Hypothetical bias in contingent valuation

The three essays in this dissertation address issues pertinent to hypothetical bias in the contingent valuation (CV) mechanism. Empirical evidence gathered through several experiments conducted at the University of Massachusetts, Amherst, is used to provide a better understanding of the cause and nature of hypothetical bias. The first essay is based upon the notion that uncertain responses in stated preference valuation are associated with hypothetical bias. While the relationship between respondent uncertainty and hypothetical bias is not well understood, calibration techniques such as the uncertainty adjustment have been developed to mitigate the bias. Hence this study seeks to fill this gap by uncovering the relationship between hypothetical bias and respondent uncertainty using induced value goods. According to the results there is no relationship between certainty of induced values and either respondent\u27s stated level of certainty or hypothetical bias. The lack of hypothetical bias with induced value goods, but its continual existence in homegrown value goods lays ground for further investigation of the bias in the second essay. By employing both induced value and homegrown value goods, this study seeks to isolate the cause of hypothetical bias. Furthermore, a within-subject design is employed to prevent the infiltration of any individual specific biases. According to the results, hypothetical bias is non-existent with induced value goods but emerges once homegrown value goods are introduced. Hence the value formation process as hypothesized by Taylor, et al. (2001) is identified as a key contributor to hypothetical bias. The third essay explores a relatively new approach to non-market valuation that is based upon a prediction format. Unlike the traditional CV format that asks individuals to state personal values and opinions, this technique inquires about their predictions of other\u27s behavior. Literature in psychology regards these estimates to be less strategic, which could potentially eliminate biases including hypothetical bias. In this study we obtain hypothetical bias in both the traditional CV and the prediction formats. Although prediction estimates were significantly lower, it was observed that individuals are able to correctly predict the magnitude of hypothetical bias in the traditional CV