CHAPTER 6. HOW TO ANALYZE QUALITATIVE DATA
6.9. Additional remarks
Analysis of interviewer/moderator infl uence and social context. In the literature on case study as qualitative research including the interviews (e.g. Yin, 2009), one point of analysis is rarely analyzed, although in the literature on qualitative interviewing in marketing research it is being one of important ones (e.g. Malhotra & Birks, 2007; Maison, 2010). This difference may be explained with different paradigms: realism in case study research and interpretative approach in marketing research, and more weight ascribed to the moderator of focus groups interviews. As mentioned in previous sections, the interviewer plays a prominent role in the process of interviewing as he or she may facilitate the openness and spontaneity of the participants or, in contrary, he or she may limit the talkativeness of the participant. Also, greater infl uence on the content of an interview was emphasized for an interviewer in individual than group interviews. At least three types of biases which are related to interviewer’s role may be observed:
• personal moderator’s bias toward own social and cultural values and attitudes,
• cognitive moderator’s biases such as heuristic of representativeness, • a need of the participant to satisfy the researcher.
All of them may be implicit or explicit. An implicit bias is mostly displayed in non verbal communication (see the next subsection), whereas an explicit bias may be observed in verbal communication, e.g., as a judgment if someone’s opinion is good or bad, in the way of asking the question. Biases may be non- intentional or intentional. In the latter case one may even say about unethical behavior (compare the guidelines of European Society for Opinion and Market Research and American Psychological Association) because the researcher consciously gathers unreliable data.
6.9. Additional remarks
Even though during the analysis it is impossible to eliminate the biases, to increase the reliability of the research, the researcher should identify how much infl uence the interviewer had on participant’s answers. It enables determining
how much trust one may have toward the participants’ statements. But –
what often is being even more important – the careful analysis of the interviewer
verbal and non-verbal behaviors may be very helpful to determine the reasons
underlying the differences between participants. For the aim of the analysis, it may be worth comparing not only answers but also the questions. Two different ways of questioning for the same issue may give partially various answers. Last but not least, the careful analysis of the interviewer’s infl uence may help to
reduce biases during further interviewing. While analyzing group interviews
one should consider also the infl uence of other opinions, of respondent
types and informational and normative infl uence of a group. More about
this topic one may fi nd in the handbook by Malhotra & Birks (2007).
To reduce the role of undesirable biases, it is worth analyzing the data within its context, compare results of two or more interviews and take into account repeated patterns of results (however without neglecting meaning of single opinions if they carry insightful impact for the research objectives). However, the best is to have high self-awareness; then the researcher may probe about given issues to check the underlying reasons in this one or in further interviews. Moreover, the increasing awareness of biases during the analysis stage may help to develop further interviewing skills.
The role of non-verbal communication. In some research topics, non-verbal communication may be an additional source of information. The intonation, paralinguistic (mhm, aha), laughing, no immediate answers, gestures, mimic and other behavioral responses may add new information or even radically modify the sense of statements (Maison, 2010), in case of both – the interviewer and the participant. For instance, the question “are you involved in this process?” may have at least two different meanings depending on the intonation – neutrally formulated it may serve as a simple fi ltering question, but with emphasis on “you” it may sound as disapproval or doubts what may cause the defensive reactions (including denying related with social risks of revealing some information). In case of professionals as participants, three roles of non-verbal communication are particularly worth of emphasis. First, the intonation and way of talking may suggest that a participant is trying to avoid revealing some information. An interviewer may deal with it during interviews by asking additional questions and eventually explicit commenting on it. But if not, it is information worth being registered because the pattern of results is not complete. Second, some non-verbal signals may suggest less or more involvement in a topic. For instance, a participant may talk about corporate values easily and with vivid engagement or with diffi culty, weighting each word. The interpretation of these behaviors in terms of employees’ familiarity and identifi cation with corporate values may
be worth of further investigation in participant other statements. Finally, some words may be said as a joke or ironically, what may totally differ their meaning as in case of the statement “it is really a good solution.”
These complimentary data about non-verbal communication may be achieved via fi eld notes made during interviewing, listening to audio-recordings and – if available – video recordings. Some pieces of this information may be also included in the transcript (e.g., that a participant was laughing, was more involved with a topic, did not answer immediately).
To interpret inconsistencies properly, it is worth remembering two rules. The consistency of verbal with non-verbal results – similarly as between implicit and explicit attitudes is a stronger base to predict further actions. Inconsistencies that may be interpreted as dual attitudes – may implicate diffi culty in predicting the direction of action. For instance, one may reveal the negative personal attitude toward the existing process of suppliers’ choice in an organization (revealed non-verbally), while evaluating it verbally in a positive way, but it may have various consequences for implementing this process (implementing it, omitting it, informally implementing other processes).
Quantitative component. In Chapter 5 “quasi-quantitative” techniques such as closed questions with a scale or ranking and Likert scale questions were mentioned. Sometimes the researcher may even conduct some statistics for example because he or she wants to confi rm if the differences really matter. In all these cases the researcher should avoid trials of statistical generalizations and should always remember not only about a small number of participants but also about the unstructured nature of interviewing. Using such statements as “majority” may be very misleading as for instance in case of 10 participants it is very subjective to say that 6 answers are majority. However, this type of answers may be very useful as a general frame for further analysis as it may help to determine the dominant patterns of results for segments, or may help to sort interviews and fi nd out further specifi c patterns.
Software-assisted analysis. In the literature one may fi nd two positions on computer-aided qualitative data analysis software (CAQDAS) applications. On the one hand, researchers are critical or at least skeptical. On the other hand, the development of opportunities in the software results in its popularization and reduction of some risks. The dictionary of qualitative management research (Thorpe & Holt, 2008) enumerates 7 advantages and 3 the most controversial ones.
CAQDAS may be advantageous unless the researcher knows its possibilities very well, knows its limitations and is able to think “qualitatively.” They may help storing, organizing, managing and searching the data (including different sources, stimuli materials, references, etc.). They may show relationships. However, they do not perform any analysis (Thorpe & Holt, 2008; Yin, 2003); they do not reveal the importance of relationships within data itself.
6.9. Additional remarks
Thus, to appreciate benefi ts of the software for a successful research, the qualitative researcher needs to develop the skills of “qualitative” thinking on the one hand, and advanced skills of software application. Without advanced skills in software service it is diffi cult to appreciate its benefi ts.
KEY POINTS
During the analysis some weight should be given to understanding the infl uence of a moderator, social context and non-verbal communication (if possible and justifi ed). The results of “quantitative” techniques (e.g., questions with scale) may be helpful to order data, but statistical generalization should be avoided. Computer-aided qualitative data analysis software may be valuable (if a researcher has enough skills), but it does not perform analysis and does not replace the researcher.
More about sorting data in tables and matrices as well as the whole process of data analysis complemented with examples can be found in the book by Matthew Miles and A. Michel Huberman (1994). Persons interested in the applying of computer software in organizational qualitative research projects may read the text by Lyn Richards (1999) and handbooks dedicated to particular software, e.g., Suzanne Frieze (2012).