Differing data between Reports on KoboToolbox and Stata dta file

I have used KoboToolbox with an XLS form. I reviewed my data on KoboToolbox Reports - Data and have noticed an inconsistency in how the data is reported.

The question is a select_one with options (label) 1, 1.5, 2, 2.5, 3, 3.5, 4, 4.5, 5; (name 1-9) - and I checked my XLS form and it only lists each value one time.

In the KoboToolBox Report page it lists the label 1.5 and 2 two times each. The rest are listed once only. See below for data report
1 n=2
1.5 n= 78
1.5 n=62
2 n=44
2.5 n= 5
2.5 n=88
3 n=0
3.5 n=19
4 n=0
4.5 n=2
5 n=0

When I export my data to Stata and use Kobo2Stata the counts for each option differ from the KoboToolBox Report page.

1 n=2
1.5 n=62
2 n=78
2.5 n=88
3 n=44
3.5 n=19
4 n=5
4.5 n=2

I assume that KoboToolbox Report Data on the website has some issue reading the data? But I want to ensure my data in STATA is correct. Please advise.

Thank you

@richards_8, would you mind sharing with the community a screenshot of your issue? Maybe that would help the community understand your issue pictorially.



These show my XLS Sheet for the choices and survey

@richards_8, when having a look at your custom report and then generating the frequency table manually through the aid of Excel, I could see this, which matches the figures from the custom reports:


Maybe there could be some issue with how you computed your data.

Thanks for your response but not sure how to proceed. Are you saying the custom report is accurate? I don’t understand if I have a select_one that the label can appear more than once in the results. How is that possible?

OK got it. So did you make redeployments by adding/editing/removing choices in your survey project for this particular variable?

I have used the same choices since the inception of the survey.

The data match perfectly between the raw xlxs file I downloaded off KoboToolbox and Stata data file.

I mostly just want to know if I can trust the raw downloaded data. I don’t really care if the report online is correct - so long as my downloaded data is trustworthy.

@richards_8, would you mind sharing the following through a private message so that I could have a closer look at it?

  • Username
  • Project name
  • Server

@richards_8, kindly please be informed that your issue has been solved:

I solved this by adding a default language as English to your survey project and then redeployed:

@richards_8, kindly please be informed that we also have a GitHub issue for this (so that the same could be fixed permanently in the system):

Hi - this is fantastic and solves the issue. Really appreciate your assistance to sort this for me. Thank you for your help.

1 Like