Essays on decision complexity, bounded rationality, and choice inconsistency

Chapter 1 studies the way in which decision complexity affects choice consistency. We use a stated preference experiment in which we define five complexity dimensions and make predictions regarding their effects on choice consistency. We use a heteroscedastic Multinomial Logit Model in which we model the scale factor as a systematically varying parameter. We test five hypotheses. We find that the complexity dimensions affect choice consistency with different magnitude and significance. In addition, choice complexity significantly distorts coefficient estimates and welfare measures.;Chapter 2 addresses two questions. First, what is the primary source of response noise: decision errors or systematic reactions? Second, can we identify some of the heuristics that individuals use? We use a heteroscedastic Multinomial Logit Model. We define two categories of complexity measures. The first category contains measures that we use in the space of variance. The second category contains measures that we use in the space of utility. We test five hypotheses. First, we find that complexity predominantly induces systematic reactions. Second, we identify four heuristics. Third, ignoring the systematic distortions caused by complexity results in biased coefficient estimates. Finally, complexity impacts respondents' welfare by inducing sub-optimal decisions.;Chapter 3 analyzes three issues. First, we study the effects of two sources of heterogeneity on choice consistency: respondents' cognitive abilities, and respondents' perceptions of complexity. Second, we analyze whether respondents' perceptions of complexity induce heuristics. Third, we identify some of the determinants of respondents' perceptions. We analyze the first two issues using a heteroscedastic Multinomial Logit Model in which we model the scale factor as a systematically varying parameter using three types of variables: an objective complexity measure, socioeconomic characteristics, and respondents' perceptions of complexity. We analyze the third issue using an Ordered Logit Model. We use three types of independent variables: objective measures of complexity, context variables, and socioeconomic characteristics. First, we find that heterogeneous cognitive abilities induce differences in choice consistency. Second, we find a strong correlation between respondents' perceptions of complexity and the adoption of heuristics. Third, we find that individuals' perceptions of complexity exhibit two features: (i) a low correlation between the objective and subjective measures of complexity, and (ii) context variables significantly impact respondents' perceptions.