Gender Prespecified Sampling for Cost Control

Abstract

Nationally representative surveys administered in the Middle East and North Africa typically are conducted using methodological techniques developed from outside the region. Sometimes these best practices require modification for local contexts, and one common—but costly—adaptation is to use teams of male and female interviewers for face-to-face surveys. We address the trade-off between costs and quality by testing a sampling method based on gender matching of interviewers and respondents. The benefits are twofold: (1) a reduction in survey costs and (2) simplified withinhousehold selection. We find a notable reduction in field costs when field-tested in Qatar. Such a sampling method could be exported to other countries where societal conditions make teams of interviewers necessary for face-to-face surveys. Social science research using nationally representative surveys has entered a period of rapid expansion in the Middle East and North Africa (MENA). Many of these surveys are conducted face-to-face using survey practices developed outside the region, and in some instances, these techniques may not be sensitive to societal conditions in these countries. Without accounting for these contextual factors, survey designers risk decreasing the representativeness of the sample through sampling and nonresponse errors. In this article, we focus on gender matching of interviewers and respondents and its implications on survey practice in the region. In the United States and Europe, it is not uncommon for interviewers and respondents to be of the opposite sex, and interviewer effects as a result of gender have been found to influence survey items based on women’s issues and gender equality (Catania, Binson, Canchola, Pollack, Hauck, 1996; Huddy et al., 1997; Kane & Macaulay, 1993). Researchers are aware of these effects and account for them when analyzing data, but in general they are not concerned about increased refusal rates or lack of cooperation based in part on an interviewer’s gender. However, this is a central concern for data collection in Muslim societies, where it is not considered appropriate for female respondents to speak with a male nonfamily member, let alone a male interviewer. Not surprisingly, a recent survey conducted in Qatar showed that a clear majority of female respondents would rather be interviewed by a female interviewer (Alemadi et al., 2010). In the same study, male respondents, in comparison, were not as bothered by the presence of a female interviewer. About 55.2% of Qatari male respondents preferred a male interviewer compared with 2.2% of Qatari female respondents. The same question was asked among nonnationals and the split is almost the same: 54.5% of males prefer a male interviewer, whereas 5.2% of women prefer a male interviewer. One way to address the issue is to hire only female interviewers as they can interview both male and female respondents. However, in several MENA countries—and particularly in the Arabian Gulf—many women refrain from driving and generally have limited access to transportation owing to cultural norms or restrictions (e.g., it is illegal for women in the Kingdom of Saudi Arabia to drive). Thus, hiring only female interviewers becomes challenging unless reliable and affordable transportation is available. An alternative fieldwork technique commonplace in the region is to have teams of interviewers in the field. The team can include one of each gender, one male and two female interviewers, or one male and three female interviewers. The male interviewer in the team is usually responsible for driving while female interviewer(s) is(are) usually responsible for conducting the interviews. For example, the Qatar Statistical Authority (QSA) usually uses a team of one male and three female interviewers in the field, and the Pan Arab Research Center (PARC) has used teams composed of one male and two female interviewers in past surveys. While this method addresses the issue, it increases significantly the field cost as more interviewers are needed during the fieldwork. In this article, we propose a sampling method with prespecified respondent gender to reduce the field costs. The basic idea is that before interviewers visit the households in the sample, we randomly specify the gender of respondent for every sampled household. Knowing the respondent gender beforehand, male interviewers (instead of teams of interviewers) can be sent to households with male respondents. The field costs for households with male respondents, which usually account for half of the sample, can be significantly reduced. In addition, the respondent will be selected from household members of the same sex (instead of all household members), as the gender of the respondent has been prespecified. This will help reduce the survey length and offers a better control of respondent gender during the within-household selection process. The new sampling method is applied to a nationally representative survey in Qatar, a small and wealthy emirate in the Arabian Gulf. We compare this survey’s operations to those of a previous survey and find a 27% reduction in field costs. We do not find any significant difference in respondents’ characteristics (e.g., gender ratio and the percentage of young people) between the two surveys. Not only is this method directly applicable to many countries in the Arabian Gulf and broader Middle East, but it is also useful in other societies where cultural or personal safety concerns make it prudent to field teams of interviewers for face-to-face surveys.