Health Education Research, Vol. 4, No. 3, 321-328, 1989
© 1989 Oxford University Press
research-article |
Comparison of different data collection methods within a study sample: telephone versus home interviews
Institut National de la Santé et de la Recherche Médicale (INSERM), Unit 21 16 ave. Paul-Vaillant-Couturier, 94807 Villejuif Cedex
1INSERM, Unit 263 2 Place Jussieu, 75251 Paris Cedex 5, France
2California State University Long Beach, CA, USA
More than one method of data collection is often necessary in order to have a sample of patients that is both representative and large enough to be meaningful. There is always the concern that different interview methods will introduce a bias in responses. In the final phase of a three-year study, 318 inner-city hypertensives were interviewed by telephone or, if that was not possible, in person. Patient-reported data were compared using discriminant function analyses to detect differences in responses by the two interview methods. Analysis showed that telephone interviews were of shorter duration than home interviews and that the combined method was less costly than an earlier home interview study of half the same cohort. No significant differences in the two interview methods were found on the basis of population distribution characteristics, completion rates, or on any of the 17 classifications analyzed. The authors' findings demonstrate that the use of more than one method of data collection with the same sample, while facilitating the augmentation of the response rate, will not necessarily bias the study results.