Self-control Running Head: SELF-CONTROL AND ACADEMIC ACHIEVEMENT

What No Child Left Behind Leaves Behind: A Comparison of the Predictive Validity of Self-Control and IQ for Standardized Test Scores and Report Card Grades Angela L. Duckwortha Patrick D. Quinna Samantha Goldmana Positive Psychology Center University of Pennsylvania 3701 Market St. Suite 200 Philadelphia, PA 19104 Phone: 215 898-1339 Fax: 215 573-2188 Primary contact email: [email protected]

a

Positive Psychology Center, University of Pennsylvania

1

Self-control

2

Abstract This three-year prospective, longitudinal study compared the predictive validity of self-control and IQ for academic outcomes in a sample of adolescents attending an urban high school. Selfcontrol predicted persistence at this high school through the end of the eleventh grade. Among students who did not drop out, more self-controlled ninth graders earned higher GPAs and scored marginally higher on standardized achievement tests at the end of eleventh grade when controlling for intelligence. In contrast, smarter students earned higher standardized test scores but not higher GPAs when controlling for self-control. The differential weighting of intelligence relative to the capacity to work hard helps explain why girls in our sample brought home better report cards but did not earn significantly higher achievement test scores. These findings suggest that public policies that rely exclusively on standardized tests as measures of student achievement differentially weight intelligence over self-control.

Keywords: Impulsivity, Self-Control, Achievement, Success, Personality

Self-control

3

1. Introduction No Child Left Behind legislation relies upon standardized achievement test scores as a gauge of student achievement. What are the unique strengths and limitations of this mode of assessment? Which capacities might inadvertently be overlooked by the exclusive reliance upon this metric for student success? The obvious advantage of standardized tests is that they enable comparison of students across classrooms and schools—they permit an apples-to-apples comparison that is impossible given the idiosyncratic grading standards of individual teachers and schools. However, there are limits to what can reliably be assessed by a multiple-choice test lasting only a few hours and designed by someone other than a student’s own teacher (Wentzel, 1991). One important difference between standardized tests and report card grades concerns novelty in both content and format. While achievement tests are intended to align with state standards, the reality is that all diverge to at least some degree from what teachers actually cover in the classroom. In contrast, teachers tailor their own assessments to the material they have recently presented (e.g., after a unit on mixed fractions, students are tested on mixed fractions). Moreover, the multiple-choice format of most standardized tests differs from the test formats used by most teachers during the school year. Thus, it is not surprising that intelligence, which includes the capacity to solve novel problems, is the best-documented individual difference predictor of standardized achievement test scores (Barton, Dielman, & Cattell, 1972; Brody, 1997; Gagné & St. Père, 2002; Gottfredson, 2004; Herrnstein & Murray, 1994; Neisser et al., 1996). Correlations between IQ and standardized achievement test scores are typically interpreted as evidence that superior cognitive ability leads to higher levels of accumulated skills and knowledge. No doubt, this causal story is true. Nevertheless, associations between IQ and

Self-control

4

achievement tests may also be spuriously inflated by, among other factors, the fact that both kinds of assessment tax the ability to solve unfamiliar problems. A second crucial difference between these two forms of assessment concerns the time period and context in which student performance is observed. Report card grades typically reflect performance on dozens of quizzes and tests, homework assignments, classroom projects, and participation in classroom discussion. To earn high report card grades, students must accomplish these tasks on a daily basis over the course of the academic year. All of these challenges require sustaining effort despite distraction, boredom, frustration, and the lure of more enjoyable but less productive alternatives to studying. Thus, report card grades reflect not only what children know and can do in the spring of an academic year, but also how consistently they were able to meet their responsibilities throughout the course of that year. Should we care about the competencies assessed by report card grades that are not well measured by standardized tests? Wechsler (1940), creator of two of the most widely used modern IQ measures, argued strenuously for the empirical study of factors such as “drive, energy, impulsiveness, etc.” that, in combination with IQ, produced adaptive behavior: “What is necessary…is the devising of test situations in which the nonintellective factors can be identified, measured, and weighed” (pp. 444-5, emphasis added). Wechsler’s mention of impulsiveness— the obverse of self-control—as worthy of further study is notable. Self-control, defined as the ability to regulate thoughts, behaviors, emotions, and attention in the service of valued goals, is arguably among the most important non-IQ capacities for success in life (Baumeister, Vohs, & Tice, 2007). Several prior studies have documented a relationship between self-control and academic competence (e.g., Duckworth & Seligman, 2005; Mischel, Shoda, & Peake, 1988; Tangney, Baumeister, & Boone, 2004; Wolfe & Johnson, 1995), and self-control is a facet of Big

Self-control

5

Five Conscientiousness (Costa & McCrae, 1992), which predicts college GPA better than does any other Big Five factor including Openness to Experience (Noftle & Robins, 2007). However, very little attention has been devoted to the differential emphasis on IQ vs. self-control in various measures of student performance. 1.1 The present study In the current three-year, longitudinal investigation, we compare the predictive validity of self-control and IQ for report card grades and standardized test scores in an urban public high school. To maximize predictive validity of self-control, we used a multi-method, multi-source approach to its measurement. Similarly, to maximize predictive validity of IQ, we individually administered a widely-used IQ test. We tested the following hypotheses: (a) Self-control measured in ninth grade predicts retention through eleventh grade; (b) among students who remain in the high school, self-control is a superior predictor than IQ of GPA at the end of eleventh grade; (c) IQ measured in eleventh grade is a superior predictor of year-end standardized achievement test scores than is self-control; and finally, (d) girls earn higher GPAs than boys, but this gender difference is less dramatic for standardized achievement tests and nonexistent for IQ tests. 2. Method 2.1 Participants Participants were 93% (N = 115) of the 124 ninth-grade students at an urban public high school in the Northeast; 56% of participants were female. About 75% of adolescents attending this school came from low-income families, as indicated by eligibility for free or reduced-priced lunch. The ethnic-racial background of students was as follows: 90% Black, 3% Hispanic, 3%

Self-control

6

Asian, and 3% other. In September 2004, when the self-control and study habit measures were administered, the mean age of the participants was 14.5 years. 2.2 Procedure Child assent and parent consent forms were mailed home and collected in early September 2004, as participants started ninth grade. In late September 2004, we collected questionnaire data from students, parents, and teachers, and administered a delay of gratification task. In January 2007, participants completed an individually administered IQ test. We recorded eleventh-grade report card grades and standardized test scores in the summer of 2007, three years after we had collected self-control data. 2.3 Measures 2.3.1 Self-control. Participants completed the Eysenck I.6 Junior Impulsiveness Subscale (EJI; Eysenck, Easting, & Pearson, 1984), a widely-used questionnaire consisting of 23 self-report yes-no items written specifically for youth (e.g., “Do you sometimes break rules on the spur of the moment?”). The observed internal reliability of the EJI was α = .76. Participants also completed the Brief Self-Control Scale (BSCS; Tangney, Baumeister, & Boone, 2004), in which individuals endorse 13 items on a 5-point scale from 5 = very much like me to 1 = not at all like me (e.g., “I have a hard time breaking bad habits.”) The observed internal reliability of the BSCS was α = .82. Parents and teachers completed informant versions of the BSCS. To minimize common source variance between self-control ratings and report card grades, ninth-grade homeroom advisors, rather than academic subject teachers, completed questionnaires. In their capacity as advisors, homeroom teachers met with students each morning and, in addition, twice weekly for

Self-control

7

an hour each of group discussion and consultation. The observed internal reliabilities of the parent and teacher versions of the BSCS were α = .76 and .91, respectively. In addition, we administered a behavioral delay of gratification task called the Choice Delay Task. Participants were given $1 in thanks for their participation in our study and told that they could either keep the money or return it for $2 one week later. We coded the choice as a dichotomous variable where 0 = $1 immediately and 1 = $2 in one week. Seven-month test-retest reliability for this task has been reported as r = .41 (Duckworth & Seligman, 2005). Intercorrelations among single measures of self-control averaged r = .21. See Table 1. To increase validity and reduce multicollinearity, we standardized and averaged measures to create a composite self-control score for each participant. Following Nunnally’s (1978) procedure for calculating the internal reliability of a linear combination of standardized scores, we found the internal reliability of the resulting composite score to be .81. To facilitate interpretation of certain analyses, we standardized scores to a mean of 0 and standard deviation of 1. 2.3.2 IQ. Trained clinical psychology graduate students administered the two-subtest form of the Wechsler Abbreviated Scale of Intelligence in the spring of 2007 as a measure of IQ (WASI; Wechsler, 1999). This individually administered test of intelligence has been shown to converge with the Wechsler Intelligence Scale for Children-Third Edition and other measures of IQ (Wechsler, 1999). The WASI IQ score is normalized according to the child’s age in months, with a mean of 100 and a standard deviation of 15. Because of the time of administration, only a subset of students (n = 69) who were still attending the high school by eleventh grade completed the WASI. 2.3.3 Academic performance.

Self-control

8

We collected from school records eleventh-grade GPA (calculated as the average of math, science, and humanities final grades for the 2006-2007 school year and converted to a ). We also collected scores on the Pennsylvania System of School Assessment (PSSA) standardized achievement test, which participants completed in the spring of 2007. Scores on the PSSA are used by the Pennsylvania Department of Education to assess yearly progress in accordance with No Child Left Behind legislation. Retention in school was coded as a dichotomous variable where 1 = retained and 0 = dropped out as of the end of eleventh grade. 2.3.4 Additional measures. In the same battery as the self-control questionnaires, participants identified their gender and reported the amount of time they typically spent on homework per day. See Table 2. 3. Results Of the 115 ninth-grade students in our initial sample, 35 had left the school by the end of eleventh grade. As hypothesized, in a binary logistic regression predicting retention, students a standard deviation higher in self-control than their peers were 65% more likely to continue in the high school, OR = 1.65, p = .01. Among students remaining in the school, self-control predicted higher eleventh-grade GPAs (r = .28, p = .01) and higher standardized test scores in the spring of 2007, r = .25, p = .03. More self-controlled students reported devoting more time to homework, r = .26, p = .008. IQ was an even more robust predictor of eleventh-grade standardized test scores (r = .59, p < .001) but was unrelated to either eleventh-grade GPAs (r = .003, p = .98) or the amount of time students spent doing homework, r = -.04, p = .73. See Figures 1 and 2. Self-control and IQ were not related, r = .04, p = .76. Thus, we were not surprised that among students of comparable intelligence, self-control continued to predict report card grades: In a simultaneous multiple regression predicting eleventh-grade GPA from self-control and IQ,

Self-control

9

self-control was a significant predictor (β = .27, p = .03), but IQ was not, β = .001, p = .99. Similarly, among students of comparable self-control, IQ was a strong predictor of achievement test scores: In a simultaneous multiple regression predicting eleventh-grade standardized test scores, IQ was a significant predictor (β = .59, p < .001), and self-control was marginally so, β = .19, p = .058. Compared to boys, girls were more self-controlled, earned higher report card grades, and spent more time on their homework; the effect sizes for these gender differences were mediumto-large. However, the female advantage on achievement test scores was half as large and not statistically significant, and boys enjoyed a very slight, statistically insignificant advantage in IQ over girls. See Table 3. Following Baron and Kenny (1986), we completed mediation analyses to test whether girls earned higher report card grades at least in part because they did more homework. Homework time predicted GPA (β = .30, p = .01), and gender also predicted GPA (β = .24, p = .03). In a simultaneous multiple regression predicting GPA, homework time was a significant predictor (β = .26, p =.03) but gender was not (β = .19, p = .11). The decrement in the coefficient for gender was marginally significant, t = 1.75, p = .08. 4. Discussion Compared to their peers, more self-controlled adolescents were less likely to drop out of high school, earned higher report card grades at the end of eleventh grade, earned higher standardized achievement test scores, and spent more time on their homework. IQ was also a robust predictor of standardized achievement test scores yet was unrelated to report card grades and homework time. Because self-control and IQ were not strongly associated, relationships with GPA and achievement test scores were not dramatically different when both self-control and IQ were considered as predictors. The differential importance of self-control and, in particular,

Self-control

10

effort on homework, to report card grades helps account for why girls in our sample earned higher GPAs but not higher achievement test scores. Which is the more valid indicator of student achievement—report card grades or standardized test scores? Certainly, both metrics depend on the skills and knowledge a student has acquired. But apart from this shared variance, there is specific variance for each mode of assessment. Report card grades and standardized test scores (e.g., the SAT) are typically correlated yet provide unique predictive validity of success in college (Willingham, 1985). This fact suggests that the more interesting question may not be which form of assessment should be taken seriously, but rather which capacities are differentially tapped by each? Report card grades reflect performance on dozens of assignments, quizzes, tests, and projects over several months. Completing all of these tasks requires self-control, including paying attention in class rather than daydreaming, listening quietly to the teacher rather than talking to a classmate, sitting down to do homework rather than playing videogames, going to bed early rather than staying up to watch television, and studying rather than chatting online. Importantly, there is no test proctor in most of these circumstances encouraging maximal performance and ensuring a quiet room with minimal distraction. Many adolescents are neither happy nor intrinsically motivated while they study; rather, most study in order to achieve longerterm goals including getting good marks (Wong & Csikszentmihalyi, 1991). Thus, outside the context of controlled testing conditions, the temptation to do more enjoyable activities is considerable. Unfortunately, report card grades are difficult to compare across teachers or schools. In contrast, standardized tests are not subject to the idiosyncratic grading practices of any particular teacher. This feature favors achievement tests for the purposes of assessing teacher or school

Self-control

11

effectiveness. However, the current study suggests that more intelligent students may do better on achievement tests than less intelligent students of equivalent knowledge and skill level. Tests designed to serve all students in a given state or nationally are intended to tap the knowledge and skills that students have been working on in the classroom, but in practice they present at least some material which students have not been taught. Even when the content covered by standardized tests overlaps with what students have been taught, the format in which the content is presented (e.g., multiple-choice questions) can differ dramatically from what are accustomed to (e.g., open-ended math problems, short answer questions). To the extent that standardized tests contain novel rather than familiar content, or present familiar material in novel formats, standardized tests such as the SAT tap IQ as well as achievement (see Frey & Detterman, 2004). Finally, because both IQ tests and standardized achievement tests last only a few hours and are conducted under conditions designed to maximize motivation and minimize distraction, they do not depend as much on the ability to effortfully control attention, emotion, and behavior. In 1916, Binet and Simon, authors of the first intelligence tests, observed that such conditions diverge from the circumstances of everyday life: It seems to us that the scholastic aptitude admits of other things than intelligence; to succeed in his studies, one must have qualities which depend especially on attention, will, and character; for example a certain docility, a regularity of habits, and especially continuity of effort. A child even if intelligent, will learn little in class if he never listens, if he spends his time in playing tricks, in giggling, in playing truant. The lack of attention, of character, of will, do not appear or scarcely so, in our tests of intelligence, the test is too short; the pupil is not left to himself sufficiently…It is not under such conditions that one can measure the ordinary power of attention of a child; it is when he is left to himself. (p. 254) We concur with the recommendation that Binet and Simon offer for educators: “And now as a pedagogical conclusion, let us say that what they should learn first is not the subjects ordinarily taught, however important they may be; they should be given lessons of will, of attention, of

Self-control

12

discipline; before exercises in grammar, they need to be exercises in mental orthopedy; in a word they must learn how to learn” (p. 257). The current study has several limitations. First, the unattenuated population correlation for IQ and GPA was r = .005, a much weaker association than usually reported in studies of IQ and report card grades. For instance, McDermott, Mordell, and Stoltzfuz (2001) find in a nationally representative sample that nonverbal IQ correlates with teacher-assigned math grades at r = .30 and that verbal IQ correlates with teacher-assigned reading grades at r = .23. This raises several concerns. It is possible that grades in this school do not accurately reflect student learning. Against this is the observed correlation between GPA and spring standardized test scores of r = .52, p < .001. A second possibility is that IQ scores were invalid because, for instance, students did not give their best performance on the test. However, as noted previously, IQ scores correlated robustly with standardized test scores. Finally, it is possible that there was restriction of range in GPA. If this were the case, then the observed correlation between GPA and self-control would have been correspondingly attenuated and, thus, this possibility does not endanger conclusions about the relative effects of self-control and IQ on achievement. A more serious concern is that ratings of self-control by parents, teachers, and the students themselves were inflated by halo. Straight-A students might be thought of, by others and even themselves, as more self-controlled than they really are. Examining individual items in the Brief Self-Control Scale and Eysenck Impulsiveness Subscale, we found none that specifically mentioned grades, homework, or preparation for class. Rather, items tapped trait-level selfcontrol: breaking bad habits, resisting temptation, thinking carefully before acting, and so on. Also, common source variance between self-control ratings by teachers and report card grades

Self-control

13

was minimized by obtaining teacher ratings of self-control from advisors who did not give students grades in academic classes. For policymakers, our findings suggest more serious consideration of the strengths and limitations of standardized tests as measures of student achievement. In the nationally representative sample studied by McDermott and colleagues mentioned above, the correlations between nonverbal and verbal IQ and student achievement measured by group-administered standardized achievement tests were twice as large as when measured by teacher-assigned grades. Consistent with the current findings, McDermott found that measures of disciplined behavior and motivation were robust predictors of teacher assigned grades but not of standardized achievement test scores. Segal (2006) has shown that eighth grade classroom behavior is as robust a predictor as achievement test scores of income in adulthood. To the extent that report card grades provide information about a child’s self-control, as manifest in classroom behavior and performance on innumerable academic tasks throughout the school year, No Child Left Behind policies may be leaving a great deal behind in its evaluation of student competence.

Self-control

14

Acknowledgements Some of the findings in this paper were previously reported in a student paper by the second author, Samantha Goldman. Her work was included in res: a journal of undergraduate research and writing, sponsored by the University of Pennsylvania. We thank Christopher Peterson for his helpful comments. This research was supported by a grant from the John Templeton Foundation, a National Science Foundation Graduate Fellowship, National Institute of Mental Health Grants MH-52270 and MH-63430, and Department of Education Grant R215S020045.

Self-control

15

References Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173-1182. Barton, K., Dielman, T.E., & Cattel, R. B. (1972). Personality and IQ measures as predictors of school achievement. Journal of Educational Psychology, 63, 398-404. Baumeister, R. F., Vohs, K. D., & Tice, D. M. (2007). The strength model of self-control. Current Directions in Psychological Science, 16, 351-355. Binet, A., Simon, T., & Kite, E. S. (1916). The development of intelligence in children (The Binet-Simon Scale). Baltimore, MD: Williams & Wilkins Co. Brody, N. (1997). Intelligence, schooling, and society. American Psychologist. Special Issue: Intelligence & Lifelong Learning, 52, 1046-1050. Costa, P. T. & McCrae, R. R. (1992). Revised NEO Personality Inventory (NEO-PI-R) and NEO Five-Factor Inventory (NEO-FFI) professional manual. Odessa, FL: Psychological Assessment Resources. Data Recognition Corporation. (2007). Technical report for the Pennsylvania System of School Assessment: 2007 reading and mathematics. Maple Grove, MN: Author. Duckworth, A. L. & Seligman, M. E. P. (2005). Self-control outdoes IQ predicting academic performance in adolescents. Psychological Science, 16, 939-944. Duckworth, A. L. & Seligman, M. E. P. (2006). Self-control gives girls the edge: Gender in selfcontrol, grades, and achievement test scores. Journal of Educational Psychology, 98, 198-208.

Self-control

16

Eysenck, S. B., Easting, G., & Pearson, P. R. (1984). Age norms for impulsiveness, venturesomeness and empathy in children. Personality & Individual Differences, 5, 315321. Frey, M. C., & Detterman, D. K. (2004). Scholastic assessment or g? The relationship between the scholastic assessment test and general cognitive ability. Psychological Science, 15, 373-378 Gagné, F., & St Père, F. (2002). When IQ is controlled, does motivation still predict achievement? Intelligence, 30, 71-100. Gottfredson, L. S. (2004). Schools and the g factor. The Wilson Quarterly, Summer, 35-45. Herrnstein, R. J., & Murray, C. A. (1994). The bell curve: Intelligence and class structure in American life. New York: Free Press. Hull, C. L. (1928). Aptitude testing. Oxford: World Book. McDermott, P. A., Mordell, M., & Stoltzfus, J. C. (2001). The organization of student performance in American schools: Discipline, motivation, verbal learning, nonverbal learning. Journal of Educational Psychology, 93, 65-76. Mischel, W., Shoda, Y., & Peake, P. K. (1988). The nature of adolescent competencies predicted by preschool delay of gratification. Journal of Personality & Social Psychology, 54, 687696. Neisser, U., Boodoo, G., Bouchard, T. J. J., Boykin, A. W., Brody, N., & Ceci, S. J. et al. (1996). Intelligence: Knowns and unknowns. American Psychologist, 51, 77-101. Noftle, E. E., & Robins, R. W. (2007). Personality predictors of academic outcomes: Big Five correlates of GPA and SAT scores. Journal of Personality and Social Psychology, 93, 116-130.

Self-control

17

Nunnally, J. C. (1978). Psychometric theory (2nd Ed.) New York: McGraw-Hill. Segal, C. (2006). Classroom behavior. Unpublished manuscript. Stanford University. Tangney, J. P., Baumeister, R. F., & Boone, A. L. (2004). High self-control predicts good adjustment, less pathology, better grades, and interpersonal success. Journal of Personality, 72, 271-322. Wechsler, D. (1940). Nonintellective factors in general intelligence. Psychological Bulletin, 37, 444-445. Wechsler, D. (1943). Non-intellective factors in general intelligence. Journal of Abnormal & Social Psychology, 38, 101-103. Wechsler, D. (1999). Wechsler Abbreviated Scale of Intelligence. San Antonio, TX: The Psychological Corporation. Wentzel, K. R. (1991). Classroom competence may require more than intellectual ability: Reply to Jussim (1991). Journal of Educational Psychology, 83, 156-158. Willingham, W. W. (1985). Success in college: The role of personal qualities and academic Ability. New York: College Entrance Examination Board. Wolfe, R. N., & Johnson, S. D. (1995). Personality as a predictor of college performance. Educational & Psychological Measurement, 55, 177-185. Wong, M. M.-h., & Csikszentmihalyi, M. (1991). Motivation and academic achievement: The effects of personality traits and the duality of experience. Journal of Personality, 59, 539574.

Self-control

18

Figure Captions Figure 1. Eleventh-grade GPA as a function of ranked quintiles of self-control and IQ. Figure 2. Eleventh-grade standardized test score as a function of ranked quintiles of self-control and IQ. Figure 3. Model of homework hours as a mediator between gender and eleventh-grade GPA. Values are standardized regression weights. Parenthetical value is the standardized weight when homework hours are included as a predictor in the regression model. * p < .05.

Self-control Table 1 Intercorrelations among Self-Control Measures Measure 1 1. Brief Self-Control Scale, self report 2. Eysenck Junior Impulsivity Subscale 3. Brief Self-Control Scale, parent report 4. Brief Self-Control Scale, teacher report 5. Delay Choice Task

2 .62*** -

3 .30* .26* -

4 .07 -.02 .30* -

5 .24* .24* .10 .01 -

Note. The Eysenck Junior Impulsivity Subscale was recoded such that higher scores indicate higher self-control. * p < .05. *** p < .001.

19

Self-control

20

Table 2 Summary Statistics for Self-Control, IQ, and Academic Performance Variables (N = 115) Observed Internal Measure Possible Range Range Consistency Brief Self-Control Scale, self report 1-5 1.69-4.69 .74 Eysenck I.6 Junior Impulsiveness Subscale 0-23 2.09-22.00 .76 Brief Self-Control Scale, parent report 1-5 1.77-5.00 .89 Brief Self-Control Scale, teacher report 1-5 1.69-5.00 .95 Delay Choice Task 0-1 0-1 IQ 55-157 65-116 Ninth-grade homework hours 0.02-4.50 0.02-4.50 Eleventh-grade GPA 0-4 1.6-4.0 Eleventh-grade PSSA scorea 825.50-1638.50 a

Mean

SD

3.36 12.85 3.74 3.99 0.84 94.70 1.41 2.79 1236.98

0.63 4.43 0.75 0.96 0.37 10.11 0.86 0.57 170.72

PSSA standardized test scores are normalized with a mean of 1300 and a standard deviation of 100 (Data Recognition Corporation, 2007).

Self-control

21

Table 3 Gender Differences in Self-Control, IQ, and Academic Performance Girls Boys Measure Mean (SD) Mean (SD) Self-Control 0.29(0.83) -0.28(1.04) IQ 94.21(9.73) 95.29(10.70) Ninth-grade homework hours 94.34(54.00) 71.73(47.21) Eleventh-grade GPA 2.88(0.58) 2.60(0.56) Eleventh-grade PSSA score 1259.63(156.85) 1206.26(177.00) * p < .05. *** p < .001.

Cohen’s d .61*** -.11 .59* .49* .32

Self-control Figure 1 3

GPA

2.9 2.8

Self-Discipline IQ

2.7 2.6 2.5 1

2

3 Quintiles

4

5

22

Self-control Figure 2

Standardized Test Score

1400

1300 Self-Discipline IQ 1200

1100 1

2

3 Quintiles

4

5

23

Self-control Figure 3

Homework hours .22*

Female Gender

.30*

GPA .24* (.19)

24