Interviewer effect

With the advent of the first mass opinion polls, researchers began to express concern about the mistakes made by interviewers in interviews. However, until the late 1920s and early 1930s, no one focused on such errors. The first scientific study on the effect of the interviewer can be considered the work of S. Rice, published in 1929 ^[1]. In it, the author clearly demonstrated the danger of biases caused by the personality of the interviewer, and drew attention to the need to neutralize them.

The second significant surge of interest in the problem under discussion dates back to the 1970s and early 1980s, when the effects of the interviewer in opinion polls after a long period of relative oblivion again began to attract the growing attention of researchers. ^{[2] [3]}
Since the second half of the 1990s. the temporary decline in research activity on the issue of the interviewer effect is replaced by a new increase in attention to it, and the special literature published since then indicates that the wave of interest in this topic does not stop. ^[four]

The effect can be manifested, on the one hand, in the positive aspects of the interview - the ability to position the respondent in a conversation and get more information on the problem being studied. However, on the other hand, the effect of the interviewer can have a significant impact on the bias of the answers and the representativeness of the data from a sociological study. ^[5]

To explain the effect of the interviewer, a “social attribution” model is used. According to this model, respondents ascribe to their interviewers certain values ​​and norms, expectations and attitudes, constructed on the basis of the minimum empirical information available to them during the survey. Sources of information are the socio-demographic characteristics of the respondents (gender, age, ethnicity, and sometimes socio-economic status and education) and some other signs associated with them (appearance, speech, emphasis, manners). These characteristics serve as distinctive clues to respondents, indicating the nature of the interviewers' expectations regarding acceptable or permissible verbal behavior. Trying to adapt to these norms and expectations, the respondents edit their answers in order to avoid discomfort and thereby protect themselves, controlling the impression made on the interviewer.
However, not only objective, but also subjective characteristics of interviewers (their opinions, attitudes and expectations) influence the behavior of respondents and determine the conformal nature of their answers. Thus, explanations of the effects of the interviewer are also based on ideas and propositions formulated within the framework of theories of integration , social desirability, self-control and conformity .

Interviewer Effect Varieties

The following types of interviewer effects are distinguished ^[5] :

• direct: related to the interpretive activities of interviewers and other types of their spontaneous behavior (subjective interpretation of the answer, “forming your own text”, various kinds of clues, comments and other additional speech inclusions of spontaneous origin);

Negative manifestations of the direct effect of the interviewer are, firstly, selective perception, misunderstanding and, therefore, inaccurate recording of the respondent's answers (the interviewer interprets the answers with a vague position as close to his own convictions and fixes the answer in this form), and secondly , “Stereotype of recognition” - after several interviews, the interviewer becomes confident that he already understands from what the respondent’s first answers what type of subjects he is, and how and this category is typically answer questions. In this case, again, the interviewer can record not so much what the respondent actually answers, but what he intended to hear in advance.

• indirect: related to the communicative nature of the interview and the influence of socio-demographic, psychological and behavioral characteristics of the interviewers on the process of forming responses by respondents. The effects caused by integration and self-presentation mechanisms are not obvious, and therefore more dangerous. ^{[6] [7]}

Some authors believe that biases associated with the influence of interviewers are in fact so “modest” and “trivial” that they are hardly worth paying serious attention to. Therefore, in everyday research practice this danger is often ignored. ^[8] Many researchers usually proceed from the implicit assumption that the effect of the interviewer does not exist. However, this source of error significantly affects the explained variance of the studied variables and the validity of the measurements. According to an experimental study by R. Touranjo, the differences between interviewers in a telephone interview "increased the variance of sample estimates by about 50%." In a personal interview, this figure can be much higher. ^[9]