The more effective educational, social, and personal treatment of "atypical" students in institutions of higher education, toward facilitating the integration of these students into the mainstream of college and American life, in modes feasible to employ in our higher education settings. The determination of better administrative and management strategies for higher education in providing for the needs of disadvantaged students as well as of other students. At the institutional level, the effective resolution of the more difficult problem of helping the traditionally black institution to become an equal partner in the national higher education enterprise. The more successful infusion of programatic solutions into a kind of American institution hide-bound, to a remarkable extent, by its belief in its tradition, self-sufficiency, independence from outside pressure, and capability to solve its own problems; and, at the same time, the better mobilization of the nevertheless rather ramarkable resources of American higher education institutions to aid in achieving the programatic and societal evolution necessary to make equal opportunity a reality. APPENDIX SUMMARY OF RESEARCH ON VALIDITY OF TESTS FOR PREDICTING ACADEMIC PERFORMANCE IN COLLEGE OF CULTURAL MINORITIES The question of the meaning of scholastic aptitude test scores for blacks or other cultural minorities is not new, and a number of relevant previous research studies are available, either as part of the published literature or as unpublished institutional studies. The writer feels it is safe to say that the following conclusions, briefly documented here, tell a major story. 1. On scholastic aptitude or achievement tests, Negroes at a point permitting the beginning of college training tend to score significantly lower than Whites. This fact is too well known to require documentation; a recent relevant statement, however, is that by S. A. Kendrick (1968), who has estimated that "not more than 15 percent and perhaps as few as 10 percent of Negro high school seniors would score 400 or more on the ver bal section of the SAT (Scholastic Aptitude Test of CEEB). Only 1 or 2 percent would be likely to score 500 or more." It is indeed this fact that is the pressure, if not the justification, behind current black student demands for abolition of test barriers. For, if tests are indeed used to screen applicants, more Negroes than Whites will be screened out. 2. Published studies of the ability of SAT to predict grades of black students in predominantly Negro colleges, however, show that SAT is as valid in this kind of situation as it is for whites in predominantly white institutions. Typical of studies reporting this finding is one by J. P. McKelpin at North Carolina College, who reported (McKelpin, 1965) in his study of SAT and high school grades for predicting (black) students' performance at his institution (italics in original): The predictive validities based on the data for commonly used preadmission variables are as high as those usually reported for college freshmen ... the SAT scores account for about 60 percent of the variation in the grades explainable by the data from the preadmission variables when first semester grades are the criterion, SAT scores give a fair appraisal of the developed ability of students entering (predominantly Negro) colleges. It is true, however (probably because of the gross differences between racial groups noted before), that the use of tests directed at lower educational levels than the entering college freshmen have seemed more useful with Negroes in some instances. For example, a recent unpublished paper by John Hills of Florida State University and Julian Stanley of Johns Hopkins (Hills and Stanley, 1968) is abstracted by the authors: The two subtests of Level 4 of the School and College Ability Tests (SCAT) for school grades 6-8, are shown to predict freshman-year grades in the three predomnantly Negro coeducational colleges of a Southern state considerably better than did the Scholastic Aptitude Test (SAT), which was too difficult for approximately one-third of the enrolled freshmen. Relative improvement in multiple correlation for SCAT compared with SAT lessened when high-school grade average became one of the three joint predictors, apparently because high school grades of SATundifferentiated students supplied some of the missing intellective components. 3. Although relatively few studies have yet been done of the validity of SAT to predict grades for black students in integrated colleges, the available evidence supports the conclusion of no difference in the levels of predictive validity of SAT for blacks vs. whites in such institutions, but also that if white-based prediction formulas are applied to blacks, these students as a group tend to perform below the predictions. In the first sophisticated study of the predictive value of SAT for Negro and white students in three integrated colleges, Cleary (1968) summarized her findings: In the two eastern colleges, no significant differences in the regression lines (SAT predicting grades, blacks vs. whites within a single institution) were found. In the one college in the southwest, significant differences were found, but it was the Negro scores which were over-predicted. Thus, in one of the three schools, the Scholastic Aptitude Test was found to be slightly biased, but biased in favor of the Negro student. The 'bias' in favor of the Negro student in the Cleary study was a result of finding, in effect, that at one of the three schools, Negro students with a given SAT score and high school rank made lower grades than white students with identical SAT scores and high school ranks. Thus, if a predicted level of performance is used in selecting among applicants, Negro applicants selected would achieve lower actual performance levels than their white counterparts, though they would more likely be admitted. A similar finding has been obtained by L. B. Wilson (1969), who has studied performance and other characteristics of black vs. white students in four College Research Center institutions. He concludes on this aspect of his data: An analysis of the relationship between Predicted Freshmen Grade (combining the Adminissions variables-SAT-V, SAT-M, Achievement Test average), indicates that traditional admissions criteria tend to be at least as correlationally valid for black students as for entering students generally. There is moreover some evidence that predictions made on the basis of standard formulae may tend to overestimate the first-year performance of black students in the several colleges studied. Even more convincing are studies within the last year by Temp (1971) and the writer (yet unpublished). Temp collected data on black vs. white students in thirteen colleges over the country, and concluded: "If prediction of (... the grade point average in college) from SAT scores is based upon prediction equations suitable for majority students, then black students, as a group, are predicted to do about as well as (or better than) they actually do." The writer obtained similar findings in six public institutions in a Southern state. A survey of the literature by Flaugher cites a review by Kendrick and Thomas (1970), and notes a host of studies-Boney, (1966): Hills, Klock, and Lewis (1963); Roberts (1962, 1964); Stanley and Porter (1967); Olsen (1967); Cleary (1968): Morgan (1968); Munday (1965); Thomas and Stanley (1969); McKelpin (1965) Funches (1967); Perlberg (1967); and Peterson (1968). These have involved SAT, tests of the American College Testing Program, and other similar college level tests-both separately and in combination with high school grades. Rather than finding in these any evidence of reduced predictive validity (than that typically found for white students) Flaugher notes that test scores predict as well for blacks, and that in a large number of instances they provide better estimates of performance in college than do high school grades-a finding that may reflect the kinds of secondary schools that as recently as several years ago most blacks attended. Flaugher also notes a number of applications to prediction of job performance-Tenopyr (1967); Grant and Bray (1970): Campbell, Pike, and Flaugher (1969)-where tests are found to overpredict, not underpredict job performance when applied to non-whites. Tenopyr (1967, p. 15) calls it "unfair discrimination (which) however, would favor, not penalize, the Negroes." Flaugher adds the explanation afforded by Rock (1970) that motivation toward achievement in college is typically a white middle-class phenomenon, and that non-whites may not be as likely "to utilize to the maximum what aptitudes they possess." There may also be problems of less adequate preparation, poorer study skills, and the intrusion of anxieties that may arise from being in a real minority in the majority college culture. All this recent evidence indicates, as in the first Cleary study, that the use of SAT or similar tests may lead to accepting Negroes who are poorer academic risks than lower-scoring whites who may be excluded if similar standards are employed. This is not to state that such admissions should not take place; rather these findings are cite to show dence for the frequent claim that tests ard again if there is a bias, it is in the social and educational system in which these students were reared. 4. If one attempts to make a case for bias in academic tests because certain subgroups of the population make lower scores than others, the evidence points to deficit as a result of cultural disadvantage rather than as a result of racial origin. Cleary and Hilton (1968) studied performance on the Preliminary Scholastic Aptitude Test for grade 12 students in integrated high schools. When blacks were compared with whites of similar socioeconomic levels, they concluded: From the bivariate plots of sums of items scores, it was apparent that there were few items producing an uncommon discrepancy between the performance of Negro and white students. It must therefore be concluded that, given the stated definition of bias, the PSAT for practical purposes is not biased for the groups studied. 5. Experience with special remedial programs for high-risk students, (e.g., students whose test scores indicate high probability of academic failure), or attempts to improve test scores (and grade performance) by special coaching, seem to indicate that at the very least unusual efforts will be needed to improve academic performance. For example, after reviewing a large body of the literature on remedial education in the community junior college, Roueche (1968) concludes: The large majority of students who enroll in remedial courses fail to complete those courses satisfactorily and are doomed to failure or are forced to terminate their education. In one typical California public junior college, of the 80 percent of the entering students who enrolled in remedial English, only 20 percent of that number continued on into regular college English classes. In a study of the effect of well-contrived and intensive instruction (though of short term duration from 4 to 6 weeks) in the kinds of cognitive tasks involved in scholastic tests, S. O. Roberts of Fisk University and D. B. Oppenheim of Educational Testing Service found (Roberts and Oppenheim, 1966) with students with inadequate instruction in the past that "it does not seem reasonable to expect that similar short term instruction given on a wide scale would be of significant benefit to disadvantaged students." 6. From institution to institution, and from year to year, the specific validities of preadmission indices will vary. This is the matter of widespread experience; it probably results from a combination of factors: differences in curricula, heterogeneity of students, difference in institutional evaluational styles, etc. (This matter assumes added importance in the present context, however, for it is reasonable to assume that if an ability-free system of instruction were developed, then our present conventional tests of ability would not be relevant to predicting success in that system.) 7. Among senior colleges and universities over the country or in many state systems, the traditionally black institutions tend to hare students of lower test and high school performance levels than are generally achieved by black students attending traditionally white institutions; on tests and performance indices however, the means of some institutional groups of black students are higher than the corresponding means of some other institutional groups of white students. No published reports of this pattern in higher education are known, although it has been noted in the public schools within state or large city systems. (Documentation for a state system of higher education will be available shortly in a report in preparation by the writer.) The major implications are two-fold: First, there are traditionally white colleges where many more black students than apply would not appear, from preadmissions measures, to be at a competitive disadvantage-although, these are smaller, frequently private, schools that are not as active as others in recruiting black students. Second, the traditionally white institutions are nevertheless eroding the top of the black talent pool, if talent is measured traditionally, and many traditionally black institutions will shortly be able to demonstrate the effect of lower levels of talent than they have previously enjoyed. What may be drawn from these conclusions, if they are indeed essentially correct? The writer feels strongly that further exhaustive study of our present tests and potentially restrictive admissions procedures take time, energy, and more importantly, attention away from the real issue. That is, we need in higher education to focus instead on the study of curricular, instructional, motivational, and total educational environment strategies that may ease or void whatever cultural disadvantage exists, and that may permit real and rewarding learning experiences for these students. Then, research involving tests should be directed not toward proving that our current measurement proce dures are biased or are not biased, but rather toward creating new measures that may facilitate their identification of what settings and what instructional strategies are effective with what kinds of students with behavioral objectives set in terms of the needs of society. not in terms of the seeming limitations of certain identifiable subgroups. BIBLIOGRAPHY Boney. J. D. Predicting the Academic Achievement of Secondary School Negro The Duke Chronicle. February 5, 1969. Duke University Afro-American Society Flaugher. Ronald L. Testing Practices, Minority Groups, and Higher Education: A Review and Discussion of the Research. Unpublished manuscripts, Educational Testing Service, 1970. Flaugher, R. L., Campbell, J. T. & Pike, L. W. Ethnic Group Membership as a Moderator of Supervisor's Ratings. Project Report 69–5. Princeton, N.J.: Educational Testing Service, 1969. Flaugher, R. L., & Rock, D. A. A Multiple Moderator Approach to the Identifica tion of Over- and Underachievers. Journal of Educational Measurement, 1969 6(4), 223-228. Frederiksen, N. & Giilbert, A. C. Replication of a Study of Differential Predictability. Educational & Psychological Measurement, 1960, 20, 759-767. |