PSYCHOMETRICS



Psychometrics



Psychometrics

Introduction

Recent years have seen the expansion of numbers of students going into higher learning at both the undergraduate and postgraduate grade. The U.K. admission benchmark up on post-graduate (graduate school) administration programs is that students must have accomplished an undergraduate or first stage, for demonstration, Batchelor of Arts (BA) / or Batchelor of Science (BSc). By contrast, in financial or work settings, human resource departments normally use variables in supplement to know-how as part of selection processes encompassing presentation on psychometric tests. The utility of measures used for selection procedures is based on the strength of relationships between those measures and subsequent performance.

In the US, standardized tests, especially the Graduate Record Examination (GRE), form very important components of admission criteria to Graduate Schools. Norcross, Hanych, and Terranova (1996) reported GRE scores are required by 93% of Doctoral programs, and by 81% of Masters' programs.

The GRE was specifically designed to measure those “abilities relevant to performance in graduate studies”. It comprises a Verbal measure (GRE-V), a Quantitative measure (GRE-Q), and an Analytical measure (GRE-A). In addition to these measures, there are GRE tests requiring knowledge of a particular subject, e.g. psychology, chemistry etc.

Numerous studies have investigated the predictive validity of GRE scores. Results show equivocal findings. A meta-analysis combining 10 studies among samples of students enrolled on psychology graduate programs (N = 963) produced a correlation of r = .15 for both the GRE-V and the GRE-Q with a criterion measure of Graduate Grade Point Average (GGPA, Goldberg & Alliger, 1992). Morrison and Morrison (1995) concluded that the GRE is a poor predictor of academic performance based on meta-analytical results showing correlations of r = .22 for GRE-Q and r = .28 for GRE-V with GGPA. In a more recent meta-analysis involving 1753 independent samples and 82,650 graduate students, predictive validity coefficients for GRE were, GRA-V r= .34, GRA-Q r = .38, GRE-A r = .36, and GRE -Subject, r = .45. Although predictive validities from meta-analyses are modest in terms of proportion of the criteria variance explained, it is argued that even small predictor- criteria correlations can have a beneficial effect on selection decisions (Taylor & Russell, 1939).

In addition to variables used for selection purposes, there is also a need to identify students at risk of failing. Recent research has investigated the relationships between perceptions of self-efficacy (Bandura, 1986, 1997) and subsequent performance. Self-efficacy is defined as the 'beliefs in one's capabilities to organize and execute the courses of action to produce given attainment' (Bandura, 1997, p.3).

An issue in the study of self-efficacy and performance is the context in which the study was conducted. Bandura (1997) stated, 'people take their self-appraisal seriously when they chose between courses of action that have significant personal consequence' (p.68). If research findings are to be applied to practice and self-efficacy measures are to be used to assist in identifying students at risk of failing, the research should have been conducted in an ecological valid ...