Voice Coding: Higher efficiency with CATI interviews via audio-file based coding

Abstract

With conventional CATI studies, interviewers enter the respondents’ answers to open-ended questions via a computer keyboard. In this process, interviewers make mistakes and abbreviate or paraphrase responses due to time constraint. As a result, a considerable amount of information and nuance is lost in the transformation and analysis of "verbatims".

The new Voice Coding technique seeks to remedy this deficiency by solely using digital audio files containing the recorded original verbatim responses – instead of transcriptions - for coding. In addition to attaining better authenticity in the analysis, this approach should allow researchers to elicit a greater amount and diversity of responses. As no valuable call time is wasted with typing, Voice Coding interviews are likely to be shorter and, as a consequence, to yield fieldwork cost savings. In May 2005, Psyma conducted an automotive CATI study in Germany to test to what extent the assumed benefits of Voice Coding can be materialized in a real-life setting. The results confirm that Voice Coding produces a greater quantity and variety of answers to open-ended questions whereas the reduction in interview length has not been significant.

However, when Voice Coding is used more extensively, interviewers will gain experience in applying the technique so that learning curve effects will arise. Consequently, it is fair to assume that the expected quality improvements and cost reductions can be fully achieved in the long run, leveraging a substantial increase in the overall efficiency of CATI interviewing.