An extension of decision field theory to constrained multiattribute choice

Roe, Busemeyer, and Townsend's (1999) multialternative decision field theory (MDFT) was extended to understand effects produced using Senter and Wedell's (1999) constraint paradigm. This paradigm studies the impact of strict presentation sequences upon choice accuracy, strategy use, and decision effort. Participants are assigned to either view each piece of information on a single alternative before viewing the next alternative (alternativewise), or view the value of each dimension across alternatives before viewing the next dimension (dimensionwise). Participants make their choices after viewing information. The choice task involves choosing one of several apartments described on four dimensions.;A key effect produced with this paradigm was higher choice accuracy and smaller viewing times produced by dimensionwise participants. To date, this effect has eluded successful explanation. Starting with MDFT, this dissertation attempted to account for seven critical effects produced using Senter and Wedell's (1999) constraint paradigm. All but one of these effects was successfully modeled. The dimensionwise advantage was attained with correlated valuation error within dimensions for dimensionwise processing, but not for alternativewise processing. The simulation study suggested that this may be a viable explanation for the dimensionwise advantage.;Experiments 1 and 2 tested this explanation. A competing hypothesis attributed source of the advantage to differences in memory retrieval. Experiments 1 and 2 manipulated the type of scale presented to participants to discriminate between these two error sources. The correlated error hypothesis predicted that the dimensionwise advantage would disappear if participants viewed a homogenous attractiveness scale instead of the usual heterogeneous scale of actual values. The memory retrieval hypothesis predicted that the dimensionwise advantage would not disappear. In Experiments 1 and 2, the dimensionwise advantage disappeared with the attractiveness scale, but neither hypothesis was fully supported or discounted, as alternativewise participants unexpectedly changed strategy with the attractiveness scale.;The MDFT extension successfully modeled data produced by participants in Experiments 1 and 2 under the correlated error hypothesis. Overall, this extension provided a theoretical framework to understand decision behavior under the constraint paradigm. Experiments 1 and 2 also implied that the value mapping process was the source of the dimensionwise advantage.