ERIC Number: ED209906
Record Type: RIE
Publication Date: 1980-Sep
Reference Count: 0
Practical Procedures for Test Length Reduction and Item Selection.
Garrison, Wayne M.; Coggiola, Deborah C.
The study involving approximately 300 deaf students entering the National Technical Institute for the Deaf was designed to determine the extent to which four cognitive measures could be reduced in overall length without serious threat of information loss. The measures investigated were the California Reading Comprehension Test, the Mathematical Skills Program, the Abstract Reasoning Test, and the Space Relations Test. In all instances, the correlation between reduced and full length tests is very high and positive, indicating that control over the distribution of the difficulty indices associated with the items in a test produces efficient tests requiring substantially fewer stimuli than those included in the original design sets. Moreover, the effects of test stimuli reduction on recomputations of the reliability estimate for each test reveals negligible decreases. In summary, the findings showed that the number of test stimuli comprising the four instruments studied could be reduced by an average of 54%. (SB)
Publication Type: Reports - Research
Education Level: N/A
Sponsor: Department of Health, Education, and Welfare, Washington, DC.
Authoring Institution: National Technical Inst. for the Deaf, Rochester, NY.
Note: For related documents, see EC 140 611-630.