Open Collections

Examinee control of item order effects on latent trait model and classical model test statistics

University of British Columbia

The purpose of this study was to determine what effect changes in the item order had on classical and on latent trait test statistics. As well, comparisons were made between students who were allowed to answer the questions in any order, and students who were required to answer the questions In the order presented in the test booklet. The results were then analyzed using the student's ability level as an additional independent factor. Four different formats of a forty item mathematics test were used with 590 students in grade eight. Half of the booklets had the items sequenced from easiest to hardest. The other booklets were sequenced from hardest to easiest. In addition, half of the tests of each sequence had special directions which prevented students from altering the given item difficulty sequence. The classroom teachers provided a rating of each student's ability in mathematics. The order of the items was found to have a significant effect. Tests which were sequenced from hard to easy had a lower mean score. Although students with test booklets with restrictive directions had lower scores on average, it was not a statistically significant difference. There were no significant interactions found. Classical and latent trait item difficulty statistics showed a high degree of correlation. It was concluded that under certain circumstances, the order of the items could effect both classical and latent trait statistics. It was also recommended that care should be taken when assumptions are made about parallel forms or local independence.

Text

2010-10-18

For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use.

http://hdl.handle.net/2429/29353

Special Education

University of British Columbia

Graduate