NotesFAQContact Us
Search Tips
ERIC Number: ED235203
Record Type: Non-Journal
Publication Date: 1983-Apr
Pages: 8
Abstractor: N/A
Reference Count: N/A
An Alternate Procedure to Obtain Ability Estimates in Latent Trait Models.
Houser, Ronald L.; And Others
This report describes a procedure that promises to improve the stability, accuracy, and efficiency of the employment of latent trait models and an application of the procedure to the Rasch model. Data were collected from the Portland Public Schools Level Tests administered to 25,740 students. Since each of the 173 items (chosen from the total number of items tested after consideration of the approximately 50% overlap of items from one level to the next) could be calibrated twice by both the linking procedure and the fixed parameter procedure, the same data set was used to compare the relative efficiency of calibration within each procedure and then to compare the efficiency between procedures. This allowed the focus of each metric to be somewhat different but still provided a test of the calibrating efficiency. The differences in the distributions between each pair of item calibrations can be viewed as the amount of error in each calibrating procedure; no statistical or practical difference between these two distributions were found. A final distribution was computed by subtracting the linking difference from the fixed parameter difference for each item. This distribution, representing the relative superiority of one calibration procedure over the other showed no significant difference between the two procedures. Thus, efficiency of calibrating items through a linking approach or through the fixed parameter procedure is essentially the same. (PN)
Publication Type: Speeches/Meeting Papers; Reports - Research
Education Level: N/A
Audience: Researchers
Language: English
Sponsor: N/A
Authoring Institution: N/A