ERIC Number: ED351381
Record Type: Non-Journal
Publication Date: 1992-Aug-10
Reference Count: N/A
Assessing Dimensionality of a Set of Items--Comparison of Different Approaches.
The performance of the following four methodologies for assessing unidimensionality was examined: (1) DIMTEST; (2) the approach of P. W. Holland and P. R. Rosenbaum; (3) linear factor analysis; and (4) non-linear factor analysis. Each method is examined and compared with other methods using simulated data sets and real data sets. Seven data sets, all with 2,000 examinees, were generated with 3 unidimensional and 4 2-dimensional data sets. Two levels of correlation between abilities were considered: p=0.3 and p=0.7. Eight real data sets were used; four were expected to be unidimensional, and the other four were expected to be two-dimensional. Findings suggest that, while the linear factor analysis often overestimated the number of underlying dimensions, the other three methods correctly confirmed unidimensionality, but differed in their ability to detect the lack of unidimensionality. DIMTEST showed excellent power in detecting the lack of unidimensionality. Holland and Rosenbaum's approach and non-linear factor analysis approaches showed good power, provided the correlation between abilities was low. Four tables present study data, and there is a 46-item list of references. (Author/SLD)
Publication Type: Reports - Evaluative
Education Level: N/A
Sponsor: Office of Naval Research, Arlington, VA.
Authoring Institution: Illinois Univ., Urbana. Dept. of Statistics.