Qualitative Analysis

Qualitative Analysis

Analysis of data in a research project involves summarizing the mass of data collected and presenting the results in a way that communicates the most important features.  In quantitative research, analysis involves things like summarizing the frequencies of variables, differences between variables, and statistical tests designed to estimate the statistical significance of the results (i.e. the probability that they did not occur by chance). All this is done basically by counting how often something appears in the data and comparing one measurement with others. At the end of the analysis, not only do we have a mass of results but we also have what we might call “the big picture”: the major findings.

In qualitative research we are also interested in discovering the big picture but use different techniques to find it. For the most part we are interesting in using the data to describe a phenomenon, to articulate what it means and to understand it. Different approaches require different types of analysis: in this introductory text we shall focus on constant comparison.

Most types of analysis involve the categorization of verbal or behavioral data, for purposes of classification, summarization and tabulation.  The content can be analyzed on two levels. The basic level of analysis is a descriptive account of the data: this is what was actually said, documented or observed with nothing read into it and nothing assumed about it. Some texts refer to this as the manifest level of analysis.  The higher level of analysis is interpretative: it is concerned with what was meant by the response, what was inferred or implied. It is sometimes called the latent level of analysis.

“Content analysis” is a phrase that is sometimes used in the literature to mean any type of analysis of the content of a transcript. However it also has a more precise use, which is in connection with a technique involving counting the frequency of occurrence of particular phrases, words, or concepts, and is probably therefore best avoided – like the term “thematic analysis” – unless the writer is specifying exactly what type of content analysis is meant.

Keeping Records and Being Organized

Whatever qualitative approach is involved, it is very important to be organized when keeping records of data or reflexive notes or memos, or documents. As in all (qualitative or quantitative) research it is crucial to maintain a good audit trail which could in theory be inspected by others. It is also important to ensure that any saved records are kept in accordance with data protection regulations. This often involves careful anonymization procedures in labeling digital or analogue recordings or documents and text. All these issues should be discussed within a research team when drawing up the initial research protocol; ethics and research governance bodies will give feedback at an early stage.

Transcribing Qualitative Data

Transcribing is the procedure for producing a written version of an interview (e.g. in narrative, or grounded theory-based research) or conversation (e.g. if using conversation analysis). It is a full “script” of the interview or conversation. Transcribing is a time consuming process. The estimated ratio of time required for transcribing interviews is about 6:1. This means that it can take six hours to transcribe a one hour interview. It also produces a lot of written text. For conversation analysis or discourse analysis, very specialised transcription is required which includes precise notation of lengths of pauses and inflections, among other features, and this type of transcription is therefore much more time-consuming.

The research team should at an early stage consider the question “who should do the transcribing?” Ideally there might be resources to pay a professional transcriber who is aware of the need for confidentiality. This is usually more cost effective than a health care professional who will take longer and is more highly paid – on the other hand some researchers find that the process of transcribing helps them to become “immersed” in the data and is therefore a useful step in the process of interpreting how the account helps in the answering of the research question. If the transcriber is unfamiliar with the terminology or language contained in the interviews this can lead to mistakes or prolong the transcribing time. All transcripts should be carefully checked by the researcher (usually the interviewer) in conjunction with the recording.

When people are in conversation only a small proportion of the message is communicated in the actual words used. A larger proportion is transmitted in the way people speak. Tone and inflection are good indicators of a whole range of feelings and meanings. When transcribing, consideration should be given to whether and how these feelings and meanings can be communicated on paper by using punctuation marks (full stops and commas will be essential for meaning, for instance). As mentioned above, discourse and conversation analysts have a repertoire of specialised conventions for inflections and pauses, and would transcribe every sound an interviewee makes. Those using constant comparison would use less intensive transcribing, concentrating on the content of what is said, and some may decide to ignore “ums” and “errs” – these issues should be decided in advance within a team. It can be useful to adopt the convention that the interviewer’s speech is upper case, and the interviewee’s speech is lower case. Some researchers may consider techniques such as underlining and emboldening: all these need care to ensure that a transcript remains as faithful as possible to the speech it transcribes – if in doubt they are best avoided in favour of listening to the original recording to clarify any potentially ambiguous statements.