Abstract: Cognitive item generation, rooted from the theory-based measurement paradigm, requires formulating an elaborated theory of the measured construct, upon which task features that are linked to cognitive variables underlying such theory can be systematically analyzed and manipulated. Cast from the perspective of problem solving as an agent – situation interactive process, both cognitive processes of algebra story problem solving and task affordance of such problems are systematically analyzed in this article. Specifically, an algebra story problem is considered as a particular quantitative structure embedded in a verbally-stated realistic situation. Three different types of representations, i.e., a textbase, a situational model and a quantitative model, need to be formed during the four phases of algebra story problem solving including translation, integration, planning and execution, which are further mediated by the schematic knowledge possessed by the problem-solver. Correspondingly, Task features that are associated with syntactic/semantic relations, quantitative elements as well as their structures, and problem situations are analyzed and synthesized based on a broad range of relevant literature. It is evident from the current analysis that subsequent research should focus on empirical studies to link the set of cognitive variables and task features for such problems, as well as to evaluate the generalizability of such linkages across the targeted domain in order for item generation of algebra story problem to lay on a solidly scientific foundation.

Key words: theory-based measurement, item generation, algebra story problem, cognitive process, task feature