ABSTRACT

This chapter discusses the types of Internet survey paradata, some of their uses, how they can be collected, and how they can be analyzed. First-level paradata describe individual events within an Internet questionnaire, for individual respondents. Second-level paradata aggregate first-level paradata across a number of actions for a single respondent. Third-level paradata aggregate data across respondents or across variables. Traditionally, a distinction has been made between server-side paradata and client-side paradata when discussing Internet survey paradata. A combined use of response latency analysis with answer changes might tap into most of the four steps of the cognitive question–answering model. Paradata should therefore be considered by survey researchers interested in assessing the quality of their survey questions and questionnaires. First-level paradata are collected in a relatively unstructured way. Data are collected by adding each action to a data string or as separate observations in a relational database.