Author: Dr Johanna Choumert-Nkolo
15 years ago, EDI Global was among the first research organisations to fully embrace electronic data collection (computer-assisted personal interviewing – CAPI). Technology has always been fundamental to our organization to deliver high quality data and research services.
One of my favourite survey experiment papers is “Improving consumption measurement and other survey data through CAPI”.[1] I use this paper whenever I need to illustrate the increased benefit of CAPI for quantitative field research. The authors conducted a randomized experiment in Tanzania comparing paper and CAPI questionnaires. They found that 21% of households in the group interviewed with a paper questionnaire had an inadequate daily calorie intake, but only 8% using electronic methods. This finding calls into question the reliability of paper survey data.
CAPI surveys are more effective for a variety of reasons. For starters, CAPI enables automated management of filters (routing) during the interview, which means that questions that do not relate to a respondent based on prior questions are automatically disabled, and the use of consistency checks, which detect missing data, unusual values, and potentially conflicting or impossible responses throughout the interview. Second, CAPI may be used to gather a collection of quantitative information about how the data were acquired (paradata), such as timestamps and respondents’ GPS positions. These paradata enable fieldwork planning and real-time monitoring, as well as data quality evaluation, as shown in my paper here. Finally, CAPI may be used to collect additional information such as photos, videos, and audio recordings, which can then be instantaneously matched to the respondent’s information.
It is important, however, to consider the forms of resistance that respondents may present to this interview mode. Although adherence to the survey process is largely dependent on the quality of communication between interviewers and survey participants regarding the objectives of the study, privacy, data protection, and so on, it is critical to evaluate the potential resistance that respondents may exhibit in response to CAPI interviews. In their study in rural Uganda, Mercader et al. (2017) tested women’s willingness to participate in a CAPI survey and showed overall acceptance of CAPI. Similarly, Paudel et al. (2013) investigated the acceptability of utilising CAPI to gather data for the Nepal Demographic and Health Survey. Respondents were more inquisitive than scared of the technology, and there was a high acceptability of CAPI. Respondents were more anxious about the possibility of audio or video recording of the interview.
Growing mobile phone penetration, even in the most remote places, undoubtedly aids survey participants in becoming acquainted with new information technology and making tablet/smartphone use less unusual. Nonetheless, the survey administration approach must be adopted with the consent of the surveyed populations and must not endanger the interviewers. The function of the human element is vital in a sector that is more geared toward the use of new technology; before collecting data with tablets or smartphones, researchers should systematically analyse any concerns populations may have.
15 years after we started using CAPI software (SurveyCTO, Surveybe, ODK, Survey Solutions, KoBoToolbox, etc.), it is exciting to see how electronic data collection has become the gold standard in the field of development, with great software products being released over the years. We are proud to have contributed to this endeavour, not only via our own research projects in East Africa, but also by providing specialised training to other research organisations across the globe, as well as graduate students.
[1] Caeyers, Bet, Neil Chalmers, and Joachim De Weerdt. “Improving consumption measurement and other survey data through CAPI: Evidence from a randomized experiment.” Journal of Development Economics 98.1 (2012): 19-33.
