Self-Efficacy Beliefs and Social Comparison Processes in the Primary Classroom

Draft paper: please contact the author before quoting

Self-Efficacy Beliefs and Social Comparison Processes in the Primary Classroom

JaneL. Webb-Williams

University of Cambridge

Faculty of Education

184 Hills Road

Cambridge

CM2 8PQ

Paper presented at the British Educational Research Association Annual Conference, University of Warwick, 6-9 September 2006

This research forms parts of a PhD currently being undertaken at the University of Cambridge, funded by ESRC.

Abstract

This paper focuses upon one of the influences that can help to create and develop an individual’s self-efficacy, that of social comparison. Building upon an earlier pilot study which established a strong, positive correlation between self-efficacy and academic performance, this study involved 301 children aged between 10 and 12 years from 2 large junior schools and 10 small primary schools located in East Anglia, UK. By using existing differently sized social comparison groups, such as those naturally found in small and large schools, the current study aimed to investigate social comparison processes in an ecologically valid way. Using a mixed methods approach, this studyfound evidence to suggest that social comparisons were automatic in nature and that pupils’ of different attainment differed in their intentional use of social comparison. Girls from large schools were found to hold the lowest correlation between self-efficacy and performance, the lowest attitudes towards science and the lowest attainment in science. The priority given by girls from large schools to interpersonal and intergroup comparisons is suggested in such findings.

Introduction and Aims

This paper concerns the psychological constructs of self-efficacy and social comparison. Self-efficacy, which plays a pivotal role in Bandura's (1986) social cognitive theory, refers to "the beliefs in one's capabilities to organise and execute the courses of action required to produce given attainments" (Bandura, 1997, p3). Self-efficacy is not concerned with the capabilities one has, but with one’s perception of these capabilities. Self-efficacy beliefs, according to Bandura (1997), determine how people feel, think, motivate themselves and behave. They constitute a principle factor of personal agency and are instrumental to the goals individuals pursue, and the control they have over their environment.

Self-efficacy, according to Bandura (1997), is influenced by four factors: mastery experience, verbal persuasion, vicarious experience and physiological/emotional state. The emphasis of this paper is vicarious experience as experienced through social comparison. Social comparison is concerned with the processes involved in comparing ourselves with others (Festinger, 1954). More recently Wood (1996) has defined social comparison as the process of thinking about information about one or more other people in relation to the self (p521). In this definition Wood (1996) has expanded the meaning of social comparison to incorporate comparisons with stereotypes and hypothetical characters. The direction of such social comparison appears to be vital.Brickam and Bulman (1977) proposed that people who seek improvement make “upward” comparisons with people superior to themselves and people that seek preservation make “downward” comparisons with inferior others. However, more recently researchers (e.g. Mussweiler, Ruter and Epstude, 2004) have suggested that social comparisons can result in positive and negative effects dependent upon whether they are seen as contrasts or assimilations. Contrasts are when individuals emphasise the differences between themselves and the comparison target. Making contrasts in downward comparisons is likely to produce feelings of superiority and confidence whereas making contrasts in upward comparisons tends to lead to feelings of jealousy and inadequacy. Assimilations can also have positive and negative consequences. When individuals make assimilations they emphasise their similarities to the comparison target. Thus upward assimilations are associated with learning, growth and inspiration as individuals aim to be more like the comparison target. However, making assimilations with downward comparisons has the opposite effect since individuals focus on the similarities between themselves and the lower comparison target leading to increased anxiety and self-doubt.

The impact of social comparison on performance has not yet been systematically studied (Huguet, Dumas, Monteil and Genestoux, 2001) which is surprising given that children pay more attention to how their performance compares to that of their peers than to how their performance compares to their own past performance (Ruble & Fleet, 1988). Moreover, although social comparison has been identified by Bandura (1997) as a major influence on self-efficacy beliefs, self-efficacy researchers have paid little systematic attention to social comparison theory. This paper addresses both these issues through its examination of self-efficacy, attainment and social comparison theory. In addition this research is one of the handful of educational based self-efficacy studies to be conducted in the UK. Despite two decades of research in the USA, the few UK studies that have been conducted have tended to involve university students, yet the construct of self-efficacy is a potentially useful explanatory construct that can be applied to many educational issues and it has significant application to the classroom since it has been found to be adaptable and open to alteration. Indeed, in contrast to individuals who doubt their capabilities, Bandura (1994) claims that people with a high sense of self-efficacy approach difficult tasks as challenges, set challenging goals for themselves, sustain effort even when faced with failure, quickly recover after setbacks, develop an intrinsic interest in activities, and attribute failure to factors which are adaptable e.g. insufficient effort or skills. Moreover, educational studies have demonstrated that self-efficacy beliefs are highly correlated with academic achievement through their influence on persistence, effort, motivation and choice (Schunk, 1981). The domain of mathematics has received the greatest attention whereas there has been little work on science education. This paper addresses these issues and focuses on the domain of science and the notions of social comparison and self-efficacy beliefs within the English primary classroom. Specifically the following aims drove the study:

• to establish whether, and to what extent, self-efficacy beliefs vary according to school size
• to explore the relationship between self-efficacy beliefs and performance in the domain of science and establish to what extent they might be influenced by size of social comparison group
• to gain an understanding of the ways in whichsocial comparison processes and effects are experienced by children from primary schools of different size.

In order to attend to the above aims, a mixed methods design was utilised with the main objective of using the qualitative findings to help to explain the quantitative results.The parallel mixed model design (Creswell, 2003; Tashakkori & Teddlie, 1998) was such that qualitative and quantitative data were collected together. The quantitative methods involved the completion of Likert scale self-efficacy questionnaires at different levels of specificity as well as science performance measures. Qualitative methods involved individual pupil interviews. Pupils from large and small schools were involved in the research so as to establish any differential effects size of social comparison group. Small schools in the UKtypically have classes of mixed year groups (e.g. year 5 and year 6 taught together in the same class). Since the year 6 children from these schools have only a small number of peers with whom to make comparisons, compared with those from large schools, it is hypothesised that they will differ according to their social comparisons and their self-efficacy beliefs.

Method

Participants and Procedures

119Year 6 pupils (56 girls, 63 boys)) from10 small primary schools (roll size < 105) and 182 Year 6 pupils (84 girls, 98 boys) from 2 large junior schools (roll size > 270) participated in the study. Thus a total of 301 pupils aged between 10 and 12 years completed Likert scale self-efficacy questionnaires at different levels of specificity as well as social comparison measures and science achievement tasks during the term prior to their transfer to secondary school (May 2005). Qualitative data were collected by way of individual interviews with 32 pupils (16 from small primary schools and 16 from large junior schools), pupils being selected for interview according to their self-efficacy and performance profiles.

Instruments

Interviews

As recommended by Tashakkori and Teddlie (1998), the funnel interview was used since it is directly applicable to mixed methods research. In this type of interview the researcher mixes open and closed-ended questions, starting with broad questions and ending with focused issues. In the current study some questions were theoretically linked to the Likert scaled items employed in the self-efficacy questionnaires and some focused on social comparison. Interviews were conducted on the school premises during school time. Interviews tended to last no longer than 20 minutes and all responses were audio taped. Using field notes and audio tapes each interview was transcribed. Transcriptions were then analysed using open coding in Atlasti to identity themes and look for patterns (Miles and Huberman, 1994). The general aim was to understand children’s ideas of their own performance and the role of social comparison.

Self-Efficacy Scales

All the self-efficacy scales employed in the study were constructed according to the framework set out by Bandura (2001). Indeed, two of the five scales were taken directly from Bandura's Children's Self-Efficacy Scale. A 7 point Likert response scale was used across all the self-efficacy measures with verbal descriptors (not well at all to very well) at points 1, 3, 5, and 7.Table 1 shows the internal consistency reliabilities.

Table 1: Self-efficacy scale reliabilities: Cronbach’s alpha

N Items / Pilot Study
N=50 / Main Study
N=301
Self-Regulated Learning / 11 / .90 / .87
General Academic Achievement / 3 / .51 / .46
General Science Attainment / 3 / .87 / .76
Science Domain Specific / 8 / .81 / .87
Scientific Enquiry Task Specific / 8 / .86 / .89

1. Self-efficacy for general academic achievement: This scale measures self-efficacy for achievement in three core subjects:English, Maths and Science e.g. “How well can you learn science?” It represents a modified shortened version of the original academic achievement scale inBandura’s Children’s Self-Efficacy scale.

1. Self-efficacy for self-regulated learning: This scale measures self-efficacy beliefs for self- regulatory processes such as time planning and management e.g. “How well can you study when there other interesting things to do?”. It was taken, without modification, from Bandura’s Children’s Self-Efficacy scale.

1. Self-efficacy for general science attainment: Items in this scale refer to one’s confidence to achieve attainment levels in science (levels 3, 4 and 5) e.g. “How confident are you that you can get a level 5 in science?”. It replicates an approach used by other self-efficacy researchers in the USA.

1. Domain-specific self-efficacy for science :This scale measures self-efficacy for a number of core topics in the sciencecurriculum e.g. “How well can you answer questions about keeping healthy?”

1. Task-specific self-efficacy for science: Items specifically regarding scientific enquiry are the focus of this measure e.g. “How well can you write a conclusion for an investigation?”

Social comparison measures

Research involving social comparison suggests that a number of methods should be employed. Wood (1996) divided these methods into three categories: the selectionapproach, the reaction approach and the narration approach. All three methods of measurement were used in this study.

1)The selection approach measuredthe information individuals sought when making social comparisons.The techniques used within this study were the rank order paradigm developed by Wheeler (1962,1966; cited in Wheeler, 1991) and the comparison target (Blanton et al, 1999).The rank order paradigm involved presenting participants with information about the rank order of a hypothetical science test and information regarding their rank. Participants were told that they occupied the middle rank and were informed of their numerical score. Participants were then asked to select one rank whose score they would like to know. In this way direction of comparison (up or down) and similarity of comparison (adjacent or extreme) was assessed.Comparison target was assessed using a free choice selection approach employed by Huguet et al (2001). Pupils were asked to nominate a student with who they typically compare their science work. Additionally pupils were asked to rate how good the comparison target was at science compared with most of their classmates in order to gauge perceptions of attainment.

2)The reaction approach to measurement involved examining the impact of social comparisons. Comparative rating was the technique used within this approach. This involved asking individuals to rate themselves in relation to others. Pupils were asked to rate on a 5 point scale (from much worse to much better) how good at science they were compared with most of their classmates.

3)The narration approach to measurement of social comparison refers to methods in which individuals report the comparisons they make during their everyday lives. This allows the analysisof the frequency, content and direction of normal everyday social comparisons. In the current study pupils were asked to comment on their engagement in social comparison within the interviews.

In addition to the above details regarding class size and year group size were gathered so as to gain a description of the social comparison group experienced by each child.

Science attainment test

A shortened version of the 2003 Key Stage 2 science SATs was utilised as a performance indicator. Pupils were given 20 minutes to complete the test.The length of this test was vital since access to schools was given on the agreement that disruption to normal lesson time was kept to a minimum. Thus, original Standard Assessment Tests (SATs), which are taken annually by all Year 6 pupils in England, were not considered since they take longer to administer. In order to ascertain whether the science performance test provided an accurate measurement of pupil performance, a comparison was made between the scores on the performance measure and the scores predicted by the class teachers. The coefficient of r = .74 (significant at the 0.01 level) demonstrates that the performance instrument was a good measure of actual performance in science. Moreover, the Cronbach’s alpha of 0.78 was good.

Analyses

All analyses were conducted using SPSS version 13 and Atlas.ti version 5.2. The quantitative analyses are presented here together with some of the qualitative analyses. The remaining qualitative analyses of the interview data are presented together with the interpretation of the quantitative findings in the discussion that follows.

The mean score on the science attainment test was 13.04 with a standard deviation of 3.86. All pupils were grouped into particular attainment categories according to the performance data obtained. Pupils who scored less than one standard deviation below the mean (9.18) where categorised as “low” attainers, pupils who scored more than one standard deviation above the mean (16.9) where categorised as “high” attainers, and pupils who scored between these scores were categorised as “medium” attainers. Of the 32 pupils interviewed 10 pupils were deemed to be in the “high” group, 11 pupils in the “low” group and 11 pupils in the “medium” group. Thus a roughly even number of pupils were interviewed at each attainment level.

Table 2: Overall mean scores for science performance and self-efficacy scales

by school size

Overall Mean (SD)
(N = 301) / Large Schools Mean (SD)
(N=182) / Small Schools Mean (SD)
(N= 119)
Science Performance / 13.04
(3.86) / 12.99
(3.96) / 13.11
(3.71)
Self-Efficacy for
Academic Achievement / 15.45 (2.73) / 15.50
(2.84) / 15.37
(2.56)
Domain-specific
Self-Efficacy / 39.55
(7.59) / 39.64
(7.56) / 39.42
(7.65)
Task-specific Self-Efficacy / 39.93
(7.88) / 40.13
(7.74) / 39.63
(8.09)
Self-Efficacy for Self-Regulated Learning / 54.51
(9.50) / 54.79
(8.90) / 54.08
(10.37)
Self-Efficacy for Science Attainment / 15.71
(4.40) / 16.13
(4.11) / 15.04
(4.76)

Table 2 shows that children from small primary schools completed the questionnaires in a similar way to children from large junior schools. The means and standard deviations were of a very similar magnitude. This suggests that as a whole pupils’ have a similar sense of self-efficacy regardless of the size of school they attend.

Table3 : Correlation between self-efficacy and performance in science

by school size

Self-Efficacy for: / Overall
N=301 / Large Junior
N=182 / Small Primary
N=119
Self-regulated learning / 0.42** / 0.34** / 0.53**
General Academic Achievement / 0.53** / 0.50** / 0.57**
General Science Attainment / 0.58** / 0.58** / 0.61**
Science - Domain specific / 0.58** / 0.55** / 0.63**
Scientific Enquiry -Task specific / 0.57** / 0.50** / 0.70**

** Significant to 0.01

It can be seen from Table 3 that all the self-efficacy measures were strongly correlated with performance and reached statistical significance at the 0.01 level. However examination of the results by gender found that the girls from large schools in particular exhibited distorted self-efficacy beliefs in relation to their attainment, with four of the five scales below a correlation of 0.50 (see Figure 1). Girls from large schools also exhibited the lowest attitudes to science, liking science and the lowest performance of the groups (see Table 4).

Figure 1 : Correlation between self-efficacy and performance in science

by gender and school size

Table 4: Attitudes and attainment in science by school size and gender

School Size / Gender / Science Attainment / Attitudes to Science / Liking Science
Small Primary Schools / Boy / 13.13 / 47.88 / 25.17
Girl / 13.09 / 42.97 / 22.62
Large Junior Schools / Boy / 13.14 / 42.03 / 21.26
Girl / 12.81 / 39.85 / 19.63

It should be noted that the science attitude scales were administered as an additional “filler” task. It was therefore expected that not all pupils would complete the scales. Indeed 75 pupils (41 boys: 34 girls) of the 119 pupils from small schools and 169 pupils (94 boys: 75 girls) of the 182 from large schools completed the scales. A correlational analysis by attainment group was not viable due to the low number of participants in each group. For example there were only 10 low attainment boys from small schools.