How to Code in Qualitative Research
In order to code in qualitative research you are faced with the task of quantifying your subjective data, and this means converting research that consists of subjective observations into an objective data set. In other words, you are required to quantify your qualitative research first in order to code in the research. In short, you will use numbers to interpret research observations that are not numbers-based.
Gather your pre-research outline as well as all of your field notes, interview question results and/or survey results. This article assumes that you have already conducted your research and are seeking to code it in for the purposes of presentation and/or publication. As converting qualitative research into quantitative results can be a capricious process, it is necessary that you assure your pre-research theory and methodology are close at hand for reference.
Quantify and tabulate your results. In order to code in qualitative research, you will have to analyze your subjective data from a mathematical viewpoint. If you are doing textual analysis, this might simply involve counting the number of times a certain word is listed--It is easier to do this in word processing applications, but you can do it by hand. Surveys are qualitative, but multiple choice questions can be coded in by percentage of response A, B, C, etc. Field notes can be coded in by finding common objective observations that you have made, e.g., "The shoppers I observed at Tesco appeared more in a hurry than those at Asda" can be coded in with 1 representing a slow pace of shopping, 2 representing a faster pace, and so on. Your numbered results will constitute the data that you will be coding in. Tabulation is simply compiling all the data together.
Enter the results into a spreadsheet. You should have at least 15 results, or you should not be thinking about "coding" anything, with the exception of a case study, in which case the 15 results would come from within the research concerning your one specific case, with each of the fifteen variables a sub-point of your subject of study. Results should be entered with the subjects' names associated with the data on one axis (spreadsheet row headers) and the numerical results pertaining to the subjects on the other axis (spreadsheet column headers).
Control for as many objective factors as possible, e.g., age, race, sex, etc., and then code in your qualitative data in the succeeding columns. Coding in your qualitative data requires you to list quantitative data that might be boring but nevertheless might also answer your research question better than your hypothesis regarding your qualitative data. So while your rows might contain only one type of data, e.g., someone's name, the column headers might contain several variables that do not constitute your qualitative data, but which represent the "control group." Coding in qualitative data without quantitative data is not a good research practice, as it leaves no controls and can invalidate your research findings.
Label the category, in the header of a column, with a word that captures the essence of what you are measuring. This is where you will need your pre-research theory and methodology notes. For empirical data, such as "age," you would of course just put in the age, but for something like "gender," which is not the same as "sex," you cannot just use the word gender, so your data set in this case should be named something that relates to the way in which you measured the more subjective variable gender. For example, "percent male" would be much more telling than simply "male" or "female" in terms of statistical comparison. This rule does not apply if you are coding in answers from a questionnaire that only allows the participant to select from two answers, in which case the coding can be simplified by assigning two different numbers, e.g., 1 and 2.
Organize your results relative to your hypothesized independent variable. This will affect the layout of your charts and graphs, but also gives you immediate insight into whether your independent variable(s) has any kind of correlation with your dependent variable(s). You do this by selecting a column that has the headers of all of the names corresponding to your coded-in qualitative data, going to "format" and selecting how the data should be organized. Once you have done this, you are ready to create graphs and charts and insert your coded-in qualitative data into your presentation or paper.
- Choose a coding method that best fits your qualitative data, or you risk losing the meaning behind your research.
- Ask yourself why you are choosing to code in your qualitative research. If you can avoid using quantitative analysis of your qualitative data, then consider coding variables that are more easily put into statistics and speak more subjectively about the qualitative research that is not best translated into empirical results.
- Do not be tempted to bias your coded data to make your hypothesis appear more right than it is; a null hypothesis is still a research result. In short, do not change your tabulation method partway through.
- Save frequently, as it takes hours to code in research results and you do not want to lose all f your data just before a deadline.
- Coding in qualitative research compromises "external validity," or the applicability to the real world, so make sure your "internal validity," or the consistency with which you measure your data, is sound.
- The more you characterize your qualitative results empirically while conducting research, the easier it will be to code it in later.
- Most variables can be coded by numbers from 1 to 10, by simply counting the number of times a given word or phrase is used, or by assigning graduated numbers which correspond to "amount of" something that is taking place.
- If you are new to coding in research, use Excel; if you are more advanced, e.g., graduate school, SPSS is the best tool.
- Make sure that your original coding of your research (before coding in) makes sense in terms of your hypothesis and independent and dependent variables.