مقایسه اجتماعی در کلاس درس: بررسی بهتر از اثر متوسط در میان کودکان مدرسه متوسطه

Abstract The better than average (BTA) effect refers to the tendency for the majority of people to rate themselves as being higher on positive attributes and lower on negative attributes than other people. The present study examined the occurrence of the BTA effect on five important characteristics among 15,806 first-year secondary school Dutch students. In addition, it explored the influence of students' gender, cultural background, and ability level on their evaluations of characteristics relative to their classmates. Results yielded small BTA effects, with the exception of the item “being eager to get high grades,” on which the effect was much larger. In addition, larger BTA effects were found among boys than girls, but this difference could not be attributed to actual differences in performance. Likewise, larger BTA effects were found among ethnic minority students from Turkish and Moroccan backgrounds than ethnic majority students, but this difference also could not be attributed to actual differences in performance. Theoretical and practical implications of these findings are discussed

1. Introduction It has long been observed that students, both those with and without learning difficulties, often hold unrealistic images of themselves, overestimating their mental and scholastic abilities (Schutte, 1929, Stone and May, 2002 and Wylie and Hutchins, 1967). An overly positive picture of their abilities may lead students to believe that they do not have to put forth much effort to succeed in school and may discourage them from completing their homework or preparing well for exams. In the social psychological literature, this type of overestimation or bias is often captured in terms of the better than average (BTA) effect. The BTA effect refers to the tendency for the majority of people to rate themselves as being higher on positive attributes and lower on negative attributes than the average or generalized other (e.g., other college students; Silvera and Seger, 2004 and Suls et al., 2002). That is, correctly or not, individuals believe that they possess more positive attributes and less negative ones than others. Most people, for instance, think they are more sensitive ( Dunning, Meyerowitz, & Holzberg, 1989) and altruistic ( Epley & Dunning, 2000) than others. It must be noted that, from the individual's point of view, the perception of being better may be completely correct. It is possible that, for instance, one is a better driver than others. However, except for attributes for which the distribution in the population is very skewed, it is not possible for the majority of people to be above average ( Dunning, Heath, & Suls, 2004). Therefore, on a group level, this perception has been called an illusion ( Suls et al., 2002). In contrast, individual evaluations of abilities or attributes relative to others are called comparative evaluations. Whether or not these evaluations are biased in nature can only be determined by relating them to factual data that serve as a “reality benchmark”. Both motivational and amotivational theories have been proposed to account for the BTA effect. For instance, from the perspective of motivational theories, it has been argued that people believe in their own superiority because positive illusions help them persist in the face of life's many frustrations, and, as a result, may promote mental health (Taylor & Brown, 1988). In contrast, from the perspective of amotivational theories, people who are unskilled in a particular domain simply lack the meta-cognitive competence to make correct evaluations of themselves relative to others (Kruger & Dunning, 1999, see Chambers and Windschitl, 2004 and Moore and Small, 2007 for reviews). Although BTA effects are almost exclusively recognized and studied in the social psychological literature, BTA effects are relevant phenomena for school psychologists, too. An important reason is that individuals' comparative evaluations of their abilities have been found to be related to their performance level. For instance, Blanton, Buunk, Gibbons, and Kuyper (1999) found students' comparative evaluations in seven courses to be predictive of their academic achievements in these courses 3 months later. As students viewed themselves as better than other students in their class, their achievements improved. Students who hold high comparative evaluations may create a self-fulfilling prophecy; because they view themselves as more capable than others, they are likely to approach tasks with a sense of self-efficacy and high performance expectations, increasing their chances of academic success (Blanton et al., 1999). In this context, it is important to note that Blanton et al. (1999) did not report differences between boys and girls, nor did they examine the potential influence of students' ethnic background. As we will discuss later on in more detail, in our opinion, these two variables—gender and culture—are important in understanding the BTA effects. Another reason to study comparative evaluations is that they are essential to the understanding of the big-fish–little-pond effect (BFLPE; Marsh, 1987; see Marsh et al., 2008 for a recent review), a phenomenon that, in educational research, has received considerable attention. The BFLPE refers to the phenomenon that equally able students have lower academic self-concepts in schools or classes where the average achievement level is high than in schools or classes where the average achievement level is low. Recently, Huguet et al. (2009) found that the BFLPE was eliminated after controlling for students' perceived standing relative to most of their classmates, showing that comparative evaluations form the roots of the BFLPE. Learning more about students' comparative evaluations therefore seems important in the light of the BFLPE. The present study set out to examine BTA effects among secondary school students, which have only scarcely been examined during this educational period. Studying BTA effects among secondary school students may generate new and exciting insights in students' learning process, classroom processes, and educational attainments.

3. Results 3.1. Mean comparative evaluations Separate analyses were conducted on the five comparative evaluation items in the selected sample of 15,806 students. There was an additional loss of cases due to missing values on the specific comparative evaluation item itself or on the associated specific predictor—especially on the physical education grade and GPA. Deleting the cases that were missing on the specific predictor hardly changed the means and standard deviations of the comparative evaluations.3Table 2 provides basic descriptive information for the five multilevel analyses. The first two rows in Table 2 show the numbers of students and classes that were available for the analysis of each comparative evaluation item. The second part of Table 2 shows how these students were divided over their classes. The distribution of the class-sizes does not deviate much from the normal distribution in each case. However, it appears that, for the first two comparative evaluations, the average class-size was lower than for the other comparative evaluations, the standard deviation was somewhat higher, the distribution was somewhat less skewed to the left and somewhat flatter. In the next part of Table 2, the descriptive information of the five comparative evaluations is shown. The means on the comparative evaluations of athleticism, ability to get high grades, likability, and attractiveness were very near the scale midpoint of 3.00, which indicates that there was hardly any BTA effect on these items. In contrast, on the item “eager to get high grades,” the mean was much higher than the scale midpoint, indicating that on this item, there was a considerable positive bias; students perceived themselves to be much more eager to get high grades than their classmates. Dividing 0.55 (the difference with 3.00) by 0.81 (the overall standard deviation) yields a value of 0.68, which can be considered the effect size (Cohen's d) of the bias ( Cohen, 1988). According to accepted norms, this is a medium effect size. The other effect sizes were far beyond the criterion of 0.20 for a small effect (0.01, 0.09, − 0.02 and 0.04, respectively). The means on the comparative evaluation items in the final selections hardly deviated from the means with maximum sample sizes. For likability and attractiveness, the distribution was very peaked. Table 2 also shows descriptive information for the specific predictor variables (i.e., the Physical education grade for the comparative evaluation of athleticism, GPA for the comparative evaluation of ability to get high grades, and academic achievement motivation for the comparative evaluation of eager to get high grades). The distribution of Physical education grade was highly peaked. Table 3 shows the means of the comparative evaluation items as a function of gender and ethnic group. In support of Hypothesis 1, the boys' mean was higher than the girls' on all five items. Furthermore, in support of Hypothesis 2, the means for Dutch students were among the lowest ones on all five items, whereas the means for Turkish and Moroccan students were consistently highest. Table 3. Uncorrected gender and ethnic group means. Statistics Athleticism Ability to get high grades Likability Eager to get high grades Attractiveness Gender Girls 2.83 3.02 2.95 3.44 2.96 Boys 3.20 3.10 3.03 3.67 3.11 Ethnic group Dutch 2.99 3.04 2.97 3.50 2.99 Indonesian 3.06 3.00 2.95 3.57 3.11 Western 3.05 3.08 2.98 3.62 3.04 SE Asian 2.99 3.10 3.14 3.78 2.99 ASA 2.99 3.15 3.14 3.80 3.25 Rest/mixed 3.06 3.18 3.11 3.78 3.16 Turkish 3.20 3.32 3.27 4.04 3.35 Moroccan 3.22 3.30 3.20 4.04 3.43 Total sample M 3.01 3.06 2.99 3.55 3.03 SD 0.86 0.69 0.60 0.81 0.74 Note. M = mean, SD = standard deviation, ASA = Antilles and Surinam, SE Asian = South-East Asian. Table options 3.2. Predicting comparative evaluations Results of the five multilevel analyses are shown in Table 4 and Table 5. Table 4 shows the results for the unconditional models (i.e., without any predictors), and Table 5 shows the results for the full models. Because the level-2 predictor prop_BOY was not significant in any analysis, we did not include this variable in Table 5. One conclusion from Table 4 is that the intercepts (i.e., the adjusted grand means) are almost equal to the overall means that were shown in Table 2. The second conclusion is that only a very small amount of the total variance is found at the class level. The percentages of total variance at the class level, as measured by intraclass correlations, varied from 0.1% for the comparative evaluation on athleticism to 5.6% for the comparative evaluation on “eager to get high grades”. Table 5 shows the results for the full models. The regression coefficients in these models represent unique effects (i.e., the effect of each variable controlled for the effect of all other variables). Table 4. Unconditional multilevel models: intercepts, variance components, and standard errors (between parentheses) and intraclass correlation coefficients. Statistics Athleticism Ability to get high grades Likability Eager to get high grades Attractiveness Intercept 3.01 (0.01) 3.06 (0.01) 3.00 (0.01) 3.57 (0.01) 3.04 (0.01) Variance estimates Class 0.00 (0.00) 0.01 (0.00) 0.01 (0.00) 0.04 (0.00) 0.02 (0.00) Student 0.74 (0.01) 0.47 (0.01) 0.35 (0.00) 0.62 (0.01) 0.53 (0.01) Total 0.74 0.48 0.36 0.66 0.55 ICC ⁎ 100 0.14 1.15 2.22 5.62 3.29 Index of fit Deviance 31,458.00 27,051.44 28,321.19 37,178.94 34,343.33 Note. ICC = intraclass correlation. Table options Table 5. Full multilevel models: regression coefficients, standard errors (between parentheses), and variance components. Statistics Athleticism Ability to get high grades Likability Eager to get high grades Attractiveness Fixed effects Student level Intercept 2.83 (0.01) 2.97 (0.012) 2.91 (0.01) 3.36 (0.01) 2.90 (0.01) Specific predictor 0.35 (0.01)⁎⁎⁎ 0.23 (0.01)⁎⁎⁎ – – 0.24 (0.01)⁎⁎⁎ – – Achievement − 0.07 (0.01)⁎⁎⁎ 0.09 (0.01)⁎⁎⁎ − 0.06 (0.01)⁎⁎⁎ − 0.03 (0.01)⁎⁎ − 0.14 (0.01)⁎⁎⁎ Gender (boys) 0.28 (0.01)⁎⁎⁎ 0.13 (0.01)⁎⁎⁎ 0.08 (0.01)⁎⁎⁎ 0.22 (0.01)⁎⁎⁎ 0.15 (0.01)⁎⁎⁎ Turkish 0.21 (0.05)⁎⁎⁎ 0.27 (0.04)⁎⁎⁎ 0.24 (0.03)⁎⁎⁎ 0.22 (0.04)⁎⁎⁎ 0.25 (0.04)⁎⁎⁎ Moroccan 0.14 (0.05)⁎⁎ 0.22 (0.04)⁎⁎⁎ 0.17 (0.03)⁎⁎⁎ 0.21 (0.04)⁎⁎⁎ 0.30 (0.04)⁎⁎⁎ ASA − 0.02 (0.04) 0.08 (0.04)⁎ 0.13 (0.03)⁎⁎⁎ 0.08 (0.04)⁎ 0.17 (0.037)⁎⁎⁎ Indonesian 0.07 (0.05) − 0.08 (0.04) − 0.02 (0.04) 0.05 (0.05) 0.11 (0.04)⁎⁎ SE Asian − 0.04 (0.06) 0.02 (0.05) 0.16 (0.04)⁎⁎⁎ 0.19 (0.05)⁎⁎⁎ − 0.03 (0.05) Western 0.04 (0.03) 0.04 (0.03) 0.00 (0.02) 0.06 (0.03)⁎ 0.02 (0.03) Rest/mixed 0.07 (0.04) 0.12 (0.03)⁎⁎⁎ 0.11 (0.03)⁎⁎⁎ 0.11 (0.04)⁎⁎⁎ 0.10 (0.03)⁎⁎ Class level ATS_class 0.10 (0.02)⁎⁎⁎ − 0.08 (0.01)⁎⁎⁎ 0.03 (0.01)⁎⁎ − 0.07 (0.01)⁎⁎⁎ 0.06 (0.01)⁎⁎⁎ Prop_EM 0.10 (0.05)⁎ 0.17 (0.04)⁎⁎⁎ 0.09 (0.04)⁎ 0.44 (0.05)⁎⁎⁎ 0.13 (0.04)⁎⁎⁎ Variance estimates Class 0.01 (0.00) 0.01 (0.00) 0.00 (0.00) 0.01 (0.00) 0.01 (0.00) Student 0.58 (0.01) 0.40 (0.01) 0.35 (0.00) 0.55 (0.01) 0.51 (0.01) Total 0.59 0.41 0.35 0.56 0.52 ICC ⁎ 100 1.20 1.71 1.14 2.50 1.54 Index of fit Deviance 28,563.54 25,138.61 27,992.59 34,946.54 33,681.80 Note. ASA = Antilles and Surinam, SE Asian = South-East Asian, ATS_class = class average on achievement test, Prop_EM = proportion of ethnic minority children in class, ICC = intraclass correlation. ⁎⁎⁎ p < .001. ⁎⁎ p < .01. ⁎ p < .05. Table options 3.2.1. Athleticism The Physical education grade (i.e., the reality benchmark predictor) appeared to be an important predictor of students' comparative evaluation of their athleticism (β = .35). As can be expected, students with higher grades within their class tended to evaluate themselves as more athletic than their classmates. Of the other level-1 predictors, the effects of achievement, gender, and two of the ethnic minority groups reached significance. The coefficient for achievement was negative (β = −.07). Thus, higher scoring students tended to evaluate themselves as less athletic. Boys (β = .28) as well as Turkish (β = .21) and Moroccan (β = .14) students evaluated themselves as more athletic. Two of the three level-2 (context) predictors were significant, with the exception of the proportion of boys. In classes where the average level of achievement was relatively high, and in classes with a relatively high proportion of ethnic minority students, students tended to evaluate themselves as more athletic (both βs = .10). The predictors accounted for 20.7% of the initial total variance. 4 3.2.2. Ability to get high grades Again, the specific predictor, in this case GPA, had a significant effect. Students with a higher GPA within their class (β = .23) evaluated their ability to get high grades as relatively high. The same was true for students with higher achievement (β = .09); for boys (β = .13); for Turkish (β = .27), Moroccan (β = .21) and Surinam/Antillean (β = .08) students; and for students in the “rest/mixed” category (β = .12). Two of the three level-2 (context) predictors were significant, with the exception of the proportion of boys. The effect of the class achievement level was negative (β = −.08); the comparative evaluation of the ability to get high grades was higher in classes with a low achievement level. In classes with a relatively high proportion of ethnic minority students, students evaluated themselves as more able to get high grades (β = .17). The predictors accounted for 13.7% of the initial total variance. 3.2.3. Likability There was no specific (i.e., “reality benchmark”) predictor. Of the level-1 predictors, the effects of achievement, gender, and five of the ethnic minority groups reached significance. Students with a higher achievement tended to evaluate themselves as less likable (β = −.06). Boys (β = .08) as well as Turkish (β = .24), Moroccan (β = .17), Surinam/Antillean (β = .13), South-East Asian (β = .16) and “rest/mixed” students (β = .11) evaluated themselves as more likable. Two of the three level-2 variables were significant. The comparative evaluation of likability tended to be higher in classes with a high achievement level (β = .03) and in classes with a relatively high proportion of ethnic minority students (β = .09). The predictors accounted for 2.5% of the initial total variance. 3.2.4. Eagerness to get high grades As can be expected, students who had higher scores on academic achievement motivation (i.e., the specific predictor variable) evaluated themselves as more eager to get high grades (β = .24). Students who had higher achievement tended to evaluate themselves as less eager to get high grades (β = −.03). Of the other level-1 predictors, the effects of gender and six of the seven ethnic minority groups were significant, with the exception of Indonesian students. Boys (β = .22) as well as Turkish (β = .22), Moroccan (β = .21), Surinam/Antillean (β = .08), South-East Asian (β = .19), Western (β = .06), and “rest/mixed” (β = .11) students evaluated themselves as more eager to get high grades. Two of the three level-2 predictors were significant. The comparative evaluation of the item “eager to get high grades” was higher in classes with a low achievement level (β = −.07) and/or in classes with a relatively high proportion of ethnic minority students (β = .44). The predictors accounted for 14.7% of the initial total variance. 3.2.5. Attractiveness There was no specific (i.e., “reality benchmark”) predictor. Of the level-1 predictors, the effects of achievement, gender, and five of the ethnic minority groups were significant. Students with high achievement evaluated themselves as less attractive (β = −.14). Boys (β = .15) as well as Turkish (β = .25), Moroccan (β = .30), Surinam/Antillean (β = .17), Indonesian (β = .11), and “rest/mixed” (β = .10) students evaluated themselves as more attractive. Two of the three level-2 predictors were significant. The comparative evaluation of attractiveness was higher in classes with a high achievement level (β = .06) and in classes with a relatively high proportion of ethnic minority students (β = .13). The predictors accounted for 4.8% of the initial total variance. Table 5 also shows that the intraclass correlations were very low. The percentage of unexplained variance at the class level range from 1.1% (likability) to 2.5% (eager to get high grades). For eager to get high grades, the percentage of variance at the class level in the unconditional model ( Table 4) was relatively high (5.6%). In the full model for this variable, more than 60% of the class-level variance was explained.