorphosed with changes in the cultural context. The differences in self-
efficacy beliefs between the East and West Berlin schoolchildren (Oettingen, 1995) also
lends credence to the possibility of a shifting sense of self, and resulting efficacy beliefs,
over a relatively brief period of time, in this case due to political/cultural variations.
35
Does collective efficacy replace self-efficacy in some contexts? Few studies have
compared the roles played by collective and self-efficacy beliefs from a cross-cultural
perspective, and none in the field of education. The three studies found in this review
were situated in a business context. Earleys 1994 study showed that group-level training
was most effective for improving expectations, effort and performance in managers with a
collectivist orientation. Managers from an individualist cultural orientation benefited
primarily from individual-level instruction. Schaubroeck et al. (2000) found bank tellers in
Hong Kong to draw on group-oriented efficacy beliefs when facing job stress, in contrast
to American bank-tellers who relied most heavily on beliefs of self-efficacy. Earley et al.
(1999) supported the notion that while individualists benefit primarily from individual-
referenced instruction, collectivists benefit from either group- or individual-focused
instruction. Banduras claim that Belief in ones ability to produce desired effects fosters
accomplishments in all cultures (1997, p. 32) may not tell the whole story. For
collectivists, group or collective beliefs also appear to be key motivational components
that foster accomplishment.
Several of the studies included in this literature review that did not explore
collective efficacy nevertheless concluded that the efficacy of the group or collective
might be especially relevant for individuals from non-Western cultures: It is possible
that the collective-oriented culture may have influenced Taiwanese students to rely less
on individual abilities than on group efforts (Mau, 2000, p.374). Eaton and Dembo
concluded Asian Americans are socialized to feel responsible to their family and
community, whose needs and expectations prevail over existing desires (p. 438). In
36
terms of conception of self, East Asian American high school students presumably view
themselves as more closely connected to their in-group and respond to different
motivational factors than do non-Asian students. Expectations from the in-group and
concerns over family or community reaction to achievement level appeared to be more
significant than levels of individual efficacy.
If Triandis (1989) sampling probability model holds true, self-efficacy, or
sampling from a private self, might be replaced for some groups by collective efficacy, or
sampling from a collective self. Thus, a student or worker in a collectivist setting might
ask Am I expected (by my ingroup) to attempt this task? and Am I (or Is my
ingroup) capable of completing this task? The findings from this review offer
preliminary support for this view. This perspective is of importance in building
understanding of student learning in a multicultural context. Little is known about how
collective or group efficacy beliefs are formed, or how classroom variables such as group-
focused versus individual-focused instruction affect efficacy beliefs. For those working in
multicultural settings, a deeper understanding of these issues may help inform
instructional practice.
Methodological Problems
Recent concern (Fiske, 2002; Oyserman et al., 2002) has been expressed about
amalgamating ethnic groups under one cultural banner without first examining similarities
and differences on the dimensions that presumably connect them. In this review, many of
the studies combined ethnic groups, for example, under the heading of Asian American.
Great diversity is found among Asian cultures; the label Asian includes people culturally
37
identified with thousands of cultures comprising more than half the worlds population
(Fiske, 2002). Although there are doubtless similarities among many of these cultures on
some dimensions, there are significant differences on the individualism/collectivism
dimension between Chinese, Japanese, Korean, and Indian samples (Fiske, 2002). For
cross-cultural research in educational psychology, it is essential to measure the cultural
dimension that presumably separates comparison groups rather than merely assume that
participants from different national backgrounds necessarily share a common cultural
composition. This problem is exacerbated in the make-up of comparison groups: the
non-Hispanic comparison group in Bongs 1999 study included Asian American,
European American, African American, and Native American students. The ethnic
youth comparison group in Nesdale and Pinters (2000) study consisted of immigrants
from both Western (i.e., New Zealand, United Kingdom) and non-Western (Vietnamese,
Chinese) cultures. Although cultural difference is not the only viable means of dividing
groups for comparison, ignoring important cultural differences within groups may
confound findings.
Another difficulty encountered in empirical studies investigating efficacy beliefs
from a cross-cultural perspective is the initial assumption that self-efficacy is a valid
motivational factor for all groups. Self-efficacy might be considered an imposed etic (Ho,
1994); that is, it is assumed to be a universal theorybut does it have a sufficient multi-
cultural empirical foundation from which to make that claim? The predictive power of
self-efficacy, in comparison to other motivational factors, is well documented, at least for
populations with an individualist orientation. Pajares and Graham (1999) found task-
38
specific self-efficacy to be a stronger predictor of math performance than previous math
grade, math self-concept, anxiety, or self-efficacy for self-regulation. As reviewed here,
most cross-cultural self-efficacy studies have not included contrasting motivational
variables, and other variables, such as fear of failure (Eaton & Dembo, 1997) and
perceptions of parents value of academics (Fulgini, 1997) have been found to be strong
predictors of performance for non-Western groups, but less so for Western research
participants. Including only a self-efficacy measure in a study investigating cross-cultural
motivational beliefs guarantees some finding about the importance of efficacy beliefs,
whether or not these beliefs in fact play a central role in the actual cognitive process of
the individuals and groups being measured.
For example, Schwarzer et al. (1997) compared the self-efficacy beliefs of three
different cultural groups believed to vary on the individualism/collectivism dimension.
The 10 items of the Generalized Self-Efficacy Scale were all composed of some type of
I statement (e.g., If someone opposes me, I can find means and ways to get what I
want, It is easy for me to stick to my aims and accomplish my goals) with a four-
point rating scale. Based on the idea that collectivists tend to sample aspects of self
differently than individualists, one might predict that collective efficacy or another type
of motivation might be more relevant. However, no other construct was explored, and the
possibility remains that whatever the self-efficacy-based findings might suggest, other
motivational variables might play a more important role for the more collectivist
participants. In terms of Triandis sampling probability theory (1989), respondents were
more-or-less forced to sample from their private self, whether or not this was their
39
natural inclination. Bempechat and Drago-Severson (1999) recommend that qualitative
and multiple methods be used to help validate self-efficacy theory in different cultural
contexts. Structured interviews, focus-group interviews, think-aloud protocols, or any of
a myriad of other qualitative methodological strategies might prove constructive in
exploring and explaining the motivational conceptions of non-Western groups.
A third difficulty was found in the reviewed literature. Although self-efficacy has
a strong theoretical base and has been well explored in a Western cultural setting, the
construct loses explanatory power when it is ill defined in practice (Klassen, 2002b;
Pajares, 1997). Other motivational beliefsself-concept of ability, self-esteem, and
competence beliefsare similar in some ways, but self-efficacy is concerned with
judgments of personal capability in a specific context. Some of the self-efficacy items
encountered in the studies in this review strayed a considerable distance from
theoretically derived definitions of self-efficacy. Consider the following items from
measures labeled as self-efficacy measures: I have devoted much effort to looking for
jobs, and I am very proud of my job skills and abilities. These items may be important
questions to ask in some context, but they bear little relation to self-efficacy as it is
typically defined. Finally, several of the studies measured general self-efficacy beliefs,
which juxtaposition some theorists would consider contradictory. Bandura, for example
warned, It is no more informative to speak of self-efficacy in general terms than to speak
of nonspecific social behavior (1997, p. 14). Removing the context from self- or
collective efficacy assessment removes much of the relevance and predictiveness of the
construct.
40
Summary of Important Findings from the Literature Review
One of the findings from this literature review is that students from some non-
Western (typically East Asian) cultural backgrounds rate their efficacy beliefs lower than
do their Western counterparts, even when performance levels are equivalent or higher. In
spite of this, as seen in a few of the studies, for collectivists the relationship between
efficacy and performance might, in fact, be stronger and more predictive than it is for
those from more individualist cultures. For some students from non-Western cultures, a
sense of realism about ones efficacy beliefs is functional, and may be a result of cultural
background rather than evidence of motivational shortcomings. Cultural background, level
of connectedness with others, and current setting may have an impact on students
efficacy beliefs and performance. There is some evidence that efficacy beliefs and
performance are differentially influenced by group- and individual-focused instruction
according to culture. However, because cross-cultural research to this point has not been
conducted in many settings, little is known about how efficacy might operate with (a)
students from non-Western, non-East Asian cultural backgrounds (for example, students
from South Asian cultural backgrounds), or (b) students below the college or senior high
school age range. Furthermore, there is a need for increased understanding of how self-
efficacy might operate in comparison to other relevant motivation constructs in diverse
settings (Pajares & Graham, 1999), and how collective efficacy might operate across
cultural settings.
41
South Asian (Indo-Canadian) Immigrants
Relative to East Asians, South Asians have not been well represented in the
academic motivation and self-efficacy literature. As discussed, there is evidence that
people from an East Asian cultural background might display lower and more accurate
efficacy ratings than do students from Western backgrounds, but researchers that make
claims about Asians or non-Westerners based on these findings may be over-
generalizing their conclusions (Fiske et al., 1998). Little or no cross-cultural self-efficacy
research has been conducted involving students from an immigrant South Asian cultural
background, in spite of the fact that in Canada, Indo-Canadians are the second largest
immigrant group (Kwak & Berry, 2001).
In an American context, Gibson studied high school students from Indian families
in a California farming community (Gibson 1988, Gibson & Bhachu, 1991). In that
setting, like in British Columbia (Correia, 2000), almost all of the immigrants were from
the state of Punjab, and 90% were Sikhs. In comparison to many other cultural groups,
Indian Sikhs tend to maintain a separate identity and resist assimilation into the majority
cultural group: Punjabis actively resist the influence of the dominant culture, fearing that
such influence will erode values they perceive as fundamental to the persistence of the
Punjabi identity (Gibson, 1988, p. 25). Gibson collected individualism scores of two
groups of Punjabi Sikhs: those raised in the U.S. and those who had arrived recently, and
compared these to Valleysiders (non-Hispanic white population). Of the recent
arrivals, 35% of females, and 24% of males selected individualist, rather than a family
orientation, while for the Sikhs raised in the U.S. 67% of females and 50% of males chose
42
the individualist orientation. Of the Valleysiders, 78% of females and 63% of males
chose the individualist orientation.
Although Hofstede (1980) characterized India as relatively low in individualism,
and Triandis (1996) claims India is a vertical collectivist culture, some researchers and
theorists disagree with that assessment. Sinha and Tripathi (1994) assert that Indian
culture supports an unusually (by Western standards), complex and contextual
coexistence of contradictions whereby Indian culture and psyche are neither
predominantly collectivist nor individualist in orientation. Their distinguishing feature is
that they incorporate elements of both orientations. Therefore if it is collectivist at
all, there is still plenty of individualism to be found there (p. 136). In Sinha and
Tripathis survey of cultural orientation (1994), the majority of participants (86.6%)
displayed a mixed orientation as a first choice, and only 1% chose a collectivist
orientation. Similarly, Mishra (1994) found mixed orientations in both younger and older
upper-caste Hindu participants, and concluded The Indian form of collectivism seems to
contain some streaks of individualism (p. 238). In a study examining the cultural
orientation of Indian university students, Sinha, Sinha, Verma and Sinha (2001) found that
despite a general priority for collectivist over individualist orientation, the most common
response to a variety of situations was an admixture of individualism and collectivism
depending on desh (place), kaal (time), and patra (person). According to Sinha et al.
(2001), individualistic behaviour is frequently pursued in order to serve the interests of
the ingroup; conversely, collectivist behaviour may at times conceal individualist
intentions.
43
It may be, then, that although there is a current of collectivism running through
Indian culture, the cultural dimension that is likely to come to mind (Oyserman et al.,
2002) is less predictable than in some cultures, and depends on the setting and the
individual. Research investigating Indian cultural dimensions is more explanatory when it
is designed so that the complexity of this culture might be captured and reflected.
Multimethod research using focus groups has been recommended as a way of providing a
more complete picture of the cultural and motivational functioning of culturally complex
groups (Calderon, Baker, & Wolf, 2000; Matsumoto, 1999; McInerney, 1998).
Multimethod Research and Focus Groups
Mixed method (or multimethod) designs provide triangulation (convergence),
complementarity (overlapping or different views), or development (extension of findings
from first method) beyond what is available from single method designs (Tashakkori &
Teddlie, 1998). Linking qualitative with quantitative data allows for enhanced validity,
interpretation, clarification, and illustration of key findings (Miles & Huberman, 1994).
Although multimethod designs are less widespread than single method designs, possibly
due to longstanding paradigm wars (Tashakkori & Teddlie, 1998), this approach to
explaining phenomena is not newalmost 60 years ago Lazarsfeld (1944) recommended
open-ended interviews to interpret and refine statistical inter-relationships (p.47).
Here the (open-ended interview) is in its most legitimate place (p.59). The present study
incorporates a multimethod approach: it is an etic/emic study that addresses students
own idiosyncratic, emic concepts of learning (Bempechat & Boulay, 2001, p. 32) as well
as the etic applicability of self-efficacy to a non-Western group.
44
Outline of Thesis
This thesis is designed to put self-efficacy to the test in terms of its
generalizability to a South Asian, specifically Indian Sikh, immigrant population. Pajares
and Graham (1999) found self-efficacy to be the only motivational variable that predicted
the math performance of American adolescent students. In addition to investigating the
role of self- and collective efficacy beliefs, other motivational constructs previously found
to be predictive of performancemath self-concept (Marsh, 1992), fear of failure (Eaton
& Dembo, 1997), and perceived parents value of academics (Fulgini, 1997)are included
to determine the relative contribution of the efficacy beliefs. Self-concept is one of the
most-studied self-constructs in academic motivation research (Bong & Clark, 1999)
whereas fear of failure and perceived parents value of academics have been shown to be
important predictors of academic performance in immigrant populations in cross-cultural
investigations (Eaton & Dembo, 1997; Fulgini, 1997).
The first (and principal) section of this projectwhich investigates self-efficacy,
collective efficacy, other secondary motivational beliefs, sources of self-efficacy and
calibration of Indo-Canadian Punjabi Sikhs and Anglo-Canadiansmight be considered an
(imposed) etic approach to the construct. An etic approach to cross-cultural research
applies (typically) Western-based constructs to non-Western groups under the
assumption that the constructs under investigation are universally relevant. Conversely,
an emic approach explores concepts that emerge from investigations of a particular
culture-group (Watkins, 1996). This project also addresses the paucity of qualitative
investigations of motivation (Bempechat & Boulay, 2001).
45
As a further exploration of self-efficacy beliefs, the second, complementary,
section of this study incorporates semi-structured focus group interviews designed to
explain and extend the data obtained through quantitative methods. The design of the
study is explanatory (Cresswell, 2002) with follow-up qualitative elements helping to
interpret results in a principally quantitative study (Miles & Huberman, 1994; Morgan,
1998). The quantitative data was collected first, and followed by the collection of
qualitative focus group data intended to refine and situate the quantitative findings. Focus
groups have been suggested as an especially appropriate methodology with children and
adolescent research participants, who may be reluctant to engage in conversation in
individual interviews (Hoppe & Wells, 1995; Jamieson et al., 2002). In contrast to other
methodologies, focus group research is hallmarked by the interaction of the participants:
group members influence each other and change and shape opinions of other members, in
a way that reflects natural processes in many social settings (Krueger, 1994).
How does this Thesis Contribute to Knowledge about Self-Efficacy?
Indo-Canadian sample: cross-cultural approaches to research in self-efficacy have
been recommended (Pajares, 1997) but infrequently conducted. Studies investigating
the academic self-efficacy of children of immigrant Indian parents have not yet been
conducted.
Comparison of the role of self-efficacy and other motivational beliefs shown to be
relevant in cross-cultural settings illuminates how self-efficacy operates.
Sources of self-efficacy beliefs: a very few studies have investigated the sources of
self-efficacy beliefs for specific populations and in specific domains (Hampton, 1998;
46
Lent, et al., 1991; Matsui, et al., 1990) but no studies have examined the sources of
academic self-efficacy from a cross-cultural perspective.
Calibration inquiry: previous research has examined the academic performance and self
beliefs of students from non-Western cultural backgrounds, but the implications of
calibration issues have less often been specifically addressed.
Research investigating collective efficacy for academic achievement using a cross-
cultural framework has rarely, if ever, been conducted.
Qualitative approach: as recommended by Bempechat and Boulay (2001) and
McInerney and McInerney (1998), a qualitative approach to investigation of
motivation and self beliefs will provide enhanced description and explanation over and
above that provided by quantitative investigation alone.
Research Questions
How do self-efficacy beliefs operate for students from a non-Western, Indo-Canadian
cultural background? Do students, in fact, attribute academic success and failure to
their efficacy beliefs, and does this vary within and across cultures?
Is self-efficacy a relevant predictor in the face of other motivational beliefs?
How do the sources of academic self-efficacy vary for students from these different
cultural backgrounds?
Are there differences in collective efficacy across these cultures?
Are the cultural dimensions of individualism and collectivism helpful in explaining
differences in self-efficacy among these groups?
47
CHAPTER 2
Method
Overview
This study was designed to examine and compare the efficacy beliefs of students
from contrasting cultural backgrounds, with a particular focus on comparisons between
students from Indo-Canadian and Anglo-Canadian backgrounds. One aspect of this
investigation of efficacy beliefs was a differential comparison of the role that other
motivation beliefsself-concept, perceived parents value of academics, and fear of
failureshown to be salient in previous cross-cultural investigations, might play in the
cultural groups included in this study. The following section explains the methodology
employed in this study.
Participants
Participants were 383 (174 male, 209 female) grade 7 students from 8 public and
two private elementary schools in a large urban/suburban community in Western Canada.
Data were collected in the spring of 2002. Research participants were volunteers; they
received a small token of appreciation at the completion of data collection. The age of the
participants ranged from 11 to 14, with a mean age of 12.38 years. Demographic
information collected included country of birth, language(s) spoken at home, fathers level
of education, fathers country of birth, and previous grade in math. Participants were
categorized according to fathers country of birth and language(s) spoken in the home.
48
Students born in Canada or the United States, and whose father was born in
Canada or the United States, and who spoke only English at home were classified as
Anglo-Canadian (44 male, 68 female, n = 112). Students who were born in India or whose
father was born in India, and who spoke Punjabi, or English and Punjabi at home were
grouped as Indo-Canadian (74 male, 84 female, n = 158). According to school officials, the
parents of the Indo-Canadian students in this community are almost exclusively Sikhs
from the Punjab region in northern India (K. Johal, personal communication, February 22,
2002).
In light of findings that home language has a strong influence on development of
self-concept and perceptions of others (e.g., Gudykunst & Bond, 1997), language spoken
at home was used as a marker of cultural background: one student whose father was born
in India, but who reported speaking only English at home was not included in the
analyses. Three students who reported fathers born in Canada, but who spoke East Asian
languages in the home (e.g., Chinese or Korean) were not included in the Anglo-Canadian
group. Of the 383 participants, 104 students did not fall into either of these two groups
(fathers country of birth East Asia, Africa, Eastern Europe, the Middle East, or South
America), and were not included in any of the analyses. As stated below, 5 Indo-
Canadians and 4 Anglo-Canadians students were excluded due to status as multivariate
outliers.
Anglo-Canadian group. The mean age for this group of 112 students was 12.43
(SD = .51), mean math grade (D = 1 through to A = 7; see Measures for complete list)
was 5.24 (SD = 1.34) and mean fathers level of education (ranging from 1 = didnt
49
attend high school through to 5 = completed college or university; see Measures for
complete list) was 3.83 (SD = 1.02). Table 1 displays the means and standard deviations
for age, previous math grade, and fathers level of education.
50
Table 1
Means and Standard Deviations for Age, Previous Math Grade, and Fathers Level
of Education
Group 1
Anglo-Canadians
(n = 112)
Group 2
Indo-Canadians
(n = 158)
M SD M SD
Age 12.43 .51 12.30 .51
Math grade 5.24 1.34 5.10 1.34
Fathers education 3.81 1.06 3.63 1.24
Note. Math grade was self-reported with 1 = D, 7 = A. Fathers level of education ratings
included the following: 1 (Didnt finish high school), 2 (Some high school), 3 (Finished
high school), 4 (Some college), and 5 (Finished college or university).
51
Indo-Canadian group. The 158 students in this sample had a mean age of 12.30
(SD = .51), and an average previous math grade of 5.10 (SD = 1.34). The average level of
fathers education for this group was 3.63 (SD = 1.24). There were no significant
differences between the two cultural groups for any of the demographic variables.
Because a sizable proportion (39%) of the Indo-Canadian participants were selected from
a private Sikh school, comparisons were made between Indo-Canadian students from the
private school and Indo-Canadian students from the public schools. Apart from previous
math grade (significantly higher for students in the private school) there were no
significant differences on any of the motivational, performance, cultural dimension (i.e.,
Individualism/Collectivism) or demographic variables.
The principals of twelve elementary schools with high proportions of Punjabi-
speaking homes (based on a school district Language Report) were contacted to
participate in the study. Of the 12 principals contacted, 10 agreed to allow the primary
investigator to contact grade 7 teachers in the school. The schools were all in the same
working class/lower-middle class area within the city. Three classes of exclusively Indo-
Canadian students from an independent Sikh school participated in the research; one class
of students (primarily Anglo-Canadian backgrounds) from an independent Christian
school participated. Teachers from 16 classes agreed to allow the researcher to come and
explain the project to the class, and ask for participants. The project was explained to the
students, and parental permission forms were handed out, and collected by the classroom
teacher. Approximately 85 % of students contacted agreed to participate in the research
project and brought back signed parental permission forms
52
Procedure
A pilot study was undertaken with one class of 21 students in which the
procedures and measures were trialed. Some items were dropped from the original math
measure due to overly high or low item difficulty levels, and the number of items in the
word problems measure was also reduced. Some of the verbal and written instructions
were changed to enhance clarity. Data from the pilot study were not included in the
analysis for this study.
The participating students completed the research measures in the classroom
setting; students who elected not to participate, or who had not returned signed
permission forms were taken by the teachers to other areas. Each data collection session
took approximately 1 hour and 15 minutes, and included all of the quantitative measures.
A sample research protocol (with item/variable identification included) is included as
Appendix A.
For the group activities, students formed their own groups of three or four
students. Some of the resulting groups were culturally homogeneous; others were
culturally mixed. The order of administration was as follows: demographic information;
parental importance measure; the I/C, math self-concept, and fear of failure combined
measure; the sources of self-efficacy measure; math self-efficacy and math measures; the
self- and collective efficacy and word problem measures; followed by the measure of
group functioning. For half of the classes, the order of administration was altered with
demographic information followed by the word problem/group measures and then the
parental importance, I/C measures, etc. All items (except for the math word problems
53
measures, and the corresponding self- and collective efficacy scales) were read aloud to
the students, who also had a copy of the questionnaire in front of them. At the end of the
session, students were thanked for their participation and given a small token of
appreciation.
Measures
Fathers level of education. As commonly included in social science research
(Jeynes, 2002) a measure of parental level of education was included as an index of
socioeconomic status (SES). Students were informed that they could rate their mothers
level of education if they were in a one-parent family. The ratings included: 1 didnt go
to high school, 2 some high school, 3 completed high school, 4 some college, and 5
finished college or university.
Previous math grade. The ratings on this self-report measure ranged from 1 for
D (or E or F) to 7 for A. All schools used the same grading symbols (A, B+, B, C+,
C, C-, D or E or F for failure) to represent achievement.
Math measure (performance task). The 22-item math measure was created in
cooperation with school district staff to reflect some of the content (basic arithmetic
skills, decimals, fractions, simple algebra) covered in the grade 7 math program
(Cronbachs a = .80). The mean score for the whole sample was 12.97 (SD = 3.79).
Self-efficacy for math. Participants completed a 22-item self-efficacy measure that
used a 0 10 scale with three descriptors: No chance (at 0), Maybe can do (at 4-
6), and Certain can do (at 10). Students were given a practice item on an overhead
projector with the following oral and written instructions: I want to know about your
54
confidence to solve these 22 math questions on your own in 15 minutes. Read each
question, and before you actually do the question, rate how sure you are that you can
solve the question. Rate your degree of confidence by recording a number from 0 to 10 in
each of the blanks using the scale given below. Lets try one as an example.
The 22 items of the math measure were briefly explained, and students were
instructed to complete each self-efficacy rating before completing the corresponding math
item. Students were given 15 minutes to complete the self-efficacy rating and the math
measure. Self-efficacy for math scores was calculated as the mean efficacy rating for each
student. Similar direct approaches to mathematics self-efficacy measurement have been
used with good reliability by other self-efficacy researchers (e.g., Pajares & Graham,
1999; Pajares & Miller, 1997; Schunk & Cox, 1986). Bandura (1997) posits that self-
efficacy measures will be most predictive when they are particularized in a specific
domain, such as math functioning. Cronbachs alpha for this measure was .91, which is
consistent with reliability coefficients obtained in similar studies. The mean score for the
whole sample was 7.55 (SD = 1.65).
Fear of academic failure. This 6-item scale was adapted from Eaton and Dembo
(1997). Eaton and Dembo note that the fear of failure construct is conceptually similar to
the ego/performance orientation of Ames and Archer (1988). Modifications to this
measure consisted of the inclusion of the 0 10 scale used throughout this project.
Students were asked How true are the following statements of you? and responded
with 0 (Not true at all) to 10 (Very true). The internal consistency reliability
55
coefficient obtained for this measure was .72; the whole sample mean score was 7.86 (SD
= 1.70).
Math self-concept. The 6-item math self-concept scale was taken from Marsh
(1990) who investigated students self-concepts in 16 academic areas. Students responded
using a 0 to 10 scale. The reliability coefficient for this measure (Cronbachs a) was .90,
consistent with the range of .885 to .949 given in Marsh (1990). The mean score for the
whole sample was 6.93 (SD = 1.96).
Perceived parental value of academics. Fulgini (1997) found students perception
of their parents value of academic success to be a strong predictor of academic success
for children from immigrant families. The 6-item scale was completed using the 0 to 10
scale used in this study. Internal consistency, measured using Cronbachs alpha, was
measured at .80. The mean score for the whole sample was 7.75 (SD = 1.66).
Calibration. Calibration was first defined as mean accuracy (see Pajares &
Graham, 1999) and calculated in the following manner. First, correct responses on the
math measure were scored as 10, and incorrect scores as 0 (corresponding with the
efficacy measure). These scores were subtracted from the efficacy judgment score, and the
absolute score was taken as an indicator of the magnitude of mis-calibration. To ease with
interpretation (and departing from the method used by Pajares and Graham, 1999), the
absolute score was then transformed by subtracting it from 10, resulting in a more easily
interpretable result. For example, if a student estimated her efficacy for an item as 8, and
correctly solved the item, she would receive a mean accuracy score of 2 (the absolute
value of 8 10) that would be transformed into a score of 8. If she had responded
56
incorrectly, she would receive a calibration score of 2. Higher scores mean more accurate
calibration; lower scores mean less accurate calibration. Accuracy scores were averaged
across all 22 self-efficacy/math items. The reliability coefficient obtained for this measure
was .70.
In the same fashion, mean bias was calculated by first transforming responses on
the math measure to dichotomous responses of 0 or 10. These scores were subtracted
from the efficacy score, with the direction of the resulting number indicating the direction
of the mis-calibration. The mean of these scores was calculated, resulting in the mean bias
score. Reliability for this method was .73.
Calibration was calculated in two additional fashions. First, the proportion of
students labeled as overestimators, underestimators, and accurate estimators was
calculated. Underestimation was defined as a mean bias score below 0 (20 % of the
sample), while the overestimators were divided into two groups: moderate overestimators
(40 % of the sample) scored between 0 and 1.65 and severe overestimators (40 % of the
sample) scored above 1.65. Second, the Pearson product-moment correlation coefficient
between performance and efficacy was used as an indicator of the calibration between the
variables.
Sources of self-efficacy. The 20 item sources of self-efficacy scale included 5 items
for each of the self-efficacy sources hypothesized by Bandura (1997)past performance,
vicarious learning, social persuasion, and emotional arousal. The scale was adapted from
the measure developed by Matsui et al., (1990) and the sources of mathematics self-
efficacy scale used in several studies by Lent and colleagues (Lent et al., 1991; Lent,
57
Lopez, Brown, & Gore, 1996; Lopez & Lent, 1992): three of the sub-scales (vicarious
learning, verbal persuasion, and emotional arousal) were taken from the Matsui et al.
(1990) measure (which used students self-report of achievement as an index of past
performance) whereas the fourth sub-scale (past performance) was taken from the scale
provided by R.W. Lent (personal correspondence, October 15, 2001). Modifications to
the measures included revised wording to enhance understanding for the grade 7 students,
and the inclusion of the 0 10 scale which was used for several of the measures in this
study. The reliability coefficient for the entire scale was .82, with the following alpha
levels obtained for each source: past performance .71, social persuasion .87, vicarious
experience.73, and emotional arousal .76.
Group taskword problems. A selection of 18 word and logic puzzles/problems was
chosen from grade seven-level commercial puzzle books and text books to provide a group-
level performance task (2 items were dropped from the initial 20 items after the pilot study
due to very high or very low item difficulty levels). The reliability coefficient (Cronbachs a)
for this measure of group problem solving was .81.
Collective efficacy for group task. Collective efficacy was assessed on two levels:
first, at the item level with a high level of specificity and correspondence (i.e., How
confident are you that your group can solve this problem?). After being formed into
small groups, students rated their groups collective efficacy for each item on a scale
ranging from 0 (No chance) to 10 (Certain can do). Students were instructed to (a)
read the item, (b) estimate their own confidence, (c) estimate their confidence in their
group, and (d) solve the item. Cronbachs alpha for this measure was .86.
58
Self-efficacy for group task. Students were also asked for their estimates of their
own confidence to solve each of the 18 word and logic problems (i.e., How confident are
you that you yourself can solve these puzzles and problems in 15 minutes?). Coefficient
alpha for this measure was .90.
Calibration between collective efficacy and performance. Collective efficacy
calibration was defined as mean accuracy and mean bias (see Pajares & Graham, 1999),
and calculated as in the calibration measure for self-efficacy discussed above. The
reliability coefficient (Cronbachs a) obtained for mean accuracy was .81, and for mean
bias, .83.
Global or generalized collective functioning. A measure of global collective
efficacy (item example The group I work with is better than average) consisting of 8
items was adapted from a study of job-related collective efficacy conducted by Riggs et
al., (1994). Global collective efficacy is conceptually distinct from task or item-level
collective efficacy; the former pertains to an individuals assessment of generalized group
functioning, while the latter is an individuals assessment of a groups capability to
complete a particular task (Bandura, 2001). Adaptations to the Riggs et al.s items were
changing the domain from vocational to educational functioning, and the scale to the 0
10 scale used throughout this study. Verbal and written instructions presented to the
students were as follows: Now think about the group youve been working with. Rate
how true the following statements are for you. Remember, you can use any number from
0 to 10. The reliability coefficient (Cronbachs a) for this measure was calculated at .68.
59
Individualism/collectivism. The I/C scale included in this study was adapted from
the 16-item scale created through factor analysis by Triandis and Gelfand (1998) and
further investigated by Soh and Leong (2002). The primary dimensions of individualism
and collectivism were further subdivided into horizontal (emphasizing equality) and
vertical (emphasizing hierarchy) components, resulting in the four dimensions of vertical
and horizontal collectivism, and vertical and horizontal individualism. Adaptations to the
original measure included revised wording to reflect the target age group and setting, as
well as the addition of the 0 10 scale used throughout most of this study. Reliability
coefficient (Cronbachs a) for the whole scale was .76, and for the four sub-dimensions
as follows: horizontal individualism .61, vertical individualism .73, horizontal collectivism
.57, and vertical collectivism .68.
Screening
Prior to analysis, the data were screened for normality, linearity, and homogeneity
of variance. Two of the variablespast performance (source of self-efficacy), and social
persuasion (source of self-efficacy)showed a significant degree of leptokurtosis (i.e.,
above 3.0 when the expected value is 0) and were transformed (reflection and square
root). However, the results of subsequent analyses (analyses of variance and multiple
regression) were not affected, and so the variables were maintained as is to improve
interpretability. Nine students (5 Indo-Canadians and 4 Anglo-Canadians) were excluded
from the study as they were multivariate outliers detected through analysis of
standardized residuals in multiple regression.
Qualitative Data Gathering
60
After the quantitative data collection portion of the study was completed,
teachers were again contacted and asked to nominate small groups of students who would
be likely to participate actively in focus group discussions. As suggested by Krueger
(1994), the focus groups were made up of homogeneous cultural groups. Five separate
groups were nominated from two schoolstwo groups of Indo-Canadian students (11
students total) and three groups of Anglo-Canadians (13 students total). Note that these
students had previously participated in the quantitative data collection process. Students
who were nominated by the teachers, and who agreed to participate, were given consent
forms to sign, and seen for a single 30-45 minute session that took place at the students
school. The two groups of Indo-Canadian students were selected from a private Sikh
school; the three Anglo-Canadian groups were selected from a private Christian school.
For the Indo-Canadian students, the sessions took place in a school boardroom, with
students seated around a rectangular table, and the facilitator, with tape recorder, at the
head of the table. The three Anglo-Canadian groups met in an empty classroom, sitting in
a loosely arranged circle.
To begin each session, students were reminded that their participation in the focus
group session was voluntary, and were assured that their responses would be kept
anonymous. Subsequently, the facilitator explained the procedure for the session (i.e.,
facilitator-posed question, followed by individual responses and group discussion). Initial
rapport was established through a brief introductory section in which each student and
the facilitator introduced him- or her-self and noted either a favourite food, or a career
aspiration. A short discussion followed in which students were asked about their opinion
61
of the value of school in general, and math class in particular. A script (Appendix B)
based on the projects main research questions was loosely followed during the sessions.
The set questions were posed to the students who were encouraged to answer as fully as
possible, and to think back about their own school experiences from the last year. The
tone of the group sessions was informal, and the students appeared to comfortably and
freely discuss the questions and topics raised. The audiotapes of the sessions were
transcribed verbatim.
Analysis of the qualitative data was based on procedures suggested by Miles and
Huberman (1994) and Krueger (1994) and focused on searching for trends and patterns
that occurred within and across groups. After the transcription of the audiotapes, a
conceptually ordered display matrix (Miles & Huberman, 1994) was developed to
facilitate data analysis. Data from the transcription were coded according to pre-existing
codes developed from the quantitative analysis. Code categories included role of self-
efficacy in academic performance (unsolicited and solicited comments); sources of self-
efficacy; preferences for group and individual learning structures; and relative influence of
family, peers, and teachers in fostering self-efficacy and academic confidence. Discussion
of calibration arose spontaneously in the groups; therefore, it was included as a category
in the data matrix. The data matrix was examined for patterns, themes, and unexpected
responses within and across cultural groups; results are presented in Chapter 4.
62
CHAPTER 3
Quantitative Results
Overview
This chapter presents the results of the quantitative analyses conducted in the
study. The research questions, as introduced in Chapter 1, are used to structure the
results. In general, comparisons of means for the two groups are presented first, followed
by further investigation of the correlations and predictiveness of the relevant variables. To
begin, the differences in levels of the motivational factors between the groups are
examined, followed by presentation of the relationships among the variables. An
examination of the sources of self-efficacy is presented next: first, a comparison of levels
between the groups, and second, an exploration of how the four sources differentially
predict efficacy beliefs for the two groups. Differences in how the groups calibrate their
efficacy with performance are then investigated, followed by an investigation of collective
efficacy and group functioning. The final quantitative analysis presented examines the
cultural dimensions of horizontal and vertical individualism and collectivism, and how
they do or do not explain differences in self-efficacy beliefs in this sample. The results
from analysis of the qualitative data collection are presented in Chapter 4.
Cross-Cultural Comparison of Self-Efficacy and other Motivation Beliefs
Previous research has found cross-cultural differences in a number of motivational
beliefs, including self-efficacy. Other studies (e.g., Fulgini, 1997; Pajares & Graham, 1999;
Pajares & Miller, 1997) have found significant sex differences in the self-efficacy and
63
other motivational beliefs of adolescents. To examine the relationship between the
independent variables of cultural group and sex, and the dependent variables of math
performance, and the four motivational variables (i.e., math self-efficacy and the three
secondary motivational variables of self-concept, fear of failure, and perceived parental
value of academics), a 2 X 2 between groups multivariate analysis of variance, using SPSS
MANOVA, was performed. Using Wilks criterion, the combined DVs were not affected
by the interaction of cultural group and sex, F(1, 266) = 1.90, p = .09. The relationship
between sex and the dependent variables was significant F(1, 266) = 4.78, p < .01,
h
2
= .08. Follow-up univariate tests showed a significant sex difference for math self-
concept, F(1, 266) = 11.17, p < .01, but effect size was low, h
2
= .04 . Although there
were mean self-concept differences between the sexes (male 7.37, female 6.59), there was
no interaction between sex and cultural group. A significant relationship between cultural
group and the dependent variables was found F(1, 266) = 38.87, p < .01. This reflects a
strong association between cultural group and the combined dependent variables, h
2
= .43.
As a follow-up to the multivariate analysis, univariate comparisons were made. In
Table 2, mean scores and standard deviations for math performance, self-efficacy, fear of
failure, math self-concept, and perceived parental view of math for the two groups are
presented. The Indo-Canadian group scored significantly higher than the Anglo-Canadian
group (mean scores of 13.41 and 12.37, respectively) on the math performance variable,
F(1, 266) = 4.53, p =.03, h
2
= .02.
64
Table 2
Means and Standard Deviations for Math Performance, Motivational Variables and
Calibration
Anglo-
Canadian
Group
(n = 112)
Indo-Canadian Group
(n = 158)
M SD M SD
h
2
Math
performance
*
12.37 4.02 13.41 3.56 .02
Math self-
efficacy
**
7.21 1.72 7.79 1.56 .03
Math self-
concept
6.75 2.19 7.06 1.77 <.01
Fear of
failure
**
6.78 1.74 8.61 1.18 .28
Parental
importance
**
6.50 1.65 8.64 .92 .41
Mean
accuracy
7.17 1.17 7.15 1.14 <.01
Mean bias 1.29 1.45 1.38 1.41 <.01
*
p < .05,
**
p < .01
65
Among the motivation variables, there were significant differences in self-efficacy,
F (1, 266) = 8.78, p < .01, h
2
= .03; fear of failure F(1, 266) = 102.74, p < .01, h
2
= .28;
and perceived parents value of academics, F(1, 266) = 172.85, p < .01, h
2
= .41, with all
means higher in the Indo-Canadian group.
Prediction of Performance
A primary research question in this study is the differential predictive
contributions of self-efficacy and other motivational variables to math performance for
Anglo-Canadian and Indo-Canadian early adolescents. Within-group correlations of math
performance, previous performance (math grade), self-efficacy, self-concept, fear of
failure, and parents value of academic success are presented in Table 3. For the Anglo-
Canadian group, math self-efficacy showed the strongest relationship with math
performance (r = .62), followed by previous math grade (r = .50) and math self-concept
(r = .48). In the case of the Indo-Canadian students, math self-efficacy was most closely
correlated with math performance (r = .64), followed by math self-concept (r = .53) and
previous math grade (r = .46).
Sequential (hierarchical) multiple regression was used to determine if self-efficacy
improved prediction of math performance beyond that afforded by previous math
performance and the secondary motivation variables of math self-concept, fear of failure,
and parents perceived value of academic success. Table 4 presents the results from
sequential multiple regression analysesone for each of the two cultural groups. For the
Anglo-Canadian group, the entry of previous performance (math grade) on step 1 resulted
66
Table 3
Within-Group Correlations for Math Performance, Previous Math Grade, and
Motivation Variables
1 2 3 4 5 6
1. Math
performance
_ .46
**
.64
**
.53
**
.08 .22
*
2. Previous
math grade
.50
**
_ .36
**
.60
**
-.04 .00
3. Math self-
efficacy
.62
**
.44
**
_ .41
**
.21
*
.33
**
4. Math self-
concept
.48
**
.66
**
.54
**
_ .25
*
.21
*
5. Fear of failure .20
*
.19
*
.22
*
.34
**
_ .47
**
6. Parental
Importance
.27
*
.16 .32
*
.29
*
.58
**
_
Note. Correlations for Anglo-Canadians appear below the diagonal; correlations for Indo-
Canadians appear above the diagonal.
*
p < .05,
**
p < .01
67
Table 4
Final Summary of Sequential (Hierarchical) Multiple Regression for Variables
Predicting Math Performance
Variable
B
SE B R R
2
Anglo-Canadians
.68
**
.46
Self-efficacy 1.09 .21 .47
**
Math grade .76 .29 .25
*
Parental importance .18 .22 .08
Math self-concept .08 .19 .04
Fear of failure -.02 .21 -.01
Indo-Canadians
.71
**
.51
Self-efficacy 1.16 .15 .51
**
Math self-concept .57 .16 .28
**
Math grade .29 .20 .11
Fear of failure -.23 .20 -.08
Parental importance -.11 .26 -.03
*
p < .05,
**
p < .01
68
in an R
2
of .25, p < .01. With the entry on step 2 of the secondary motivation variables of
fear of failure, math self-concept, and perceived parents value of academic success, the R
2
increased to .31, p < .01. Finally, the entry of self-efficacy significantly raised the R
2
to
.46, F(5, 106), p < .01. After entry of all of the variables, self-efficacy was the strongest
predictor of math performance, = .47, p < .01, followed by math grade, = .25, p < .01.
Math self-concept, parental importance, and fear of failure did not contribute significantly
to the regression equation.
For the Indo-Canadian group, the entry of previous performance (math grade) on
step 1 resulted in an R
2
of .21, p < .01. With the entry on step 2 of the secondary
motivation variables of fear of failure, math self-concept, and perceived parents value of
academic success, the R
2
increased to .32, p < .01. Finally, the entry of self-efficacy
significantly raised the R
2
to .51, F(5, 152), p < .01. After entry of all of the variables,
self-efficacy was the strongest predictor of math performance, = .51, p < .01, followed
by math self-concept, = .28, p < .01. Previous math grade, parental importance, and fear
of failure did not contribute significantly to the regression equation.
Sources of Self-Efficacy
Results from one-way ANOVAs comparing each of the four sources of self-
efficacy across the two groups are presented in Table 5. There were significant differences
for the first variable, past performance, F(1, 268) = 17.91, p < .01, with the Indo-
Canadian group rating the variable higher (M = 8.09) than the Anglo-Canadian group
69
Table 5
Sources of Self-Efficacy: Means and Standard Deviations
Anglo-Canadian Group
(n = 112)
Indo-Canadian Group
(n = 158)
M SD M SD
h
2
Past performance
**
7.28 1.89 8.09 1.22 .06
Vicarious learning
**
4.93 2.08 5.84 1.92 .05
Social persuasion
**
7.64 2.17 8.66 1.30 .09
Emotional arousal 4.65 2.23 4.53 2.24 <.01
**
p < .01
70
(M = 7.28). The strength of association was modest, h
2
= .06. Comparisons for the
vicarious experience variable also revealed significant differences, F(1, 268) = 13.96, p <
.01. This second source of self-efficacyvicarious learningwas again rated lower by the
Anglo-Canadians than by the other group. The effect size was modest, h
2
= .06.
For the third variablesocial persuasionthe Indo-Canadian group showed higher
ratings, F(1, 268) = 22.92, p < .01, h
2
= .09. For the fourth variableemotional
arousalthere were no significant differences between the groups, F(2, 302) = .94, p =
.39.
Correlations between self-efficacy and the four efficacy sources followed the same
pattern for each cultural group. All four variables were significantly related (p < .01) with
self-efficacy. For the whole sample, past performance was the most strongly related (r =
.35), followed by vicarious experience(r = .28), social persuasion (r = .26), and emotional
arousal (r = -.24). The negative correlation between emotional arousal and self-efficacy is
reflective of the primarily negative interpretation of emotional arousal used in this study;
i.e., similar to anxiety or fear. The variable is not reversed in order to keep interpretation
consistent with typical definitions of emotional arousal in the self-efficacy literature.
A standard multiple regression was run (Table 6) using SPSS REGRESSION with
math self-efficacy as the dependent variable, and the four sources of efficacy
beliefspast performance, vicarious learning, social persuasion, and emotional
arousalas the four independent variables. The independent variables were entered
71
Table 6
Summary of Standard Multiple Regression for Variables Predicting Self-Efficacy
Variable
B
SE B R R
2
Anglo-Canadians
.50 .25
Emotional arousal -.26 .07 -.34
**
Past performance .28 .11 .31
*
Vicarious experience .09 .08 .11
Social persuasion .003 .10 .004
Indo-Canadians
.39 .15
Emotional arousal -.14 .05 -.21
**
Past performance .25 .13 .19
Vicarious experience .13 .07 .17
*
Social persuasion .09 .12 .07
*
p < .05,
**
p < .01
72
simultaneously. For the Anglo-Canadian group R for regression (R = .50, R
2
= .25) was
significant, F(4, 107) = 8.82, p < .01. Two of the four IVs were significant predictors of
self-efficacy: emotional arousal, = -.34, p < .01; and past performance, = .31, p = .02.
The non-significant -weights for vicarious experience and social persuasion were .11, and
.004, respectively.
For the Indo-Canadian sample, R was significantly different from zero (R = .39, R
2
= .15, F(4, 153) = 6.91, p < .01). Two of the four IVs were significant. Emotional arousal
was a significant predictor, = -.21, p = .006, as was vicarious learning, = .17, p = .05.
The variable of past performance approached significance, = .19, p = .06, whereas social
persuasion was not significant, = .07, p = .45.
Calibration of Efficacy Beliefs
Calibration was assessed using several methods, including mean accuracy and
mean bias (see Pajares & Graham, 1999). For mean accuracy, responses on the math
measure were scored dichotomously as 10 (correct response) or 0 (incorrect response),
and then subtracted from the efficacy judgment score that ranged from 0 to 10. Next, the
absolute value of the difference was subtracted from 10. The resulting number is an
indicator of the magnitude of calibration, with higher scores indicating better calibration
between efficacy and performance on the math measure. For the mean bias method, the
dichotomous score from the math measure was subtracted from the self-efficacy score,
with the results ranging from 10 through to +10. The arithmetic mean of these scores
reveals the direction of mis-calibration. There were no differences between cultural groups
for either mean accuracy or mean bias (see Table 2).
73
The complete sample was divided into three calibration groups: under-estimators,
moderate over-estimators, and severe over-estimators. Twenty percent of the sample was
classified as under-estimators, which was defined as mean bias score below 0. The
remainder of the sample was divided into two groups, with moderate over-estimators (40
%) scoring between 0 and 1.65 on the mean bias measure, and severe over-estimators (40
%) scoring above 1.65 on the mean bias scale.
The three calibration groups (under-estimators, moderate over-estimators, and
severe over-estimators) were compared for performance on the math task. The under-
estimators (M = 14.30, SD = 3.53), and the moderate over-estimators (M = 14.27, SD =
3.54) scored significantly higher than the severe over-estimators (M = 11.41, SD = 3.50),
F(2, 267), p < .01, h
2
= .14. There was no significant interaction between the three
calibration groups and the two cultural groups.
Finally, the bivariate correlations between self-efficacy and performance can be
viewed as an indicator of calibration. The Pearson product-moment correlation coefficient
for self-efficacy and performance was r = .62 for the Anglo-Canadian group, and r = .64
for the Indo-Canadian group (not significantly different).
Collective Efficacy and the Group Task
Collective efficacyitem-level and globalwas assessed (along with self-
efficacy) for a group task that involved solving word and logic puzzles in small groups.
Table 7 displays the means and standard deviations for performance, global collective
74
Table 7
Group Task: Performance Score, Self- and Collective Efficacy, and Calibration
Anglo-Canadian Group
(n = 112)
Indo-Canadian Group
(n = 158)
M SD M SD
h
2
Group task score 9.79 4.18 8.96 3.56 .01
Self-efficacy 8.55 1.57 8.21 1.41 .01
Collective efficacy
*
9.29 1.07 9.03 .93 .02
Global collective efficacy 7.80 1.33 8.11 1.25 .01
Mean accuracy 3.97 2.04 4.18 1.65 <.01
Mean bias 2.78 2.56 2.69 2.28 <.01
*
p < .05
75
efficacy, self-efficacy, and item-level collective efficacy, as well as mean accuracy and
mean bias. The global collective efficacy measure (adapted from Riggs et al., 1994)
focused on perceptions of group functioning, and included items such as The group I
work with is better than average, and I try to avoid disagreements when working in a
group. For the self-efficacy variable, students were asked, How confident are you that
you yourself can solve this question? Item-level collective efficacy was measured in the
same way as self-efficacy, i.e., a 0-10 scale was used for each of the 18 word puzzles, and
students were asked, How confident are you that your group can solve this question?
One-way ANOVAs were conducted to assess differences between groups for
performance, collective and self-efficacy, and the calibration variables of mean accuracy
and mean bias.
There were no significant differences between the two groups for the performance
score, self-efficacy, global collective efficacy, or calibration. A statistically significant
between-group difference was found for item-level collective efficacy, F(1, 266) = 4.90, p
= .03, with the Anglo-Canadian group rating their collective efficacy higher (M = 9.29)
than the Indo-Canadian group (M = 9.03). The strength of association, or effect size, was
small, h
2
= .04.
In contrast to the item-level collective efficacy, which corresponded very closely
to the performance task, the global collective efficacy measure assessed an individuals
generalized perception of group functioning. There were no significant differences found
between the groups for this variable. However, an analysis of item-level differences
76
revealed some significant differences between groups. Contrary to expectations, the
Anglo-Canadian group rated I enjoy working in a group significantly higher than did the
Indo-Canadian group F(1, 266) = 6.86, p = .01, h
2
= .07. The groups rated I would
rather work alone than in a group differently, with the Indo-Canadian group rating the
item higher than the Anglo-Canadian group, F(1, 266) = 10.78, p < .01, h
2
= .06.
Bivariate Correlations and Multiple Regression Analyses
Within-group correlations for the group task, the three efficacy variables, previous
math grade, and math self-concept (the only significant predictor of math performance of
the three secondary motivation variables) are presented for the two groups in Table 8.
For the Anglo-Canadians, variables that showed significant correlations with group
performance included previous math grade (r = .50), followed by collective efficacy (r =
.40), self-concept (r = .36), and global collective efficacy (r = .28). For the Indo-Canadian
group, the variables that were significantly correlated with group performance included
previous math grade (r = .42), self-concept (r = .33), and collective efficacy (r = .27).
Self-efficacy for the group task and global collective efficacy were not significantly
correlated with performance.
Table 9 presents the results of standard multiple regression analyses that were
conducted to explore the relative contributions of item-level collective and self-efficacy,
global collective efficacy, math self-concept, and previous math grade to the DV,
performance score on the group task. Independent variables were entered simultaneously.
For the Anglo-Canadian group, R
2
after the entry of all variables was .41, p < .01.
Three of the five predictors were significant (and positive), with item-level
77
Table 8
Within-Group Correlations for Group Task Performance and Motivation
Variables
1 2 3 4 5 6
1. Group task
performance
.42
**
.27
*
.16 .08 .33
**
2. Math grade .50
**
.06 .28
**
-.11 .60
**
3. Collective efficacy .40
**
.25
*
.55
**
.15 .14
4. Self-efficacy .15 .28
*
.67
**
.13 .25
*
5. Global collective
efficacy
.28
*
.17 .07 -.04 .11
6. Math self-concept .36
**
.66
**
.27
*
.30
*
.18
Note. Correlations for Anglo-Canadians appear below the diagonal; correlations for the
Indo-Canadians appear above the diagonal.
*
p < .05,
**
p < .01
78
Table 9
Summary of Standard Multiple Regression for Variables Predicting Group Performance
Variable
B
SE B R R
2
Anglo-Canadians
.64
**
.41
Collective efficacy 1.80 .40 .46
**
Math grade 1.38 .32 .44
**
Self-efficacy -.73 .28 -.27
*
Global group efficacy .51 .24 .16
*
Math self-concept .005 .19 .004
Indo-Canadians
.51
**
.26
Math grade 1.10 .26 .40
**
Collective efficacy 1.14 .34 .30
**
Self-efficacy -.36 .26 -.13
Global group efficacy .27 .21 .09
Math self-concept .12 .19 .06
*
p < .05,
**
p < .01
79
collective efficacy the strongest predictor, = .46, p < .01. Math grade was the next
strongest predictor ( = .44), followed by global collective efficacy ( = .16). Self-concept
did not predict group performance (although when math grade was not included in the
regression, math self-concept was a strong predictor). Self-efficacy was a negative
predictor, and apparently acts as a suppressor variable in this regression. An impure
suppressor variable is slightly correlated with the dependent variable and strongly
correlated with another independent variablein this case, collective efficacy (Woolley,
1997; see also Cohen & Cohen, 1983, Pedhazur, 1982). The suppressor variable typically
increases the predictive validity of another predictor variable by suppressing irrelevant
variance, while remaining largely unrelated to the dependent variable (Tabachnick &
Fidell, 1996).
For the Indo-Canadian group, R
2
was .51, p < .01. Two of the predictors were
significant, with math grade ( = .40) and collective efficacy ( = .30) both predicting
math performance. Self-efficacy was a non-significant, negative predictor (again, a
suppressor variable), while global efficacy and math self-concept did not predict group
performance.
Individualism, Collectivism and Self-Efficacy across Cultural Groups
The dimensions of individualism and collectivism have been used to help with the
understanding of differences in perceptions of separateness and connectedness of
individuals and cultural groups. As suggested by Triandis and Gelfand (1998), in this
80
study these two dimensions have been further subdivided into horizontal and vertical
dimensions, with the four resulting sub-dimensions of horizontal and vertical
collectivism, and horizontal and vertical individualism. Table 10 displays the means and
standard deviations for the two groups.
Between-group analysis of variance (ANOVA) was used to explore differences in
these four dimensions. No statistically significant between-group difference was found for
horizontal collectivism or horizontal individualism. A statistically significant between-
group difference was found for vertical collectivism, F(2, 268) = 68.85, p < .01, h
2
= .20.
A statistically significant between-group difference was also found for vertical
individualism, F(1, 268) = 29.44, p < .01, h
2
= .10. As seen in Table10, the Indo-Canadian
group showed significantly higher means for both vertical collectivism and vertical
individualism.
Because there is frequently as much variation within groups as between group in
terms of individualism and collectivism (Oyserman et al., 2002), and because one question
of interest in this study pertained to the across-culture relationships between I/C and the
motivation variables, correlations between the four dimensions of I/C and the efficacy
variables (math self-efficacy, item-level collective efficacy, and global collective efficacy)
are presented for the entire sample in Table 11. Math self-efficacy was significantly but
weakly correlated with horizontal individualism (r = .17), vertical collectivism (r = .18)
and vertical individualism (r = .16). Item-level collective efficacy was not significantly
correlated with any of the four cultural dimensions, whereas global collective efficacy was
significantly correlated with horizontal (r = .33) and vertical (r = .28) collectivism.
81
Table 10
Horizontal and Vertical Individualism and Collectivism
Anglo-Canadian Group
(n = 112)
Indo-Canadian Group
(n = 158)
M SD M SD
h
2
Horizontal
collectivism
6.54 1.60 6.87 1.56 .01
Vertical
collectivism
**
6.81 1.86 8.40 1.29 .20
Horizontal
individualism
6.65 1.66 6.10 1.59 .02
Vertical
individualism
**
3.41 2.14 4.88 2.22 .10
**
p < .01
82
Table 11
Between-Group Correlations between Horizontal and Vertical Individualism and
Collectivism, and Self- and Collective Efficacy Variables
1 2 3 4 5 6 7
1. Horizontal
collectivism
.14
*
.49
**
-.04 .10 .04 .33
**
2. Horizontal
individualism
.26
**
.35
**
.17
*
-.01 .05
3. Vertical
collectivism
.23
**
.18
*
.02 .28
**
4. Vertical
individualism
.16
*
-.04 .08
5. Math self-
efficacy
.32
**
.12
6. Item-level
collective efficacy
.09
7. Global collective
efficacy

*
p < .05,
**
p < .01
83
In order to examine the utility of the four I/C dimensions to predict cultural group
membership, direct logistic regression analysis was performed using SPSS LOGISITIC
REGRESSION to predict membership in either the Anglo-Canadian or Indo-Canadian
cultural groups on the basis of the four dimensions of I/C: horizontal and vertical
collectivism, and horizontal and vertical individualism. The full model, tested against a
constant-only model, was statistically reliable, c
2
(4, N = 270) = 80.43, p < .01,
suggesting that the predictors, as a set, reliably distinguished between Anglo-Canadians
and Indo-Canadians. The overall correct prediction rate was 70%, with success rates of
56.25% for the Anglo-Canadian group, compared to 79.75% for the Indo-Canadian group.
See Figure 1 for a histogram of observed groups and predicted probabilities. A look at the
Wald criterion and levels of significance shows only vertical collectivism (p < .01) and
vertical individualism (p < .01) as variables that reliably predict cultural group. However,
as recommended by Tabachnick and Fidell (1996), a less stringent criterion for inclusion,
such as .15, would allow recognition of horizontal collectivism as a further reliable
predictor variable.
Response Bias
The Indo-Canadian group rated most self-report variables higher than did the
Anglo-Canadians groups. Of the 4 efficacy variables, 3 secondary motivation variables,
4 sources of self-efficacy, and 4 dimensions of I/C, the Indo-Canadian group rated 8 (of
15) significantly higher, while the Anglo-Canadian group rated 1 variable significantly
higher. It may be that the Indo-Canadian group responses were skewed by a response
84
bias, such as an extreme (positive) response style, sometimes found in cross-cultural
comparisons (Chun, Campbell, & Yoo, 1974; Grimm & Church, 1999).
Correction for response style was considered in this study; however, a number of
arguments can be made for not carrying out a corrective procedure. First, there is little or no
previous research that discusses response biases of similar research participants adolescent
children of Indian Sikh parents. It is difficult to separate content variance from response
style variance (Grimm & Church, 1999), especially when little is known about the typical
response style (or psychological traits) of a group. Second, the Indo-Canadian group scored
higher on the math performance task. One might predict higher motivation variables in this
case. Indeed, on the group performance task, the Anglo-Canadian group scored higher, and
displayed significantly higher collective efficacy, and (non-significantly) higher self-efficacy.
Finally, an examination of extreme response styles in this study was inconclusive. For the
sources of self-efficacy variables, which were chosen for this purpose because they should not
be affected by math performance, the Indo-Canadian students chose the extreme positive
response about as often as did the Anglo-Canadian students. In the end, it was felt that
correcting for response bias was not warranted, especially given the equivocal results of
previous studies that have attempted to correct for this problem (see Grimm & Church,
1999). Nevertheless, some caution should be used in interpreting the findings from this study.
85
CHAPTER 4
Qualitative Results
Overview
Qualitative research can be used to add depth to quantitative research results
(McInerney, 1998). In this thesis, the relationship between the qualitative and
quantitative components falls into the dominant/less dominant model explained by
Cresswell (2002), in which priority is placed on the quantitative data collection and
analysis, and the smaller qualitative section is used to explain or elaborate the quantitative
results (Cresswell, 2002).
Five focus group sessionsthree Anglo-Canadian groups (total = 13 participants)
and two Indo-Canadian groups (total = 11 participants)were conducted to further
explore cross-cultural dimensions of efficacy beliefs in academic functioning.
After an initial rapport-building discussion about interests, school and math class
(see Chapter 2), a series of questions was asked to elicit discussion about:
why some students do better than others (even when there are no differences in prior
knowledge),
what factors affect academic confidence to do well (sources of self-efficacy),
parental, peer and teacher influence on academic functioning,
preferences for group over individual functioning.
In the primary quantitative section of this study, considerable information was
gathered about how efficacy beliefs function with early adolescents from two cultures.
For this secondary qualitative section, students were first asked, with no reference to
86
confidence or self-efficacy from the researcher, to discuss the motivational factors they
considered relevant to academic, specifically math, performance. The first question, then,
underlying this qualitative investigation pertained to students actual unsolicited beliefs
about self-efficacy: Does some form of self-efficacy come to mind when early adolescents
are asked about factors that contribute to their performance?
Self-Efficacy
The Indo-Canadian students made several unsolicited allusions to self-efficacy,
typically couched under the term confidence. Representative comments included:
Some students have less self-confidence - maybe they dont believe they could pass it and
just give up, so they dont try.... You might leave out some questions, because you think
you dont know how to do some questions, but you really do.... For me, if you study it and
you know everything on the test youll be more confident that youll at least pass it,
probably.... It (performance on a math test) actually depends on how much confidence you
have - so if you have lots of confidence you can probably pass the test.... If you have no
confidence in your self, you say No, Im not going to do good, so you just forget it, so you
dont try as hard. More commonly, students referred to differences in ability: Not
everyones brain functions the same way.... Some people might understand another subject
better and some might understand math better.... Maybe they have more experience.
The Anglo-Canadian group also made unsolicited references to self-efficacy beliefs
when asked Why do some students do better in math than others, even when theyre
equally prepared? Apart from comments referring to differences in ability (Theyre just
born smarter.... Theyre mechanically brained.... Theyre born to write tests), students
87
also made comments about self-efficacy, although with less frequency than the Indo-
Canadian students: Its like, Im not going to do good!....If you dont feel youre going
to do it, youre not going to do it.... If you feel that youre prepared and confident youll do
well, youll do better.
Calibration
When asked specifically about the role of confidence, students from both cultural
groups gave mixed responses, and alluded to the issue of calibration: If you have lots of
confidence, it makes you think the test is easier and you pass it.... But it wont make that
much difference - if you think youre not going to do good, you wont study.... If youre
confident that youll pass, youll probably pass, but if youre not so sure about it then
theres a more likely chance you wont get as good marks as others. Banduras (1997)
hypothesis that high levels of efficacy promote performance was supported by some
comments, e.g., It actually depends on the confidence you have - so if you have lots of
confidence you can probably pass your tests.... If you have lots of confidence, it makes you
think the test is easier and you pass it.
Mis-calibration of efficacy and performance was a theme noted by a few focus
group participants in both cultural groups: Sometimes people are very too much
confident in themselves, and they think they know it all, and they end up knowing
nothing....And some people think like they have no confidence, and they go No, I wont
pass, and they get such a good mark.... Maybe because the other person thinks theyre
better, so they dont study.... If you have less confidence, youll do better - if you have less
88
confidence, youll study, and youll get better (but) if you have more confidence, youll
already know that you dont have to study.
Sources of Self-Efficacy
When asked about the factors influencing confidence (sources of self-efficacy
beliefs), participants mentioned all four of the hypothesized sources. For the Indo-
Canadian group, emotional arousal was the most frequently mentioned source (comprising
33 % of responses that were categorized under sources of self-efficacy), followed by
vicarious experience (30 %), past performance (18 %), and social persuasion (18 %). For
the Anglo-Canadian group, emotional arousal and past performance each comprised 33 %
of the responses, followed by vicarious experience (22 %) and social persuasion (11 %).
For both cultural groups, emotional arousal was typically defined as nervousness or not
being able to think clearly: If I get nervous, I get scared that the test is going to be hard,
and my confidence goes down.... Yeah, you get nervous sometimes, and it adds a lot of
pressure. If youre under a lot of pressure, your mind goes blank, like five seconds ago
you knew everything, but you get the test and you blank out.... If youre too nervous and
youre thinking about it and at the end you dont have enough time to finish it and you just
rush through everything because youre too nervous.... Some people get nervous and it
blocks out your brain, and they get nervous and they dont really think as much.... The
nervous thing - it blocks your brain. Youre too busy thinking, Ive got to do well, Ive got
to do well, Ive got to do well! And then you do badly.... It makes you feel more energetic
if you have more confidence, if you dont have confidence, you drag yourself around...
89
Past performance was typically referenced as previous grades: Like, you know if
you have a good grade in the past, sometimes you think, You know what? I might get a
good grade again.... So if you get a bad mark last time, you think Id better study
harder.... How youve done in the past affects you - you know youre going to do well.
Sources of social persuasion included peers, family, and teachers: When people in
school tell me Im good at it, then I want to get better.... Theyre (my parents) always
saying that Im good at this. They say always, no matter what happens, it doesnt matter,
as long as youve tried your hardest, but maybe its not the best thing for you to do.... My
parents and my teacher encourage me.... My parents tell me Im doing well - because like
my dads a teacher, and hes really good at math; also, my brother and sisters sometimes
tell me Im doing well... Not my teacher, but friends and family.
The reference group for vicarious experience was typically peers in the classroom,
and most comments included aspects of social comparison: When other kids do well, it
kind of challenges me to do well... When the people around me do better I want to try
harder and do better - it becomes a challenge.... If you see others do well, you think Oh,
this guy can do it, and so I can too, and youre kind of motivated to do as well as the guy
beside you.
Both groups noted that the sources of self-efficacy could be positive as well as
negative (i.e., inverse) contributors to confidence: I think if you had better marks in the
past youll have poor marks right now because youll think I did really good in the past
and I dont need to study now (past performance).... Sometimes if other kids do really
well, I give up (vicarious experience).... If your teachers or parents say youre really good,
90
you wont try as hard, and you say Yeah, Im good at it, so Im just going to do whatever
(social persuasion).... Friends have a lot to do with it (confidence), because if theyre your
age, say if they do better than you, you lose your confidence.
Family, Friends, Teachers
Two of the four sources of efficacy beliefssocial persuasion and vicarious
experiencerely on exposure to significant others. The question of cultural differences in
the relative influence of peers, family, and teachers on the formation of efficacy beliefs
was posed to each of the groups. For the Indo-Canadian students, family was the most-
mentioned influence on efficacy beliefs (70 %), followed by peers (25 %) and teachers (5
%). Some examples of family responses included Family is important, because if my
brother does good on tests, my parents want me to do well.... My family gives me the
courage to do well, and then I do well.... Family gives you confidence in doing good, and
they help you study.... My parents compare me to my cousins, like they go Why is Amerjit
doing good, and why arent you?
The Anglo-Canadian group also made more references to family as a primary
source of efficacy beliefs than to peers or teachers (63 % family, 21% peers, and 16 %
teachers). Comments included If your parents want you to do well, and theyre really
pushing you to do well, and youre kind of like I better do well on this test, or theyll get
mad,.... I dont mind if my parents are confident in me - thats OK - but if they force me to
do it, then it just gets you frustrated on the test.... Friends and family - you spend a lot of
time at home with your family and they ask you questions, but your friends affect your
school life.... Not my teacher (What affects your confidence?), but both friends and
91
family.... If my parents think Im good at something, they keep telling you that, and you
think you can do well, and you do better.
When asked about the influence of cultural background on their confidence, few
students in the Anglo-Canadian group attributed any weight to their or their parents
cultural background. Only one student made reference to his own background: No, it
doesnt matter if Im like Scottish like my Dad, or Dutch, like my Mom. Several students
discussed the confidence and performance of other cultural groups, specifically Asian:
For some students, like if theyre Asian, if they dont get A+, theyre really sad.... Some
students, like Asians, are too hard on themselves and they lose their confidence - like they
get As and Bs, but they want like 99 %.
The Indo-Canadian students made more frequent reference to their parents
cultural background and educational experiences: My parents think school is everything -
if you dont have school you dont have a living.... It (school) was really hard there (India)
and they think that our work is easy, and if you fail they get mad. They hit you with sticks
there (India).... They compare, they compare.... Well, my family came here 10 years ago,
and my uncle did really well in school and he expects the same from us.... They want us to
become more than they did. They dont want us to feel the same way they do now.... Its
like in our family, some people say that people from India are better in math. My Dad says
its because they were there and its pretty much what they studied most. But I dont know,
Im not sure about it. I think its pretty much the same for us, we were born here, and we
only go to India for visits, its not like we studied there....Sometimes they (parents) think
that now theres a lot of East Indians here, and we have a lot of confidence in ourselves
92
now, but before when there were only a few, they probably thought themselves as being
less than others. And they probably thought they needed to work even harder, because they
werent as smart as the other people here, but nowadays its not that big of a deal,
because its pretty much the same.
Group versus Individual Functioning
Students were asked how classroom structure (groupwork versus individual work)
affected their confidence. The proportion of students favouring group work over
individual work was identical for each cultural group (50%). The Indo-Canadians (40 %)
favoured individual work more than the Anglo-Canadians (20 %). The proportion of
students whose responses included both individual and group work was 10 % for the
Indo-Canadians and 30 % for the Anglo-Canadians. Responses from the Indo-Canadian
focus groups included It kind of makes my confidence the same, because the other people
might not be that good in math, so I have to do everything.... Working in groups makes
my confidence go up, because if sometimes I dont get something by myself, the others
help.... Im kind of a group person, because the people you talk to during math, or during
lunch, or hang out with, they can help you during class.... It doesnt matter to me, youre
going to have to do the work by yourself, anyway.... More individual, I prefer like in math,
I think its better to work individually, but occasionally you want to work in a group just
for fun.... Its better to do everything yourself because youll really get what youre doing,
and you wont be relying on everyone else.
Among the Anglo-Canadians, comments about the relative influence of group and
individual functioning on confidence included Groups make me feel more confident,
93
because youre around people you can talk to and you can compare answers and figure
out stuff.... It (confidence) sort of stays the same.... Im more confident in groups, because
if you dont know that subject well, you can learn from the others.... If they dont like your
ideas, or if youre not accepted, you dont feel part of it.... I like working by myself, but its
fun to ask others to help you... I get more confident in a smaller group - I consider myself
a small-group person, like two or three people.
94
CHAPTER 5
Discussion
Overview
In this section, I discuss and interpret the research findings presented in Chapters
3 and 4 in the context of this study and previous related research. I then consider the
implications of these findings in terms of educational practice and future research, and
elaborate the limitations of this study.
Academic Performance
The academic performance of the Indo-Canadian group was impressive, with
children of Indian-born Punjabi Sikhs scoring, on average, higher than students from non-
immigrant parents. There was no significant difference in terms of reported math grades.
There is little previous research about the academic performance of children of immigrant
Indian parents, but considerable evidence (e.g., Fulgini, 1997) showing that children from
East Asian (as opposed to South Asian) immigrant backgrounds out-perform children of
native-born parents in the United States. Gibson (1988) in her sociological investigation
of Indian Sikhs (Punjabis) who had immigrated to California, found high school seniors
who were children of Punjabi parents received better grades than non-Punjabi classmates,
and were less often placed in remedial math classes (or in advanced math classes). In the
current study, children of immigrant Sikhs scored as well or better than children of non-
immigrant, English-speaking Canadians on a math task, and reported math grades that
were not significantly different from their Anglo-Canadian peers. It is not likely that the
95
differences in math task performance shown in this study are due to SES characteristics,
with no difference between the two groups for fathers level of education.
Self-Efficacy
The self-efficacy ratings of the Indo-Canadian students were significantly higher
than for the Anglo-Canadian students, although effect size was small. On average, Indo-
Canadians rated their efficacy to complete math tasks at 7.79 (on a 0-10 scale) whereas
the Anglo-Canadians displayed mean ratings of 7.21. Performance on the criterial
taskthe 22-item math testfollowed a similar pattern. The Indo-Canadian students
earned a mean score of 13.41, whereas the Anglo-Canadians earned a mean score of 12.37.
There were no main or interaction effects for gender. Self-efficacy ratings, then, paralleled
the performance scores.
Although a certain amount of research has explored the self-efficacy beliefs of East
Asian students and children of East Asian immigrants to North America, little is known
about how students from a South Asian background might rate their efficacy. Most
research of groups from East Asian backgrounds converges on the findings that in general,
East Asians display relatively higher academic performance, but more modest efficacy
beliefs (e.g., Eaton & Dembo, 1997; Salili et al., 2001). From this study, as will be seen, it
might be concluded that self-efficacy beliefs do not follow this pattern for students from a
South Asian cultural background.
Self-efficacy was strongly correlated with math performance for both groups, and
was the strongest predictor of math performance for both the Anglo-Canadian and Indo-
Canadian students, even after the entry of past performance (math grades) and other
96
motivational variables in a regression analysis. In this sample of early adolescents, self-
efficacy was a strong and relevant predictor of performance, regardless of cultural
background. Although a good deal of criticism has been leveled at motivation research due
to its relatively narrow (in cultural terms) research base, the results from this study
incrementally add to our cross-cultural knowledge base, and specifically to our
understanding of how self-efficacy operates in this particular non-Western culture.
Results from the qualitative data support the quantitative findings. Self-efficacy
(or at least aspects of efficacy, like confidence) was a valid and meaningful concept for
students in both cultural groups: It (performance) actually depends on the confidence you
have so if you have lots of confidence you can probably pass your tests. When students
were asked to speculate about which factors might contribute to performance differences
in similarly skilled individuals, responses included references to innate ability, but also to
self-beliefs about capabilities. In the same way that math self-efficacy was the primary
factor predicting academic performance in the quantitative segment of this study, so too
were efficacy or confidence beliefs predominant in a qualitative assessment of students
beliefs about factors contributing to math performance.
Calibration
Most of the students (80 %) in this study were, on average, somewhat or very
optimistic about their performance, whereas only 20 % tended to underestimate their
ability to complete specific math items. This unexpected tendency of optimism and/or
accuracy of calibration for both cultural groups may be a reflection of the chosen
domainmathand the self-efficacy measures used in this study. Using item-level
97
efficacy measures ensures a high degree of specificity and correspondence, as articulated
by Pajares (1996b), and a strong relationship between beliefs and performance. The level
of specificity possible in math efficacy measurement is considerably higher than in other
academic domains, such as writing. Consider the difference in specificity between How
confident are you that you can do well in math? and How confident are you that you
can solve this question: 71 18 = x?
There were no differences between the Indo-Canadian and Anglo-Canadian groups
in terms of calibration mean accuracy or mean bias. In other words, these two groups
displayed a tendency to correlate their efficacy beliefs and performance at about the same
level. Bivariate correlations (another measure of calibration) reveal that self-efficacy was
strongly related to subsequent performance for both cultural groups; in fact, self-efficacy
was more closely associated with current performance than was previous math grade, or
any of the other three motivation variables included in this study.
Students who under-estimated or moderately overestimated their math functioning
performed best on the criterial task. Bandura (1997) suggests that a certain amount of
optimism fosters optimal performance; in this study, the most optimistic students
(across cultural groups) performed at a lower level than moderately optimistic or
pessimistic students. Consistent with Banduras warning (1995) that gross
miscalculation can get one into trouble (p. 12), results from this study indicate that
students who drastically overestimate their capabilities tend to perform poorly, regardless
of cultural background.
98
Not all students agreed that confidence aided performance: If you have less
confidence, youll do better but this comment referred to pre-performance preparation,
rather than task-related efficacy. In the qualitative investigation, a few students made
references to mis-calibrationboth overconfidence and underconfidenceof efficacy and
performance: Sometimes people are very too much confident. Some people think like
they have no confidence and they get such a good mark. Students utterances about
calibration or mis-calibration were limited, and no cross-cultural differences were noted.
Students in both cultural groups tended to agree that a certain level of confidence
improved performance, and that strong, but not unrealistic, confidence was optimal.
Secondary Motivation Variables
There were differences in levels of two of the secondary motivational
beliefsperceived parental attitudes and fear of failure. Comparisons between the two
principal groups reveal that the Indo-Canadian group labeled both of these variables
higher than did the Anglo-Canadian group. Fulgini (1997) found that the academic success
of first and second generation immigrant students was correlated with the students
perception that their parents placed a strong emphasis on educational attainment. In
contrast, children of native-born parents did not possess as strong a perception of
parental value of education. In the current study, children of fathers born in India, and
who spoke Punjabi in the home, also perceived their parents to hold higher expectations
for their educational attainment than did children of those born in Canada or the United
States, and who reported speaking only English in the home. The correlations between
perception of parents view of education and performance were about equal for the two
99
cultural groups; in terms of prediction of math functioning, parents value of education
was not a significant predictor for either group, although the beta-weight was stronger for
the Anglo-Canadian students.
In the qualitative segment of data collection, Indo-Canadian students made
reference to culturally influenced parents value of education whereas the Anglo-Canadian
students did not. For the children of immigrant Punjabi Sikhs, their perception of their
parents value of education was strong and pronouncedMy parents think school is
everything if you dont have school, you dont have a living. They want us to become
more than they did they dont want us to feel the same way they do now. When the
Anglo-Canadian students mentioned family influences, the comments were more typically
reflective of support and encouragement, i.e., My parents tell me Im doing well. Fulgini
(1997) also found that American children of Asian immigrants perceived their parents to
value academic success more than did the students of European background. In the current
study, as in Fulgini (1997), student perceptions of a familys values do not directly
predict performance, but might be channeled through the individual attitudes, work
habits, and beliefs.
Fear of failure, too, was rated more highly by Indo-Canadian students than by
Anglo-Canadian students. For the Asian American ninth-grade students in the study
conducted by Eaton and Dembo (1997), fear of failure was the variable most highly
correlated with achievement; for the non-Asian group, fear of failure was not significantly
correlated with achievement. In this study, although fear of failure was rated significantly
higher by the Indo-Canadian group than by the Anglo-Canadian group, it was less highly
100
correlated with math achievement or with previous math grade. Indo-Canadians in this
study tended to agree more strongly that Doing badly in school will hurt my chances for
a good job in the future, but these beliefs were not associated with success on a math
task or with previous math performance. Again, as with perceived parental value of
academic success, the fear of failure variable might be reflected in the more proximal
beliefs and attitudeslike self-efficacy and self-conceptthat do directly predict
performance.
Prediction of Performance
Comparisons between the two groups revealed significant differences in the levels
of self-efficacy, fear of failure, and perceived parental attitudes. Beyond investigating
these mean differences, this study explored how these variables (along with previous
math grade and self-concept) differentially predicted current math performance.
Results from this study reveal that self-efficacy makes a significant independent
contribution to math performance for Anglo-Canadian and Indo-Canadian early
adolescents, even when other variables shown to predict academic performance in other
cross-cultural contexts are included in the regression equation. Self-efficacy was, in fact,
the only variable that significantly predicted performance in both groups. Standard
multiple regression analyses revealed differences between the two principal groups in the
order of significant variables contributing to math scores. For the Anglo-Canadian group,
only self-efficacy and previous math grade reached significance, with perceived parental
value of academics, math self-concept, and fear of failure failing to contribute. In the case
of the Indo-Canadian group, math self-efficacy was once again the strongest predictor, but
101
self-concept, instead of previous performance was the other significant contributor to
prediction of math performance.
The motivational variable of math self-concept was shown to be a significant
predictor of performance for the Indo-Canadian, but not for the Anglo-Canadian students.
Previous researchers have suggested that self-concept measures are more likely to assess
affective as well as cognitive facets of functioning, and are heavily influenced by social
comparison (Bong & Clark, 1999; Marsh 1992). In contrast, efficacy self-appraisals
center on cognitive appraisals of judgments of ones capabilities (Bong & Clark, 1999).
As suggested by Oettingen (1995) cultural dimensions influence how students form their
self-appraisals. It may well be that for the Indo-Canadian students social comparison is a
more salient feature than it is for Anglo-Canadian students.
These results strongly support the predictive power of efficacy beliefs in cross-
cultural settings. Although Eaton and Dembo (1997) found both self-efficacy and fear of
failure to predict performance in their (East) Asian American and non-Asian American 9
th
grade students, in this study, self-efficacy was found to be the only common predictor,
with fear of failure a non-significant predictor for both groups. Eaton and Dembo
speculated that (East) Asian American students might be more prone to view efficacy on
a group level based on a sense of shared responsibility to achieve (p. 438). Here,
however, both cultural groups appear to draw primarily on their individual self-beliefs (as
opposed to a more group-oriented motivation variable like fear of failure or perception of
parents value of education) when faced with individual tasks. The relative differences in
the significance of the next strongest predictorsprevious math grade (a form of enactive
102
experience), for the Anglo-Canadians, and math self-concept (which includes a social
comparison aspect), for the Indo-Canadianslead to questions and discussion about the
relative importance of the sources of self-efficacy.
Sources of Self-Efficacy
The sources of self-efficacy beliefs have not previously been investigated from a
cross-cultural perspective. Bandura (1997) proposed four sources of self-efficacy
beliefspast performance, vicarious experience, social persuasion, and
emotional/physiological arousalof which past performance is posited to be the most
influential source. In this study, the multiple regression analyses conducted for the Anglo-
Canadian and Indo-Canadian groups suggest that emotional arousal has a strong (inverse)
relationship with math self-efficacy beliefs for early adolescents from immigrant and non-
immigrant families. (To measure emotional arousal, students were asked how certain
emotional/physiological events caused them to lose confidence. As a result, these scores
predict self-efficacy inversely). The four sources of efficacy beliefs measured in this
study explained more of the self-efficacy variance for the Anglo-Canadian group than for
the Indo-Canadian group (R
2
= .25 for Anglo-Canadians, .15 for Indo-Canadians).
For the Anglo-Canadian students, past performance was the second strongest
significant predictor of efficacy beliefs, whereas for the Indo-Canadians, vicarious
experience was the next strongest significant predictor. This finding may be consistent
with differences between the groups seen in prediction of math performance. For the
Anglo-Canadians, self-efficacy and previous math grade were the only two significant
predictors of math performance. For the Indo-Canadians, self-efficacy and self-concept
103
were the only two variables that significantly predicted math performance. Similarly, in
the case of variables predicting self-efficacy beliefs, after the initial influence of emotional
arousal, the Anglo-Canadians relied on past performance to instill efficacy beliefs;
however, for the Indo-Canadians, vicarious experiencea variable that includes aspects of
social comparison (as does self-concept)was the next most influential significant
predictor. The self-efficacy beliefs of both groups of early adolescents appear to be
influenced by a substantial affective/emotional component, contrary to previous
theoretical suppositions (e.g., Bong & Clark, 1999) that focused on the cognitive aspects
of efficacy beliefs. Also of interest is the finding that Indo-Canadian early adolescents
appear to embrace aspects of social comparisonboth in the formation of efficacy beliefs
and in the prediction of performancemore readily than do the Anglo-Canadian early
adolescents.
Are these findings of a strong emotional component in the make-up of efficacy
beliefs supported by previous research? Anderson and Betz (2001) found emotional
arousal and past performance to be the two strongest predictors of social self-efficacy
among university undergraduates in the U.S. Likewise, Lopez and Lent (1992) found past
performance and emotional arousal as the two strongest predictors of the math self-
efficacy of American high school students. In a study of Japanese undergraduates, Matsui
et al. (1990) found three of four efficacy sourcespast performance, emotional arousal,
and modeling (vicarious experience)to significantly predict math functioning. However,
Hampton (1998) did not find that physical (emotional) arousal and social persuasion
contributed to the prediction of academic self-efficacy for older high school students.
104
From these studies, it can be seen that past performance and emotional arousal are the
two efficacy sources with the most empirical support; support for the other two sources
is less clear and appears to shift depending on the context and the measures used.
In this study, the variance of math self-efficacy accounted for by the four self-
efficacy sources was moderate (25 % and 15 %, for the Anglo-Canadians and Indo-
Canadians, respectively). Bandura (1997) hypothesizes that the self-efficacy source
vicarious experience includes aspects of both modeling and social comparison. It may be
that an aspect of comparison with peers is lacking from the way in which vicarious
experience was operationalized in this study. Consider the difference between I feel
confident when one of my friends does well in math (which includes the modeling aspect
of vicarious experience) and Compared to others my age I am good at math (which
includes the social comparison aspect of math self-concept). Bong and Clark (1999)
suggested that for younger children, the lines between self-efficacy and self-concept are
blurred. For this sample of early adolescents, and perhaps especially for the Indo-
Canadian students, the sources of self-efficacy might ideally have included a heavier
emphasis on the social comparison aspect of vicarious experience, as opposed to an
emphasis on social modeling; indeed, for the sample as a whole, the addition of self-
concept as a variable predicting self-efficacy boosted the explained variance from R
2
= .21
to R
2
= .30.
In focus group discussion, both Indo-Canadian and Anglo-Canadian students
attributed their efficacy beliefs to each of the four sources of efficacy, with emotional
arousal acknowledged as the strongest predictor of efficacy beliefs. This qualitative
105
finding supports the quantitative finding of emotional arousal contributing more strongly
to efficacy beliefs than past performance. The role of anxiety in math was much
discussed, with both groups referring to problems with their thinking/brains/minds being
frozen or blocked or blanked. Students did not recognize that a certain level of
emotional arousal might be beneficialall comments referred to physiological indices that
hampered performance. The other three sources of efficacy were occasionally viewed
with ambiguity; i.e., a few students noted that vicarious observation of similar others
success could be detrimentalSometimes if other kids do really well, I give up. So too
with social persuasionIf your teachers or parents say youre really good, you wont try
as hard and past performanceYoull think I did really well in the past and I dont
need to study now.
Bandura (1997) discusses how the discerning, weighing, and integrating of efficacy
information changes with development. Self-appraisal and self-knowledge develop over
time from evaluation of beliefs and subsequent outcomes: for early adolescents, it may be
that more attention is paid to somatic information than other sources of efficacy. The data
from this study suggest that in terms of perceptions of math functioning, early
adolescents of both Anglo-Canadian and Indo-Canadian backgrounds appear to first and
foremost evaluate their emotional status, and only after this initial anxiety check do
they evaluate other sources of efficacy. Contrary to the beliefs of many teachers, verbal
persuasion does not appear to have much of an impact on self-efficacy for either of the
two cultural groups investigated in this study. The participants in this research also
appear to be well engaged in the process of social comparison as a way of appraising
106
personal capabilities. In the future, measurement of vicarious experience should
encompass aspects of comparison as well as modeling.
Collective Efficacy and Group Functioning across Cultures
Item-level collective efficacy ratings paralleled the performance scores on the
group task, with the Anglo-Canadian group showing higher group performance along with
higher item-level collective efficacy beliefs. Item-level collective efficacy and performance
were significantly correlated for both groups; however, for the Indo-Canadian group the
relationship between self-concept and performance was higher than for collective efficacy
and performance. Global efficacy was significantly correlated with performance for the
Anglo-Canadian group, but not for the Indo-Canadian group. Self-efficacy for the group
task (Rate your confidence to solve these puzzles and problems on your own in 15
minutes) was not significantly correlated with group performance for either group.
In the multiple regression analyses, item-level collective efficacy and previous
math grade were the strongest predictors of group performance for both cultural groups.
Self-efficacy was an inverse predictor for both groups: it appears to operate as a
suppressor variable (see Chapter 3) that is strongly correlated with another independent
variable (item-level collective efficacy) but weakly related to the dependent variable.
Global collective efficacy significantly predicted performance for the Anglo-Canadians,
but not for the Indo-Canadians; math self-concept was not a significant predictor for
either group.
Previous math performance and collective efficacy were the strongest predictors
of performance, while self-efficacy and self-concept were not predictive. It may be that
107
the nature of the group task contributed to this difference. The word problems/puzzles
used in this study were somewhat ambiguous and were difficult to evaluate on first
reading. Self-efficacy for the group task was measured through the direct method
students read the word problems and then estimated their own capability to solve the
task, followed by their groups capability of solving the task. Self-efficacy beliefs are
highly dependent on appraisal of the criterial task. Early adolescent students in this study
may have been unsure of what the criterial task entailed: Im not really sure what this
task is maybe I can solve it, whereas for collective efficacy, perhaps an appraisal of the
qualities of the group (a combined internal/external frame of reference both other- and
self-related) was more salient: my group seems pretty smart Im sure we can solve
whatever this task is.
The specificity of item-level self-efficacy measurement is a double-edged sword
(Marsh, Roche, Pajares, & Miller, 1997) that, although powerfully predictive, limits the
generalizability of efficacy beliefs to broader criteria. Marsh et al. (1997) argue that
whereas efficacy beliefs that are measured at the item level predict performance
effectively, it is unlikely that the self-efficacy will have any meaningful causal influence
on the performance test scores (p. 375). In this study, students collective efficacy was a
significant predictor for both individual and group tasks; item-level self-efficacy did
predict individual math performance, but failed to predict performance on a group task.
Collective efficacy appears to be a relevant predictor of early adolescents group
functioning in cross-cultural educational settings. Similar to the case with self-efficacy and
individual math functioning, collective efficacy was a strong and significant predictor of
108
functioning across cultural settings, although perhaps surprisingly in light of the assumed
collectivism of the Indo-Canadian group, collective efficacy was a stronger predictor of
group functioning for the Anglo-Canadian students. In both cultural groupings of early
adolescents, the judgment concerning confidence in the groups capability to solve a task
predicted level of achievement whereas judgments of self-capability did not. Self- and
collective efficacy were strongly correlated for both groups, but self-efficacy appeared to
act as a suppressor variable and did not predict performance. As proposed by Bandura
(1997) it appears that group or collective efficacy operates as an emergent property that
is the product of the interactive and coordinative dynamics of its members (p. 477);
that is, it continues to influence performance even after past performance and individual
efficacy beliefs are accounted for.
The qualitative investigation found that more Indo-Canadians (40 %) favored
individual work than Anglo-Canadians (20 %) whereas proportion of students choosing
group work was the same (50 %) for both groups. However, the themes of working
alone versus working together were found in both cultural groups: Working in groups
makes my confidence go up. Its better to do everything yourself because youll really
get what youre doing, and you wont be relying on everyone else (Indo-Canadian
students), and Im more confident in groups. I like working by myself (Anglo-
Canadian students). This finding of the Asian or non-Western students preferring
individual work to group work is not necessarily unexpected. The assumed collectivist
nature of Asian students does not necessarily translate into a preference for academic
settings that are group-oriented. Volet (2001) found that Asian international students
109
studying in Australia did not display more positive appraisals of group assignments than
Australian students. In this study, Indo-Canadian early adolescents do not appear to
prefer group work. It has been seen that self- and collective efficacy beliefs are important
variables that explain performance for both cultural groups; however, what is not clear is
how the cultural dimensions of individualism and collectivism might influence the
formation of these beliefs.
Individualism, Collectivism, and Efficacy Beliefs
Indo-Canadian students exhibited significantly higher vertical collectivism and
vertical individualism scores than the Anglo-Canadian students. There were no significant
differences between the two groups on the horizontal dimensions of individualism or
collectivism. Correlations (for the combined cultural groups) between the four cultural
dimensions and the three efficacy beliefs revealed significant, positive relationships
between both horizontal and vertical collectivism and global collective efficacy (r = .33
and r = .28, respectively), suggesting that students who highly rated elements of
sociability, family integrity, cooperation and interdependence, also rated highly aspects of
the desirability and value of group work. More moderate correlations (r = .16 to .18) were
found between the three cultural dimensions of horizontal collectivism, vertical
collectivism, and vertical individualism and the motivation variable of math self-efficacy.
Perhaps the strongest conclusion that can be drawn from these results is that students of
any culture who are allocentric (i.e., the individual corollary to group collectivism) report
higher satisfaction and efficacy with the groups in which they work.
110
Several hypotheses might help explain the absence of horizontal collectivism and
presence of vertical individualism in the Indo-Canadian sample, that might have been
expected based on Hofstedes work (1980). First, the dimensions of I/C might not be
universally useful in explaining human functioning. Although Hofstede reported findings
for several dozen countries, including India, in his 1980 landmark study, the conclusions
he drew from his sample of IBM employees may not generalize well to other segments of
the population. Second, little is known about developmental aspects of I/C, and early
adolescents may not reflect these dimensions in the same way as adults. Third, Punjabi
Sikhs might not reflect the cultural groups of other dominant Indian cultures. As stated
previously, Gibson (1988) found American-born Punjabi Sikh students to choose an
individualist orientation almost as often as other American students. Other researchers
(Agarwal & Misra, 1986; Mishra, 1994; Sinha et al., 2001; Sinha & Tripathi, 1994) note
the coexistence of opposites (Sinha & Tripathi, 1994) in terms of the I/C dimensions.
Mishras analysis of generational and urban/rural differences in I/C revealed the existence
of both individualist and collectivist values among people at the same time, though the
former were held somewhat more strongly than the latter (p. 236). The results from the
current study suggest a similar coexistence of opposite beliefs, although with the added
overlay of information about the verticalness and horizontalness of Indo-Canadian
students.
The choice of cultural lens (e.g., I/C) might be invoked according to contextual
clues. Students in an academic setting may choose a primarily individualist, competitive
lens with which to view their personal and group functioning, even when predominantly
111
collectivist in other settings. Triandis (1989) postulated that people sample different
aspects of self depending on cultural background and current setting. Life changes, such as
immigration (especially rural to urban migration), influence a person to look inward to the
private self, where previously the tendency might have been to sample the collective self
(Triandis, 1989). Similarly, Hong and Chiu (2001) suggest multicultural individuals can
spontaneously change the cultural lenses that are available to them through multicultural
learning, depending on which cultural theories are activated by contextual clues (p. 193).
Indian selfhood, according to Sinha and Tripathi (1994) is contextual, and is conditioned
by the exigencies of the situation (p. 124). The Indo-Canadian students sense of self in
this specific context was not solely interdependent or collectivist.
More strongly, the cultural dichotomies of individualism and collectivism may be
insufficiently developed to meet the challenges of a multicultural society. Hermans and
Kempen (1998) proposed that particular attention should be paid to contact zones,
such as are engendered through immigration, where cultures meet and interact with
resulting shifts in self and identity. Certainly, schools such as those included in this study
can be considered as existing in the contact zones between cultures; Indo-Canadian
students in these schools live in two different cultures, and are doubtless influenced by
contrasting sets of beliefs and values. A dichotomous measure of individualism and
collectivism may be inadequate to capture the nuances of these students beliefs. In this
study, the vertical and horizontal elements proved useful to help explain differences
between the two groups in cultural beliefs.
112
The results from the logistic regression suggest that the vertical cultural
dimensions are reasonably good predictors of group membership. For the most part,
Indo-Canadian students were more vertical, whereas (and this is less well-supported)
Anglo-Canadians were more horizontal. In their discussion of the vertical and horizontal
dimensions of individualism and collectivism, Triandis and Gelfand (1998) suggest that
the vertical component primarily emphasizes hierarchy and authoritarianism. They
describe three elements that would be found in common in the vertical dimensions of both
individualism and collectivism: self different from others, authority ranking, and low
equality. In vertical individualism, people competitively strive for high status and are self-
reliant; vertical collectivism is found in traditional communal societies with strong leaders
that stress sociability, authoritarianism, and importance/integrity of family. Angelo
(1997), in his work describing both tradition and change in Sikh immigrant
communities, notes that Indians increase in individualism and independence upon
immigration to North America. Traditional Sikh culture is a hierarchical society (in
which) relationships are not based upon egalitarian precepts but on a vertical basis of
authority and obedience (Angelo, 1997, p. 119). Gibson (1988) emphasizes the
importance of family integrity to Sikh immigrants, but also of family status, and of
maintaining and improving their familys good name (p. 110). Likewise, Kwak and
Berry (2001) found parental authority to be more strongly endorsed by Indo-Canadian
parents than by Anglo-Celtic, Korean Canadian, or Vietnamese Canadian parents.
Immigration to North America provides the impetus for greater individualism for Indians
who are leaving a largely collectivist environment: both time and distance cause Indians
113
to become more individualistic and independent after coming to America (Angelo, 1997,
p. 126). In sum, there is considerable support for the immigrant Sikhs concurrent cultural
beliefs of vertical individualism and vertical collectivism.
Conclusions
This thesis has examined the role of efficacy beliefs in two different cultural
groupsAnglo-Canadian and Indo-Canadian early adolescents. As is often the case in
cross-cultural research (McInerney, 1998; McInerney, Hinkley, Dowson, & Van Etten,
1998), there are more similarities than differences between the contrasted cultural groups.
Self-efficacy is a strong predictor of academic achievement in both groups; in fact, it has
been seen that self-efficacy is a better predictor of math performance than previous math
grade, or than the motivation constructs of self-concept, fear of failure, or perceived
parental value of academics. From the qualitative investigation, it has been shown that
both cultural groups spontaneously recognize the role that self-efficacy beliefs (viewed as
confidence) play in influencing academic outcomes. Both cultural groups tend to
calibrate their efficacy beliefs and performance at about the same level, with grossly
optimistic students performing poorly in both groups.
Notwithstanding the important similarities found between these two groups of
early adolescents, there were also a number of significant differences found in the way
that efficacy beliefs operate between these two cultural groups. It has been seen that
students from an Indo-Canadian cultural background may view the world in somewhat
different ways than students from an Anglo-Canadian cultural background. Indo-Canadian
early adolescents appear to be more vertical in their perceived relationships with others,
114
and exhibit higher levels of both vertical individualism and vertical collectivism.
Differences were found in the relevance of the secondary motivational constructs;
specifically, math self-concept significantly contributed to the math performance of Indo-
Canadians but not to the performance of Anglo-Canadian students. Social comparison (in
math self-concept), consistent perhaps, with the cultural dimension of verticalness,
appears to play a stronger role among Indo-Canadian students than among Anglo-
Canadian students. Although the Indo-Canadian students rated their fear of failure and
perceived parental value of academics more highly than did the Anglo-Canadians, there
were no differences in the predictive roles of these variables.
Differences in the patterns of the sources of self-efficacy were seen. For both
groups of students, emotional arousal was the strongest predictor of self-efficacy beliefs.
However, past performance contributed significantly to the self-efficacy of Anglo-
Canadians, but not for Indo-Canadians. Among the four sources of self-efficacy, verbal
persuasion played little part in predicting efficacy beliefs for either group. As has been
discussed, emphasizing the social comparison aspect of vicarious experience (in addition
to the aspect of social modeling) might result in a significant increase in the amount of
variance explained by the efficacy sources.
On the group task, collective efficacy paralleled the groups performance, with
higher efficacy and performance for the Anglo-Canadian group, and lower efficacy and
performance for the Indo-Canadian group. Collective efficacy was a relevant predictor of
small-group functioning for both cultural groups. The correlation between performance
and collective efficacy was significant for both groups, but individual past performance in
115
the form of previous math grade was more strongly related to group performance for both
groups than collective efficacy. Global collective efficacy was considerably less predictive
of performance than item-level collective efficacy, although global efficacy was predictive
of performance for the Anglo-Canadian students, but not for the Indo-Canadians.
A central question addressed in this thesis is If conception of self is markedly
different in non-Western cultures, and self-beliefs reflect these differences, are self-efficacy
beliefs relevant in all cultures? Specifically, do efficacy beliefsself- and
collectiveoperate differently in Indo-Canadian as compared to Anglo-Canadian early
adolescents? Numerous cross-cultural investigations in educational psychology have
explored differences between Western and non-Western individuals, or between
Asian and non-Asian students, with typical conclusions that Asian or Asian
American students display more modest self-beliefs in the face of superior performance.
Previous researchers have questioned the applicability of various self-beliefs to Asian
populations. Whang and Hancock (1994), for example, proposed theories of
achievement motivation are rooted in individualism and may have validity primarily for
American and other similar cultures (p. 315). Results from this study call into question
the practice of forming generalizations about people and immigrants from non-Western
populations. As noted by Fiske (2002), there is enormous diversity among Asian
cultures, and broad conclusions about the psychological functioning and self-beliefs of
non-Westerners or Asians may be misleading. The results from this study of Indo-
Canadian Punjabi Sikh and Anglo-Canadian adolescents suggest that self- and collective
efficacy appear to operate in a way that is generally similar across both groups, and
116
furthermore, strongly predict performance in both groups. Although the Indo-Canadian
students might view the world through a different cultural lens, with a stronger emphasis
on social comparison and social hierarchy, these students appear to hold self-efficacy and
collective efficacy beliefs in a way that is not unlike that of students from a Western
culture.
Triandis sampling probability model (1989) may help with conceptualizing how
an individual might differentially sample aspects of self over time, in diverse contexts, and
with life changes. According to Triandis and others (e.g., Markus and Kitayama, 1991)
who argue that people differentially sample their social environment according to cultural
background, the self is mutable, or changeable; at the very least, the ways in which
various aspects of self are sampled might change according to context. Markus and
Kitayama cite work by Mauss (1938/1985) who claimed that the self is a delicate social
category subject to significant, if not infinite, variation (p. 226). Fay (1996) suggests
that the self is a relational entitynot a noun but a verb, and not a fixed entity with
definite boundaries but a process whose nature (is) fluid and changeable (p. 39). In like
fashion, Martin and Sugarman (2001) conceive of the self as an ever changing, dynamic
process of understanding particular being (p. 107). Viewing the self as a verb (Fay,
1996) or as an understanding (Martin & Sugarman, 2001) helps to address how the self
(and consequently, self-efficacy beliefs) actively interacts with culture (Fiske et al.,
1998).
117
Future research
Future exploration of self- and collective efficacy might profitably examine the
processes through which recently arrived immigrants acquire the self-understanding and
motivational beliefs of the majority group. The Indo-Canadian group in this study clearly
functions within a contact zone (Hermans & Kempen, 1998) in which Punjabi Sikh
culture and majority Anglo-Canadian culture intersect. For immigrant Asian groups, views
of a variety of social structuresparental authority, childrens rights, childrens
obligations, and cultural traditionschange with degree of acculturation (Kwak & Berry,
2001). Ghuman (1994), in his study of the acculturation process of Indo-Canadian high
school students, found the mostly Punjabi Sikh adolescents to largely reject the inward
looking attitudes of their parents (not mixing with whites, sticking to their food, living in
the areas where the community has settled, rejecting traditional arranged marriages) (p.
239) but to remain positive about their religion and home language. Measurement tools
exist to assess degree of cultural assimilation, but measuring the process of change in self-
beliefs about motivational constructs may prove more difficult. In the context of the
current study, an exploration of changes over time in the vertical cultural dimensions of
I/C may help researchers to understand one aspect of self-understanding. Longitudinal
research investigating changes in the sources/formation of self-efficacy (including social
comparison) beliefs may also help with understanding of changes in motivational beliefs.
Another consideration for future research is investigation into how the
operationalization of self-efficacy, particularly with regards to degree of specificity,
influences research results. Previous research has shown that the way in which self-
118
efficacy is measured influences how it operates in predicting performance (Pajares,
Hartley, & Valiante, 2001). In this study for example, collective efficacy was measured at
the item level as well as at the more generalized global level. The two collective efficacy
measures operated quite independently: although both forms of efficacy were
significantly correlated with group performance for the Anglo-Canadian students, only
item-level collective efficacy was significantly correlated with the group task performance
for the Indo-Canadian students. On the other hand, item-level collective efficacy showed
no significant correlations with the four cultural dimensions explored, whereas global
collective efficacy was significantly correlated with both the vertical and horizontal forms
of collectivism.
The different findings from the two contrasting measures of collective efficacy
highlight the importance of theoretically grounding efficacy measurements. Both measures
used in this study purport to measure collective efficacy, yet their conceptual foundation
and the ways in which they operated were very different. The specificity of efficacy
measurement can vary enormously, but greater specificity of measurement is not always
desirable. Marsh et al. (1997) discuss The downside of standing too close to the trees in
item-level efficacy assessment. Using the exact same items to measure self-efficacy and
performance (as was done in the present study) may result in inflated correlations, and at
the same time negatively bias the role played by related constructs, like self-concept.
When self-efficacy is measured at a very specific level, and with very close
correspondence with the criterial task, it loses relevance outside of that specific context.
In this study, math self-conceptmeasured at the domain levelproved the more robust
119
variable, being significantly correlated with individual performance and group
performance, while item-level self-efficacy was significantly correlated only with
individual performance. Furthermore, because the self-efficacy measures were based on
specific math items, they possess very little external relevance in and of themselves,
whereas the broader math self-concept is shown to be both internally and externally valid
(from its use in previous research).
Although some previous work has been conducted on the formation and sources
of self-efficacy, little if any previous work has examined the sources of collective efficacy
beliefs. Bandura (1997) proposes that collective efficacy is rooted in self-efficacy, and
that a collection of inveterate self-doubters is not easily forged into a collectively
efficacious force (p. 480). Nonetheless, collective efficacy is made up of more than an
aggregation of self-efficacy beliefs, and the four self-efficacy sourcesenactive
performance, vicarious experience, social persuasion, and emotional arousaldo not all
map well on to appraisals of group functioning. From this current study, it can be
concluded that collective efficacy operates in somewhat similar ways for both Indo-
Canadian and Anglo-Canadian students, but little is known about its formation at this
point.
Implications for Education
Modern education needs cross-cultural psychology (Triandis, 2001b) in order
to foster understanding of the implications of living in a multicultural world. This study
has examined one aspect of human functioningefficacy beliefsin a restricted
groupearly adolescent Indo-Canadian and Anglo-Canadian grade 7 students. In many
120
ways, efficacy beliefs proved to be a potent and predictive motivational variable and
operated similarly across both cultural groups. However, self-efficacy for the Indo-
Canadian group appears to operate somewhat differently than the self-efficacy of East
Asian students as described in the literature. Whereas modest efficacy beliefs may
obscure high levels of functioning with East Asian students, for South Asian Indo-
Canadians, efficacy beliefs and performance appear to be more accurately calibrated.
Social persuasion in the form of positive affirmations of ability does not appear to
effectively increase self-efficacy in most students. In light of the finding that emotional
arousal has a strong negative effect on efficacy beliefs, teachers may wish to focus
attention on test-taking strategies designed to reduce anxiety. The Indo-Canadian students
in this study displayed tendencies of both individualism and collectivism, but with a
particular emphasis on the vertical aspects of both of these dimensions. Those working
with this population would be well served to understand the basis for the perhaps
unanticipated concern with social comparison and awareness of social hierarchy. At the
same time, neither the qualitative nor quantitative investigations in this study found
gender differences in confidence or self-efficacy beliefs, suggesting that again there are
more cross-cultural similarities than differences in this regard.
Limitations of the Study
One of the difficulties in this study was investigating the functioning of the in-
group or collective in multicultural settings. It may be that Indo-Canadian students are
more collectivist when working or socializing with their cultural in-group. The
multicultural nature of most school settings and many of the student-formed small groups
121
in this study perhaps reduced students feelings of in-group collectiveness. However, in
the culturally homogeneous private Sikh school, Indo-Canadian students did not report
higher levels of group confidence or item-level collective efficacy than the Indo-Canadians
in the culturally mixed public schools. Another difficulty of this study might be that the
group task was somewhat opaque for the students, and made for difficulties in
estimating efficacy beliefs. If efficacy beliefs are formed through an appraisal of task and
personal performance attributes (Oettingen, 1995), students in this study may have had
difficulty coming to an accurate representation of task demands. There are limitations to
examining group functioning using short-term groups (Earley, 1999) in which the cultural
dimension of collectiveness might be poorly primed (i.e., brought to mind). Also,
Prussia and Kinicki (1996) note that individual perceptions of group functioning are only
surrogates of a consensual assessment of group functioning. However, Bandura (1997)
cautions that assessing group functioning through group consensus may result in
seriously flawed conclusions that are swayed by the most influential group members. The
thoughtful collaborative critiques by self-concept researcher Marsh and self-efficacy
researcher Pajares and colleagues (1997) advise moving away from the previously
recommended practice of constructing efficacy measures from the exact same material
used in the performance task. It may be that the role played by self-efficacy in this study
was artificially inflated due to that practice. As recommended by Marsh et al. (1997)
future research, based on intertheoretical cross talk, may help clarify the respective
motivational roles played by the related constructs of self-efficacy and self-concept.