PERSONAL AND TELEPHONE INTERVIEWS: COMPARISON FOR TREND DATA (SURVEY, TREND, METHODS)

Telephone interviews have become a popular method for conducting attitudinal surveys and are now commonly substituted for personal interviews. Although there is a growing body of literature comparing the two methods, there is very little research which compares the two methods with respect to trend data. The problem at hand relates to a specific body of trend data collected annually through personal interviews over a 15 year period. Those responsible for the research needed to answer three questions. First, what are the relative advantages and disadvantages of the telephone and personal interview methods? Second, what is the best way to adapt questions previously designed for and asked in personal interviews for use in telephone studies. Third, can subsequent trend data be collected through telephone interviews?; Two simultaneous national studies were conducted in the spring of 1981--one used personal interviews while the second used telephone interviews. In terms of the relative advantages/disadvantages of the two methods, both techniques were compared on 23 factors. The results indicate that the two methods are comparable with respect to nine factors, telephone surveys have the advantage on seven factors, and personal interviews have the advantage with respect to seven factors. For two of four factors generally considered very important in any study--cost and researcher control--telephone surveys have a clear advantage over personal interview surveys. With respect to the other two important factors, sample coverage and response rates, the two methods are rated equally. In terms of adapting personal interview questions to the telephone, it is shown that there are no insurmountable problems for any of the eight question types tested. The question types include: open-end, visual six point scale, visual four point scale, visual three point scale, visual three part statement, rank order, short answer same context, and short answer different context. Finally, it is concluded that although there is a great deal of similarity between the two methods, it is not acceptable to report trend data across the two different survey methods. Rather, it is suggested that separate trend lines should be established in each survey mode.