ABSTRACT
The Graduate Record Examinations (GRE) revised General Test launched in August 2011 is a high-stakes adaptive test delivered on demand and is used for graduate admissions worldwide. It is composed of three independent measures: Analytical Writing, Verbal Reasoning, and Quantitative Reasoning. The two writing prompts are delivered first and are followed by two verbal sections, two quantitative sections, and a single unscored section.1 Each verbal and quantitative section includes twenty items and is limited to thirty and thirty-five minutes, respectively (Educational Testing Service 2012a). The verbal and quantitative measures follow the same multistage adaptive testing (MST) design (see Figure 21.1). Scores are determined based on the number of items correctly answered (NC) following the well-known item response theory (IRT) true score equating approach described by Kolen and Brennan (2004) and Lord and Wingersky (1984). The verbal and quantitative reporting scales have been redefined to enhance score interpretation by producing relatively symmetric, centered, and smooth distributions with little ceiling toward the top of the scale. As part of the redefinition, the two 130-to 170-integer scales were aligned so that the expected mean and standard deviation for the total test-taking population would be approximately the same for both the verbal and quantitative measures. After one year of testing, the observed verbal and quantitative mean and standard deviation were 150.8 and 8.5 and 151.3 and 8.7, respectively (Educational Testing Service 2012b, Table 1).
