Hello
I've collected socio-economic data using ODK that I'm analyzing using R software. I have a multiple choice question with several modalities that I would like to have the frequency of each modality.
I tried to convert all modalities of the variable into boolean variables using this code:
qcm$q0_cacao
Dear Jean-Luc,
I think the contents of the following links (and the last line's suggested book) can help you to cope with the issue:
[1] https://www.researchgate.net/post/How_to_perform_data_analysis_for_multiple_choice_multiple_answers_questions2
[2] https://stackoverflow.com/questions/2896742/datamining-analyzing-responses-to-multiple-choice-questions-in-a-survey?rq=1
[3] https://stackoverflow.com/questions/38675151/how-to-use-the-r-survey-package-to-analyze-multiple-response-questions-in-a-weig
[4] Introduction to Scientific Programming and Simulation Using R, Second Edition (Chapman & Hall/CRC The R Series), https://www.amazon.com/dp/1466569999/ref=rdr_ext_tmb
Regards,
Alireza
Dear Alireza
Thanks for your suggestions. I read all of them but I did find a real answer to my issue.
Regards
Hi Jean-Luc,
I can't access Dropbox, but you can use the 'table' function to make tables in R. For example:
dat
Hi Matthew
Thanks for your insights. You could find attached my dataset (qcm) and the expect output (result) in percent.
model.matrix is perfect to create dummy variables :)
Hi Jean-Luc,
Now I understand. That is definitely more complicated. However, this might work. This first step sums up the crops in each village.
out
Hi Matthiew
Back with some issues,
1. How to have the proportion by total individuals per village (total count per locality) [Tiassale=6, Abigui = 6; Man = 5] not by the total like above? The total proportion by village may be more et less than 1 (not equal to 1).
2. Any chance to avoid converting multiple choice questions into dummy variables first (by extracting directly count values of each choice) ?
Hi Jean-Luc
You can get individuals per village with
indiv.per.village
Hi Matthew
Thanks for your support and ok for your suggestion.
I would like to calculate the global proportion.
out1
Hi Jean-Luc,
I think this is not valid, so I am not sure what you mean to do.
out1
Matthew
I'm trying to have the global proportion of each choice!
village_name
It is not obvious to me what "global" means and I don't understand the first three lines of code. If you want to total something (if that is what your first line of code means), 'sum' is useful. sum(indiv.per.village) will yield the 17 respondants in your data. sum(out) will yield the 64 total entries under "farm_type" (but only if you make dummy variables for every crop, which I did not do).
You will only be able to divide "dat0u" and "indiv" if they are the same size, or an even multiple. If I wanted to divide "out" by indiv.per.village and sum(indiv.per.village) I would use two lines:
out[,-1]/indiv.per.village
out[,-1]/sum(indiv.per.village)
If that is what you want to do (I'm not sure), then you can't do it on one line.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Statistical Software
- R Statistical Package