Differential Validity, Differential Prediction, and College Admission Testing: A Comprehensive Review and Analysis. Research Report No. 2001-6.

https://onlinelibrary.wiley.com/doi/pdf/10.1002/j.2333-8504.1978.tb01166.x

Population validity and college entrance measures

..............................................................1

/pdf/differential-validity-differential-prediction-and-college-3zdna93otl.pdf

Differential Validity, Differential Prediction,and College Admission Testing: A Comprehensive Review and Analysis

This nonexperimental study explored the predictive strength of English proficiency levels on academic achievement of middle school students in a sample of 17,470 native English-speaking (NES) students, 558 English language learners (current ELLs), and 500 redesignated fluent English proficient students (former ELLs). Results of multilevel analyses indicated that after controlling for relevant student- and school-level characteristics, former ELLs significantly outperformed current ELL and NES students in reading (effect sizes: 1.07 and 0.52) and mathematics (effect sizes: 0.86 and 0.42). The results support Cummins's (1979, 2000) lower level threshold hypothesis predicting that upon reaching adequate proficiency in the language of schooling and testing, ELLs would no longer experience academic disadvantages. Refinements for the theory and directions for future research are discussed.

http://eri.louisville.edu/papers_pres/Tretter_Language_Learning_2012.pdf

English Language Learners and Academic Achievement: Revisiting the Threshold Hypothesis

Predictive Validity of the SAT in Black and White Institutions The controversy over whether standardized tests like the SAT and the ACT are fair to ethnic minorities continues to rage. There is considerable opinion that standardized tests are unfair to African Americans and other minorities. Proponents of standardized tests argue that they offer an objective, common yardstick that helps identify capable students who come from various backgrounds and grading systems. Thus, they prevent discrimination against able minority candidates (College Board, 1983). The National Research Council of the National Academy of Sciences reports that SAT scores predict performance as accurately for blacks as they do for majority applicants. Though the report acknowledges that average scores for whites are higher than those for blacks, just as they are for men compared to women, the averages do not reveal the different abilities of individuals within those groups, which is what the test is intended to identify. Opponents of standardized tests allege that they are inherently unfair to disadvantaged minorities because they are culturally and educationally inappropriate (Hilliard, 1990; Carty-Bennia, 1989), because such tests are frequently wrong in assessing the potential of minorities (Crouse & Trusheim, 1988), and because wide variation in predictive validity suggests unfairness (see Wilson, 1981; Houston, 1983; Temp, 1971). The evidence for predictive validity of the SAT for white students is consistent (see Table 1). A study by Cameron Fincher (1974) is representative of many large-scale efforts that validate the effectiveness of the SAT. Fincher's work was carried out in 29 institutions of the university system of Georgia over a 13-year period. The average correlation obtained was 0.490 (or 24% of the variance in grades). Ford and Campos (1977) similarly reported a summary of validity data gathered over 10 years from 1964 to 1974, and found the average correlation with freshman grades for the verbal SAT and the math SAT to be 0.370 and 0.320, respectively (13.7% and 10.2% of the variance). This study also found that SAT scores were predicting as well in more recent years as in earlier years, but that the validity of the high-school record and multiple correlations had declined slightly. A study by Ramist, Lewis, and McCamley-Jenkins (1994) reported validity coefficients for students in 11 colleges. The average correlation of the SAT with freshman grades was 0.420 (17.6% of the variance). Willingham and Breland (1982), in a study of 18 independent colleges, found correlations averaging 0.420 (17.6% of the variance). Note that validity coefficients are rarely considered alone but in combination with the high-school record, which produces substantially higher multiple correlations. Several smaller studies rely on data from a single college and also find results similar to the large-scale efforts. For instance, Wilson (1980) found that at a state university the math and verbal SATs correlated 0.388 and 0.392, respectively, with grades (15.5% and 15.1% of the variance). Wilson (1981) also reported data at a highly selective college and found the math and verbal SAT correlations to be somewhat lower, that is, 0.301 and 0.338 (9.1% and 11.4%). At New York University, Fleming & DuBois (1981) found the SAT score to be the best predictor of college grades (r = 0.410; 16.8%) in a study that considered several motivational variables. Other studies, both by ETS researchers and by independent researchers, report similar results (Morgan, 1990; Dittmar, 1977; Goldman & Hewitt, 1976; Goldman & Richards, 1974). For all reviewed studies combined, the average correlation for white students was 0.342, accounting for an average of 11.7% of the variance in grades. A study by Pennock-Roman (1990) is the only one reporting relatively low average correlations of 0.243 (5.9% of the variance) for white students in three public and three private institutions. …

https://www.jstor.org/stable/pdfplus/2649106.pdf

Are Standardized Tests Fair to African Americans? Predictive Validity of the SAT in Black and White Institutions.

On five occasions spanning a thirteen year period, data were gathered to examine the predictive validity of the Scholastic Aptitude Test (SAT) and high school achievement (HSA), when predicting college freshman grades. Results replicated an earlier study by Michael and Jones in two respects: (a) with regard to academic achievement, females were more predictable than males; (b) the ability to predict college grades from SAT and HSA seems to be decreasing.

A Decline in the Predictive Validity of the Sat and High School Achievement

In order to determine the concurrent validity of the Language Assessment Scale I as a measure of English language proficiency test scores were compared to teacher judgment as reflected in a rating scale. It was found that the Language Assessment Scale I had moderately high concurrent validity. Scores correlated .73 with a quantitative measure of teacher judgment. All students who scored low on the test were judged by teachers to need special instructional help because of their English language deficiency.

Starrett Dalton

Papers

A Decline in the Predictive Validity of the Sat and High School Achievement

Validation of the Language Assessment Scales.