Читать книгу Researching in the Age of COVID-19 Vol 1 - Группа авторов - Страница 15

The authors recognize that these preliminary findings have certain limitations, but also highlight some interesting results. Contrary to protocols prescribed by prior studies, completion rates peak between the third or fourth call attempt and yield marginal benefit thereafter. Furthermore, the efforts taken in making nine attempts for each respondent is not commensurate with the completion rates of surveys. This finding is especially salient for studies in which time, personnel and financial resources are key constraints. If the call attempt protocol was recalibrated to move on to the next respondent after the third call, the host studies would have completed the same number of surveys by making 14 per cent fewer calls.

The study shows that follow-​up calls, even after accounting for study fixed effects, yield 25.6 per cent higher response rates than first-​round surveys. Intuitively, this could be explained by the fact that, in the instance of a follow-​up call, a degree of goodwill and trust has already been established between respondent and researchers. Additionally, in the case of follow-​up calls the risk of calling invalid numbers is greatly reduced. This finding suggests that establishing some form of contact with the respondents before the CATI survey (for example, an introduction from a local or trusted institution) could increase response rates. This hypothesis will require further research to be confirmed. Finally, it is worth noting that, if a survey is sensitive in nature, this effect could be null (for example, GBV), signalling understandable reluctance to answer certain questions over the phone. This trumps the positive effects of any pre-​established rapport –​ something that needs to be carefully considered when choosing the survey methodology.

The study also reveals that audio audits can stifle survey completion rates, due to a higher refusal rate in comparison to manual back-​checks. This result remains consistent and significant after accounting for state and enumerator fixed effects, further reinforcing the importance of enumerator training in rapport building while conducting surveys.