ERIC Number: ED190622
Record Type: RIE
Publication Date: 1980-Apr
Reference Count: 0
Further Work Developing Parallel Tests by P-Value Item Substitution.
Nassif, Paula M.; And Others
The statistical parallelism of a pair of tests equated by the P-value method was investigated, as well as the relationship between their raw scores and scores transformed by linear equating. Data were obtained from eight sections of the 1980 New Jersey Minimum Basic Skills Program; specifically, the reading and mathematics tests for each of grades 3, 6, 9, and 11. Twenty-five equating test items were selected intact, and 25 were selected by p-value item substitution, as the parallel forms. Both the Wilks test and linear score equating supported the parallelism hypothesis for the two subject areas and four grade levels, except for the grade 12 mathematics test. Thus, the p-value method would benefit testing programs, where two or more parallel forms are administered annually for practical and statistical reasons. First, p-value requires a smaller sample size than linear or equipercentile score equating. Second, the technique of adding pilot items to existing forms can be used to generate, during an actual administration, the item statistics (p-value and point biserials) needed for item-substitution. (CP)
Publication Type: Speeches/Meeting Papers; Reports - Research
Education Level: N/A
Authoring Institution: N/A
Identifiers: Linear Equating Method; New Jersey Minimum Basic Skills Programs; Parallel Forms Reliability
Note: Paper presented at the Annual Meeting of the National Council on Measurement in Education (Boston, MA, April 8-10, 1980).