It is noted that the ideal design (Fully interpenetrated) provides for the random assignment of the sampled individuals to the interviewers. It is difficult to implement, more often a drawing is used quasi-interpenetrated, which provides for the assignment of interviews to active operators at a given moment. In the investigation face-to-face generally the areas are assigned to the interviewers, making it impossible to distinguish the effect of the area itself from the effect of the interviewer. To estimate the interviewer's effect, adjacent areas are randomly assigned to 2 interviewers (partially-interpenetrated designs).
Population coverage
it is noted that interviewers often confirm the addresses present in a list, without adding or removing new names. Moreover, difficult contacts are more likely to be classified as ineligible by interviewers.
Response rate
Socio-demographic characteristics
Interviewers with another job tend to work more at weekends and have a higher response rate.
The majority of the studies do not detect differences due to sex, a small minority identifies a higher response rate among the interviewers. The same result applies to the age in which only a minority of studies find a positive effect.
On the contrary the matching between the socio-demographic characteristics of the respondent and the interviewer it works but is difficult to implement in large studios. Interviewers who are perceived as belonging to a non-disadvantaged class struggle to gather information on income while good-looking interviewers induce women not to declare psychiatric symptoms.
Interviewers behavior
The response rate is sensitive to how respondents stand, in particular those who show self-confidence and / or positive attitudes are more successful than others. Those who manifest a wide-mindedness unexpectedly are less successful.
It should be noted that the interviewers who introduce themselves at the beginning of the interview and make a concise introduction, using "kind" output formulas ("I'm sorry I'm not interested, can I recall later?" For example), which they adapt their approach to the respondent have a higher response rate. The effect of the experience is not clear, the results are conflicting in particular because it is measured unevenly in the studies examined.
Measurement error
Among the factors that can influence the response is the ethnicity of the interviewer (particularly relevant to sensitive questions, respondents avoid offending the interviewer with their answers). The perception of the interviewer of the color of the skin also varies according to his ethnicity.
Il matching between the characteristics of the respondent and the characteristics of the interviewer, positive in terms of response rate, does not have a clear effect in terms of measurement error. Men interviewers are more suitable to collect data in the medical field because they are perceived closer to the authority of a doctor or, for research on drugs, more predisposed to use them.
The effect of age on measurement error is debated: the acquired experience and the authority on the one hand, the difficulties of understanding greater on the other contribute to muddy the results.
A more personalized interview has an effect on the answers but can be beneficial when it is introduced. In general the authors, to reduce the distortion in the answers, recommend using a standardized style but using a more flexible approach, with more explanations only for questions related to complex factual characteristics. In interviews face-to-face the answers can also be influenced by the physical presence of the interviewers alone
Length of the questionnaire
Young, male or more experienced interviewers tend to finish the interview in less time; giving preference to the number of answers on the quality of the interviews leads interviewers to suggest to respondents shortcuts (how to respond less times in an affirmative way) to conclude the interview faster and collect more interviews
Finally it is noted that a more experienced interviewer tends to provide richer answers in open questions.
Taken from:
Explaining Interviewer Effects: A Research Synthesis
In the Journal of Survey Statistics and Methodology, volume 5, number 2, June 2017 West and Blom make an analysis of articles from the literature on the effects of interviewers in telephone surveys.
