NotesFAQContact Us
Search Tips
ERIC Number: ED346130
Record Type: RIE
Publication Date: 1992
Pages: 12
Abstractor: N/A
Reference Count: N/A
Comparison of the Chi-Square Procedure with the Symmetric Regression Procedure for Developing a Common Metric in Item Response Theory.
Abdel-fattah, Abdel-fattah A.
A modified regression procedure for finding a common metric for item response theory (IRT) parameter estimates and for total test scores is proposed and compared to the chi-square method of equating. The chi-square method was chosen after literature review for the currently used scaling procedures. The modification used with the regression analysis was to account for its asymmetry. The new method was compared to the chi-square procedure and was found to produce similar transformation constants. The new method is simpler and more flexible than the chi-square procedure, and it is inexpensive and conceptually valid. It has been implemented in a computer program that equates the scores of two tests. Parameter estimates from different calibrations or different tests can also be scaled to a common metric by using this program. It can be used for linear, non-linear, raw score, or true score equating. The equating can be performed by using the ordinary least squares, the weighted least squares, or the maximum likelihood estimation procedure. The weighting is used to account for outliers. Weighting can also be performed by using the reciprocal of the average standard deviations of the two variables to be equated. The program used in this study is available in both the Statistical Analysis System (SAS) and FORTRAN. One table and 12 graphs present study data, and there is a 10-item list of references. (Author/SLD)
Publication Type: Reports - Evaluative; Speeches/Meeting Papers
Education Level: N/A
Audience: N/A
Language: English
Sponsor: N/A
Authoring Institution: N/A
Identifiers: Linear Equating Method; Outliers; Symmetric Regression Procedure
Note: Paper presented at the Joint Annual Meeting of the Psychometric Society and the Classification Society (New Brunswick, NJ, 1991).