NotesFAQContact Us
Collection
Advanced
Search Tips
ERIC Number: ED442818
Record Type: Non-Journal
Publication Date: 2000-Apr
Pages: 44
Abstractor: N/A
Reference Count: N/A
ISBN: N/A
ISSN: N/A
Using Latent Class Analysis To Set Academic Performance Standards.
Brown, Richard S.
The use of latent class analysis for establishing student performance standards was studied. Latent class analysis (LCA) is an established procedure for investigating the latent structure of a set of data. LCA presumes that groups, classes, or respondents differ qualitatively from one another, and that these differences account for all of the relationships in the data. A student assessment instrument was developed consisting of a 10-item multiple choice component and 2 performance assessment tasks. The instrument was completed by 191 seventh and eighth grade mathematics students. The latent class procedures provided information regarding the qualitative differences among the student responses. A series of latent class models was explored using binary and continuous indicators. For this sample of student responses, the LCA procedure indicated that a two-group structure was the most appropriate model for explaining the differences in student performance. Perhaps the most important finding from the study deals with the extent to which the varying standard-setting procedures rendered comparable conclusions regarding what constitutes proficient performance. Another implication of the findings is that if empirical methods can be shown to render determinations regarding student proficiency that are comparable to the more common, and more costly, judgmental approaches, these methods could be used more often to support the determinations made with the judgmental approaches. An appendix contains the developed student assessment instrument. (Contains 2 figures, 10 tables, and 70 references.) (SLD)
Publication Type: Reports - Research; Speeches/Meeting Papers; Tests/Questionnaires
Education Level: N/A
Audience: N/A
Language: English
Sponsor: N/A
Authoring Institution: N/A