Processes used in the formulation of behavioral frequency reports in surveys
Information on consumers' frequencies of behavior is commonly desired in marketing research. This research concerns processes used by survey respondents in formulating responses to behavioral frequency questions. This research rejects the common presumption that respondents always recall and enumerate behavioral events and examines the processes reported by survey respondents in formulating frequency estimates, the impact of task manipulations on reported processes, and the relationships of task variables and response formulation processes with response accuracy. Hypotheses were tested in three studies. Results from each of the three studies indicate that a variety of processes are used to formulate responses to questions about behavioral frequency in survey settings. Task characteristics that affected reported processes included the number of events of interest, question time frame, and time given for response. The behavior itself also appeared to be a task variable which might affect response formulation processes. Overall, results indicated that the processes used by consumers to answer questions about behavioral frequency are likely to differ for durable and non-durable goods. The conventional model, focusing on errors of omission and telescoping, seems appropriate for durable goods purchases. For consumer non-durables, an alternative model is needed. The size of response errors at the individual level are large, emphasizing the need for this alternative model and demonstrating the advantages of using mechanical data sources (such as scanners) to measure purchase frequency when possible.