نظریه هوش و رفتارهای روزانه ناتوان سازی دانش آموزان

Abstract The current study sought to examine the relationship between students' theory of intelligence and daily self-handicapping behaviors. Ninth grade students completed a background survey with an eight-item measure assessing one's theory of intelligence (Dweck, 1999) and global measures of procrastination and self-handicapping. Participants then completed daily surveys for 2 weeks in which they reported how much homework they had, perceived school difficulty, time spent studying and in other domains, and how much effort they spent on their homework/studying. Results revealed that the strength of one's entity theory of intelligence was positively associated with self-handicapping and procrastination, replicating past findings. It was also found that entity theories of intelligence were associated with reduced responsiveness to daily school demands when compared to incremental theories. Not only do these results demonstrate an association between theory of intelligence and maladaptive school behaviors, but they show how these behaviors manifest on a daily basis.

1. Introduction Previous research has demonstrated that people can have one of two implicit theories of intelligence (Dweck, 1999). Those holding an entity theory believe that intelligence is a fixed, stable quality. However, those with an incremental theory believe that intelligence is malleable or changeable with effort. Past research has shown that entity theories of intelligence are related to negative academic behaviors, including self-handicapping and procrastination (e.g., Howell and Buro, 2009, Ommundsen, 2001, Rhodewalt, 1994 and Shih, 2011), as well as negative feelings toward school (e.g., King et al., 2012 and Robins and Pals, 2002). These negative outcomes likely emerge because of entity theorists' attributions to failure and success. When entity theorists fail, they form a global, stable belief of lack ability. Even when they succeed, entity theorists are still concerned with possible future failure if their ability is not adequate for future success (Dweck et al., 1993, Hong et al., 1999 and Siegle et al., 2010). Entity theorists' view of their intelligence is thus constantly in a state of vulnerability, particularly when confronted with challenge. Therefore, they tend to respond to challenges with helplessness and a lack of self-regulation, rather than an effort-driven approach (Burnette et al., 2013, Diener and Dweck, 1978 and Robins and Pals, 2002). On the other hand, incremental theorists are more likely to form a strong belief in the importance of effort, whether in the face of failure or success (Dweck et al., 1993 and Hong et al., 1999). While these associations have been identified between individuals, they have not been explored within individual students to examine how they manifest themselves in terms of specific behaviors toward schoolwork and feelings about school as they occur in students' lives. The goal of the current study is therefore to examine different patterns of school behaviors and feelings about school on a daily basis according to students' individual theories of intelligence. 1.1. Theories of intelligence and school behaviors Previous studies have found many associations between entity theories of intelligence and students' self-reports of self-handicapping and procrastination (e.g., Howell and Buro, 2009, Ommundsen, 2001, Rhodewalt, 1994 and Shih, 2011). Self-handicapping is defined as the creation of obstacles to compensate for possible future poor performance, thus allowing the self-handicapper to externalize the cause of failure (Midgley et al., 2000, Ommundsen, 2001 and Rhodewalt, 1994). Procrastination is the tendency to delay or avoid a task due to a lack of self-regulation (Howell and Buro, 2009 and Tuckman, 1991). In the current study, we conceptualize procrastination as a method of self-handicapping: the self-handicapper uses procrastination as a reason for possible failure or poor performance. Those with entity theories are more likely to self-handicap because doing so provides them with an explanation for possible poor performance that does not reflect on their intelligence, or lack of intelligence, allowing for an attribution to something other than one's intelligence. They are motivated to do so because they believe that failure to demonstrate intelligence indicates a lack of intelligence, which they perceive to be unchangeable (Ommundsen, 2001 and Rhodewalt, 1994). However, because less effort is put into studying and schoolwork, those who self-handicap do not do as well and are not as successful in school (Tice & Baumeister, 1997). Even when entity theorists are successful though, they attribute this success to external factors, such as luck, and not to internal factors, such as intelligence (Robins & Pals, 2002). In addition to self-handicapping as a means for having an excuse for potential failure, entity theorists may also put less effort into their schoolwork because they believe that trying hard could be perceived as demonstrating a lack of intelligence (Hong et al., 1999 and Robins and Pals, 2002). Entity theorists therefore also avoid challenges whenever possible (Dweck & Leggett, 1988). The result of this is that students with entity theories tend to be less academically successful as a school year progresses, with decreasing grades over time (Blackwell et al., 2007 and Haimovitz et al., 2011). On the other hand, previous research has shown that incremental theories of intelligence are associated with more academically adaptive behaviors. For example, incremental theories have been negatively correlated with procrastination and self-handicapping (Howell and Buro, 2009 and Ommundsen, 2001), and positively correlated with increased effort and persistence in response to theoretical, academic challenge response statements (e.g., “When I encounter difficulties completing academic assignments and want to give up, I always tell myself to keep persisting,” Diener and Dweck, 1978, Robins and Pals, 2002 and Shih, 2011). Blackwell et al. (2007) found that junior high school students who held an incremental theory of intelligence were less likely to respond to failure with helpless attributions and more likely to respond with positive self-regulation strategies such as increased effort in studying. They found that students with an incremental theory of intelligence at the beginning of junior high school had higher math grades two years later than students with an entity theory of intelligence (Blackwell et al., 2007). However, previous work is limited in that it assesses students' self-reports of their likelihood of self-handicapping, rather than their specific self-handicapping behaviors. Using self-report measures of likelihood of procrastination in general has previously been found to be less accurate in predicting students' academic performance than the use of behavioral measurements of procrastination (Moon and Illingworth, 2005 and Steel et al., 2001). This is because such self-report measures do not take into account fluctuations in academic deadlines and other pressures and resulting changes in procrastination behaviors. For example, in a study by Howell and Buro (2009), students filled out a procrastination survey measuring their likelihood to delay starting or completing a task. Similarly, Rhodewalt (1994) had students fill out a self-handicapping scale rating themselves on lack of effort and procrastination. By asking students explicitly whether or not they procrastinate or self-handicap, many may give socially desirable answers instead of the truth. Students may also simply be unaware of their behaviors and thus unable to answer these kinds of questions correctly. In general, daily reports provide greater ecological validity than one-time measures of behavior and are also less likely to suffer from problems of retrospective bias (Hurlburt & Melancon, 1987). While previous research has explored associations between individuals' theories of intelligence and their perceptions of their procrastination or self-handicapping behaviors (Howell and Buro, 2009, Ommundsen, 2001, Rhodewalt, 1994 and Shih, 2011), the goal of the current study is to explore what self-handicapping look like in terms of the actual behaviors students engage in on a daily basis. Therefore, in addition to asking general questions about whether or not they tend to procrastinate and self-handicap, we also explore students' daily behaviors such as time use and effort on school work, and whether these vary on a daily basis according to school challenges. Rather than asking retrospective questions targeted at the previous week, month, or school year, our measures specifically assess students' behaviors for that current day so as to eliminate any memory loss or confusion from long-term recall, or re-interpretation of behaviors (Hurlburt & Melancon, 1987). 1.2. Theories of intelligence and school feelings In addition to academic behaviors, theories of intelligence have also been found to be associated with feelings toward school. For example, for junior high school students, entity theories of intelligence have been found to predict negative feelings toward school including anger, anxiety, shame, boredom, and hopelessness (King et al., 2012 and Robins and Pals, 2002). Entity theorists are also more likely to have declining motivation as the school year progresses (Haimovitz et al., 2011). In contrast, incremental theorists are more likely to have positive emotions toward school, such as determination, enthusiasm, and inspiration (Robins and Pals, 2002 and Shih, 2011). Again, however, this work is limited in that it relies on students' global reports of school feelings. Given entity theorists' particular concern with challenge and failure, it is likely that feelings about school will fluctuate according to these experiences on a regular basis. For example, entity theorists may be less likely to report liking school on days in which they find schoolwork difficult since this difficulty might challenge their view of their own. However, entity theorists may be more likely to report liking school on days in which their schoolwork is not difficult, thus reaffirming their confidence in their intelligence. 1.3. The current study The purpose of this study is to examine whether global associations between theories of intelligence and school behaviors are replicated when students' actual daily behaviors are examined. First, before exploring daily behaviors, we attempt to replicate associations between theories of intelligence and global measures of self-handicapping and procrastination. We hypothesize that an entity theory of intelligence will be positively correlated with self-handicapping and procrastination. Next, we will examine associations between theories of intelligence and averages of daily school experiences and feelings. We hypothesize that an entity theory of intelligence will be negatively associated with positive feelings toward school and feelings of being a good student, averaged across the days of the study. Finally, we will explore how self-handicapping manifests on a daily basis by examining whether students with incremental and entity theories of intelligence respond differently to daily changes in academic pressure, such as having a lot of homework or feeling like school is hard, in terms of their time use and effort on schoolwork. We hypothesize that entity theorists will spend less time and effort studying on days in which they report having more or challenging homework, while spending more time engaged in non-school related activities. We also explore gender differences, given mixed findings from previous research in both mean levels of theories of intelligence and associations between theories of intelligence and behavior. For example, Howell and Buro (2009) found that undergraduate women scored lower than men on incremental beliefs but no differences were found for entity beliefs or procrastination. In contrast, Rhodewalt (1994) found that females believed effort had greater impact on ability than did males (supporting an incremental theory). Stipek and Gralinski (1996) found that, on average, third through sixth grade boys scored higher than girls on an entity scale, and Ommundsen (2001) found that ninth grade boys were more likely to have entity theory than girls. However, Ommundsen (2001) also found that girls reported more self-handicapping behavior than boys. Given these inconsistencies in previous work, no specific hypotheses regarding gender are proposed.

3. Results 3.1. Preliminary analyses Given that a number of participants were excluded from the current analyses due to completing few of the daily reports, we first explored the extent to which participants included in our sample varied from those who were excluded for not completing at least six of the school-day daily report forms. There was no difference in theory of intelligence between those who were included and those who were excluded, t(204) = .47, n.s. There was also no gender difference between the current sample and the larger sample, χ2 (1) = .64, n.s., and no difference in GPA, t(198) = .30, n.s. We also explored whether participants' theory of intelligence differed based on whether they completed the survey online or on paper, by mail. Participants who completed the surveys on paper were more likely to hold an entity theory of intelligence (M = 3.02, SD = .92) than those who completed them online (M = 2.62, SD = 1.04), t(140) = − 2.33, p = .021. Given this difference, method of completion was explored as a control variable in the primary analyses described below. Final preliminary analyses were conducted to explore gender differences in mean levels of theories of intelligence. Girls' theories were found to be more entity focused (M = 2.95, SD = .94) than were those of boys (M = 2.55, SD = 1.07), t(140) = 2.35, p < .05. Descriptive statistics for all variables are presented in Table 1. As shown, participants responded using nearly or all of the full range of the measures. Bivariate correlations between all variables are presented in Table 2. Table 1. Descriptive statistics for all variables. M SD Min Max Theory of intelligence 2.78 1.01 1.00 5.38 Global self-handicapping 2.18 .95 1.00 5.00 Global procrastination 2.35 .61 1.19 3.94 GPA 3.13 .82 .64 4.00 Have a lot of HW/studying⁎ 0.44 .28 .00 1.00 Difficulty of HW/studying⁎ 2.49 .78 1.00 4.71 Study time⁎ .90 .73 .00 5.79 Time on non-school activities and 1.25 .76 .11 4.49 Effort on schoolwork⁎ 2.87 .82 1.00 5.00 Feel like a good student⁎ 4.69 1.24 1.00 7.00 Positive school feelings⁎ 3.95 1.36 1.00 7.00 ⁎ These variables were averaged across the daily surveys to create mean, standard deviation, minimum, and maximum values. + Summed across the non-school activities (i.e., with friends outside of school, participating in extracurricular activities outside of school, and watching television or movies). Table options Table 2. Bivariate correlations. 1 2 3 4 5 6 7 8 9 10 11 1. Theory – 2. Self-hand .16+ – 3. Procrast .24⁎⁎ .59⁎⁎⁎ – 4. GPA − .13 − .20⁎ − .08 – 5. Lot of HW⁎ − .17⁎ − .03 .01 .30⁎⁎⁎ – 6. Hard HW⁎ − .12 − .05 − .19⁎ .07 .40⁎⁎⁎ – 7. Study time⁎ − .20⁎ − .24⁎⁎ − .28⁎ .39⁎⁎⁎ .43⁎⁎⁎ .32⁎⁎⁎ – 8. Non-school time# .05 .00 .13 .13 − .13 − .18⁎ − .03 – 9. Effort⁎ − .23⁎⁎ − .16+ − .40⁎⁎⁎ .17⁎ .22⁎⁎ .55⁎⁎⁎ .49⁎⁎⁎ − .10 – 10. Good stud⁎ − .24⁎⁎ − .17⁎ − .34⁎⁎⁎ .18⁎ .08 .27⁎⁎⁎ .26⁎⁎ .12 .56⁎⁎⁎ – 11. Pos sch feelings⁎ − .28⁎⁎⁎ − .16+ − .40⁎⁎⁎ .14+ − .02 .24⁎⁎ .24⁎⁎ .04 .48⁎⁎⁎ .64⁎⁎⁎ – Note.+p < .10; ⁎p < .05; ⁎⁎p < .01; ⁎⁎⁎p < .001. ⁎ These variables were averaged across the daily surveys to create mean, standard deviation, minimum, and maximum values. # Summed across the non-school activities (i.e., with friends outside of school, participating in extracurricular activities outside of school, and watching television or movies). Table options 3.2. Associations between theory of intelligence and global measures of self-handicapping, procrastination, and achievement Our first goal was to replicate previous research demonstrating an association between one's theory of intelligence and global measures of self-handicapping and procrastination, as well as GPA. We analyzed a series of regression equations in which each outcome variable was predicted by theory of intelligence (a single indicator with higher scores indicating an entity theory and lower scores indicating an incremental theory), gender, and an interaction between theory of intelligence and gender. As an additional control variable, method of completion of survey (i.e., online vs. paper) was also included in the models. The interaction term with gender was not significant in any of the models and was thus dropped from the final analyses, as was the variable indicating method of completion, given that it did not change any of the other results. As shown in Table 3, the strength of one's entity theory of intelligence was positively associated with global self-reports of self-handicapping and procrastination, but was not associated with GPA. Gender was not significantly associated with any of the three outcomes. Table 3. Associations between theory of intelligence and global self-handicapping, procrastination, and GPA. Self-handicapping Procrastination GPA B SE B β t B SE B β t B SE B β t Constant 2.28 .13 18.19⁎⁎⁎ 2.34 .08 29.19⁎⁎⁎ 3.01 .12 28.36⁎⁎⁎ Theory .17 .08 .18 2.12⁎ .14 .05 .23 2.75⁎⁎ − .12 .07 − .15 − 1.74 Gender − .17 .17 − .09 − 1.02 .02 .11 .02 .20 .20 .14 .12 1.39 Note. Gender was coded as males = 0 and females = 1. ⁎ p < .05. ⁎⁎ p < .01. ⁎⁎⁎ p < .001. Table options 3.3. Associations between theory of intelligence and average daily school experiences and feelings The second goal of the study was to examine whether theories of intelligence predicted averages of students' daily experiences. We tested this in a similar way to the global measures, but using Hierarchical Linear Modeling (HLM; Bryk & Raudenbush, 1992) instead of regression. This method is particularly well-suited to handling missing data, allowing us to make estimates which were weighted according to how many days of information each participant provided. The following equations were tested: equation(1) View the MathML sourceStudyVariableij=b0j+eij Turn MathJax on equation(2) b0j=c00+c01(Theory)+c02(Gender)+c03(Theory×Gender)+u0j.b0j=c00+c01Theory+c02Gender+c03Theory×Gender+u0j. Turn MathJax on Eq. (1) represents adolescents' reports of their daily school experiences and feelings, across the school days of the study. Eq. (2) shows how the average score was modeled as a function of theory of intelligence, gender, and their interaction. All of the predictors were grand mean centered so that the intercept could be interpreted as the mean of the sample. Together, these equations are conceptually similar to a regression model in which the daily reports of the outcome variables are averaged across the days of the study. The key difference is that HLM differentially weights the model based on the number of daily reports from which each participant's outcome variable is based. As with the global measures, the interaction term was not significant for any of the outcomes and was dropped from the final models. As shown in Table 4, the strength of one's entity theory of intelligence was negatively associated with positive feelings about school, feelings of being a good student, study time, effort, and experiences of challenge on homework, here defined as having a lot of homework/studying and the difficulty of homework/studying. Compared to boys, girls reported spending more time studying and had a higher percentage of days in which they reported having a lot of homework or studying to do. A supplementary analysis was conducted, in which method of completion was added as an additional level 2 variable. The key results did not change with this addition. Table 4. Hierarchical linear models predicting average daily school experiences and feelings. Pos school feelings b (SE) Feel like a good student b (SE) Study time b (SE) Effort on schoolwork b (SE) Difficulty of HW/studying b (SE) Have a lot of HW/studying b (SE) Level 1 Intercept 4.09 (.11)⁎⁎⁎ 4.98 (.10)⁎⁎⁎ .94 (.06)⁎⁎⁎ 3.02 (.07)⁎⁎⁎ 2.57 (.07)⁎⁎⁎ .44 (.02)⁎⁎⁎ Level 2 Theory − .40 (.11)⁎⁎⁎ − .29 (.10)⁎⁎ − .18 (.07)⁎⁎ − .22 (.07)⁎⁎⁎ − .12 (.06)⁎ − .06 (.02)⁎⁎ Gender .07 (.23) − .16 (.20) .34 (.13)⁎ .09 (.14) .01 (.14) .10 (.05)⁎ Note. Gender was coded as males = 0 and females = 1. Both theory of intelligence and gender were grand mean centered. ⁎ p < .05. ⁎⁎ p < .01. ⁎⁎⁎ p < .001. Table options To verify that the associations with theory of intelligence were not caused by differences in GPA, cumulative GPA was added to the models as additional predictor. While GPA was significantly associated with study time (b = .30, p < .001) and having a lot of homework/studying (b = .08, p < .01), the addition of GPA did not change any of the findings with regard to theory of intelligence described above. 3.4. Differences according to theory of intelligence in daily associations between school challenges and school feelings, effort, and time use The final goal of the study was to examine the extent to which theory of intelligence was associated with differences in school feelings, effort, and time use in response to school challenges on a daily basis. The two daily challenges that were examined were adolescents' reports of the difficulty of their homework or studying and whether they had a lot of homework or studying. Thus, these two variables were used as independent variables in these analyses, rather than dependent variables as in the analyses presented in Table 4. We examined the association between each of these predictors and positive feelings toward school, feelings of being a good student, effort on homework, and time spent studying. We also examined time use in a variety of domains outside of school (e.g., with friends, participating in extracurricular activities, and watching TV or movies), as these are activities students may engage in if they are avoiding studying. The general form of the model used for these analyses was as follows: equation(3) View the MathML sourceFeelings/Effort/TimeUseij=b0j+b1jSchoolChallenge+eij Turn MathJax on equation(4) b0j=c00+c01(Theory)+c02(Gender)+c03(Theory×Gender)+u0jb0j=c00+c01Theory+c02Gender+c03Theory×Gender+u0j Turn MathJax on equation(5) b1j=c10+c11(Theory)+c12(Gender)+c13(Theory×Gender)+u1j.b1j=c10+c11Theory+c12Gender+c13Theory×Gender+u1j. Turn MathJax on Eq. (3) shows how adolescents' feelings, effort, or time use on a particular day (i) for a particular individual (j) was modeled as a function of the individual's average feeling, effort, or time use (b0j) and school challenge (i.e., difficulty of homework or studying, or whether or not the participant had a lot of homework or studying). Given that the variable assessing whether or not the participant had a lot of homework had a meaningful 0 value, this variable was uncentered while difficulty of homework was group mean centered. As earlier, all level 2 variables were grand mean centered, such that level 1 association can be interpreted for the mean of the sample. Eq. (4) shows how the average feeling, effort, or time use was modeled as a function of theory of intelligence, gender, and their interaction. Finally, Eq. (5) shows how the association between school challenge and feeling, effort, or time use was modeled as a function of theory of intelligence, gender, and their interaction. Together, these equations allow us to explore the ways in which one's global level of theory of intelligence moderates the daily association between school challenge and each outcome variable. As shown in Table 5, at the mean level of homework difficulty, there was no association between theory of intelligence and any of the outcome variables. With increasing levels of homework difficulty, adolescents on average exerted more effort on their homework and spent more time studying. While this did not vary according to a main effect of either theory of intelligence or gender, it did vary according to the interaction between theory of intelligence and gender. An interaction between theory of intelligence and gender also moderated the association between homework difficulty and positive feelings toward school. Further analyses examined these associations separately by gender and found that theory of intelligence did not moderate the association between homework difficulty and any of the outcome variables for boys. However, for girls with a stronger entity theory of intelligence, homework difficulty was associated with reduced positive feelings toward school (b = − .13, p < .05), effort on homework (b = − .18, p < .01) and time studying (b = − .20 p < .001), compared to those with a more incremental theory of intelligence. As an example, the gender difference in the association between school difficulty and study time as a function of theory of intelligence is shown in Fig. 1. Neither theory of intelligence nor the interaction between theory and gender moderated the association between difficulty of homework or studying and any of the non-school time use domains (time with friends: b = − .05, p = .53; time in extracurricular activities: b = − .08, p = .18; time watching TV or movies: b = .07, p = .20). Table 5. Hierarchical linear models predicting daily associations between reports of difficulty of studying or homework and effort and time use. Pos school feelings b (SE) Feel like a good student b (SE) Effort on schoolwork b (SE) Study time b (SE) Level 1 Intercept 4.13 (.11)⁎⁎⁎ 5.00 (.10)⁎⁎⁎ 3.06 (.07)⁎⁎⁎ .98 (.06)⁎⁎⁎ Level 2 Theory − .24 (.16) − .23 (.14) − .15 (.08) − .07 (.08) Gender .05 (.23) − .14 (.20) .11 (.14) .34 (.13)⁎⁎ Theory × Gender − .29 (.22) − .12 (.20) − .12 (.13) − .21 (.13) Level 1 Difficulty of HW/studying − .01 (.03) .01 (.03) .53 (.04)⁎⁎⁎ .39 (.04)⁎⁎⁎ Level 2 Theory − .02 (.05) .05 (.04) .03 (.06) − .01 (.04) Gender .03 (.07) − .02 (.07) − .01 (.09) .08 (.08) Theory × Gender − .15 (.07)⁎ − .09 (.06) − .20 (.08)⁎ − .19 (.07)⁎⁎ Note. Gender was coded as males = 0 and females = 1. Difficulty of studying or homework was group mean centered. All level 2 variables were grand mean centered. ⁎ p < .05. ⁎⁎ p < .01. ⁎⁎⁎ p < .001. Table options Gender difference in the association between school difficulty and study time as ... Fig. 1. Gender difference in the association between school difficulty and study time as a function of theory of intelligence. Girls are shown in the top panel and boys in the bottom. Figure options For the analyses examining the daily challenge of having a lot of homework, the interaction between theory of intelligence and gender was not significant for any of the outcome variables and was dropped from the final models. As shown in Table 6, an entity theory of intelligence was associated with less positive feelings about school and of feeling like a good student. Neither of these varied according to whether or not a participant reported having a lot of homework or studying. However, on days in which adolescents reported having a lot of homework, they reported exerting more effort on their homework and spent more time studying on average. This was moderated by theory of intelligence such that adolescents with an incremental theory (i.e., low entity theory) increased their effort and study time the most, while those with an entity theory increased their effort and study time less. This was consistent with the behavior of girls in response to difficulty of homework described above. All of the associations between having a lot of homework or studying and time spent on non-school activities were non-significant (time with friends: b = .15, p = .47; time in extracurricular activities: b = .20, p = .52; time watching TV or movies: b = .04, p = .76). Table 6. Hierarchical linear models predicting daily associations between reports of having a lot of homework or studying and effort and time use. Pos school feelings b (SE) Feel like a good student b (SE) Effort on schoolwork b (SE) Study time b (SE) Level 1 Intercept 4.11 (.11)⁎⁎⁎ 5.00 (.10)⁎⁎⁎ 2.82 (.07)⁎⁎⁎ .63 (.05)⁎⁎⁎ Level 2 Theory − .37 (.11)⁎⁎⁎ − .25 (.10)⁎ − .12 (.07) − .04 (.06) Gender .18 (.23) − .09 (.20) .15 (.15) .21 (.11) Level 1 Have a lot of HW/studying − .00 (.06) − .05 (.08) .43 (.08)⁎⁎⁎ .66 (.08)⁎⁎⁎ Level 2 Theory − .05 (.08) − .09 (.08) − .16 (.08)⁎ − .21 (.08)⁎ Gender − .25 (.14) − .16 (.17) − .25 (.18) .11 (.16) Note. Gender was coded as males = 0 and females = 1. Not having a lot of homework or studying was coded as 0 and having a lot of homework or studying was coded as 1. This variable was uncentered. All level 2 variables were grand mean centered. ⁎p < .05. ⁎⁎p < .01. ⁎⁎⁎p < .001. Table options As before, adding a level 2 variable indicating method of completion did not change any of the key results for either of these models. Similarly, cumulative GPA was added as an additional predictor to the models. Again, GPA did not change any of the findings with regard to theory of intelligence. In only one case, GPA itself was also a significant moderator such that adolescents with high GPAs responded with more study time to homework difficulty than did those with lower GPAs, b = .11, p < .01.