Do C-Tests with Different Number of Gaps Measure the Same Construct?
|
| Title | Do C-Tests with Different Number of Gaps Measure the Same Construct? |
| Authors | |
| Abstract | Constructing C-Tests with 20-25 blanks in each passage is the standard practice among C-Test users and researchers. The present study explores if C-Tests with different number of blanks measure the same construct as those with standard 25 gaps. Out of data from a four-passage C-Test with 40 blanks in each passage, two other datasets with 10 and 25 blanks in each passage were constructed. The three datasets were separately analysed with the Rasch model and the psychometric characteristics of the three test forms were compared. Results showed that the three forms have similar fit indices and are equally unidimensional. But C-Tests with more gaps turned out to be more reliable. Cross plotting students’ ability measures from pairs of C-Test forms on x and y axes revealed that test-takers have identical ability measures regardless of the number of gaps in the C-Tests. This was interpreted as evidence of the invariance of the C-Test construct when the number of gaps in the passages is altered. Implications of the study for C-Test applications with an important word of caution are given. |
| Publisher | ACADEMY PUBLISHER |
| Date | 2011-06-01 |
| Source | Theory and Practice in Language Studies Vol 1, No 6 (2011) |
| Rights | Copyright © ACADEMY PUBLISHER - All Rights Reserved.To request permission, please check out URL: http://www.academypublisher.com/copyrightpermission.html. |