Unstructured interviews take the form of conversations where there are no predetermined questions. Although unstructured interviews allow for more valid data by analyzing sensitive topics and complex issue in depth, they may not readily translate into statistical data unlike structured interview in which answers can be expressed quantitatively.
The issue addressed in the question refers to a situation when the researcher inadvertently biases the results; a common problem with both structured and unstructured interview. The idea of ‘social construction’ is that interviewer conducts the interview in such a way so as to generate the answers he/she is looking for. This may be intentional or more commonly, unintentional. As the selection of respondents and responses are usually chosen to fit the initial hypothesis of the interviewer, so much so as the wording of the questions will affect the response. Firstly, language is an imperfect means of communication. There is also an added risk of the use of ‘more aggressive conversational tactics’ as advocated by Howard Becker. For example, the interviewer may bully the interviewee into providing a satisfactory answer by constantly rephrasing the question.
Since the interviewer chooses the questions to be asked in the first place he/she may have already imposed his/her views on the respondent’s answer. The interviewer also decides what is not asked and may therefore overlook certain issues. If responses do not comply with the initial hypothesis how much weight will the interviewer give to those responses? In this way the interviewer is ignoring subtle nuances, which largely affect the validity of the data, as it isn’t an exact account of what the interviewee has said or has intended to say.
The ethnicity, background and gender of the interviewer have often been shown to have a dramatic affect on the interviewee. Women may not feel comfortable talking to male interviewers about certain issues e.g. pregnancy. Likewise, an interviewer from a middle class may affect the response of a criminal from a lower class. The criminal may respond in a way to comply with the interviewer’s value thus triggering interviewer bias (prejudice on part of the interviewer whether intentional or unintentional, thus affecting the outcome of the interview) Interview bias is illustrated in Stuart A. Rice’s study in 1914. Two thousand destitute men were asked to explain their situation. Those interviewed by a supporter of prohibition were more likely to blame their problems on alcohol, whereas those interviewed by a committed socialist were more likely to explain their plight in terms of their industrial situations. Respondents may also mistrust the interviewer or lie. It is therefore important to establish a rapport with respondents before conducting the interview.
Perhaps more a more informal interviewing technique can be adopted with a non-directive approach by refraining from offering opinions or expressions of approval or disapproval. Informal interviews can address very sensitive issues and can lead to discoveries into new areas that the interviewer may have not taken notice of before the interview was conducted. Such interviews ensure honesty and validity. Another alternative can be the approach adopted by the feminist researcher Ann Oakley (1981) in her study of childbirth who believed that the interviewers should be empathetic towards the interviewees.
The usefulness of interviews is undermined by the fact that they are socially constructed. As interviews are based on social interaction it is hard to completely overcome this problem. Although most bias is unconscious, one way of overcoming it is by not seeking answers which sociologist wish to obtain. The usefulness of interviews is largely affected by the questions asked and interpretation on part of the sociologist. It is therefore important that interviewers conduct the interview, which total objectivity so that the validity of responses is not affected by outside sources. For this reason, well-trained interviewers should conduct interviews. Perhaps even a combination of methodologies can overcome this problem e.g. a combination of participant observation and unstructured interviews. .