Handout #12: - Data Analysis
and Interpretation
Analysis of qualitative data is a long-winded and difficult process. Ensure that comments are not quoted out of context. One approach is to examine the interview scripts, question by question, noting all similarities and dissimilarities then compare the answers to specific questions given by a range of interviewees looking for common themes and any others that come up.
Other approaches include:
- Analysis of tape recordings or notes taken directly, holding on to the process of looking for significant statements and comparing what was said in different interviews.
- If you have recorded and produced transcripts of interviews, and have the facilities put these on to a computer (software packages exist to assist with such analysis). The text of the interviews can be coded or marked either during or after input. Some programmes will allow you to search transcripts for particular codes or words.
Quantified forms of observation lend themselves to fairly routine forms of analyses, which can be very powerful in getting across particular issues in tabular diagrammatic form.
Although questionnaires can provide both qualitative and quantitative data, they lend themselves more to quantitative forms of analyses, partly because they are designed to collect many discrete items of information i.e. numbers or words which can be coded and represented as numbers and also because they are used in larger scale surveys, where the common focus is representation which encourages a numerical/near-numerical summary of the results.
Analysis of questionnaires calls for the use of statistics. Descriptive statistics is used in most small-scale research studies and makes extensive use of proportions, percentages, central tendencies (mean, median and mode) and variations/dispersions (range, standard deviation etc).
i.e. it is adequate to say:
“20% or 10 out of 50 respondents answered yes….”
The need may arise, however, to go beyond and make use of inferential statistics which is typically used to compare the data/measurements collected from your sample for a particular variable, with another sample or population in order that judgement can be made about how similar or dissimilar the groups are as well to test the significance of the results. Examples of inferential statistics include chi-squared (χ2), student’s t-test and z-score, correlation, and regression.
Inferential statistics makes certain assumptions about the nature of the data and how it was collected, such as randomness, and should not be used if the assumptions do not hold.
In this section you summarise and put your own meanings to the data collected and analysed, and compare the meanings with those advanced by others.
This involves expressing in words, what the tables, graphs, averages, percentages, etc are saying
e.g. 1. Among the 16 students who studied more than the average amount of the time, 11 (68.8%) received an above average result in the history examination while among those 14 students who studied less than the average amount of time, 2 (14.3%) received an above-average result…
e.g. 2. Sixty three percent (63%) of the respondents were of the view that products/services offered have increased. However, only 17% of them thought that profits have increased.
The first step in drawing conclusions from your findings is to consider the research question. Other questions to be asked include:
- What did I find?
- What is significant?
- What does it suggest?
- How could this study be developed further?
e.g. …businesses in Ocho Rios are not experiencing any increase in profits. It should be noted that the majority of persons indicated that more products and services of higher quality were being offered for sale. However, the sales and profits of the businesses have not increased. This indicates that small businesses are not doing very well as stated by Johnson (2001)…
Remember, your own role and position must be explicit within your research. This is partly about asserting ownership and partly about recognising the possible limitations, influences and biases of your own perspective.
Although, it is paramount that your own perspectives are included, it is also very important that you do not get too embedded and bound up in this view. You could take some time out, perhaps a week or two before going back to your analysis.