Analysis of Individual Two-tier Items from a Probabilistic Modeling Approach

Abstract

The use of the two-tier item format as a way to assess students' learning of science concept has gained popularity in recent years, as evidenced by a steady increase in the number of studies utilizing this approach (Treagust, 1988; Treagust & Smith, 1989; Tan & Treagust, 1999). A typical two-tier item is composed of two portions, usually a factual and a reason portion, both in multiple choices format. The purpose of the first portion is to assess if students can provide factual or other declarative knowledge regarding a certain phenomenon that is stated in the item stem. The purpose of the second portion is to assess, in addition, if students can demonstrate knowledge regarding the reason behind the given phenomenon. While the traditional multiple-choice format usually suffer from the constraint of assessing students' knowledge in a piecemeal manner, the combined assessment of the two-tier format allows for a deeper insight into their knowledge structure behind some science concepts. Though useful as this format may appear to be, the kind of analysis that one could perform on the data thus collected is still open to further research. Since it is relatively new to most pure and applied researchers alike, so part of the talk will be an introduction to this item format and to bring out some technical issues about it. The rest of the talk will report the result of a preliminary study that concerned itself with the analysis of two-tier items by means of a probabilistic modeling approach, through which some useful parameters can be estimated.