THE IMPACT OF NUMBER OF CALLS AND PARTNERS PRESENCE IN TELEPHONE SURVEYS

Our analysis of data from the Survey on Sexual Behaviour in France (AC SF) is focused on two possible sources of bias in replies to sensitive questions put in telephone surveys. How many times should respondents be called to limit the bias caused by the tendency of members of a pa rticular group being.difficult to contact? Respondents who were contac ted after several calls tend to belong to groups in which multiple sex ual partners are more common, and where the number of sexual partners is larger. In the case of such respondents, dialling more than 12 time s to collect information has proved valuable. We also aimed to determi ne whether the presence of the respondent's sexual partner at the time the call is made influences the responses. In these circumstances, dr ug use tends to be acknowledged less frequently, and the declared numb er of past sexual partners is smaller. Responses in telephone surveys are, therefore, affected by the partner's presence, when the questions relate to longer periods, or to periods which preceded the partnershi p.