skip to main content skip to footer

Assessing the Dimensionality of the GMAT Verbal and Quantitative Measures Using Full Information Factor Analysis GMAC IRT

Author(s):
Kingston, Neal M.
Publication Year:
1986
Report Number:
RR-86-13
Source:
ETS Research Report
Document Type:
Report
Page Count:
19
Subject/Key Words:
Graduate Management Admissions Council, Factor Analysis, Graduate Management Admission Test (GMAT), Item Response Theory (IRT), Maximum Likelihood Statistics, Statistical Analysis, TESTFACT

Abstract

The dimensionality of the Graduate Management Admission Test Verbal and Quantitative measures was assessed using full information factor analysis (FIFA). FIFA, as implemented by TESTFACT, uses marginal maximum likelihood to estimate (reparameterized) discrimination and difficulty parameters of multidimensional item response theory models. The lower asymptote for each item is treated as a known constant whose value is input by the program user. A stepwise FIFA with orthogonal and Promax rotations was run for each measure. In each case a three factor solution was statistically significant. Higher order models were not run, but additional factors might have been significant. For the Verbal measure, the percent variance explained by the three factors in the orthogonal solution was 24.0, 3.7, and 2.6, respectively. For the Quantitative measure, the variance explained by the three factors was 33.7, 3.5, and 1.3. In each case, the first factor is clearly dominant. (19pp.)

Read More