Options
Using a Rasch model to analyse parallel forms of two achievement tests on biology
Citation
Chew, L. C. (1999). Using a Rasch model to analyse parallel forms of two achievement tests on biology. In S. P. Loo (Ed.), Proceeds of the MERA-ERA Joint conference 1999: Educational Challenges in the New Millennium (pp. 88-94). Malacca, Malaysia.
Abstract
Achievement tests on biology were constructed as part of an item pool development for a
larger study on computerised adaptive testing. A total of 220 four-alternative multiple-choice test
items were targeted. A test specification, modelled after the 1995 GCE ‘O’ level syllabus, was
used to guide the writing and selection of test items aimed at assessing learning outcomes of
biological concepts and principles. Two tests were planned; each with parallel forms based on
content consistency. In addition, ‘anchor’ test items were placed on the two forms. The test forms
were spiralled (i.e. arranged in a fixed alternate order) in the data collection from a representative
sample of about 2000 students from a population of Singapore students who were candidates for
the 1994 GCE ‘O’ level biology examination. This paper reports on an application of the Rasch
model to analyse the test items on parallel forms of the two tests, and to assess the model-fit of the
items.
Date Issued
December 1999
Description
This paper was published in the 1999 Proceedings of the MERA-ERA Joint Conference held at Malacca, Malaysia from 1-3 December 1999