College Board Scholastic Aptitude Test Research Articles

Psychological profiles of the mathematically talented: some sex differences and evidence supporting their biological basis.

For over 20 years, above-level testing with the College Board Scholastic Aptitude Test (SAT) has been used to assess the abilities of well over 1,000,000 highly able 12-13-year-olds (students in the top 3% in intellectual ability). In this population, the predictive validity of the mathematical part of the SAT, SAT-M, for academic and vocational criteria has been demonstrated over 10-year gaps. Here, we document aspects of the psychological and achievement profiles of these highly able students, paying particular attention to sex differences. Males score higher on SAT-M (i.e., mathematical reasoning ability) than females; this difference is accompanied by differences between the sexes in spatial-mechanical reasoning abilities and in a number of lifestyle and vocational preferences. Collectively, these attributes appear to play a key role in structuring male-female disparities in pursuing advanced educational credentials and careers in the physical sciences. After profiling a number of the behavioural characteristics of the highly able, we examine some underlying biological correlates of these phenotypic manifestations. These include hormonal influences, medical and bodily conditions and enhanced right hemispheric activation.

A HISTORICAL PERSPECTIVE ON THE CONTENT OF THE SAT®

ABSTRACTThis paper provides a historical perspective on the content of the SAT®. The review begins at the beginning, when the first College Board SAT (the “Scholastic Aptitude Test”) was administered to 8,040 students on June 23, 1926. At that time, the SAT consisted of nine subtests: definitions, arithmetical problems, classification, artificial language, antonyms, number series, analogies, logical inference, and paragraph reading. Over the years, the SAT has evolved in the way it measures what is now referred to as verbal and mathematical “reasoning.” With each redesign of the SAT, a variety of considerations were taken into account, including fairness issues, scaling issues, cost, public perception, face validity, changes in the test‐taking population, changes in patterns of test preparation, and changes in the college admissions process. This paper describes the reasons for the various changes while emphasizing that the value of SAT scores rests on the test's high technical quality, and on the assumption that scores would maintain their meaning over time.

Long-term effects of acceleration on the social-emotional adjustment of mathematically precocious youths.

The Study of Mathematically Precocious Youth (SMPY) identified over 2,000 12-14 year-olds who scored as well as a random sample of high school females on the College Board Scholastic Aptitude Test. SMPY encouraged these students to accelerate their education; over 50% did. Their social development at age 18 and at age 23 was then assessed. We investigated the effects of amount and type of educational acceleration (grade skipping and subject matter) on psychosocial indices (self-esteem, locus of control, self-acceptance/identity, and social interaction)

EFFECTS OF COACHING ON THE VALIDITY OF THE SAT: A SIMULATION STUDY

ABSTRACT The effects of student coaching in preparation for the College Board Scholastic Aptitude Test (SAT) on the predictive validity of this test for freshman year performance were studied using data on 1985 freshman year students from four colleges. After the validity of the SAT was estimated for each school, a given proportion of students was picked, and a given magnitude of score gains was added to the observed SAT verbal and mathematics scores. The validity of the new SAT scores was then estimated and compared to the observed validity measures. Each simulation provides an estimate of the change in velidity resulting from a hypothetical coaching effect. Sample sizes from the 4 colleges were: (1) 996 students from a highly selective school; (2) 1,346 from a moderately selective school; (3) 386 from a small, less selective school; and (4) 203 from a small religiously affiliated college. Results indicated that the validity indicators of the SAT in highly selective colleges are more contaminated by coaching effects than in less selective colleges. Results also indicate that changes in the levels of coaching do not seem large enough to account for the observed decline in the validity of the SAT over the past 10 years. Appendix A provides a detailed discussion of the second college, and Appendix B describes decrements in multiple partial correlation coefficients. (Contains 5 figures, 14 tables, and 19 references.) (SLD)

Enhanced right hemisphere involvement during cognitive processing may relate to intellectual precocity

In the present study, intellectually precocious and average ability youths performed a dichotic listening task (Experiment 1) and a free-vision chimeric face task (Experiment 2). Patterns of hemispheric lateralization and the relative involvement of the left and right hemispheres during cognitive processing were assessed. In Experiment 1, the average ability youths demonstrated a right ear/left hemisphere (re/LH) superiority for identification of CV syllables, while the gifted subjects failed to show any ear/hemisphere advantage. In Experiment 2, members of both groups tended to judge the leftside smile/rightside neutral half-faces as “happier”, a pattern indicative of enhanced right hemisphere (RH) arousal. Notably, the degree of RH involvement was significantly greater in the gifted as compared to average ability youths. Moreover, laterality scores of the precocious on the chimeric face task correlated with their performance on the College Board Scholastic Aptitude test (SAT), i.e. the greater the leftward bias, the higher the SAT score. These findings, taken in composite, suggest that a high level of RH involvement during cognitive processing may be related to intellectual precocity.

EQUATING THE SCORES OF THE PRUEBA DE APTITUD ACADÉMICA™ AND THE SCHOLASTIC APTITUDE TEST®

ABSTRACTThe present study is a replication, in certain important respects, of an earlier study conducted by Angoff and Modu (1973) to develop algorithms for converting scores expressed on the College Board Scholastic Aptitude Test (sat) scale to scores expressed on the College Board Prueba de Aptitud Académica (paa) scale, and vice versa. Because the purpose and the design of the studies, though not all of the psychometric procedures, were identical in the two studies, the language of this report often duplicates that of the earlier study. The differences in procedure, however, are worth noting, and it is hoped that this study will contribute in substance and method to the solution of this important problem.The study described in this report was undertaken in an effort to establish score equivalences between two College Board tests—the Scholastic Aptitude Test (sat) and its Spanish‐language equivalent, the Prueba de Aptitud Académica (paa). The method involved two phases: (1) the selection of test items equally appropriate and useful for English‐ and Spanish‐speaking students for use as an anchor test in equating the two tests; and (2) the equating analysis itself. The first phase called for choosing a set of items in each of the two languages, translating each item into the other language, “back‐translating” independently into the original language, and comparing the twice‐translated versions with their originals. This process led to the adjustment of the translations in several instances and, in other instances, to the elimination of some items considered too difficult to be translated adequately. At this point both sets of “equivalent” items, each in its original language mode, were administered as pretests, chiefly to determine whether the two response functions for each item were sufficiently similar for the items to be considered equivalent.On the basis of these analyses two sets of items—one verbal and the other mathematical—were selected for use as anchor items for equating. These were administered again (in the appropriate language) at regularly scheduled administrations of the sat and the paa. An item response theory (irt) model was used to equate the paa to the sat, with the anchor items serving as the link in the equating process.The equating itself showed definite curvilinear relationships in both verbal and mathematical tests, indicating in this instance that both sections of the paa are easier than the corresponding sat sections. The results also showed good agreement between the current conversions and the 1973 Angoff‐Modu conversions for the mathematical tests, but not so close agreement for the verbal tests. The reasons for the difference are (speculatively) attributed to improved methodology in the present study, especially for the more difficult verbal equating, and to the possibility of scale drift in one or the other test (or both tests) over the intervening 12 to 15 years since the last study.

The Predictive Validity of a Study Habits Inventory for First Semester Undergraduates

In an effort to provide additional information concerning the relationship between study habits and college grade-point average (GPA), the College Adjustment and Study Skills Inventory (CASSI), high school rank (HSR), and the Verbal as well as the Mathematics scores on the College Board Scholastic Aptitude Test (SAT-V and SAT-M) were intercorrelated. Three stepwise multiple regression analyses were computed, with GPA as the criterion measure. One multiple regression equation used as predictors six subscales of the CASSI; another employed HSR, SAT-V and SAT-M; and the third, used a combination of HSR and the Time Distribution scale of the CASSI. Also, differences between means of males and females on all measures were calculated. Results showed that all the academic variables and all but one of the CASSI variables were correlated significantly with GPA; females were significantly higher than males on HSR and GPA; and males were significantly higher than females on SAT-M. The combination of SAT-V, SAT-M, HSR, and Time Distribution yielded a coefficient of multiple determination (R2) that was 16% higher than that obtained from HSR alone.

SPECIFYING THE CHARACTERISTICS OF LINKING ITEMS USED FOR ITEM RESPONSE THEORY ITEM CALIBRATION1,2

ABSTRACTThe purpose of this study was to investigate the effects on IRT true‐score equating results of the characteristics of the linking items that are used to place parameter estimates for items contained in two test editions on the same scale prior to equating. The study was carried out using the three‐parameter logistic item response theory model and Monte Carlo procedures. So that the simulated data reflected actual test data, the true item parameters were taken from the estimated parameters obtained from LOGIST calibrations of item responses obtained from selected administrations of the verbal sections of the College Board Scholastic Aptitude Test (SAT‐V).The study was conducted in two phases, with the second phase based on results derived in the first phase. In the first phase, the effects of the characteristics and number of linking items were investigated using two of the common linking or scaling designs: concurrent calibration and the characteristic curve transformation method. The research extended previous research by systematically investigating the number of linking items needed to insure adequate equating results. In addition, the effects on equating of two different characteristics of the parameter estimates of the linking items were investigated: 1) items with parameter estimates having standard errors of estimation similar to those in typical SAT‐V common item sections, and 2) items with parameter estimates which have small standard errors of estimation. Finally, the effect on true‐score equating results of using peaked and uniform distributions of abilities for groups used to estimate the parameters for the items in the test editions to be equated was investigated.In the second phase of the study, the effects on equating results of common item sections that contained a few items whose item response functions were not well fit by the three‐parameter logistic model and of common item sections that contained a few items on which the two groups taking the common item section responded differently were systematically investigated. These effects were investigated using both the concurrent calibration and the characteristic curve transformation linking procedures for varying numbers of common items. The effects of these common item sections were studied using only, in most cases, a uniform ability distribution and characteristics of the parameter estimates selected as a result of the first phase of the study.The results of Phase I and II of the study indicate that, for data such as those simulated for this study, improved equating results are obtained if longer linking tests are used with a uniform distribution of examinee ability.The results of the study that investigated the quality of the linking item parameter estimates were confounded by the total test representation of the various linking tests. One possible conclusion from this aspect of the study might be that it is more important to evaluate the relative efficiency of the linking items, with respect to the total test, than it is to evaluate the size of the standard errors of estimation of the linking item parameter estimates.Finally, the study indicated that equating results, particularly those obtained using a characteristic curve transformation design, are affected by the presence of linking items that function differently for the two groups used to provide data for the equating and calibration. This is particularly true for shorter linking tests. It appears that the quality of an equating depends to some extent on prior screening of linking tests and removal of items that function differently for the two groups.

The Relationship of Laterality to Academic Aptitude

Self-report subscale scores from the Herrmann Brain Dominance Survey and Torrance and Reynolds' Your Style of Learning and Thinking were correlated with the College Board Scholastic Aptitude Test (SAT) Verbal (V), Mathematics (M), and Total (T) scores and with the grade point averages of 40 junior female occupational therapy students. Findings included (a) a strong relationship between the Herrmann right cerebral and total cerebral scores with the SAT V, M, and T scores, (b) modest relationships between the Herrmann limbic scores and the SAT M scores, (c) no meaningful relationships between the Torrance and Reynolds instrument scores and SAT scores, and (d) no consistent relationships between the measures of laterality and grade point average.

Student Perceptions as Predictors of Academic Performance in College Developmental Studies

The investigation examined (a) perceptions held by students of their preparation to do college-level work, (b) attitudes towards the Developmental Studies program and the university, and (c) knowledge of Developmental Studies as predictors of academic performance. In addition, the high school grade point average (GPA) and the College Board Scholastic Aptitude Test scores were employed as predictors in order to provide a standard by which to compare the predictive efficiency of the student-perception variables. The academic variables, in comparison with those associated with student perceptions, accounted for more variance in first-quarter GPA. Of the student-perception variables, those related to a student's estimation of preparation for college-level courses appeared to have the most direct relationships to first-quarter GPA.

SEX DIFFERENCES IN THE ACADEMIC PERFORMANCE OF SCHOLASTIC APTITUDE TEST TAKERS

ABSTRACTThe number of women college students has increased dramatically over the past fifteen years; in this same time period, the average College Board Scholastic Aptitude Test (SAT) scores for women have declined more than the scores for men. This study summarizes recent evidence concerning the academic performance of women and men by examining sex differences among (1) all SAT takers, (2) test takers grouped by anticipated major field of study, and (3) college freshman year courses and grades.Three major research questions guided this study: Are there consistent differences in the intellectual abilities of men and women? Do precollege admissions variables predict college performance with equal accuracy for women and men? Have the contents or structure of the SAT contributed to observed sex differences in performance on the test?Though the men and women who take the SAT differ in high school courses and grade point average, major field and career interests, and socioeconomic backgrounds, neither these differences nor research evidence on differential cognitive functioning were found sufficient to account for all of the observed sex differences in performance on the SAT. Some but not all of the male‐female differences in scores on the mathematics portion of the SAT were accounted for by differences in the depth and recency of course preparation. The observed‐first‐year college grades of women were slightly underpredicted by test scores and high school grades, but the magnitude of the underprediction was very small. While specific changes in the contents or structure of the SAT could not be associated with score changes, it seems likely that numerous small adjustments have played some part in recent shifts among the average test scores of women and men.Consistent with recent literature on sex differences in cognitive performance, this study concludes that sex‐related SAT differences are very small relative to the generally similar levels of performance by men and women, and that using both test scores and high school records to predict first‐year college grades continues to work reasonably well for both sexes.

Sex differences in mathematical reasoning ability: more facts.

Almost 40,000 selected seventh-grade students from the Middle Atlantic region of the United States took the College Board Scholastic Aptitude Test as part of the Johns Hopkins regional talent search in 1980, 1981, and 1982. A separate nationwide talent search was conducted in which any student under age 13 who was willing to take the test was eligible. The results obtained by both procedures establish that by age 13 a large sex difference in mathematical reasoning ability exists and that it is especially pronounced at the high end of the distribution: among students who scored greater than or equal to 700, boys outnumbered girls 13 to 1. Some hypothesized explanations of such differences were not supported by the data.

The Comparative Predictive Validities of High School Rank, the Ammons Quick Test, and Two Scholastic Aptitude Test Measures for a Sample of Black Female College Students

In the prediction of college grade-point averages (GPA) for minority college students, researchers often have reported relatively low predictive validity for both the Mathematics and Verbal sections of the College Board Scholastic Aptitude Test (SAT-M and SAT-V) in comparison with that obtained for high school class rank (HSR). This circumstance has suggested the need for alternative predictors to replace SAT-V and SAT-M as admissions indicators, particularly in the instance of minority college applicants. One such potential replacement was thought to be the Ammons Quick Test (AQT), which requires ony a short time to administer (10 to 12 minutes). By means of a multiple regression analysis with data from a sample of 63 specially admitted Black female undergraduates, the relationship between the optimally weighted composite of predictors (SAT-V, SAT-M, HSR and AQT) and the criterion (GPA) was evaluated. The HSR, SAT-V, SAT-M, and AQT afforded predictive validity coefficients of .52 (p < .001), .28 (p < .05), .22 (p > 05) and .19 (p > .05), respectively. In the stepwise procedure, there was an R2 of .36. HSR contributed .27 to the variance in the criterion measure, SAT-V added .08, AQT yielded only an additional .006, and SAT-M contributed virtually zero. The AQT did not appear to be a valid predictor. It can be anticipated that the large number of predictors relative to the smallness of the sample probably led to an inflated estimate of the contribution of each predictor variable to the variance in the criterion measure, although the relative importance of each predictor in all likelihood was approximated fairly accurately.

Bias in the Validity of Predicted College Grades in Four Ethnic Minority Groups

Measures of high school achievement and scholastic aptitude are often used to predict college freshman grades. Are these predictions biased for members of ethnic minorities? Majority freshmen in a large state university in one year were used to develop a regression equation for predicting first-semester grade point average from high school grade point average and College Board Scholastic Aptitude Test scores. This majority-based equation was used to make predictions in ethnic minority groups of Asian, Hispanic, black, and American Indian students in two successive freshman classes. If unbiased, the predictions would have been (1) at the correct level of achievement in each group, and (2) of equal accuracy in each group. Predictive bias in the form of small but significant over prediction did occur, except among American Indians. Possible bias in predictive accuracy was made difficult to detect because the prediction error variances were moderately high in all ethnic groups in both years. Total error variances were slightly larger in the black and Hispanic minority groups than in the other minority groups, although only a single significant difference was found.

Long-Range Predictive and Differential Validities of the Scholastic Aptitude Test in Medical School

With regard to the recent controversy about the predictive validity of the College Board Scholastic Aptitude Test (SAT), the present study was designed to investigate the long-range predictive and differential validities of this measure. Data for 1284 students who entered the Jefferson Medical College from 1965 through 1974 were analyzed by dividing them into four quarter groups from high to low based upon their scores on the Verbal and Mathematics portions of the SAT. By applying analysis of variance, statistically significant differences were found among high scorers (examinees in the top quarter) and low scorers (examinees in the bottom quarter) of the SAT relative to the yearly grade point averages (GPAs) and average standing on comprehensive examinations in medical school in the favor of high SAT scorers. The convergent and divergent validities of the SAT over an extended range of time were supported.

The Nelson-Denny Reading Test as a Predictor of College Freshman Grades

This study represents an attempt to determine the difficulty level, the internal consistency, and the usefulness of Form C of the Nelson-Denny Reading Test (NDRT) as a predictor of freshman grades at a public, midwestern university. The 2,122 entering freshmen (37.7% male) produced a high school mean grade point average (GPA) of 3.16, a college freshmen GPA of 2.63, an American College Testing Program (ACT) Composite score of 20.6 and a NDRT of 82.8--7 points above the NDRT norm group. The internal consistency coefficients (Kuder Richardson formula 21) of the NDRT were .90 for the 100-item vocabulary section, .69 for the 36-item comprehension section, and .91 for the combined score. The NDRT was correlated positively with the verbal portion of the College Board Scholastic Aptitude Test (SAT-V) (.79), the ACT Composite score (.71), high school GPA (.39), and college freshman GPA (.33), as well as with other indicators of academic ability. High school GPA was the single most valid predictor of college freshman GPA (.56), but combinations of high school GPA and the NDRT increased the prediction of college GPA to .58. Combining high school GPA and ACT composite scores produced a predictive validity of .60, whereas high school GPA and SAT-V and SAT-Mathematics produced a predictive validity of .61. Adding the NDRT to a prediction equation already containing high school GPA and either ACT or SAT scores did not improve the prediction. College grades in large freshman courses were similarly, but less effectively, predictable.

The Watson-Glaser Critical Thinking Appraisal as a Predictor of Performance in a Critical Thinking Course

The validity of the Watson-Glaser Critical Thinking Appraisal (WGCTA) measure (Watson and Glaser, 1964) was examined in an academic setting emphasizing the use of critical thinking with an accelerated group of students who were pursuing a combined baccalaureate and professional program of medicine. Using a criterion of grades from a specially designed physics course intended to stress critical thinking, validity coefficients of .54 ( p < .00002) and of .45 ( p < .0007), respectively, were observed for each of two predictors: (a) total scores on the College Board Scholastic Aptitude Test (SAT) and (b) total scores on the WGCTA measure. The WGCTA did not add much to the total prediction and should probably not be used as a substitute for regular entrance examinations in a college setting. However, the fact that it significantly predicted standing in a specialized course at the college level lends some credence to its criterion-related validity.

Relative Effectiveness of the California Achievement Tests in Comparison With the Act Assessment, College Board Scholastic Aptitude Test, and High School Grade Point Average in Predicting College Grade Point Average

Studied was the relative effectiveness of the. California Achievement Tests (CAT), the ACT Assessment (Academic Tests) of the American College Testing Program (ACT), the College Board Scholastic Aptitude Test (SAT), and high school grade point average (GPA) in predicting college freshman GPA. The incremental and differential incremental effectiveness of the CAT, ACT, and SAT in addition to high school GPA were also studied. Although high school GPA was the best single predictor, the CAT was as effective a predictor as was the ACT or the SAT. Use of either the ACT, SAT, or CAT resulted in an 18.47% increase in predictive etBciency over that obtained by using high school GPA alone. As the increase in predictive efficiency was very nearly the same (within rounding error) for the three tests (ACT, SAT, and CAT), they failed to demonstrate differential incremental validity.

Predicting Academic Achievement Among Specially Admitted Black Female College Students

Reports have indicated that minority students generally perform well in predominantly white colleges, in spite of low scholastic apti tude test scores. Although the cost of underpredicting their achieve ment has been discussed, the actual incidence of this under prediction seldom has been investigated. By means of stepwise multiple regression analysis, the relationship between college grade- point average (GPA) as the dependent variable and selection predic tor variables—high school rank, the Verbal score and Mathematics score on the College Board Scholastic Aptitude Test (SAT-V and SAT-M)—was evaluated in a sample of black female under graduates who initially had not been accepted by but later were ad mitted to a predominantly white liberal arts college. Results showed that after 8 semesters, 19 percent of the students withdrew or were transferred to another institution, 52% were graduated and 27% were dismissed for academic reasons. Stepwise regression analysis, using SAT-M, SAT-V and high school rank as predictors revealed a coef ficient of multiple determination ( R2) of .22. High school rank con tributed the greatest proportion of the variance to R2 (.14). SAT-V and SAT-M scores furnished increments of .07 and .01 respectively. The need to identify predictors other than high school rank and SAT-V or SAT-M scores is stressed and the injustice of under prediction is emphasized.

National Educational Development Tests as a Predictor of College Entrance Examination Board Scholastic Aptitude Test Scores

The purpose of this study was to determine the predictive validity of composite scores on the National Educational Development Tests (NEDT-C) scores in the prediction of College Board Scholastic Aptitude Test (CEEB-SAT) total (T) scores. The subjects were 69 students in a private college preparatory school for girls. The data consisted of NEDT-C scores obtained during the subjects' ninth grade year and SAT-T scores obtained during the subjects' twelfth grade year. The NEDT-C yielded a correlation of .76 with the SAT-T. This correlation was statistically significant at the .01 level.