Predicting the Response Rates of a Survey from the type of Presentation Made by the Interviewers

 How does the behavior of a CATI operator at the beginning of a call influence the response rate of the respondents?

The article we are considering today analyzes part of the data from the Wisconsin Longitudinal Study (WLS, 2005) of the University of Wisconsin at Madison, which since 1992 has collected audio data from thousands of telephone interviews (CATI). Specifically, the article focuses on 257 pairs * of audio tracks corresponding to the year 2004, studying the relationships between the interviewer's and respondent's approaches to predict the response rate of a survey.

The starting assumption is that it is possible to predict participation rates in a telephone survey by evaluating the interviewer's approach in the first moments of a call, both in terms of interaction between the caller and the responder, and in terms of form and nature of the prosodic and acoustic elements on both sides.

* The study has isolated two different approaches of the telephone operator, one defined "canonical" and the other "efficient".

Results of the study

The analysis of the long historical series of the WLS has allowed us to isolate some elements that allow the implementation of training courses for telephone operators with the aim of producing greater awareness of the behavior to keep in the first moments of a telephone call: the first sentences pronounced by the interviewer have indeed a great impact on the response rate of the survey, as well as length and type of respondent's greeting are able to provide elemnti for the prediction about the participation of the same to the survey.

Calls have been classified according to whether they start in an "efficient" or "canonical" manner.
As canon the interviewer introduces himself and introduces the topic of the investigation to the person who answers the call. If the respondent is not among the eligible units, the interviewer will then have to repeat the presentations to a second subject (if present in the contacted house) who possibly falls within the sample. On the contrary, in the method efficient the interviewer sends back the presentations and assures himself before speaking with an eligible subject, presenting himself, therefore, only once.

In spite of what the definition would imply, the probabilities of participation are substantially and significantly lower adopting the "efficient" method, and this for the reason that in the case of a canonical presentation the identification and cognition of the object of the call takes place on a basis of reciprocity, while in the case of an effective approach it takes place on the basis of a difference in level between interviewee and interviewer (against the interviewee). Therefore a canonical approach is to be preferred even if it is good for the interviewer to present himself before purchasing, that the respondent asks who is calling him. In an optimal introduction, in fact, the respondent should be led to ask questions about the subject of the questionnaire or its length and not the reasons for which the questionnaire is given (the "why" questions in English), questions, this last, which reduce participation rates.

The effectiveness of different levels of courtesy in the beginning of a call and it has been possible to verify how only very polite introductions bring an increase in the rate of replies, while presentations that are too informal and friendly involve decreases of the respsonse rate. In fact, formulas like "hello" result more effective respect to greetings such as "hi", because the first is perceived as more courteous and formal (in some ways the equivalent in Italian of "salve" and "hello"). In the same way, too long and detailed presentations decrease the rates of participation in the survey.

Again: the chances of participating are more if during the initial bars of the telephone dialogue the interviewer adopts a cadenced speech, which leaves room for short breaks in the speech, allowing the interviewee to process it and assimilate it better.

Finally, with regard to the behavior of the respondent, the greater the duration of the greeting of the sample subject more are the possibilities that he participates in the survey, meaning only the first word (hello).

Our revision from the article:  Greeting and response: predicting participation from the call opening
by Nora Cate Schaeffer, Bo Hee Min, Thomas Purnell, Dana Garbarski, Jennifer Dykema

Leave a comment