THE IMPACT OF VARYING REFERENCE PERIODS IN SURVEY QUESTIONS ABOUT MEDIA USE

Data from a representative, national sample of American adults are used to examine the impact of varying reference periods (the ''past week'' as opposed to a ''typical week'') in survey questions about mass media use. Results from a split-sample experiment comparing the alternative question wordings suggest that providing respondents with the more specific and recent time period (i.e., the '';ast week'') results in significantly lower overall reports of usage ''ross a variety of media. Results further suggest the potential atypicality of the narrower time period, which might in principle adversely affect the validity of responses, is not a serious concern.