Validation of the Frey and Detterman (2004) IQ prediction equations using the Reynolds Intellectual Assessment Scales

Abstract

This study sought to assess the ability of the equations Frey and Detterman (2004) presented to assess IQ in the Reynolds Intellectual Assessment Scales (RIAS). The study confirmed Frey and Detterman’s equation (2) best predicted IQ from recentered SAT scores. Nonetheless, both of the Frey and Detterman’s equations did not match the optimal model found in the current data, namely that SAT Total or SAT Verbal, alone, best predicted IQ as measured by the RIAS. Implications from this study are much the same as those stated by Frey and Detterman, namely, that SAT appears to be a measure of general intelligence and is a useful tool in predicting cognitive functioning. Nonetheless, future research is needed with a wider range of IQ instruments to assess which SAT variables are the best predictors, and what weight each should be ascribed.

Introduction

In 2004, Frey and Detterman reported that the Scholastic Aptitude Test (SAT) was, in large part, a measure of Spearman’s (1904) g. Moreover, in the same article they developed two sets of regression equations to predict future IQ scores based on scores from the SAT (either the Total score, or a linear combination of the Math and Verbal sections). When such a regression model is developed in order to predict observations, it needs testing to assess its accuracy as a measure of quality—a process that requires observations obtained outside of the original sample (Camstra and Boomsma, 1992, Neter et al., 1996, Snee, 1977). Consequently, the purpose of this study is to assess the two regression equations given by Frey and Detterman.

The IQ score used for this study was the Composite Intelligence Index (CIX) from the Reynolds Intellectual Assessment Scales (RIAS; Reynolds & Kamphaus, 2003), a brief, individually-administered psychometric instrument that measures both general cognitive ability and general memory. There are six subtests given on the RIAS, four of which compose the CIX, with the others composing the Composite Memory Index. (For a listing of the subtests that make up the CIX, see Table 1.) The subtests that make up the CIX require both verbal and nonverbal reasoning. As it is a new instrument, few published studies are available in the professional literature, although the technical manual reports that the four subtests that make up the CIX all have moderate to high g loadings, ranging from .57 to .83 (Reynolds & Kamphaus, 2003, p. 98). In addition, the manual reports that the CIX highly correlates with the Wechsler Adult Intelligence Scale-Third Edition Full Scale IQ (.75; p. 107).