Professional Documents
Culture Documents
Denis Kolev
/1703578/
Dimitar Panchev
/1708387/
Mariya Yotova
/1717305/
Stanimira Marinova
/1719350/
Summary
In the following applied research project our group investigates if a difference exists
in the level of satisfaction with life among immigrants and native- born. This analysis
is based on all countries from the European Social Survey dataset, except Israel
because we want to examine the topic in question across European countries as a
whole. Through the observation of two factors, our report compares the extent to
which the immigrants and the native-born citizens in Europe differ in their level of
satisfaction with life. Using such a broad scope, we believe that this is a large enough
sample to highlight the differences within immigrants and natives in Europe as a
whole. We suppose that the tendency in various levels of satisfaction with life is built
up from different variables defining the countries. For this reason, our group decided
to choose eight different variables which we suppose are highly correlated to the
level of satisfaction with life. As a next step we grouped the chosen variables into two
factors which are as follows: satisfaction with public institutions and satisfaction with
personal life. Using the factors stated, we performed various analyses, in order to
compare the variation in their mean results. As a result, we observed that with
respect to satisfaction with public institution there is a statistically significant
difference among immigrants and natives. On the contrary, our analysis showed no
evidence of a difference between immigrants and natives when considering the
satisfaction with their personal life.
the established better conditions on the continent would improve immigrants life by
bringing them satisfaction with life. In considering that issue we must pay close
attention to the anti-discrimination law that states: people must be dealt with on an
equal basis regardless of age, sex, race, ethnicity, nationality, sexual orientation,
gender identity and sometimes religious and political opinions. .The promotion of
equality between people has managed to encourage progress in dealing with the
prejudices of Europeans that we have already talked about despite the fact that
discrimination is still present across countries. To explore that issue our goal is to find
whether the immigrants are more satisfied with the life they have in the country they
have established given their social, cultural and economical origin compared to the
native born population of the country. Our assumption is that immigrants have left
countries with lower quality of life in search of better conditions. For this reason we
expect them to be more satisfied with the country they live in than the native-born
who dont have a basis for comparison and are used to living in these conditions.
Nevertheless, the immigrants may feel cultural and social differences from their
native country which may affect their satisfaction with life. Furthermore, immigrants
may feel left out of the society, because of the social and ethnic prejudices.
According to Shin and Johnson (1978) life satisfaction is defined as a global
assessment of a persons quality of life according to his chosen criteria. While
keeping in mind these abstract ideas weve already discussed, we have decided to
perform a more concentrated study by using the data available to set up factors and
obtain results on the basis of which we could make conclusion for the aspect we are
exploring. To do so, we focus on 8 variables which we group into two factors. The
first factor we investigate is Satisfaction with public institutions. We regard this
factor as having much explanatory power to overall satisfaction with life. What we are
trying to explore here is whether our assumption, that institutions in the countries that
immigrants have left were less efficient, is correct and if that would be a reason for
them to judge their satisfaction with public institutions of the current country at a
higher level. Our second factor concentrates on Satisfaction with personal life as we
are interested in understanding how individuals asses their own life satisfaction in the
two groups. We expect that these results would improve our understanding of
whether Europeans have combated to a certain extent their prejudices and have
managed to accomplish the important social goal of sharing values like tolerance,
equality and acceptance of different people.
The focus of our paper is on an important aspect of contemporary Europe. It
concerns immigration, one of the biggest problems it faces nowadays. We would like
to offer a way to check whether the continent (using as an illustration both EU and
non-EU countries) does indeed bring better life for immigrants by observing their
satisfaction with life. This research gives answers to some of the question raised by
the application of the following techniques. First of all, Factor Analysis is used to
group the variables into the factors: Satisfaction with public institutions and
Satisfaction with personal life. As a second step, we use Analysis of Variance to
test for differences between immigrants and native residents for each of the factors.
We believe our research will serve as a very convenient first step in distinguishing
between the immigrants and natives with respect to two different aspects of
satisfaction with life. Nevertheless, it is up to future research to track these
differences.
Data Description
The data set used for this analysis is drawn from Round 6 of the European Social
Survey conducted in 2012 as being the most transparent and varied source on which
we could focus our research. Since the inquiry is focused on the examination of the
difference among immigrants and native-born in European countries as a whole, we
use all the countries from the dataset while excluding Israel for obvious reasons. We
define 8 variables that will be grouped into two factors later. For defining the two
groups in our analysis- immigrants and native-born, we have transformed the
Variable Born in country into Immigrant status, assuming that immigrants are the
individuals not born in the country to which they correspond. Yet, we realize that
there could be a small percentage of people who do not correspond to this way of
classification, but we believe that this percentage would have a negligibly small effect
on our analysis.
Taking into account that data sets may contain different problems such as: outliers,
missing values and errors, we start with description of the variables we use and
comment on data screening results:
Variable
Label
cntry
Country
stfeco
stfgov
stfdem
stfedu
stfhlth
happy
sedirlf
Have a sense of
direction in your life
plinsoc
immigrstatus
Native-born or
immigrant
Scale
{AL= Albania} ;{BE=Belgium};
{BG=Bulgaria};
{CH=Switzerland}
{0,Extremely dissatisfied} ;
{1,"1'"}; {2,"2"); {3,"3"}
{0,Extremely
dissatisfied};{1,"1'"};{2,"2");
{3,"3"}
{0,Extremely
dissatisfied};{1,"1'"};{2,"2");
{3,"3"}
{0,Extremely bad}; {1,"1'"};
{2,"2"); {3,"3"};{4,"4"}
{0,Extremely bad};
{1,"1};{2,"2");{3,"3"};{4,"4"}
{0,Extremely unhappy};
{1,"1'"};{2,"2");{3,"3"};{4,"4"}
{0,Not at all};
{1,"1'"};{2,"2");{3,"3"};{4,"4"}
{0,Bottom of our society};
{1,"1'"};{2,"2");{3,"3"};{4,"4"}
{0,Immigrant};{1,native-born}
Measurement
Nominal
Ordinal
Ordinal
Ordinal
Ordinal
Ordinal
Ordinal
Ordinal
Ordinal
Nominal
Table 1
Univariate Analysis
Given that sometimes datasets present problems such as missing values, outliers
and other errors we start with screening the data as a first step. With regard to
4
No. of Extremes
Mean
Std. Deviation
Count
Percent
Low
High
stfeco
51178
4.00
2.594
987
1.9
745
stfgov
50683
3.96
2.580
1482
2.8
521
stfdem
50162
5.13
2.578
2003
3.8
stfedu
49638
5.58
2.419
2527
4.8
1845
stfhlth
51583
5.10
2.650
582
1.1
happy
51760
7.08
2.105
405
.8
1841
plinsoc
51056
5.41
1.856
1109
2.1
1479
650
sedirlf
51201
6.99
2.186
964
1.8
2172
a. Number of cases outside the range (Mean - 2*SD, Mean + 2*SD).
Table 2
Multivariate Analysis
For analyzing our research question, we use the following techniques: Factor
Analysis (FA) and one-way ANOVA.
Factor Analysis
The basis of factor analysis is correlation. This interdependency model, that analyzes
patterns in the data, is used to reduce a set of variables to a smaller set of factors.
First of all, we have to check if there is sufficient correlation between the eight
variables that we use. Pearsons correlation coefficient, r, is related to the strength of
the linear relationship that exist between two variables. In order to analyse these
coefficients we take a look at the correlation matrix that shows
a
Correlation Matrix
us that a sufficient relationship between the variables exist.
Besides, on the basis of the determinant of the correlation
a. Determinant = .063
matrix that we obtain we conclude that sufficient correlation
exists since the value we get is low. At the same time, this
value is higher than 0 indicating that there is no multicollinearity.
Table 3
We proceed with performing Kaiser-Meyer-Olkin Test and Bartletts test. We take that
step because KMO test measures whether there is a linear correlation between the
variables. Its necessary to implement that test since obtaining results equal or close
to zero signals absence of linear relation and there is no purpose with continuing the
factor analysis. The result we get is satisfactory as it is above 0.8 revealing that the
partial correlations coefficients are small, which indicates that significant correlation
between the variables exists so we can form meaningful factors.
Moreover, we run Bartletts test for sphericity that shows the approximate ChiSquare, degrees of freedom and significance level. According to theory, at
5
significance level of 0 we can reject the null hypothesis stating that there is no
correlation between the variables. In our case, the significance level is .000 so the
null hypothesis is rejected and we conclude that there is correlation between the
variables.
KMO and Bartlett's Test
Kaiser-Meyer-Olkin Measure of Sampling Adequacy.
Bartlett's Test of Sphericity
Approx. Chi-Square
df
Sig.
.836
126136.816
28
.000
Table 4
Antiimage
Correl
ation
How satisfied
with present
state of
economy in
country
How satisfied
with the
national
government
How satisfied
with the way
democracy
works in
country
State of
education in
country
nowadays
State of health
services in
country
nowadays
How happy are
you
Your place in
society
Have a sense
of direction in
your life
.847
-.445
-.183
-.035
-.114
-.091
-.097
.001
-.376
-.036
-.069
.039
-.018
-.023
-.203
-.105
-.089
-.034
-.012
-.399
-.046
.002
-.014
-.101
-.070
.046
-.231
-.293
-.216
-.445
.807
-.183
-.376
.865
-.035
-.036
-.203
.838
-.114
-.069
-.105
-.399
.850
-.091
.039
-.089
-.046
-.101
.828
-.097
-.018
-.034
.002
-.070
-.231
.854
.001
-.023
-.012
-.014
.046
-.293
-.216
.760
On the basis of the performed tests we conclude that there is sufficient correlation
and common variance in our sample dataset and therefore we proceed with the
Factor Analysis.
We are at the phase at which we have to define the number of factors. We do so by
observing Eigen-values which are equal to the sum of the squared loadings for a
given factor. These values represent the proportion of total variance explained by
each factor in the initial solution. According to the Latent root criterion or Kaisers
rule the optimal number of factors to use is those that have an eigen-value above 1.
We take into account the fact that the use of a larger number of factors would not
explain much more of the variation, because unique variance starts to dominate
common variance. By observing the scree plot a plot of the Eigen-values against
the number of factors we consider extracting two factors as an adequate decision.
Chart 1
Our goal for this factor analysis is including only those variables for which the
extracted factors explain an adequate amount of variance.For this reason we
proceed with observation of the communalities. The extraction column of the table
represents the percentage of common variance given the extracted factor model.
Typically, it should be above 0.2. In our case, variables are above 0.36 which we
believe is sufficiently high extraction so we can safely go on.
Communalities
Extraction
.585
.623
.635
.386
.423
.482
.363
.364
We also observe the total sample variance explained by the two factors
looking at the Total variance explained table. We obtain considerably satisfactory
value close to 50 %.
Factor
Initial Eigenvalues
% of Variance Cumulative %
45.330
45.330
Total
3.626
1.286
16.070
61.400
.787
9.836
71.236
.636
7.950
79.186
.557
6.965
86.151
.437
5.462
91.613
.377
4.713
96.326
.294
3.674
100.000
.716
1.280
8.950
48.262
15.994
48.262
Pattern Matrix
Factor
1
How satisfied with the national
government
.775
.769
.729
.614
.594
.645
.599
.554
On the basis of the table above we notice that each variable is strongly associated
with only one of the factors. Then, each factor would be interpreted on the basis of
the variables that load strongly onto it. The variables included in our first factor are:
stfgov, stfdem, stfeco, stfhlth, stfedu. Our second factor includes the following
variables: happy, sedirlf, plinsoc.
Factor 1 represents: Satisfaction with public institutions
Factor 2 represents: Satisfaction with personal life
After having defined the factors, we find factor scores and create a new latent
variable for each factor that is used in the analysis.
ANOVA
We implement one-way Analysis of Variance for each of our two defined factors. By
doing so, we measure the difference among the means of different sampled groups,
in our case the difference in level of satisfaction with life between immigrants and
natives.
The assumptions for ANOVA are that populations are normally distributed with equal
variances and also that samples are randomly and independently drawn. So before
we continue, we perform some tests to assure if these assumptions are met.
9
First we run the test of variance homogeneity. The result shows p-values equal to
zero for our first factor, while the value for our second factor is equal to 0.497. We
interpret these results that for our first factor the assumption for homogeneity is not
met since we reject the null-hypothesis that states that variances in the
subpopulations are the same at a 0.05 significance level. Even though ANOVA is a
fairly robust to violation of this assumption, we must interpret the results of the
analysis with care especially since our two groups are not of approximately equal
sizes.
df1
df2
Sig.
23.013
48947
.000
.461
48947
.497
Table 9
We also include Normal P-P Plots in order to check if the other assumption of
ANOVA, stating that populations are normally distributed, is satisfied:
Chart 2
From the plots above we conclude that each factor exhibits normal distribution with
very little deviations. After we have checked if the assumptions of ANOVA are met,
we would now want to proceed with testing whether there is a difference in the mean
of level of satisfaction with life between the two groups we analyse: immigrants and
natives. In order to make conclusions, we take as our null hypothesis the hypothesis
that the two means are equal, and we hope that we will have sufficient information to
reject the hypothesis, thus proving that the two means are different. The hypothesis
we want to test is:
H0: For a given factor, natives = immigrants
H1: the two means are different
10
Satisfaction with
public institutions
Satisfaction with
personal life
Between Groups
Within Groups
ANOVA
Sum of Squares
375.499
36111.246
df
1
48946
Mean Square
375.499
.738
F
508.960
Sig.
.000
Total
Between Groups
Within Groups
36486.745
2.036
29124.389
48947
1
48946
2.036
.595
3.422
.064
Total
29126.426
48947
Table 10
After performing the analysis of variance, we can see from the table above that pvalue for our first factor it 0.000 therefore we can conclude that the null hypothesis is
rejected. We interpret that result as the presence of significant difference between
satisfaction with public institutions from the point of view of natives and from that of
immigrants. On the other hand, the p-value for our second factor shows us the
evidence that we cannot reject the null hypothesis. That result gives us the
explanation that no significant difference in satisfaction with personal life exists
between immigrants and natives.
Given the results that we obtain on the first tests we perform checking the
assumptions for ANOVA in which we found that the assumption for homogeneity is
not satisfied for our first factor, we decide to confirm our result with further test. We
decide to run Robust Test of Equality of Means which does not require the
assumption for homogeneity of variances to be met. From the table below, we can
observe that we obtain similar results to those stated above regarding mean for
immigrants and mean for natives for each factor.
Robust Tests of Equality of Means
a
Statistic
df1
df2
Sig.
Welch
462.282
4783.758
.000
Brown-Forsythe
462.282
4783.758
.000
Welch
3.382
4859.988
.066
Brown-Forsythe
3.382
4859.988
.066
a. Asymptotically F distributed.
Table 11
Conclusion
The analysis performed aims at investigating whether indeed a difference exists
between how satisfied with his or her life would an immigrant be with respect to the
native population of any country. In the course of our work, we distinguished two
factors that we examined, considering them to have much explanatory power to the
topic in question.
We explored the degree of satisfaction with life more in-depth by considering the
results for Satisfaction with public institutions and Satisfaction with personal life
as being our two factors. How people answer questions regarding the state of country
and its public institutions together with peoples perception for their own life are
diverse factors affecting the same general topic. As far as the former factor is
considered, our analysis showed evidence of the existence of a difference between
immigrants and natives satisfaction with public institutions. One of the main reasons
11
for immigration is the search for better life. Thus, we believe that this difference
stems from the fact that immigrants come from countries with worse conditions so
their perception of public institutions is higher than that of the natives who are used to
be living at that quality. With regard to the latter factor, results were just the opposite
it showed no sign of a statistically significant difference. However, we must pay
close attention to the interpretation of the result obtained since the data we used for
analysis is quite subjective.
In conclusion, we believe that our analysis has presented some interesting
observations. Nevertheless, the topic under research is quite complex and we would
suggest a more in-depth research taking into account the results we already
introduced.
References:
1. Power point lectures and slides
2. Shin & Johnson (1978)
3. Oxford Dictionaries Immigration
4. Wikipedia Anti-discrimination law
5. Europes Immigrant Problem Robert Weissberg
12