The key long-term objective of this application is to examine an overlooked source of measurement error in less developed countries (LDCs). The standard norm in contemporary data collection and data editing practices in LDCs is to privilege """"""""strangers"""""""" over """"""""insiders"""""""". That is, data are collected and edited by persons who have no prior knowledge of respondents, their families or their communities. Contrary to this approach, this study tests a simple hypothesis that is deeply rooted in different areas of social theory. Specifically, insiders collect more valid data, especially in developing country and middle income settings. There is already intriguing evidence that data quality could be improved in these settings if researchers were to rely more on """"""""insiders"""""""". This study will systematically evaluate this hypothesis for the first time by carrying out a carefully designed survey data collection and data editing experiment in the Dominican Republic. The proposed research design experimentally assigns female respondents to be interviewed by female interviewers whose level of familiarity or """"""""insiderness"""""""" with the respondent is known a priori. A face-to-face interview is followed by a self-administered questionnaire on sensitive topics. In addition, data collected from a number of sources-the project's own measurements, government-issued documentation and administrative records-will allow the researchers to assess the validity and reliability of survey responses across a range of topics. The statistical analyses will: (a) test the relationship between insiderness and the validity and reliability of survey responses for multiple substantive variables considered important in international social research, including child school enrollment, anthropometrics and vaccination records, family planning, standard of living, and economic flows;(b) evaluate whether the effect of insiderness on both the validity and reliability of survey responses depends on the degree of sensitivity of the questions;(c) assess whether insiderness effects on data quality are reduced when self-administered questionnaires are used to increase confidence in response confidentiality;and (d) explore how characteristics associated with respondents, interviewers, and the interview process itself affect our measurement of response validity and reliability. The design also includes an experiment that will make it possible to evaluate the relationship between insiderness and the ability to accurately solve two types of data problems: inconsistencies and missing values. In view of our preliminary studies, the present application offers distinct promise that by leveraging the insiderness of local interviewers the quality of health-related data-and socioeconomic and demographic data more generally-collected in LDCs can be improved. Since in most LDCs health statistics depend on social surveys, this research could lead t very significant improvements in public health surveillance, planning and program evaluation.
In many countries, particularly poorer and middle-income ones, information about the general population's health or socioeconomic status, as well as information used to evaluate the effectiveness of new policy interventions, is primarily collected through social surveys. Our proposed project experimentally evaluates, for the first time, a method of data collection which could significantly improve the quality of survey data: using local interviewers who have a preexisting relationship with respondents instead of outsiders. If our hypotheses are proved correct, this method will provide a simple way to increase data quality in many areas of the world, a fact which, in turn, should significantly improve the design and evaluation of policy, including health policy.
| Sana, Mariano; Stecklov, Guy; Weinreb, Alexander A (2016) A test of the stranger-interviewer norm in the Dominican Republic. Popul Stud (Camb) 70:73-92 |
| Stecklov, Guy; Weinreb, Alexander A; Sana, Mariano (2015) Family Planning for Strangers: An Experiment on the Validity of Reported Contraceptive Use. PLoS One 10:e0136972 |
