Hello fellow researchers! Newbie here :)
I created a survey with open-ended and Likert-type questions. The survey was administered nine times to the same people using google forms. I plan to use R for analyzing the quantitative data and NVivo for the qualitative data.
I would like to know the best way to combine the 9 forms. Currently, I have two options:
A. Add a column (Week number/Lesson number) which will be numbered from 1-9. It will contain the order in which the survey was administered (e.g., 3rd out of 9, or 8nd out of 9). And then merge all records vertically.
B. Add columns for each item of each survey and add a week or lesson number (e.g., w1Q1, w2Q1, w3Q1, etc.) and merge them horizontally.
'A' might have been the more obvious choice, but I'm not quite sure about its implications when doing repeated ANOVA analysis in R or coding in NVivo.
Can someone point me to some resources on how these are properly done for this case?
Thanks in advance!
I personally wouldn't worry too much about NVivo, as what you import as a dataset are basically just transcripts. So it's probably easier to go with what is best for your quantitative analysis, and then create separate transcripts to analyse in NVivo.
Hello Roberto,
As far as the quantitative portion goes, the answer depends on what specific research question(s) you are trying to answer. That wasn't clear from your query.
I would imagine that "wide" format (e.g., all responses, from time 1 to time 9 for a given participant, in the same row) would likely be best for a number of common analyses one might perform using such data. That is consistent with your proposal, "B." Proposal "A" is so-called "long" format.
Concerning the open-ended responses, again, the answer will hinge on what research question(s) are of greatest interest to you. For example, if you believe that thematic changes might take place over the 9 (!) measurements, then it might make more sense to analyze the early sets independently of the later sets. However, as Nicolò Zarotti has noted, the open-ended responses will all be entered as a set of transcripts. It's up to you to identify any useful patterns, themes, trends, and so forth, that may be present. To the extent that coding the specific respondent, time, question, and so on helps you to do so, then include these as part of the data set.
Good luck with your work.
R isn't very consistent with how data should be organized, but I would say in most cases the long format (option A) is what is wanted. This way, you will have the variable Time (or Week or Lesson) to enter into the model.
R does have some tools to transform data between long and wide formats. The tidyr package is helpful here.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA