NotesFAQContact Us
Search Tips
ERIC Number: ED441808
Record Type: Non-Journal
Publication Date: 2000-Apr
Pages: 32
Abstractor: N/A
Reference Count: N/A
Assessing Dimensionality of a Set of Language Test Data with Simulated and Real Data Sets.
Pyo, Kyong Hyon
The primary purpose of this study was to compare the performance of three procedures to assess dimensionality that were investigated by R. Nandakumar (1994) at different test conditions to reflect the characteristics of language test data. Procedures investigated were nonlinear factor analysis, the procedure of P. Holland and P. Rosenbaum, and W. Stout's nonparametric approach. To serve this purpose, simulated data consisting of 2,400 sets of unidimensional and multidimensional test data were used. The procedures were repeated with real language test data from the Test of English as a Foreign Language for 300, 500, and 1,000 examinees. Speculation of the real data structures was made on the basis of the findings from the simulation. Findings show that all three methodologies are able to discriminate between one- and two-dimensional test data. For simulated unidimensional test data, all three procedures were able to confirm unidimensionality, but for two-dimensional test data, the three procedures differed in their ability to detect the lack of unidimensionality. As statistics for assessing the degree of multidimensionality, both the approximate chi-square statistic and the DIMTEST T values were considered useful. An appendix contains six tables of study data. (Contains 6 tables and 36 references.) (SLD)
Publication Type: Reports - Evaluative; Speeches/Meeting Papers
Education Level: N/A
Audience: N/A
Language: English
Sponsor: N/A
Authoring Institution: N/A