NotesFAQContact Us
Collection
Advanced
Search Tips
ERIC Number: ED393914
Record Type: RIE
Publication Date: 1995-Nov
Pages: 35
Abstractor: N/A
Reference Count: N/A
ISBN: N/A
ISSN: N/A
Smoothed Standardization Assessment of Testlet Level DIF on a Math Free-Response Item Type.
Lyu, C. Felicia; And Others
A smoothed version of standardization, which merges kernel smoothing with the traditional standardization differential item functioning (DIF) approach, was used to examine DIF for student-produced response (SPR) items on the Scholastic Assessment Test (SAT) I mathematics test at both the item and testlet levels. This nonparametric technique avoids model misspecification problems. It also has fewer sampling errors because of smoothing. Results from the smoothed item-level DIF analysis showed that regular multiple choice items have more variability in DIF values than SPRs. The testlet DIF analysis indicated that results from White examinees may exceed comparable African-American examinees by 14-19 scaled score units on average. Differences between the smoothed standardization and the traditional standardization are small and most likely due to the use of the true score as a matching variable in the smoothed standardization approach. (Contains 2 tables, 4 figures, and 16 references.) (Author/SLD)
Publication Type: Reports - Evaluative; Speeches/Meeting Papers
Education Level: N/A
Audience: N/A
Language: English
Sponsor: N/A
Authoring Institution: Educational Testing Service, Princeton, NJ.
Identifiers: Item Bias Detection; Kernel Method; Scholastic Assessment Tests; Smoothing Methods; Specification Error; Standardization; Testlets; Variability
Note: Paper presented at the Annual Meeting of the National Council on Measurement in Education (San Francisco, CA, April 19-21, 1995).