Telephone versus postal surveys of general practitioners: methodological considerations.

BACKGROUND

High response rates to surveys help to maintain the representativeness of the sample.

AIM

In the course of a wider investigation into counselling services within general practice it was decided to assess the feasibility of increasing the response rate by telephone follow up of non-respondents to a postal survey.

METHOD

A postal survey was undertaken of a random sample of 1732 general practitioners followed by telephone administration of the questionnaire to non-respondents. The identical questionnaire was administered by telephone to a separate random sample of 206 general practitioners.

RESULTS

Of 1732 general practitioners first approached by mail, 1683 were still in post of whom 881 (52%) completed the postal questionnaire and a further 494 (29%) the telephone interview. Of 206 general practitioners first contacted by telephone, 197 were still in post of whom 167 (85%) completed interviews. Compared with doctors first approached by mail, those first approached by telephone were significantly more likely to report having a partner with a special interest in psychiatry (P < 0.01); and a general practitioner, practice nurse or health visitor who worked as a counsellor (P < 0.01 in each case). A comparison of doctors first approached by telephone with those who completed telephone interviews after failing to respond to the postal questionnaire showed that postal non-respondents were significantly less likely to report having a general practitioner, practice nurse, health visitor or community psychiatric nurse who worked as a counsellor (P < 0.01 in each case).

CONCLUSION

These findings suggest that non-response to the postal survey was associated with lack of activity in the study area. Telephone administration of questionnaires to postal non-respondents increased response rates to above 80% but, as telephone administration enhanced the reporting of counsellors, a social desirability bias may have been introduced.