Unintended behavioral consequences due to repeated measurement

This dissertation deals with the various ways how retrieving information from respondents in surveys may affect their future reporting behavior and/or actual behavior. Such effects, called respondent conditioning, can take various forms. For example, respondents of a panel survey may recall having answered the same questions before and report differently in future waves of the same survey. Similarly, responding to questions about certain behaviors may work as a stimulus that encourages some respondents to change their behavior. I provide a framework for the analysis of such conditioning effects and study their various forms in several social science surveys