Comparison of average values from one survey execution to another without keeping single responses | XM Community
Skip to main content

We have just executed a survey which is planned to be repeated each quarter with a similar target group. The results are displayed as average values and scores in a dashboard. 

Unfortunatelly, we need to delete all old survey data before the survey is executed the next time due to data privacy requirements. 

Any suggestions on how we would still be able to have a comparison between each execution to identify any changes?

Of course we could export the data from the dashboard each time and have the comparison externally, however, we would like to take advantage of Qualtrics dashboards for proper display. We could also create a “Comparison Project” where we create one response for each iteration of the survey and track the average values in that response and compare each response in a dashboard. But that would be quite some additional effort… 

After discussions with the Qualtrics support, I will answer the question myself: Unfortunatelly, there is no easy way to keep only average values and delete all single responses. From data privacy perspective, it would be fine to keep all single responses if everything is anonymized. Therefore, we are thinking about some way of further anonymization, e.g.: 

  • Argument on “pseudoanonymization”: For most organizations this is a sufficient way of anonymization as all personal data is deleted from the response and the token to map a pseudoanonymized response with a token can only be used by involving Qualtrics and client legal teams. This is very unlikely. 
  • Clear personal data via API: The update response API could be used to clear specific fields (including external reference). However, the Qualtrics internal contact ID probably remains. 
  • Delete response and create a new one: One approach would also be to delete the original response and just create a new one via API that contains all the answers but none of the personal data. 

Our preferred approach is option 1. The pseudoanonymization was introduced based on the feedback of industry leaders. For most of the large organizations the pseudoanonymization is sufficient. 

Leave a Reply


Terms & Conditions