Arrow et al. (1993) issued guidelines for contingent valuation, one of which was the recommendation that valuation questions include a ‘no-answer’ option such as ‘I prefer not to vote (PNV)’ in addition to the typical ‘yes’ and ‘no’ options. However, they did not give further guidelines on what to do with such responses, and there are various opinions in the literature. The objective of this thesis was to identify factors that affect the probability of a respondent choosing the PNV option in stated preference surveys. This thesis identified a positive relationship between offered bid for the proposed environmental project and the probability of respondents choosing PNV and a negative relationship between perceived consequentiality of the survey and the probability of respondents choosing PNV. From the findings, this thesis suggests possible solutions to reduce the frequency of such responses in order to increase accuracy of welfare estimates and cost efficiency.