Not infrequently, investigators assume that reliability for groups is greater than reliability for persons, or that the error variance for groups is less than that for persons. Using generalizability theory, it is shown that this "conventional wisdom" is not necessarily true. Examples are provided from the course-evaluation and the…

The simulated oral proficiency interview (SOPI) is a semi-direct speaking test that models the format of the oral proficiency interview (OPI). The OPI is a method of assessing general speaking proficiency in a second language. The SOPI is a tape-recorded test consisting of six parts: simple personal background questions posed in a simulated…

A study answered the question: are there significant differences in scores for two forms of the Gates-MacGinitie Reading Tests, Third Edition? Subjects, 23 fifth graders, were given Forms K and L, Level 5/6, of the Gates-MacGinitie Reading Test. The tests were administered by the regular classroom teacher in two testing sessions. Students were…

Reliability and concurrent validity data for the (SCII) were developed for a sample of 180 students. Two-week test-retest reliability correlations were approximately .90. Agreement between high scores and self-reported curriculum and occupational preferences ranged from 32 percent to 60 percent. Scoring on all Occupational scales produces…

The American College Testing Program (ACT) developed tests as a result of the College Outcome Measures Project (COMP). These instruments were intended for evaluation of nontraditional and traditional postsecondary education programs. They measure skills rather than information. The study was designed to check on several aspects of use of the COMP…

Three equating methods were compared in terms of magnitude of scale drift: equipercentile equating, linear equating, and item response theory (IRT) equating. A sample of approximately 2670 cases was selected for each pairing of a form of the Scholastic Aptitude Tests (SAT) and an anchor test. Of the two conventional equating methods,…

A live tailored testing study was conducted to compare the results of using either the one-parameter logistic model or the three-parameter logistic model to measure the performance of college students on multiple choice vocabulary items. The results of the study showed the three-parameter tailored testing procedure to be superior to the…

As alternatives to the objectives-based approach to specifying content domains for test construction purposes, six strategies are proposed: (1) amplified objectives; (2) Instructional Objectives Exchange (IOX) test specifications; (3) item transformations; (4) item forms; (5) algorithms; and (6) mapping sentences. Their effectiveness is assessed…

Rasch and classical test analysis methods were compared with respect to their similarities and differences in the identification of noninformative items and implausible person records. Using computer simulated data with known parameters, each model was evaluated in terms of its effectiveness in: (1) identifying noninformative or "bad"…

The Alpert-Haber Achievement Anxiety Test was developed to measure the extent to which individuals experience test anxiety. In at least two published studies, the authors claim to have used the test when in fact the response format was changed from that used in the original instrument and the "buffer" items were omitted. To investigate…

Two forms of the CEEB English Composition Test and four tests constructed by the University of Illinois Rhetoric department were compared as final examinations for the freshman English composition course. Results from 2545 students indicate that while the CEEB tests discriminate more between students and are more reliable, the departmental tests…

Cognitive style research is hindered by a variety of poorly described measures and style classifications and the use of small samples. This study describes and compares three measures of three postulated dimensions of cognitive style: impulsivity/reflectivity; internal/external locus of control; and field dependence/field independence to determine…

Results suggest that different groups of judges do set similar examination standards when using the same procedure, and that the average of individual judgments does not differ significantly from group consensus judgments. Significant differences were found, however, between the standards set by the two procedures employed. (RC)

WISC and WISC-R IQs of two groups of normal 10-year-old children from divergent socioeconomic backgrounds were compared in a counter-balanced research design. Significantly higher WISC IQs were obtained on Verbal and Full Scales of low socioeconomic group and on Performance and Full Scales of high socioeconomic group. (Author)