AbstractPrevious research demonstrated that respondents assume that the range of precoded response alternatives reflects the researcher's knowledge of the distribution of opinions or behaviours in the population. This assumption may influence respondents' reports in two ways: respondents may either use the range of the response alternatives as a frame of reference in estimating their own behavioural frequencies, or they may be reluctant to report frequencies that appear extreme in the context of the scale. Three experiments using reports of mundane behaviours, namely watching TV and drinking beer, were conducted to differentiate between the frame of reference and the self‐presentation hypothesis. The results of all studies favour the frame of reference hypothesis, and suggest that the impact of response alternatives is the more pronounced the less episodic information about the behaviour is accessible in memory. Specifically, proxy‐reports were found to be more affected by the range of response alternatives than self‐reports (Experiments 1 and 2), and respondents with dispositionally low access to self‐related information were found to be more affected than respondents with dispositionally high access to self‐related information (Experiment 3). Implications for questionnaire construction are