You are on page 1of 30

Journal of Mathematical Behavior 21 (2002) 371400

Collaborative inuences on emergent statistical thinking a case study


Helen L. Chick a, , Jane M. Watson b
a

Department of Science and Mathematics Education, University of Melbourne, Melbourne 3010, Vic., Australia b University of Tasmania, Hobart, Tasmania, Australia

Abstract The purpose of this case study is to examine how collaboration affects the emergent statistical thinking of a group of three Grade 6 boys. Results of previous studies of students in Grades 3, 6, and 9 suggested that (a) when nding and justifying associations in data sets students working in groups may produce higher level outcomes than those working individually, and (b) there are numerous factors that inuence the success or otherwise of collaborative activity. The current study, based on detailed analysis of video tape and transcripts of a group working collaboratively on a data handling task, documents various factors that affect collaboration and how these contribute to the attainment of desirable cognitive outcomes in terms of the task set. These outcomes are classied by emergent statistical themes and insight is gained into how na ve statistical thinking begins to develop during the collaborative process. Implications for educators and researchers are considered. 2002 Elsevier Science Inc. All rights reserved.
Keywords: Emergent statistical thinking; Collaboration; Elementary students; Data handling; Cognitive change

Collaboration while completing mathematical tasks has been the subject of numerous studies and debates by mathematics educators. Many authors (e.g., Davidson, 1985; Lindquist, 1989; Phelps & Damon, 1989; Yackel, Cobb, Wood, Wheatley, & Merkel, 1990) have found evidence that it can be a successful approach to mathematical learning. They highlight the benets that come from providing students with opportunities for verbalization, allowing students to resolve conicts in understanding, and giving students increased responsibility. Collaboration allows a diversity of problem solving techniques and has great scope for accommodating students needs and interests. In contrast, Stacey (1992) gives some examples of collaborative problem solving episodes that show that there can be a tendency among groups to choose ideas and approaches that are easily accessible, but not necessarily appropriate or correct, thus showing that a collaborative environment need not lead to successful conceptual development.
Corresponding author. Tel.: +61-3-8344-8324; fax: +61-3-8344-8739. E-mail address: h.chick@unimelb.edu.au (H.L. Chick).

0732-3123/02/$ see front matter 2002 Elsevier Science Inc. All rights reserved. PII: S 0 7 3 2 - 3 1 2 3 ( 0 2 ) 0 0 1 3 5 - 9

372

H.L. Chick, J.M. Watson / Journal of Mathematical Behavior 21 (2002) 371400

The main difculty associated with trying to draw conclusions about the value of collaboration is that there are many factors involved. Although acknowledging that cooperative learning can increase student learning, Slavin (1989/1990) calls for the identication of the specic conditions required to achieve it. Various authors (e.g., Good, Mulryan, & McCaslin, 1992; Johnson & Johnson, 1985; Noddings, 1989) propose a number of important variables, such as the nature and purpose of the task, the learning strategies employed, the occurrence of academic disagreement among group members, the time spent on task, the support and feedback of peers in the group, the ability levels of group members, the composition of the group, the role of the teacher in preparing students for small group activities, classroom management, the age of the students, and the role of explanations given and received within the group, but acknowledge the difculty in isolating these factors for study. One report (Good et al., 1992, p. 190) emphasized that the way that students experience group work is so varied that it is virtually impossible to predict the outcomes of a collaborative event. Furthermore, although it has been suggested that certain statistics-oriented activities be conducted in a collaborative setting (see, e.g., Russell & Corwin, 1989), there have been only a few studies of the benets to the learner when encountering statistical and probabilistic ideas in that setting. There are reports of some successes but the data are by no means complete or conclusive (Carvalho & Csar, 2000; Lajoie & Lavigne, 1994; Lajoie, Jacobs, & Lavigne, 1995). Lehrer and Romberg (1996) examined the choices children make about how to transform observed phenomena into data and construct suitable representations from which they can make meaningful statistical inferences, but group work per se was not considered by those researchers. There are thus two questions of interest for the current study. One concerns the effectiveness of collaborative activity in the statistics part of the mathematics curriculum. This is particularly relevant at the stage of development where students are moving from what Shaughnessy (1992) termed na ve-statistical conceptions to emergent statistical ones. As part of an earlier study that investigated how higher order thinking in statistics can be recognized and evaluated, Watson, Collis, Callingham, and Moritz (1995) speculated that students working in groups while interpreting data seemed to produce higher level outcomes than individuals. For students in the middle school years, this study provides the opportunity to consider in detail the types of emergent statistical insights that arise in a collaborative setting. The second question involves the determination of factors that inuence collaborative outcomes. This has already been addressed in some detail by the present authors (Watson & Chick, 2001a) in a precursor to the current study. Three levels of cognitive outcome were observed there during sessions with eight groups of three students. Lifting was the term used when collaboration was seen to raise the cognitive levels of one or more group members. The term falling was used when the outcome of collaboration was a reduction in the level of cognitive functioning. Finally, the term hovering was used to describe the outcome of a collaborative event that neither lifted the group, nor caused it to fall. It is important to note that lifting and falling are associated with a change in cognitive functioning as a result of the collaboration for one or more of the group members. Sometimes this is seen in terms of relative change or in sensible decision-making, for example, which may not lead directly to optimal performance but is benecial at least in the short-term. Further to the observation of collaborative outcomes, Watson and Chick (2001a) documented 17 factors in the group setting that appeared to inuence directly the observed outcomes. These are summarized in Table 1 under three general headings. The rst nine are cognitive factors, associated with aspects of cognition and metacognition that can inuence problem-solving outcomes. The next ve are social or interpersonal factors that can have an effect on collaborative outcomes. Finally, there are three external

H.L. Chick, J.M. Watson / Journal of Mathematical Behavior 21 (2002) 371400 Table 1 Factors associated with collaborative outcomes Cognitive factors Cognitive ability: The intellectual capability of an individual or the group; the inuence of this factor depends on the level of ability and leadership skills, and can lead to lifting or falling Previous experience: Understanding from another setting that is applied to the current task; outcome depends on nature of understanding and persuasiveness of person who brings it Cognitive disagreement: Cognitive dissonance that, if not resolved, may result in hovering or falling; outcome depends on nature of arguments and sometimes who puts them Doubt: Re-evaluation of current understanding based on new evidence from the task or group; outcome depends on condence of student and quality of arguments presented Misunderstanding: Ideas that are misheard, not clearly expressed by speaker, or misinterpreted by hearer; outcome is often negative if not resolved quickly Tenacity of ideas: An obsession for an idea, brought up repeatedly until acknowledged by the group; outcome depends on quality of the idea and whether the group takes it on The big picture: Ability to keep in mind the overall goal of the group; outcome may be negative if group is distracted from overall objective or positive if goal can be kept in sight Picking the easiest ideas: Making unsophisticated choices of strategy or technique when better ones are available; outcome is usually negative Organizational collaboration: Collaboration that is on-task but not directly cognitive. It is usually concerned with the mechanics of assigning tasks and is deemed to be metacognitive as it requires an overview of the task and knowledge of the strengths of group members Social or interpersonal factors Leadership factors: Ability or otherwise of a group member to lead the group and convey ideas; outcomes depend on both the leadership skills and the worth of the ideas espoused Social disagreement: Social dissonance that may affect further interactions within the group Egocentrism: Being locked in ones own world-view and unable to listen to others or refusing to be inuenced by others opinions; outcome usually negative (or non-positive) Social collaboration: Off-task collaboration in which the group still functions as a group but is not engaged in the task, or collaboration based on social conventions rather than cognition; the outcome is often negative distraction Other social factors: Pre-existing friendships, gender balance, location of interviewer, etc.; outcome depends on factor External factors Task factors: Use of materials, conceptions of expectations of the task, open/closed nature of task, etc.; outcome depends on factor Outsider: Inuential person from outside the group; such a person may be a teacher or interviewer or a fellow student; outcome depends on nature of intrusion from this person Environment: External physical factors such as classroom noise, physical positioning in group, outside interruptions, etc.; outcome depends on factor

373

factors that may contribute to outcomes. The rst two sets of factors represent inuences of individuals in the group in their attempts to construct ways of handling the data presented and of the group in devising mathematical or statistical practices appropriate for the task. These factors are consistent with the emergent perspective suggested by Cobb and Yackel (1996), combining aspects of a socio-cultural perspective and a constructivist perspective. In the study in which these factors were described (Watson & Chick, 2001a), these outcomes and inuences were observed in eight collaborative groups, each involving three students, from Grade 3 (two groups), Grade 6 (four groups), and Grade 9 (two groups). Although the groups worked on tasks from the chance and data part of the curriculum, the prime focus of that study was on collaboration and the factors that inuenced it, not on the mathematical outcomes of the tasks.

374

H.L. Chick, J.M. Watson / Journal of Mathematical Behavior 21 (2002) 371400

In that study the groups varied in the degree of collaboration and the levels of outcomes, with both often disappointing. One of the groups observed, however, appeared to engage in more collaborative behavior than the others, and also attempted to deal with quite sophisticated statistical ideas for their age. This group is the subject of the current study, verifying that successful outcomes can occur in a group setting and that difcult statistical ideas can be grappled with by relatively inexperienced students.

1. Research aims In contrast to most previous research on collaboration, this study considered (i) a task that was not well-dened; (ii) a task from the probability and statistics section of the mathematics curriculum; and (iii) the interactions that appeared to cause cognitive change to take place, recorded as they happened by means of a fully transcribed video recording. This case study investigated issues raised in the literature (cf. Watson & Chick, 2001a) by observing a group of three students working in a collaborative environment. The fully transcribed video recording provided a nearly complete picture of exactly what took place in interactions to produce cognitive change. This was essential, as the open-ended task produced some dynamic interactions. There were two closely interrelated aims of the research. The rst was to corroborate the existence of the factors and collaborative phenomena identied by Watson and Chick (2001a) that appeared to have a direct inuence on cognitive outcomes. Of particular interest were the changes that took place in the level of cognition of the group: lifting, falling, and hovering. This will go further to address Slavins (1989/1990) concern to document specic conditions contributing to collaborative outcomes. The second aim within this collaborative framework was to identify the emergent statistical ideas and outcomes of na ve but motivated students and how these were inuenced by the collaborative phenomena. For statistics educators this may inform decision-making in terms of teaching methodology and the curriculum.

2. Methodology 2.1. Procedure This case study considers the collaborative group work of three Grade 6 boys, Sam, Mark, and Luke (names have been changed). Their group was one of eight groups of students from Grades 3, 6, and 9 who worked in group settings on two protocols from the chance and data section of the mathematics curriculum (see Watson & Chick, 2001a; this group was Group 6 in that study). The boys worked together on the protocol described below for two sessions separated by 2 days; the rst was 40 min long and the second 75 min. The boys were selected by their teacher, who was asked for three above-average students who would work well together on a cooperative task. The boys had previously participated in collaborative classroom activities, but not necessarily with each other. The sessions took place in a quiet room, removed from the classroom, in the presence of a researcher, and both sessions were video-taped. Complete transcripts were prepared for both sessions. The transcripts were annotated by the transcriber, the researcher who had attended the sessions, and the two authors. The researcher who worked with the boys was a classroom teacher who was working with the second author as a research fellow at the time the sessions took place. Her role was to explain the task, facilitate

H.L. Chick, J.M. Watson / Journal of Mathematical Behavior 21 (2002) 371400

375

the boys work on it, and answer their questions. The sessions were not intended to provide instruction nor were they in the style of a traditional interview. Most of the time the boys worked without intervention, but with the researcher in the room. This implies that the setting for this collaborative work might be regarded as ideal, in that the presence of the researcher ensured that the group remained on task for the entire period. Her presence alone seemed enough to achieve this as she did not have to remind the students to focus on the task. On one occasion when she left the room, however, the boys immediately stopped work and shared some sweets among themselves, seemingly oblivious to the video camera. This indicates that the students did not necessarily limit their off-task behavior for fear of having such misbehavior recorded; thus the use of a video camera probably did not distort or inhibit the students natural conduct. Since the students were focused for virtually the entire activity, time-on-task is not a variable of interest in this study. It should also be pointed out that this group required no special encouragement to work collaboratively nor was there provision of explicit instruction about the social behavior that might be expected to facilitate successful collaboration. The desire to mimic a classroom environment, but one where high quality data collection could occur, determined the conditions as described above. The interrelationship of the collaborative setting, the task set, and the cognitive and social skills brought by the boys, was complex and ts within the framework described by Yin (1994) as appropriate for employing a case study. The beginning propositions for the case study were based on the 17 factors identied by Watson and Chick (2001a) and associated with cognitive outcomes reecting lifting, hovering, or falling. The data collected on the unit of study the group consisted of the complete identication of the factors as well as the statistical concepts employed and developed as part of the collaborative activity. Although a body of statistical concepts existed that could be expected to be used by a statistician (e.g., Moore, 1991), these were not delineated a priori because the students had had no formal exposure to them. The identication of the emergent statistical concepts used by the students took place as part of the data analysis, as did the associated judgment of whether the outcomes improved, diminished, or remained unchanged. This design is consistent with that of Yin (1994), where for this study the ndings are interpreted in the light of the relationships among collaborative factors and cognitive outcomes, the expectations for meaningful statistical inquiry, and the implications for the classroom. 2.2. Data cards protocol The students were supplied with a set of 16 cards, each containing information such as age, weight, and food consumption for an individual. An example of one of the cards is shown in Fig. 1 and the entire data set is produced in Appendix A. The data provided were designed to describe 16 children ranging in age from 8 to 18 years, with plausible favorite activities, and weights that were within the normal range as published by the National Health and Medical Research Council (1957). At the commencement of the session the students were given time to examine the data cards and were invited to produce their own cards with information in the same categories about themselves, thus producing a total of 19 data

Fig. 1. Example card from the concrete materials for data cards interview items.

376

H.L. Chick, J.M. Watson / Journal of Mathematical Behavior 21 (2002) 371400

cards for analysis. The students were then asked to nd any interesting connections among the variables represented on the cards and justify these for the researcher. Although the sample size arising from the set of data cards would not be considered appropriate for a statistician to carry out an analysis, it was considered adequate for students of this age to discover relationships using the concrete methods available to them. This sample size is consistent with those used by other statistics educators with school students somewhat older than those in this study (e.g., Estepa, Batanero, & Sanchez, 1999; Ross & Cousins, 1993). A larger sample size would have required tools needing additional explanation (e.g., a spreadsheet) or would have involved greater computational complexity. The resultant level of abstraction may also have interfered with students ability to demonstrate understanding. The desire was to observe the students data handling skills in a context they had not met before with a manageable data set. The seven variables (including gender) for the data cards set were chosen for several reasons. Because the original study in which the cards were used involved students in Grades 39, it was necessary to have variables that young children could easily interpret. It was also important to have categorical as well as measurement variables, both to assist younger children if they decided to use counting techniques and to allow for a variety of methods of comparison for older students. From a statistical education point of view, it is of interest to observe which variables the students of different ages focus on. Eye color, although of no interest to a statistician, was found by Watson et al. (1995) to be of interest to many Grade 3 and 6 students. Gender was intended as an implicit variable that could be determined by consideration of the names, and many groups in earlier studies compared boys and girls on the weight and fast food variables. Students were provided with materials to produce graphical representations of the data: pens, rulers, paper, glue, scissors, blocks, and counters. To give the students an opportunity to demonstrate their knowledge or understanding of a variety graphical forms, they were invited to contribute their own ideas. Near the beginning of the second session, students were also shown prepared examples of data representations such as tables, pie charts, line graphs, scattergrams, and column graphs, and were asked if they had used any of these before. They were then given the opportunity to use one or more of them to relate variables from the data. Further description of the protocol is available in Watson et al. (1995), with examples of responses largely from individual students in Lidster, Chick, and Watson (1997), and from groups in an earlier study in Watson and Callingham (1997). 2.3. Transcription of the sessions Both sessions were video-taped, then fully transcribed and annotated in an attempt to provide both an accurate account of what was said and to illustrate the interactions and activity that were taking place. On some occasions, the annotator (occasionally the transcriber or researcher at the sessions, but more usually one of the authors) has speculated about the exact nature of the observed behavior; the fact that such conjecturing may be inaccurate is indicated by question marks or other comments when it occurs. In the excerpts from the transcripts the annotators comments are within square brackets; a [. . . ] in the transcript means that material has been edited from it; and an ellipsis with no enclosing brackets means a conversational pause. The researcher present at the sessions is identied by R. 2.4. Synopsis of the interview sessions The following summary is provided to assist with comprehension of the tasks attempted and their place in the time-frame of the sessions. In the rst session the boys initially spent time examining the cards and

H.L. Chick, J.M. Watson / Journal of Mathematical Behavior 21 (2002) 371400

377

creating data cards bearing information about themselves, before deciding, after some discussion, to graph average weight against age. After completing this task the researcher suggested that they examine some other aspect, which led to a discussion about why graphs are good ways to summarize information. The group then decided to examine the relationship between fast food and weight, a task made complicated by the fact that they wanted to incorporate some notion of fat, despite the lack of data for it, and by the decision to represent the details with towers of Unix blocks. The struggle to reconcile various difculties continued in the second session, until the researcher helped them to focus on simpler ideas. This was achieved when she presented various forms of graphical representation (such as bar charts, pie charts, and scattergrams) that the boys interpreted, before they decided to produce their own bar charts to show, separately, favorite activity and dominant eye color. Having completed these they produced a graph showing the relationship between average fast food consumption and age, and then redid a graph of average weight for age, a task previously done in the rst session. The session closed with a discussion on working collaboratively led by the researcher. 2.5. Analysis The transcripts and the video recording were examined in order to determine the cognitive changes that took place leading to better, worse, or unchanged performance and the factors that inuenced them. This was done during the analysis for the larger study (Watson & Chick, 2001a) by means of a type of clustering process (Miles & Huberman, 1994). Transcripts were successively analyzed and similar types of collaborative events were grouped together and given succinct descriptors. The nature of the outcome, whether positive, negative, or neutral, was also recorded. Validation of the decisions regarding these classications occurred through a triangulation process (Maxwell, 1996), involving discussions among the authors, the researcher at the sessions, and the transcriber. The complete transcript for this group of boys was then analyzed in detail to identify collaborative factors and outcomes associated with identiable emergent statistical concepts. Extracts were selected and clustered to provide evidence of these.

3. Results The open-ended task chosen for this study was selected to allow for a range of outcomes and to provide an opportunity for students to display high order statistical reasoning. As in the study of other Grade 6 students by Watson et al. (1995), no formal training in analyzing complex data sets or in providing evidence to support hypotheses had been provided for the students. It was hence of interest to observe the level of outcome associated with na ve attempts to analyze the information on the data cards. The achievement observed for the group in this study equated with the highest level observed by groups in Watson et al. This level was characterized by consolidation and application of concepts in a manner that integrated understanding of the relationships between the different aspects of elements involved in the task (p. 250). Although it is acknowledged that the boys in this group did not always perform at this high level, the opportunity to observe the groups behavior over a relatively long period of time provides evidence of the variation in performance even for those who do achieve high level outcomes. Such an analysis was not possible in the Watson et al. study because only nal outcomes were analyzed. Hence false starts or dead ends, such as seen when the boys struggle with percent in Vignette 6, were not documented in the earlier work. Although statisticians might criticize the analysis carried out by this

378

H.L. Chick, J.M. Watson / Journal of Mathematical Behavior 21 (2002) 371400

Table 2 Emergent statistical thinking in the Vignettes Statistical concept Association (Vignette 1) Average (Vignette 2) Graphing (Vignette 3) Evidence (Vignette 4) Part/whole (Vignette 5) Mathematical tools (Vignette 6) Limitations (Vignette 7) Brief description Consideration of two or more variables and possible causeeffect relationships Idea of average as: typical, middle, a summary value. Ability to calculate with meaning Use of graphing skills to tell a story Realization of the need to prove suppositions, i.e., to nd and support implications from the data Recognition of part/whole relationships in data sets Trying available mathematical tools to tackle a problem, e.g., percent, totals Recognition that ideas and conclusions are limited by the information at hand

group, it represents a high level of performance for students of this age. Table 2 contains a summary of the emergent statistical thinking that occurred during the groups collaboration and the results are presented in seven Vignettes that each illustrate a different statistical theme. The order of these themes was chosen to reect the chronology of the sessions as far as was possible. A nal Vignette contains the students beliefs about working in groups. The results of the analysis of collaboration are summarized in Table 3, which indicates the relationship of the factors identied as inuencing the outcomes of collaboration and the type of cognitive change observed (lifting, hovering, or falling). The statistical concepts associated with the outcomes are noted
Table 3 Vignettes illustrating collaborative outcomes and inuencing factors Inuential factors Outcomes Lifting Cognitive factors Cognitive ability Previous experience Cognitive disagreement Doubt Misunderstanding Tenacity of ideas The big picture Picking the easiest ideas Organizational collaboration Social or interpersonal factors Leadership factors Egocentrism Social collaboration Other social factors External factors Task factors Outsider 1, 2, 3, 4 1, 2, 3 3, 4, 5, 6 3, 4, 6 3 1, 3, 4 2, 3, 4, 5 3 2, 3 1, 2, 3 Hovering 1, 5, 7 1, 3, 6, 7 1, 4, 7 2, 3, 5, 6, 7 3, 5, 6 1, 2, 3, 6, 7 3 1, 3, 6 1, 6 6 2 5 2, 3, 5, 7 6 6 Falling

6 6, 7

6, 7 6

3, 4, 8 3, 4 1, 3, 4

Note: Vignette 1 concerns association, 2 = average, 3 = graphing, 4 = evidence, 5 = part/whole, 6 = mathematical tools, 7 = limitations, 8 = beliefs. Social disagreement and environmental factors were not observed for this group of students.

H.L. Chick, J.M. Watson / Journal of Mathematical Behavior 21 (2002) 371400

379

in the table in the cells whose characteristics they display. Most of the inuencing factors described do not exist in isolation, and, indeed, interact extensively; for example, doubt can lead to conict that may then produce lifting or falling. Furthermore, the group and the individuals in it are not always consistent in their levels of functioning, as exemplied for average when Luke rst teaches Sam how to calculate averages, but then later encounters difculties himself. 3.1. Vignette 1: association [cognitive ability, previous experience, cognitive disagreement, tenacity of ideas, the big picture, organizational collaboration, leadership factors, outsider] The following extract is from early in the rst session, immediately after the researcher has introduced the boys to the problem, shown them the materials, and invited them to work together. It gives an example of the usefulness of the task in providing a context with which the boys can immediately engage, exploring ideas of association. Mark: Why dont we check out the favorite activities rst and check out the weight and everything [after]. [Leadership factors and an attempt at organizational collaboration. Outcome hovers for a while.] Luke: Like if the favorite activity is something lazy like TV. Mark: Yes, theyd have a big weight. [Probably previous experience, which will lift them.] Luke: And then check out what the weight is . . . . Sam: Shall we make a graph? [This query has no impact initially.] Luke: Some people can be heavy but its really muscle. So it depends on favorite activity. [More previous experience.] Sam: Most of the people who weigh a lot like watching TV 74, 60 . . . . [An early sense of the big picture that remains undeveloped, implying hovering.] Luke: Or it could be his height because Zack is about 50. Mark: Cos the weight and swimming . . . 15 . . . hed have strong arms . . . and the netball . . . Luke: Hes got two fast foods a week that might add on . . . . [Cognitive ability from all the boys showing recognition of multiple variables and causeeffect relationships.] Sam: How can we say that in a graph? [Tenacity for graph idea.] Luke: We could put ages and put . . . . Sam: . . . the weight for their ages . . . . The students are getting a feel for the data and the task. Sams sense of the big picture is such that he believes that drawing a graph is a relevant way of summarizing the data they have, and he is able to guide the group to consider it. This is an early indication that Sam may be a charismatic intellectual for the group, that is, someone with cognitive ability and good leadership skills who can direct the group productively. Later, after the boys have completed their work on weight and age, the researcher prompts further consideration. R: What sort of things were you hoping to show when you were looking at fast food and weight? [The students had only suggested this; they had not studied it.] Sam: The ones that have a lot of fast food are . . . theyd have a higher weight than people who dont. [Previous experience and cognitive ability.] [. . . ]

380

H.L. Chick, J.M. Watson / Journal of Mathematical Behavior 21 (2002) 371400

R: So if someone came through the door you could say, Look here, this is what we have found out about this group of people. [Mark starts to sort the cards into age order.] Luke: Say the average of . . . shall we focus on the fast food map [his term for graph?], the weight or what? Sam: Well . . . weight and fast food I think theyre in the same one . . . somehow connected . . . . [Nascent idea of correlation.] Luke: Mmm? Sam: Well, the more fast food you eat the fatter you get so . . . . [Correlation idea claried.] Mark: Well, if youve got a fast metabolism it may not make the difference. [Cognitive disagreement based on previous experience.] Sam: Yes, but if the person was 74, the average we got was 70 so . . . . [Cognitive ability.] It is possible that Sam may be thinking about people being above average and below average, and suspecting that this is an indication of fatness. Unfortunately Sam does not get a chance to pursue this connection because Luke proposes an idea that moves the group in another direction. At one point the group, while struggling to consider several variables at once, actually realizes that it is falling. Sam, perhaps in an attempt to regain a view of the big picture, suggests that at least they do not have to worry about eye color and Mark agrees. The researcher the outsider nally intervenes in an attempt to help the group refocus on fewer variables at once. Sam: Luke: R: Luke: Weve done weight and age already so . . . weight and activity? Or we could do activity and food. You could do activity and fast food . . . that would be another interesting one wouldnt it? We could do weight/activity, weight/food because activity and food both goes with weight and we could do two charts like that. [Again the ability to consider complex relationships is shown.]

A confounding difculty is that by this stage the boys are dealing with data that they have grouped according to age, and the weights they are working with are average weights. This may well make it more difcult for them to achieve a useful representation and keep the big picture in sight. The researchers intervention lifts them to focus on a narrower, more straightforward task. 3.2. Vignette 2: average [cognitive ability, previous experience, doubt, the big picture, organizational collaboration, leadership factors, social collaboration, task factors] There are several extracts from the sessions illustrating the boys previous experience with average. The following is from early in the interview when the boys are examining the cards; at this stage no averages of any kind have been calculated or even mentioned. Luke: With the guy who watches TV and has 12 fast foods a week, I mean 12 is way over average. Luke uses the expression way over average quite condently. There is also a hint of the same idea in the following aside from later in the interview. Mark: Richards 65 [kg] . . . and hes only 10 [years old]. Sam: Richard? [It would seem that Richard is a classmate who is a bit large.] Luke: Yeah, but hes above average. [All dissolve into ts of giggles.]

H.L. Chick, J.M. Watson / Journal of Mathematical Behavior 21 (2002) 371400

381

Lukes original remark may well have inuenced their decision to start looking at average weights for age, thus lifting them. The boys are quite comfortable with the mechanics and the concept of average, showing their cognitive ability. At one point later in the second session they are sidetracked discussing a recent mathematics test, which also illustrates social collaboration and hovering because no change in cognitive functioning occurs. Mark: Sam: Mark: Sam: Luke: Mark: Sam: Luke: Mark: What did you get in that maths test? Which one? The one that we did with the worksheet . . . . 10. 11. I got 12 . . . James got 13! I know. 10, 11, 12, 13 . . . Im the average! [With dry good humor] Thank you Luke!

Although Lukes last statement above is not strictly correct, he has appropriate understanding, and it is his previous experience and knowledge of average that lifts the group in the following extract. In order to work out average, Luke reminds Sam of the process involved (interestingly, Luke does not just take over the calculations), and then, after collaboratively deciding on the organization of the graph, it is Mark who guides and supervises Luke in its production. They want to work out age-based weight averages. Sam: We could work out the average of what their weight is for their age. [. . . ] Ill just write down all the ages. [Sam directs proceedings here, exhibiting organizational collaboration and leadership factors. He makes a list of all the age groups and they then ll in all the weights by the ages. Mark organizes the pile of cards and Luke calls out the values.] Sam: [Uncertain] We add them up then? And . . . you work out . . . . Luke: The average . . . . Sam: Yeah, the average . . . but theres more people for some of them. [Sam shows he is aware of what an average is supposed to take into account, because he has more people in some age groups. There is cognitive ability here, even if he does not know or remember how to calculate the mean.] Luke: You know how you add those [points to the list of 8-year-olds, of which there are two] . . . you only divide by two [continues down the list], divide by two, divide by two, divide by one, [reaches the four 11-year-olds] then you divide by four . . . so it would be different . . . averages. [Lukes cognitive ability and previous experience with average computation lifts the group.] Sam: So if you take the rst one, which is 54 [the sum of the weights of the two 8-year-olds] . . . divide it by two . . . 27. The average for eight is 27. . . . So, Mark, do you want to take a piece of graph paper and put the ages along the bottom and put up the side, the weights up the side. [Sams organizing skills lead to positive cognitive outcomes for the group at this point. Mark and Luke work on their assigned tasks, while Sam continues the calculations with a calculator.] [. . . ] 29.5 . . . does that seem right? Yup, 29.5 is OK for age nine. [Self-corrected doubt.] So you get a pencil and put them on the graph with the rst one 27 and 29.5 is the average for the people. [Organizational collaboration.] Luke: Oh, so this chart will show the averages of weight, will it? [Doubt here as he tries to get a sense of the big picture. Mark is doing most of the preparation for graphing and so, with Sam doing the

382

H.L. Chick, J.M. Watson / Journal of Mathematical Behavior 21 (2002) 371400

calculations, Luke is a little unoccupied at present. This task factor may contribute to hovering for Luke.] Although Sam needs Lukes advice on the actual procedure for calculating the averages it is Sam who decides that it is averages that they want to have on their graph, perhaps inuenced by the earlier comment of Lukes. Thus, one member of the group has helped to lift the group by proposing a task that will represent the data in which they are interested, whereas another has lifted the group by providing the mechanics of a particular subtask. It is interesting to note that whereas Luke helps Sam here, it is Sam who helps Luke in a similar situation described in the next Vignette. In this case, organizational ability has helped the group, too, as Sam provides directions that will enable the group to achieve its objectives. Sams dominant role provides more evidence that he is the charismatic intellectual. Towards the end of the excerpt, Luke does seem to have some doubt about what they are doing; perhaps he has temporarily lost sight of the big picture, or, indeed, has not quite seen it at all yet. Shortly after, as he realizes the point, he is lifted to a higher level of functioning. 3.3. Vignette 3: graphing [cognitive ability, previous experience, cognitive disagreement, doubt, misunderstanding, tenacity of ideas, the big picture, picking the easiest ideas, organizational collaboration, leadership factors, other social factors, task factors, outsider] There are several issues associated with graphing that arise for the boys during the sessions. These range from decisions on which type of graph to use, to assisting each other represent the information correctly and to scale. In the following short extract lifting or falling is not at issue but the group is hovering as groundwork is being laid for later experiences. As they continue to work on the graph of average weight against age, Luke and Mark bring in some ideas for some alternative representations. The mention of pie charts is likely to be based on previous experience. Luke: We could do a pie graph. [Previous experience. Continues to work on chart, with Mark overseeing his work.] Sam: Should we put . . . look this ones 50 so its obviously below . . . . [Sam continues to work out averages on his calculator. The others continue to add the weights to the chart: organizational collaboration.] Luke: 61?! Its gone down. [He seems to have noticed an increasing trend of weight against age, and this decrease surprises him. He has some sense of the big picture.] Mark: How can we work out the graph with blocks? Sam: How about the average fast food? Mark: Yeah, we could do that . . . . Luke: Or pie charts? [Tenacity.] Mark: Or we could use the blocks . . . . [More tenacity.] Marks tenacity for using the blocks may provide an impetus for their use when the boys later try to deal with the issue of fat (Vignette 6), but the evidence is not clear. The tenacity for the idea of pie graphs reappears 2 days later, although it is Mark and Sam who propose it then and Luke who suggests that they do not use it. At other points, Luke also suggests they use a logic graph and a tree graph, again using his previous experience. Both times these ideas lapse.

H.L. Chick, J.M. Watson / Journal of Mathematical Behavior 21 (2002) 371400

383

Prior to the comments elicited below the researcher has shown the boys a series of graphs such as a pie chart, bar graph, and scatter graph, intended to illustrate different kinds of data representation. The researcher then gives them the chance to consider the data again using some of the new ideas. The extract contains three illustrations of picking the easiest ideas. First, there is the choice to graph eye color, which only had three categories; second, there is the rather ippant remark by Sam about avoiding graphs altogether; and, third, there is the choice of a bar graph as easier to construct than a pie graph. This last suggestion is likely based on previous experience. These and the other seven factors noted throughout the extract produce a rapid alternation of lifting and hovering. Luke: Maybe if we had this sort of graph with ages [along the top] to kilograms [down]. Mark: I reckon what we could do is with the sport they do we could like . . . we could have a total of how many people play football . . . . Sam: But what does that mean? . . . . [Doubt.] Mark: Ball games . . . whats most popular . . . . [Cognitive ability.] Sam: Yeah . . . suppose so . . . we could do what the most popular eye color is. Ss: Oh, yeah! [Here they pick the easiest idea, although perhaps only because they have not considered eye color yet.] Luke: We havent really thought about eye color. Sam: Do you have to make the most . . . . Mark: Yeah, do eye color . . . . Sam: . . . like the question you asked like what was the most . . . [to researcher]. Luke: How we can tell them . . . [. . . ] Sam: Why cant we just show them the cards? [Doubt and tenacity from earlier when he made the same suggestion (see Vignette 5), and perhaps a desire to pick the easiest idea.] R: Well if you can imagine you were producing a newspaper or what have you, you dont print all the information thats on . . . . Sam: Its got to be in a graph . . . . R: Yes, an efcient way of showing out what youve found out . . . it summarizes the information. [Conversation continues about graphing. Note the outsiders inuence. Other social factors may also affect what happens, such as a desire to satisfy the researcher.] Sam: OK, Mark, do you want to do their favorite activity over all? Luke and I will do eye coloring over all . . . and then well set it out in a graph. [Leadership factors and organizational collaboration.] [. . . ] Sam: So Ill call out the eye color rst. [Reads out list of students and the others write the details.] Mark: What kind . . . pie? [Previous experience.] Sam: Pies pretty good. Luke: Three sorts of colors, so its a bit hard . . . . Sam: [Looking at Marks lists of favorite activities.] These seem pretty close . . . the favorite seems to be . . . . Mark: TV! Luke: The one you can sort of read very well is you know . . . bar graph because its easy to read. In pie graph you have to guess your percentage and we dont really have the equipment to do it properly. [Picking the easiest ideas based on previous experience.] Sam: So we do a bar graph. Do we have squared paper?

384

H.L. Chick, J.M. Watson / Journal of Mathematical Behavior 21 (2002) 371400

The most obvious aspect of their collaboration illustrating the phenomenon of picking the easiest ideas is their choice of doing a graph of eye color, which reduces the task to a univariate problem that requires only simple tallying. The group has, of course, already considered much more complicated situations and, in fact, it appears that eye color and favorite activities are really only considered because they have run out of other ideas. At the end of the extract above the boys reject the possibility of doing a pie graph, because it is too hard for them given their resources. This might be regarded as a task factor. In a sense the resulting choice of a bar graph is picking the easier option, too, although they present cogent reasons for that choice. The boys ability to collaborate without social conict in correcting errors is shown on several occasions. The following occurs when they are plotting averages for each age group, with Sam and Mark paying attention to what Luke is graphing. Luke: Right, Ill just mark it with this and then color it in different colors. So if the average for 8 is 27. Average for 9, 29.5. [Luke enters this into chart.] Mark: Wrong one . . . thats about the 27 . . . [Cognitive disagreement.] Luke: Oh yes, I see now. [Adjusts the graph with Marks help.] Sam: [Working on averages] See it goes up . . . [The big picture and cognitive ability: Sam seems to have expected the graph to show this.] The following extract provides another good example of collaborative behavior, as the boys discuss the way that they are going to present their graphs. Sam appears to lift Luke to choose a good scale, although it should be recognized that perhaps Luke has perfectly good ideas about the scale anyway this is, in fact, the crux of the misunderstanding. It is not clear, however, whether Luke is content with the decision because he genuinely believes that Sam has made a good choice of scale, or because he is just happy to let Sam decide. Nevertheless, Luke can see the reason behind Marks proposed choice of colors for the bars in the eye color bar chart. Sam: [Takes paper and begins.] OK, this is for eye color . . . What was the favorite eye color? . . . Brown . . . next one, green and . . . What do we make zero? . . . . So should it go up one or should it go up half sort of . . . cos there are only ve six seven . . . ? [Organizational collaboration necessitated by the task factor of the size of the grid on the supplied graph paper.] Luke: Yup . . . I think it should go up 2s . . . . [Has Luke misunderstood what Sam meant?] Sam: Then youre going to have a tiny graph though. [Perhaps, in fact, Sam misunderstands Luke: it is possible that Luke meant that each unit should be represented by two squares on the graph paper. Sam makes a decision and there is no further disagreement.] Go up ones . . . whats the highest number we had? . . . . Seven. What was your second highest number, Mark? [Sensible graph-making decisions rely on this sense of the big picture.] Mark: They all tied, four of them. [Mark is working on favorite activities.] Sam: What were the lowest . . . swimming and computers? Mark: One . . . Yup. Shall we color it in different colors? You can do brown, green and blue. [Color choice based on a sense of the big picture.] Luke: Im going to do brown, red and green . . . . Mark: No, blue, because theyre blue [eye colors]. [Cognitive disagreement.] Luke: Oh, yeah! [Comprehension dawns and Luke is lifted.]

H.L. Chick, J.M. Watson / Journal of Mathematical Behavior 21 (2002) 371400

385

Similar issues of computation and graphing were considered later in the sessions and the boys show successful cooperation to ensure the graph and calculations are done correctly. During the next exchange Sam and Mark watch Luke to make sure he does the right thing the previous experience of Sam and Mark with graphing and computational mathematics seem to be stronger, so they have more complete cognition of graphing processes. Mark exhibits tenacity with his name idea, although at its second mention he is not urging its use it seems that he just wants the group to acknowledge his observation. They are working on a line graph that will show average weight (y axis) against age; this is initially the task of Sam and Luke, while Mark works on a graph of average of fast foods against age. For most of the time the impression is that of lifting. Luke: Oh, so were doing a line graph and joining up the dots? [Sam goes on writing the ages across the bottom of the graph up to 18.] Well have to get the information . . . oh, youve got it. Sam: [Working up the vertical axis, which is for weights, counting] 14, 15, 16 . . . . Oh no! Well have to go up to 74! [Task factors as the graph paper again inuences what they do.] Luke: Thats a bit . . . you should have done it in . . . . Sam: Shall we do it in 2s? [Organizational collaboration, perhaps based on previous experience.] Luke: 5s. Mark: I reckon 5s. Luke: Denitely 5s. Sam: 5s it is. [. . . ] Mark: Their names dont mean anything do they? [Previous experience.] Sam: Average of names! [Chuckles.] No. Can you just get the next one Luke? [Organizational collaboration. Works out average on calculator.] 29.5. Next one is 29. [Luke enters it on the graph incorrectly and Sam intervenes (Cognitive disagreement) and corrects it.] [. . . ] Luke: Maybe I should do the adding. [Implying that Sam should do the entering on the graph. They change jobs. Again, organizational collaboration is evident as appropriate task delegation leads to better outcomes, because tasks have been delegated according to perceived strengths and weaknesses.] What about tens? Sam: Theres only one . . . and its 29. [. . . ] Mark: OK [Mark has nished his graph of average fast food consumption against age, and shows it to the others.] Write what mine is. Average for fast foods. [Writes in the title, a graphical convention based on previous experience. Meanwhile Luke is calculating.] [. . . ] Luke: What! [Surprised at the result of his calculation.] Thats 19 kg. [Doubt arising from a sense of the big picture.] Sam: Yeah . . . dont think so Luke. Luke: Ive made a boo-boo. Sam: Did you do how many two? You sure you havent done how many four? Have you cleared it? [This time Sams cognitive ability and leadership lift Luke with the average calculations.] Luke: [Luke recalculates.] 38. I must have done how many fours and got confused. Now thats 55 and thats the end. Sam: Theyre getting fatter! [Shows his sense of the big picture.] [. . . ] Mark: We cant really do anything with the age and the name! [Note Marks tenacity with names.] The previous experience the boys have had with graphing stands them in good stead as they appreciate the need for scale, labeling, and giving their graph a title. From this point the boys are tidying up and

386

H.L. Chick, J.M. Watson / Journal of Mathematical Behavior 21 (2002) 371400

checking their work. As they do so, they initially think that there are some cards that they have overlooked. They start to redo their calculations until a comparison with their previous analysis helps them to realize that the results have already been included. This again illustrates collaborative lifting with all three boys contributing as they ascertain that the details of the big picture are complete. 3.4. Vignette 4: evidence [cognitive ability, cognitive disagreement, doubt, tenacity of ideas, the big picture, other social factors, task factors, outsider] Having completed the initial graph of average weight against age, the boys are undecided about what to do next and there is some indication of cognitive disagreement. The following extract is the rst instance where the idea of proof is mentioned. It appears that although Sam does not possess the vocabulary to describe it, he is describing the dilemma of drawing conclusions from the data. His comment later about a single data card and a whole would seem to support this contention. Overall, the impression is one of lifting. Luke: Shall we make a whole lot of graphs that show fast food, weight and umm . . . ? Sam: What does it prove though? Its only facts. [Doubt.] Luke: Yeah, thats what were meant to do, I think . . . . Whats the main question? [His question is directed at the researcher who has moved to sit away from the group. Her position away from them may be a social factor that allows them to interact with greater ease. The question helps the group clarify the objectives of the task; that the task requires them to do this is a task factor.] The group interaction and questioning, and even the doubt that Sam voices about what, exactly, is required, helps them to keep the big picture in sight. The placement of the researcher possibly has an inuence on the extent of collaboration. Later Sam suggests an association of fast food and weight. R: OK, so it would be interesting if you put forward that thats what you think it is, and have a look at the information and . . . . Sam: We have to prove it. [Tenacity of ideas.] R: Could you show it in a graph or a table? [Outsider: the researchers comment almost certainly inuences what they do subsequently.] [. . . ] Sam: Somehow we have to prove our point, cos we all think that these people who eat a lot of fast food are fat and these people arent because they dont. [Tenacity of ideas for proof, together with cognitive ability and a sense of the big picture.] Sams observation here is very astute. They all seem to believe there is a relationship, but Sam is aware of the big picture and the need to prove our point. This is further evidence of Sams cognitive ability. 3.5. Vignette 5: part/whole relationship [cognitive ability, cognitive disagreement, doubt, tenacity of ideas, the big picture, other social factors, task factors] On a number of occasions the boys show understanding of the part to whole proportionality ideas associated with indicating relative frequencies. When recording favorite activities, for example, Luke comments that we should show how many this is out of . . . 19. Late in the interview the boys look

H.L. Chick, J.M. Watson / Journal of Mathematical Behavior 21 (2002) 371400

387

at a pie graph shown by the researcher utilizing data from the original 16 data cards. They endeavor to imagine what the pie graph of their own data would have looked like if they had done it. Luke: [Reading the values on the pie graph.] 38%, 31 . . . 31 . . . and theyre meant to be the same as ours but, no, brown is . . . [he notices that the supplied graph is different from what they would have had] . . . oh, yeah, but we put ourselves on . . . . [The big picture.] Ss: Yep. Mark: So what would be one persons percentage? [. . . ] Luke: But I dont get . . . blue and green tied, but if we didnt have ours on, blue and green theyd all tie wouldnt they? Mark: Hey, but what is one persons percentage? How do you work that one out? [Tenacity of ideas.] Sam: What?! [Doubt, mingled with laughter as Sam and Luke look at Mark in amazement.] Luke: You mean how many people t into 31%? Mark: Yes. Luke: Ah! Well that would be 5 according to that graph. So that would be 5, that would be 5 and that would be . . . . Sam: 6, and with us added on that would win. Luke: Yes, because there are 16 people in it. Sam: Plus us . . . oh, were not in the group. [. . . ] Mark: This all adds up to 100%! [The others agree in a friendly Yes, Mark, we understand that! tone: Cognitive ability from all of them, together with other social factors based on their friendship.] They then attempt to grapple with how to modify the researchers pie graph to reect their own data, but keeping track of both the changes in frequencies and the two different totals involved proves too difcult for them. This task factor, along with the other factors, ultimately contributes to a hovering outcome. On another occasion the boys are discussing how to represent the information on the cards. R: Luke: Mark: Luke: Sam: What could you show somebody about that group of people? [Insistently] So if we did the charts it would be showing them . . . . The weights. It would be giving them the information. But wouldnt this be giving them the information [holding up a single card; this causes a pause] . . . or do you want a whole . . . ? [Expresses some cognitive disagreement.]

This is the idea that Sam presents again, perhaps tenaciously, when later involved in graphing (reported in Vignette 3: Graphing). 3.6. Vignette 6: mathematical tools [previous experience, cognitive disagreement, doubt, tenacity of ideas, the big picture, picking the easiest ideas, organizational collaboration, leadership factors, egocentrism, task factors, outsider] The longest and most complex interaction in which the boys engage deals with the issue of fat, which is the name they use for their perception that some of the people described on the cards are overweight. As noted earlier, they are aware that there are possible connections among fast food consumption, weight, and favorite activity. Not surprisingly, however, they nd it difcult to extract the various relationships from the data. This very task factor is what makes the data cards protocol an open-ended task that is

388

H.L. Chick, J.M. Watson / Journal of Mathematical Behavior 21 (2002) 371400

appropriate for evaluating higher order functioning. In exploring the relationships there are points where it is clear that their previous experience in mathematics lets them down. This is particularly evident with respect to their concept of percent. Various collaborative phenomena are revealed as they struggle to resolve these issues. Although in the short term there are instances of lifting and hovering, overall the impression cognitively is one of falling. This extract provides an example of a case where not picking an easy idea has led to falling: the group is unable to satisfactorily resolve all the complex relationships. After some discussion of association and the need for evidence to support their claims, the boys decide to work with the small colored blocks provided, which lock together easily to form towers. Luke uses a mathematical term that sounds good to him but that is not entirely appropriate: like say . . . we work out the percentage of how much fat and the percentage of how much strength and we work it out in blocks so its like a percentage. After some confusion over whether they are representing information for the groups of different ages or for individuals, they settle on individual data to obtain a recognizable output. Sam: Weve done weight and age so we might as well do weight and food. [Organizational collaboration and leadership factors again.] What well do . . . just say were doing Mark, then well have 40 thats what he weighs [makes a tower of eight red blocks; each red block represents 5 kg] and two fat pink [places two pink blocks, which represent Marks two fast food meals per week, beside the tower of red ones]. About 2% of his 40 is fast food. This time it is Sam who tenaciously wants to keep using the blocks, and he also wants to maintain the use of different colored blocks for various aspects, in particular, the use of pink blocks for fast food, which they regard as being indicative of fat. Here the absolute number 2, representing the actual number of fast food meals is (a) being compared with blocks which represent 5 kg of weight and (b) used by Sam as a value for percent of fast food fat in the weight (at least, this seems to be how Sam sees it). Mark, as seen in the excerpt below, at least makes an attempt to reconcile problem (a), but does so erroneously; however, he convinces them, probably because there is so much to keep track of that they are all overwhelmed. The group does not identify problem (b), suggesting that the understanding of percent that they are bringing to this problem as previous experience is not fully formed. Mark: Sam: Mark: Sam: No, thats 10 [Each block represents 5 of something. Cognitive disagreement.] Sorry, 10 out of his 40 must be fast food or things like that. 30 natural fat! Dont forget there are other fats. [Tenacity as he had acknowledged this earlier when talking about 8-year-olds. It is based on previous experience and shows his understanding of the big picture.]

The boys produce more columns, consistently using a base ve system to represent weight with different colored blocks (seven red and one yellow is 36 kg) and using a pink block for each fast food meal (multiplied by 5 and treated as a percentage). Hence Sam, who is 36 kg and has three fast food meals per week, is deemed to be 15% fat. In the next exchanges doubt is expressed several times. Luke expresses doubt about Sams claim, which indicates that within this group even the person who may be considered the charismatic intellectual can be questioned. It may also be that the boys feel such a sense of mutual equality that they are not necessarily aware of Sams role. Although this expression of doubt has the potential to be good for the groups outcomes, none of the three can suggest a successful change of course.

H.L. Chick, J.M. Watson / Journal of Mathematical Behavior 21 (2002) 371400

389

Luke: Mark: Sam: Mark: Luke:

Mark:

Sam: Mark: Sam: Mark: Sam: Luke: Sam: Luke: Mark: Sam:

Luke:

Its amazing how small . . . it doesnt seem right. [Doubt.] But what if the person had like that guy with 12 [fast food meals]? That would be . . . . His fat would be nearly all of his body! Well do that 18-year-old guy rst. [They are referring to an 18-year-old in the data cards who weighs 74 kg and eats 12 fast food meals per week.] [Referring to the towers showing Sams data with eight blocks for weight and three blocks for fat.] If you say thats 100%, thats a third of his body. [He points to the small pink column with three blocks against the last red/yellow column with eight. Note that this is now a different interpretation of comparing the columns, and is correct at least as far as a proportion for comparing column heights is concerned.] There he is [the 18-year-old]. Seventy-four weight and each of them is 5. [He gets 16 reds/4 yellows which they will realize is incorrect shortly and 12 pinks and compares the columns.] Oh boy! 60%. I dont think its 60%. [Cognitive disagreement.] [Counts, 5 per pink block] 60%. The most take away [fast food] you can have is 21 [per week]. So out of his whole body . . . 60% . . . . 60% of his weight is fast food! And the rest of its like . . . some of this could be other food and the rest of its like body weight. [Previous experience.] Struth! Hey, how can it be that because that would be like 20 . . . isnt it 74? Weve got 84. [Doubt based on a sense of the big picture. Their basic mathematics skills are okay.] [Removes two red.] So now its even worse! [The columns are more alike in height, and yet none of them acknowledge the idea of percent as a proportion: as far as they are concerned the percentage fat is still 60% because the fast food column is 12 blocks high.] Lets just call him Mr Fat! [Laughter.]

In the latter half of the above excerpt there is additional evidence that the boys understanding of percent brought from previous experience is incomplete. At one point Luke gives a reasonable interpretation of proportion, when comparing Sams own data of fast food blocks against weight blocks (ignoring the fact that there is an additional complication because of the different values of blocks). Later, however, when they are making adjustments to the columns representing the heavy 18-year-old, Sam acknowledges that the proportion of fast food to weight is even worse after the correction. There is, however, no recognition of any implications for their interpretation of percent: their rule for obtaining a number for percentage appears to have nothing to do with proportion comparing fast food to weight. The diversions and dead-ends pursued by the boys in this extract are similar to what occurs for practicing statisticians when they are problem solving; it is just that statisticians are debating about more sophisticated tools. Given the stage of development of statistical ideas in these boys, their attempts to deal with comparisons of variables using percent is not totally unreasonable if they had the appropriate data. It might be considered equivalent to statisticians trying various sophisticated statistical techniques to see which one better tells the story of their data. The problem for the boys is their incomplete understanding of the technique they wish to employ. This extract illustrates the major incidence of falling experienced

390

H.L. Chick, J.M. Watson / Journal of Mathematical Behavior 21 (2002) 371400

by the boys; in fact they do not resolve it in a statistically satisfactory way but move on with the assistance of the researcher. The other mathematical tool that they suggest using during their sessions is the total. This arises several times as Mark suggests considering the data per year, instead of per week, hence multiplying by 52. Towards the end of the interview Mark has a turn at being tenacious, and, as the following excerpt shows, nearly brings Luke into sharing his obsession. The boys have been looking at and discussing another groups graph of fast food and weight. Almost out of the blue, Mark reintroduces his idea of considering total fast food consumption for a year (it had, in fact, occurred to him an hour earlier). Again the outcomes for the group are hovering. Mark: Sam: Luke: Sam: Mark: Sam: Mark: Ss: Mark: Luke: Mark: But what would be 52 times 12? [Tenacity of ideas.] What?! [Simultaneously, with incredulity] What?! What are you talking about now? 520 plus . . . . What are you doing? . . . Ah! Calculator! [They work it out] . . . Why are you working this out, Mark? [Doubt.] 624! 624 fast foods he had per year. Oh really! Who wants to know that! Yeah, well . . . . It might be good actually . . . . [Luke nearly falls, inuenced by Marks suggestion.] But if we added them up for a year we could show how much fat that would put on. [Tenacity for fat again.]

The idea of fat that inuenced so much discussion in the rst session is also still being held by Mark. They return to the task and consider another two graphs, before the researcher the outsider gives Mark another chance to bring up his idea about annual results. Here his tenacity is verging on egocentrism: he may well be insisting on the idea just for effect. R: Mark: Luke: Mark: Luke: Mark: If we had time is there anything else youd like to graph? How much they would eat in the year, thats what Id like to do. You could take that male and female thing a bit more. Oh yeah. Male and female per year. Males weight and females weight. You could do that for a year. [Tenaciously determined to get it in!]

It is interesting that these boys do not consider the possibility of a gender split earlier, as was done by other groups studied (e.g., Watson & Callingham, 1997; Watson & Chick, 2001a). 3.7. Vignette 7: limitations [cognitive ability, previous experience, cognitive disagreement, doubt, tenacity of ideas, the big picture, task factor] The issue of using blocks dominates the groups activity at the time of this extract. What to represent with them, however, is difcult to decide. Luke is the one most tenaciously focused on the fat idea and the use of blocks. As can be seen in what follows, Sam and Mark have some reservations that recognize

H.L. Chick, J.M. Watson / Journal of Mathematical Behavior 21 (2002) 371400

391

limitations in the data, but the idea seems to linger and their obsession with it pulls their performance down. Luke: Well why dont we . . . like . . . we could use red colors to show that theyre strong and say we think half of its strong and half of its fat. Pink could represent fat. [Grabs some pink and red blocks: tenacity again.] Sam: But anybody could be strong. [Doubt for the idea, revealing further cognitive ability.] Luke: Yeah, but what we think . . . . Sam: Cos if they watch TV they could still be strong. [Previous experience.] Luke: Yeah, I know . . . but if someone liked football but eats a lot of fast food that could be the fat and the strength in the weight. [Has two pink cubes with three red cubes on top in a column.] Sam: Or this could be the good in the food and that could be the bad in the food [points to each of the two colors in turn]. Luke: But how do we know what other foods they eat? [Cognitive disagreement, in which the limitations of the data are acknowledged.] [. . . ] Luke is keen to do something, perhaps because considering fat was his idea, and although Sam still has some doubt because he knows fat comes from sources other than fast food, he proposes an approach that will try to take it into account. His approach represents individuals whereas Luke is still thinking of age groups. Mark cannot follow them and asks for clarication that Luke attempts to provide. Luke: Well, say . . . this is all the fat, this is all the strength, we work out how much fat there is in the 11-year-olds, like say . . . we work out the percentage of how much fat and the percentage of how much strength and we work it out in blocks so its like a percentage. [More tenacity for the fat idea; the introduction of percentage is probably based on previous experience.] Luke uses a mathematical term here that sounds good to him but that is not entirely appropriate. It sounds convincing enough to Mark, however, although Sam still has reservations about the whole plan. Mark: That sounds pretty good. Sam: You cant do strength cos you cant work out how much strength they have. [Doubt and cognitive ability: the limitations of the data are again recognized.] Luke: This is like what we think, and what were showing . . . we dont know how much fat theyve got either . . . so, you cant really say . . . . Sam: Why not instead of calling that fat [points to the pink blocks] we call it fast food because we know how much fast food they have? [Cognitive disagreement as a reection of cognitive ability.] Sam highlights that there is a problem with this vague fat concept, which is similar to the earlier reservation of Luke about what foods people eat. Sam clearly indicates that he is aware of the need to use the data that they have, an important early statistical recognition. Two days later when the session resumes the boys continue to wrestle with the problem. Interestingly, it is Luke who now acknowledges the difculty of relating fast food consumption directly to weight. Luke: But havent you got to know the weight you get from the fast food to umm . . . . [Both Luke and Sam had mentioned this problem 2 days earlier. Tenacity and the big picture.] Sam: Yeah, we need it . . . .

392

H.L. Chick, J.M. Watson / Journal of Mathematical Behavior 21 (2002) 371400

Luke: Because thats just saying that one fast food equals one kilogram and thats not true is it? [Cognitive disagreement. The units have changed since the previous session: one fast food is no longer turned into ve but is kept as one.] Sam: No, but thats the percent of your body that has fast food in it per weight. Luke: We dont really know that do we? [Tenacity.] [. . . ] Sam: But we still wouldnt know what the weight would be. Finally, at one point the researcher asks the following question. R: What would someone think if they saw that? Luke: Whats the average weight for that age. Sam: Out of the people weve got . . . . [Sam acknowledges the fact that the data form a small sample. Although not included here, he expresses the idea on other occasions (see Watson & Chick, 2001a, Group 6): tenacity.] Sam is aware of the data set as but a small, perhaps non-representative, sample of a larger population. It is not clear, however, that this idea was not already known to Sam, nor does it appear to lift the others; hence it results in an indeterminate hovering outcome at the end. 3.8. Vignette 8: beliefs [other social factors] At the end of the second session the researcher asked the students about how they felt about working in a group. Their responses are interesting, and indicate that students are capable of determining the benets, effectiveness, and constraints of working as a collaborative team. R: One other question I wanted to ask: do you prefer to work on your own or do you prefer to work in a group? Ss: Group! [unanimous] Mark: Cos you get it done quicker . . . like Im doing the coloring in and hes doing the calculator and hes drawing up another graph [pointing to the others in turn]. Sam: As long as youve got people . . . . Mark: . . . who co-operate . . . . Sam: Yup . . . because if youve got people who just sit around . . . . Mark: . . . watching . . . . Sam: . . . or if someone tells you to do it all the time. R: So its quicker, and as long as you get on with the other people . . . . Is there any other reason you like group work? Sam: You get shared ideas and you get to listen to other peoples ideas and you probably come up with a solution which would be better . . . or better than what you were thinking . . . not all the time but some of the time you would . . . . R: Is that what you were thinking? Ss: Yup! R: Youll agree with him because youre co-operating! Ss: Yeah. [Laughs all around!]

H.L. Chick, J.M. Watson / Journal of Mathematical Behavior 21 (2002) 371400

393

It appears that these boys, due to their natural friendship (a social factor) and probably due to previous work together in the classroom, had acquired collaborative skills and recognized their benets, which may have been a signicant inuence on the overall lifting experiences of the group.

4. Discussion and conclusions The implications of the observations from this case study will be discussed from several perspectives, including what is learned about collaboration and its potential effect on cognitive outcomes especially in relation to emergent statistical concepts, the implications of these observations for educators, and suggestions for future research in this area. 4.1. Collaboration and its effect on cognitive outcomes There is ample evidence from the Vignettes to demonstrate that lifting that is, an improvement in the level of cognitive functioning can occur for individuals within the group and for the group as a whole as a result of collaboration. In contrast, it is also apparent that the level of cognitive functioning may fall or hover. Whether it is lifting, falling, or hovering that occurs depends on a number of factors that inuence the nature of the collaboration and what is achieved through that collaboration. Thus, for example, there are situations in which a students tenaciousness in bringing up a particular idea can be benecial, as in Sams persistence with the idea of doing a graph. Alternatively, it can be detrimental, as demonstrated by Luke who clings to the idea of trying to consider fat, even after he acknowledges that they do not have the data to do so. Resolving conicts can also produce varied outcomes, depending on the correctness of the arguments and the force with which they are presented. In terms of mathematical ability and the cognitive factors identied in Table 1, the boys in this group appear able and often attempt to bring previously acquired skills to this new context. Although for the concept of average they are quite successful, for the concept of percentage they are less successful and the outcomes show the potential for falling when there is not a member of the group with the previous experience to prevent it. Their communication skills, however, are such that they successfully teach each other techniques as required, as when Luke teaches Sam how to calculate average. They also often check each others work, even when they are struggling with percent. It should be noted that in all the examples of lifting illustrated here, the group lifts because one of the group members knew an appropriate strategy that could be explained to the group. An interesting special case of the factor of previous experience prevalent in this group is the use of jargon, where some individuals use terms that sound impressively technical. This may occur for a number of reasons: (a) the individual knows what the term means and that it will be useful; (b) the individual knows what the term means, but is proposing it as a straw man, unsure if the idea is worthwhile; (c) the individual is unsure of the terms meaning but thinks it may be relevant to the situation; or (d) the individual is unsure of the terms meaning, and it is said in order to impress or to create an illusion that the speaker is knowledgeable about such things. In this study, Luke uses a number of terms that are mathematical but seem to be introduced in part because they sound good. Although he seems to hope that they will be helpful, it is not immediately clear that he always knows how appropriate they are, so he probably ts into the third category above.

394

H.L. Chick, J.M. Watson / Journal of Mathematical Behavior 21 (2002) 371400

Turning to social or interpersonal factors, this case study took place with individuals who were prepared to be cooperative with the researcher and her goals, as well as with each other. The camaraderie exhibited by the boys in the group suggests they were close friends. They shared ideas effectively, tasks were assigned and undertaken readily, leadership roles were shared naturally, difcult issues were addressed without social disagreement, and ideas were treated with respect, even those regarded as incorrect. It is hypothesized that the positive relationship between the boys and their willingness to engage in the task are conducive to positive outcomes; it seems reasonable to expect that poor rapport and/or a negative attitude to the task would reduce the likelihood of improved cognition. The observations of the study were thus made in the most favorable circumstances for collaboration, and hence the presence of both helpful and hindering collaboration is signicant. It is difcult to imagine a scenario involving an extended open-ended task where both outcomes of collaboration will not occur. Finally, two of the external factors task factors and an outsider inuenced the collaborative process. Although the researcher had only 6% of the turns at talking in the extracts shown (and less than 11% overall), there were occasions when her comments appeared to cause signicant changes in approach. In particular, she encouraged the boys to consider fewer variables and sanctioned the use of graphs. In both these cases her recommendations lifted the group, and enabled the boys to achieve recognizable outcomes more efciently than they would have managed on their own. Although no comparison was carried out with other types of open-ended mathematical tasks, it appears that this task provided a motivational context in which these boys could collaborate successfully. That it was slightly above their current level of experience and expertise provided challenges that were sometimes met (leading to lifting) and sometimes not (leading to hovering and falling). In generalizing the ndings of this study, the importance of four groups of factors are highlighted in Fig. 2. These were identied individually or in clusters in Table 1. First there are the two sets of factors that are dependent on the members of the collaborative group. These individuals bring a variety of cognitive and social attributes to the collaborative event, which vary from individual to individual and hence from

Fig. 2. Key factors that inuence cognitive outcomes during a collaborative event.

H.L. Chick, J.M. Watson / Journal of Mathematical Behavior 21 (2002) 371400

395

group to group. Some of these factors may be more inuential than others cognitive ability is an obvious example but there is no guarantee that these factors will come into play in a predictable way. Secondly, there seem to be two particularly important factors that are outside the control of the group: the task itself and the inuence of an outsider. In most school-based events, external inuences govern the choice of task, but any task has attributes that may make it more or less conducive to use in a collaborative situation. In the present case, the data cards task was sufciently open-ended to stimulate productive statistical collaboration. The other factor the outsider, who was the interviewer in this study but who is more usually the teacher in classroom situations plays a signicant role in regulating the course of the collaboration. In the present study, if the researcher had not intervened, particularly in the fat episode (see Vignette 6), the outcomes would have been less satisfactory. This highlights the importance of appropriate intervention by teachers, not necessarily to instruct but to steer the group towards fruitful avenues. What is striking in this study is that it seems to be impossible to predict the outcomes from the interplay of these factors. The point made by Good et al. (1992) about the difculty of predicting collaborative outcomes is borne out clearly: lifting, hovering, and falling may all occur as a result of collaboration, and almost any of the factors may lead to any of these cognitive outcomes (see Table 3). The framework provided in Fig. 2, however, may provide a useful starting point for future research. 4.2. Mathematical and statistical cognition As far as statistical cognition is concerned, it is appropriate at this point to summarize this groups overall performance in the context set by the data cards protocol. The use of this task supports the views of Phelps and Damon (1989) about the benets of using collaborative activities requiring higher order reasoning skills. From the starting point of their previous experience these three boys achieved high level cognitive outcomes for students in Grade 6 (cf. the study of Watson et al., 1995). This may not be surprising in the light of their ease of collaboration, apparent ability levels, and the time devoted to the task. In particular, as highlighted by Table 2 and the Vignettes, there were a number of statistical concepts of which the boys were cognizant even if they had limited strategies for dealing with them. The concept of average appeared well-understood throughout the extracts. Although the boys strategy of computing average weights for age would not have satised a statistician due to the small number in each age group, it overcame the difculty of repeated ages encountered by some groups in the study of Watson and Chick (2001a). One Grade 9 group in that study, for example, opted for an awkward doubling up scheme when plotting bars for two weights associated with one age. The ideas of causeeffect relationships and correlation were evident, with the boys previous experiences suggesting some of the relationships among variables. They also had reasonable expectations for what some of the two-variable relationships would look like when depicted graphically (e.g., weight against age). The limitations of the data, both because of the variables involved and the size of the set as a sample, were recognized by the boys. Their graphing skills were, in general, sufcient for their needs, and graphing conventions were employed without any prompting. That the techniques used and outcomes achieved may not be considered the best by practicing statisticians does not detract from the boys higher order reasoning. These are emerging skills, not yet rened ones. Had the researcher intervened in a teaching mode it is likely that she could have helped the boys achieve more statistically satisfying outcomes. In this study of collaboration, however, teaching was not meant to be a factor. Having recognized their high level of achievement, it is also appropriate to note the many false starts and less statistically viable options the boys considered along the way, particularly their struggle with the

396

H.L. Chick, J.M. Watson / Journal of Mathematical Behavior 21 (2002) 371400

part/whole aspects of percent. There were hence many opportunities to go astray and many factors that inuenced the ultimate better choices made. This case study illustrates the extremely complex processes occurring even when overall outcomes are considered to be favorable mathematically and hence some caution must be exercised before totally accepting the views of Phelps and Damon (1989). 4.3. Implications for the classroom This study highlights a number of issues that are relevant to the use of collaborative activities in the classroom, not least in helping teachers and students become aware of the factors that will inuence outcomes. Virtually all of the factors can give rise to any of lifting, falling, or hovering. As pointed out in Watson and Chick (2001a) teachers may be able to model appropriate behavior for dealing with the factors to help students reduce the occurrence of falling and increase the occurrence of lifting. Teachers can also be aware that the performance of groups will depend on their composition, and whether or not their members have sufcient cognitive ability, leadership skills, and appropriate previous experience for the tasks assigned. The task chosen for the group to consider also inuences the extent and success of collaboration. Open-ended tasks, such as the data exploration considered here, allow greater scope for discussion of concepts than do closed or routine activities. The task should engage the interest of the group members, and, if possible, should be such that it can only be dealt with in a collaborative way. There may be particular benets if the task develops and extends previously obtained knowledge. In this specic case, if students previously had done some work with data representation (e.g., being asked to produce a bar graph for some simple data), then the data cards protocol would have extended this by forcing them to make choices about how to represent more complicated relationships. Such an extension would help students to apply knowledge in a broader context. Finally, there are some caveats for teachers. First, when working on long or demanding tasks, student fatigue can lead to a drop in cognitive functioning. It may be less demanding for the group to choose an easy option rather than a better but more challenging one, or students may cease to persist with an idea that they have had, or conict may not be fully resolved as to do so would be too taxing. Second, whereas it can be useful for a group to have a charismatic intellectual, it is certainly not true that bright people always have bright ideas. Since students do identify individuals operating at higher levels it follows that a group may be led astray by a less satisfactory idea simply because it came from the student that the other group members look to for intellectual leadership. Third, although the delegation of subtasks according to strengths and weaknesses allows the group to achieve probably the best possible outcome, such delegation does not allow those students with particular weaknesses to have experiences that will help them to develop facility in that area. Finally, it may be difcult for a teacher to decide when to intervene in a falling situation such as occurred when the boys were struggling with percent. Should the group be given a mini-lesson on percent or led in the direction of using averages or a scattergram? Sometimes the teacher may have to intervene when it appears that a group cannot get back on an upward track on its own. 4.4. Future research As highlighted earlier, the setting for this study was conducive to optimal collaborative behavior: a group of above average students with good rapport, isolated from the rest of the class, in the presence

H.L. Chick, J.M. Watson / Journal of Mathematical Behavior 21 (2002) 371400

397

of a researcher, and working on a stimulating open-ended task that extended their previous experiences. A study of the excerpts makes it apparent that worthwhile collaborative behavior was achieved under these conditions. It is natural to consider what would happen in less than ideal conditions. Aspects of the following questions have been considered in the literature, but there is still scope for future research as indicated. 1. What aspects of students emergent statistical understanding are particularly worth fostering in collaborative settings? This research suggests that using contexts where students can bring their out-of-school previous experience to bear, for example in recognizing potential associations among variables, is one useful starting point. Are some of the emerging statistical ideas sufciently sophisticated that teacher intervention is necessary, even in a collaborative setting? Research on the provision of help has begun in this context (Watson & Chick, 2001b). 2. Is it possible to build a theoretical framework to describe and predict how the factors that inuence collaboration will interact to produce particular cognitive outcomes? The evidence of this study and others (e.g., Chick & Watson, 2001; Good et al., 1992; Stacey, 1992; Watson & Chick, 2001a, 2002) suggest that it may be difcult, if not impossible, to achieve this. However, the framework of factors at least seems to describe satisfactorily the inuences on collaboration, and it is certainly possible to identify ways that factors can give rise to benecial outcomes. The external factors the task and the teacher were indeed signicant, for example, in directing how the cognitive and social factors contributed to the emerging statistical ideas. 3. What is the effect of individuals abilities on the groups outcomes? What takes place in the interactions if one student is markedly more or less capable than the others in the group? This, too, has been partially addressed in the literature (e.g., Cohen, 1994; Webb, 1991), but rarely for groups of three or for open-ended, ill-dened mathematical problems. Answers to this would go some way to answering question 2. 4. Can teachers help students to improve the outcomes of collaboration by discussing and modeling benecial behaviors? Students would need to be taught to recognize the factors and how to channel them in ways more likely to produce positive outcomes. (For an overview of some strategies see, for example, Watson & Chick, 2001a; Webb, Troper, & Fall, 1995.) 5. In groups of three, how is collaboration inuenced by the composition of the group? Most studies of collaboration have focussed on pairs or foursomes: in the reviews of Webb (1991) and Cohen (1994) these are the only group sizes mentioned explicitly. It would be particularly interesting to examine what happens to interactions in mixed gender threesomes, since one of the group satises a criterion that clearly identies him or her as the odd one out. 6. Can teachers improve their selection and usage of tasks for collaborative situations so that higher level thinking is developed? The areas of data handling and chance provide the opportunity for a variety of open-ended tasks, including data collection. These can be compared with more conventional mathematical problem-solving tasks in terms of the degree of collaboration engendered. 7. Can the research environment more closely mirror a real classroom? In particular, is time on task inuenced by the presence of a researcher or teacher or even a video camera? Previous research has used video cameras, but usually in a limited way by focussing on only one or two of the collaborative groups or by recording the whole classroom (e.g., Artzt & Armour-Thomas, 1997; Cobb & Whitenack, 1996; Yackel, Cobb, & Wood, 1991). Some preliminary work on this question has been done by the present authors (Chick & Watson, 2001; Watson & Chick, 2001b, 2001c).

398

H.L. Chick, J.M. Watson / Journal of Mathematical Behavior 21 (2002) 371400

It is important that future research not only considers the outcomes arising in collaborative versus individual work, but also examines the behaviors and group dynamics to ascertain those that lead to improved cognition.

Acknowledgments This research was funded by Australian Research Council Grant Number A79532539. Sharyn Lidster was the researcher present for the sessions with the boys and Sue Anderson transcribed the videos and added helpful insights. At the time most of this research was carried out Helen Chick was on the staff of the University of Tasmania.

Appendix A. Complete data set Name David Jones Brian Wong John Smith Adam Henderson Andrew Williams Peter Cooper Scott Williams Simon Khan Rosemary Black Jennifer Rado Anna Smith Kathy Roberts Mary Minski Dorothy Myers Sally Moore Janelle MacDonald Age 8 9 10 12 14 16 17 18 8 9 11 12 13 15 17 18 Favorite activity TV Football Football Football TV Board games TV TV Netball Board games Board games Netball Reading Swimming Reading Reading Eye color Blue Green Green Blue Blue Green Blue Brown Brown Green Brown Brown Green Blue Brown Blue Weight (kg) 30 26 29 45 60 54 65 74 24 33 32 32 55 50 56 66 Fast food meals per week 7 1 0 5 10 2 8 12 0 4 1 0 3 2 1 4

References
Artzt, A. F., & Armour-Thomas, E. (1997). Mathematical problem solving in small groups: exploring the interplay of students metacognitive behaviors, perceptions, and ability levels. Journal of Mathematical Behavior, 16, 6374. Carvalho, C., & Csar, M. (2000). The game of social interactions in statistics learning and in cognitive development. In: T. Nakahara, & M. Koyama (Eds.), Proceedings of the 24th Conference of the International Group for the Psychology of Mathematics Education (Vol. 2, pp. 153160). Hiroshima: PME. Chick, H. L., & Watson, J. M. (2001). Data representation and interpretation by primary school students working in groups. Mathematics Education Research Journal, 13, 91111.

H.L. Chick, J.M. Watson / Journal of Mathematical Behavior 21 (2002) 371400

399

Cobb, P., & Whitenack, J. W. (1996). A method for conducting longitudinal analyses of classroom videorecordings and transcripts. Educational Studies in Mathematics, 30, 213228. Cobb, P., & Yackel, E. (1996). Constructivist, emergent, and sociocultural perspectives in the context of developmental research. Educational Psychologist, 31, 175190. Cohen, E. G. (1994). Restructuring the classroom: conditions for productive small groups. Review of Educational Research, 64, 135. Davidson, N. (1985). Small-group learning and teaching in mathematics: a selective review of the research. In: R. Slavin, S. Sharan, S. Kagan, R. Hertz-Lazarowitz, C. Webb, & R. Schmuck (Eds.), Learning to cooperate, cooperating to learn (pp. 211230). New York, NY: Plenum Press. Estepa, A., Batanero, C., & Sanchez, F. T. (1999). Students intuitive strategies in judging association when comparing two samples. Hiroshima Journal of Mathematics Education, 7, 1730. Good, T. L., Mulryan, C., & McCaslin, M. (1992). Grouping for instruction in mathematics: a call for programmatic research on small-group processes. In: D. A. Grouws (Ed.), Handbook of research on mathematics teaching and learning (pp. 165196). New York: Macmillan Publishing Company. Johnson, D. W., & Johnson, R. T. (1985). The internal dynamics of cooperative learning groups. In: R. Slavin, S. Sharan, S. Kagan, R. Hertz-Lazarowitz, C. Webb, & R. Schmuck (Eds.), Learning to cooperate, cooperating to learn (pp. 103124). New York, NY: Plenum Press. Lajoie, S., & Lavigne, N. C. (1994). How does group composition and gender inuence the learning of statistics? Paper presented at the annual conference of the American Education Research Association in New Orleans, LA. Excerpted in NCRMSE Research Review: The Teaching and Learning of Mathematics, 3(2), 79. Lajoie, S. P., Jacobs, V. R., & Lavigne, N. C. (1995). Empowering students in the use of statistics. Journal of Mathematical Behavior, 14, 401425. Lehrer, R., & Romberg, T. (1996). Exploring childrens data modeling. Cognition and Instruction, 14, 69108. Lidster, S. T., Chick, H. L., & Watson, J. M. (1997). Developing cognition in interpreting data. In: N. Scott, & H. Hollingsworth (Eds.), Mathematics creating the future. Proceedings of the 16th Biennial Conference of the Australian Association of Mathematics Teachers (pp. 202209). Adelaide: Australian Association of Mathematics Teachers. Lindquist, M. M. (1989). Mathematics content and small-group instruction in grades four through six. Elementary School Journal, 89, 625632. Maxwell, J. A. (1996). Qualitative research design: an interactive approach. Thousand Oaks, CA: Sage. Miles, M. B., & Huberman, A. M. (1994). Qualitative data analysis: an expanded sourcebook (2nd ed.). Thousand Oaks, CA: Sage Publications. Moore, D. S. (1991). Statistics: concepts and controversies (3rd ed.). New York: W.H. Freeman. National Health and Medical Research Council (Australia). Nutrition Committee. (1957). Standard height-weight tables for Australians. Canberra: Australian Institute of Anatomy. Noddings, N. (1989). Theoretical and practical concerns about small groups in mathematics. Elementary School Journal, 89, 607623. Phelps, E., & Damon, W. (1989). Problem solving with equals: peer collaboration as a context for learning mathematics and spatial concepts. Journal of Educational Psychology, 81, 639646. Ross, J. A., & Cousins, J. B. (1993). Patterns of student growth in reasoning about correlational problems. Journal of Educational Psychology, 85, 4965. Russell, S. J., & Corwin, R. B. (1989). Used numbers: real data in the classroom. Statistics: the shape of the data. Grades 46. Palo Alto, CA: Dale Seymour Publications. Shaughnessy, J. M. (1992). Research in probability and statistics: reections and directions. In: D. A. Grouws (Ed.), Handbook of research on mathematics teaching and learning (pp. 465494). New York: National Council of Teachers of Mathematics and Macmillan. Slavin, R. E. (1989/1990). Research on cooperative learning: consensus and controversy. Educational Leadership, 47, 5254. Stacey, K. (1992). Mathematical problem solving in groups: are two heads better than one? Journal of Mathematical Behavior, 11, 261275. Watson, J. M., & Callingham, R. A. (1997). Data cards: an introduction to higher order processes in data handling. Teaching Statistics, 19, 1216. Watson, J. M., & Chick, H. L. (2001a). Factors inuencing the outcomes of collaborative mathematical problem solving an introduction. Mathematical Thinking and Learning, 3, 125173.

400

H.L. Chick, J.M. Watson / Journal of Mathematical Behavior 21 (2002) 371400

Watson, J. M., & Chick, H. L. (2001b). Does help help?: collaboration during mathematical problem solving. Hiroshima Journal of Mathematics Education, 9, 3373. Watson, J. M., & Chick, H. L. (2001c). A matter of perspective: views of collaborative work in data handling. In: M. van den Heuvel-Panhuizen (Ed.), Proceedings of the 25th Conference of the International Group for the Psychology of Mathematics Education (Vol. 4, pp. 407414). Utrecht: PME. Watson, J. M., & Chick, H. L. (2002). Factors inuencing collaborative outcomes in mathematics in near-classroom and isolated environments. Manuscript submitted for publication. Watson, J. M., Collis, K. F., Callingham, R. A., & Moritz, J. B. (1995). A model for assessing higher order thinking in statistics. Educational Research and Evaluation, 1, 247275. Webb, N. M. (1991). Task-related verbal interaction and mathematics learning in small groups. Journal for Research in Mathematics Education, 22, 366389. Webb, N. M., Troper, J. D., & Fall, R. (1995). Constructive activity and learning in collaborative small groups. Journal of Educational Psychology, 87, 406423. Yackel, E., Cobb, P., & Wood, T. (1991). Small-group interactions as a source of learning opportunities in second-grade mathematics. Journal for Research in Mathematics Education, 22, 390408. Yackel, E., Cobb, P., Wood, T., Wheatley, G., & Merkel, G. (1990). The importance of social interaction in childrens construction of mathematical knowledge. In: T. J. Cooney, & C. R. Hirsch (Eds.), Teaching and learning mathematics in the 1990s (pp. 1221). Reston, VA: National Council of Teachers of Mathematics. Yin, R. K. (1994). Case study research: design and methods (2nd ed.). Thousand Oaks, CA: Sage.

You might also like