Section IV. Interviewer Effects and Interview Context and Mode

Introduction and Background

Surveys that collect sensitive information often require the interview to be conducted in private; that is, only between the respondent and the interviewer. Having a third party present during the interview may affect the response process, leading to measurement error within and across samples (Aquilino 1997; Mneimneh, et al. 2015). Numerous investigations have looked at the effect of third-party presence on reporting sensitive information. Some studies found that third-party presence reduces the reporting of undesirable outcomes (Aquilino 1993; Aquilino, Wright, and Supple 2000; Moskowitz 2004), others observed increased reporting of such outcomes (Bulck 1999; Edwards, Slattery, and Ma 1998; Hoyt and Chaloupka 1994), and some showed no association between third-party presence and reports of sensitive information (Aquilino 1997; Pollner and Adams 1997). Recent work has revealed that the effect of third-party presence is not systematic; rather, it depends on the cultural background of the respondent and his or her need for social conformity (Mneimneh, et al. 2015). As Aquilino (1997) proposed in his framework on third- party effects, such variations in the effect are theorized to be driven by the nature of the information held by the third party. According to Aquilino's (1997) framework, the effect of third-party presence depends on whether the bystander is already knowledgeable about the information requested during the interview, and on the likelihood that the respondent might experience negative consequences by revealing new and unwelcome information to the bystander. Whether the respondent has disclosed such sensitive information to the bystander depends on the type of information (e.g. behavior, attitude, observable, not observable), the relationship between the third person and the respondent, and the respondent's need for social conformity.

There is also extensive literature examining the association between respondent and household characteristics and third-party presence. Male, older, married, and unemployed respondents who live in larger households are more likely to have a third person present during an interview (Aquilino 1993, 1997; Hartmann 1995; Mneimneh, et al. 2018a), and respondents with lower levels of education and income are more likely to have a spouse present (Aquilino 1993, 1997; Mneimneh, et al. 2018a). The level of income of the country where the interviews are conducted also seems to be related to the interview setting; non-private interviews are more common in low- and middle-income countries than in high-income countries (Mneimneh, et al. 2018a). However, very little research discusses the role of the interviewer in establishing a private setting. In this chapter, we focus on the interviewers and their contribution to achieving interview privacy.

Interviewers are key in enforcing the privacy requirement that is part of the protocol for many surveys. Yet achieving and maintaining privacy can be difficult. Interviewers are essentially guests, and household members do not normally expect them to dictate the interview setting. In spite of the difficulty of the task, many study protocols only require interviewers to conduct the interview in private without providing interviewers with specific instructions or details training material on how to request and achieve interview privacy (Aquilino 1993; Mneimneh, et al. 2018b; Smith 1997; Taietz 1962). Thus, interviewers have to apply their own judgment and skills in enforcing the privacy requirement for the interview.

With such minimal training, one would expect interviewers to vary in their attitudes toward the importance of establishing and maintaining privacy and their ability to do so. In fact, a recent publication (Mneimneh, et al. 2018a) found the estimated between-inter- viewer variance in interview privacy to be significant. Yet, very little is known about what interviewer characteristics predict the privacy setting of the interview. Only one known study has investigated the relationship between interviewer characteristics (age, gender, and years of experience) and interview privacy, showing that more experienced interviewers achieved higher rates of interview privacy than less experienced interviewers in three out of the five surveys analyzed from different countries (Lau, et al. 2017). However, the study did not assess between-interviewer variance in interview privacy or the contribution of interviewer characteristics to such variations. Moreover, we know of no studies that investigate whether an interviewer's own opinions and attitudes toward privacy contribute to the between-interviewer variation. Measuring between-interviewer variation in interview privacy and the contribution of interviewer characteristics to such variation is essential for identifying modifiable factors for interventions aimed to reduce the presence of a third party and for designing training materials or protocols to aid interviewers in establishing privacy.

This study uses data from a national face-to-face interviewer-administered study in the Kingdom of Saudi Arabia (KSA) to: (1) investigate what interviewer characteristics predict third-party presence during the interview; (2) quantify the between-interviewer variation in third-party presence; (3) estimate the relative percentage of between-interviewer variance (in third-party presence) that is explained by different types of interviewer characteristics; and (4) investigate the effect of third-party presence on reporting desirable and undesirable attitudes and behaviors.

Given the collectivist nature of the culture in KSA where the self is defined in terms of the relationship with others and where harmony is maintained by paying close attention to others in the social context (especially among family and friends) (Smith, Bond, and Kagitciba.si 2006; Triandis 1995), investigating interview privacy predictors and the effect on reporting is highly relevant.