The main aim of our study was to develop an eective and cost-efficient calendar recall aid for a retrospective questionnaire to be used in a large online panel. Calendar instruments, such as the Event History Calendar, have been shown to improve data quality in face-to-face and telephone surveys, ensuring higher consistency, completeness, and accuracy of retrospective reports. They combine two important aided recall techniques, namely temporal bounding cues (or landmark events) and visual feedback of the respondents answers (i.e. domain timelines). Using an experimental approach we investigate which of these components contribute the most to the calendars positive eects on data quality. We optimize the calendar method by identifying its most eective elements and eliminating redundant or ineective cues, while taking into account the requirements of online survey methods in terms of clarity and user-friendliness.

In our paper, we present the results of a split-ballot experiment in which we tested three versions of a calendar recall aid and compared them to a regular standardized questionnaire. Approximately 2000 participants of an existing online household panel were randomly assigned to four conditions. Respondents in condition 1 (control) filled in a regular online retrospective questionnaire. Respondents in the experimental conditions lled in the same questionnaire with either visual feedback but no public or personalized landmarks (condition 2: visual feedback), a timeline with public and personal landmarks (condition 3: landmarks only), or a standardized questionnaire with both, visual feedback and landmark events from the reference period (condition 4: full calendar). We compare the data from the four conditions with regard to the total numbers of reported transitions, completion time, number of self-corrections, refusal and break-o rates, and respondent evaluations of the method.