On the identification of students’ misconceptions in a two-tier item

Abstract

The two-tier item format has gained some popularity in use by science educators worldwide, including those from Taiwan. Despite its broad applications, there are relatively few studies that focused themselves on extending the present methodology in analyzing this particular item format. As a result, there are still uncertainties concerning how the result in the data table from a two-tier item should be analyzed and interpreted, especially with respect to the identification of potential misconceptions held by the participating students. The usual practice in this area of research is to assume that if more than 10 % of the respondents picked a wrong combination of options across the two tiers, then that combination can be regarded as reflecting the presence of a misconception of some sort. This study argues against the use of the 10 % rule, on the ground that it is, among other reasons, subjective and lacks substantive support based on the subject matters. Instead, it is suggested that correspondence analysis can be performed in a three-stage manner. Potential types of misconceptions as held by the participating students could then be identified by means of interpreting the clusters of categories across the two tiers in the correspondence plot. This study reflects the kind of interest from local researchers on methodological issues surrounding the two-tier item format.