Furthermore, James Holstein and Jaber Gubrium argue that respondents in group interviews are more reflexive than in individual interviews due to multivocality. This is because respondent are able to think more deeply about their answers and reflect critically upon them in their responses to others’ answers. Hence, both forms of interview are seen to be high in validity.
However, it is criticized that data produced in interview may be neither valid nor reliable. This is because the responses given may not be accurate and may not reflect true behavior due to forgetfulness or dishonesty. In fact, it is argued that interview is an artificial situation. Cicourel has questioned if people will act according to their stated beliefs in the interview. There is certainly no guarantee to this because people are capable of saying one thing and doing another. Hence, data produced interview may not be valid.
Furthermore, interviewers may be deceived by the respondents especially those who are good at deception and manipulation. For example, some criminals interviewed by Laurie Taylor later claimed they had made up fanciful stories about their escapades in order to see how gullible Taylor was. This has resulted in data less valid than that of participant observation.
Nonetheless, theoretically, the data collected in interview is still considered more valid than that of questionnaire. This is because, through face to face interaction, interviewer may be able to detect lies by judging the facial and body language of respondents. This would prevent respondent from lying or misleading the researcher. Hence, interview is seen to be more valid than questionnaire.
Besides that, interview has the advantage in the sense that the interviewer could probe deeper to gain in-depth information, especially through unstructured interview. This can be seen in Betty Friedon’s study of women’s satisfaction as housewives. Though, this would not be the case for structured interview as the interviewer is not allowed to deviate from questions provided. Hence, data collected in unstructured interview is more valid than of structured interview.
Moreover, the fact that in interview, interviewer is presence to prompt replies and explain answers, clarification of the concepts and words is possible. This can prevent misinterpretation of questions or answers that would otherwise happen in questionnaires. Despite this, researcher impositions are still unavoidable in structured interviews as interviewer has set limits to what respondent can say by having fixed choice answers. This in turn lowers the validity of structured interview.
On the other hand, the objectivity of data collected in interview can be affected by interviewer effect. The answers given may be influenced by the way the interviewer define the situation. This can be seen in William Labov where young black American children responded differently when interviewed in different context.
A further problem with unstructured interviews where there is more opportunity for the interviewer to direct the interviewee towards giving certain types of response. Consciously or unconsciously, respondents might give answers that they believe the interviewer wants to hear, rather than saying what they truly believe. This is known as interviewer bias. Therefore, it is argued that interview produce results of less objective as compared to questionnaires.
Besides, although the numerical data collected in structured interview is easy to be quantified, it is not that reliable as compared to questionnaire and experiment. This is because interview is an interacting process which cannot be recreated. Hence, the data collected would be unreliable.
Nonetheless, interview has the practical advantage in that larger sample size can easily be acquired in a shorter time as compared to participant observation. For example, howard s. becker’s study of marijuana consists of 50 samples. Besides, there is much higher response rate in interview than in questionnaire as it is much more personal and it is difficult to reject interviewer when approached politely. Hence, generalization and representativeness of data collected in interview are more justified.
Even so, interview has the practical disadvantage of language barriers. Although this can be solved by hiring trained interviewer, it would be rather costly.
In addtion, interview has less ethical issue than experiment or covert participation as consent is usually obtained before the interview can take place. Confidentiality and privacy of the interviews can also be protected. Informed consent from participant would also prevent misleading the participant of the nature of research. Unlike covert participant observation, both interviewer and respondent are unlikely to be harm or placed in danger in interview as respondent is aware of interviewer identity and purpose. Hence, interview is seen as a rather ethical research method.
Withal, interview produces results which are less valid than participant observation but more valid and less reliable than questionnaire and experiment. Although interview tends to be limited to small scale study which has low representativeness and generalizability, interview is useful to study attitudes and emotions which cannot be observed directly. These limitations can be overcome by using method triangulation by Norman, which allows crosschecks of finding produced by other method, as to increase the validity and reliability of the findings.