Handle ordered factor when multiple choice question uses recoded variable names #343
Conversation
…use the choiceText.
|
@jmobrien do you mind taking a look at this, to see what you think? I don't think I've done much with |
|
If it helps, here's some more details about the error. I'm fetching a survey that has a multiple choice question with recoding as described in the issue. I use the following code, which results in the following warning.
The values of the variable all are set to NA because the factor levels do not match the values returned by the survey API. Here's a picture of the problems output,
The expected value is "Tenure or Tenure-track Professor", but the actual value returned from the API is "tenure or tenure-track professor". I think you'll be able to re-create the issue with a single question survey that uses recoding and variable names, but let me know if you need any more details. |
|
Sure. I glanced at it earlier--this makes sense and could likely be a good
change, but I should do some testing across a few setups. I seem to recall
some idiosyncrasies around this part of the metadata() endpoint. Can't
recall exactly, but I think it might have had something to do with
whether/how it respected the checkbox for these. So, there are some use
cases we'll want to check first (like if people populate these fields, but
then turn them off)
…On Wed, May 8, 2024 at 8:06 PM Julia Silge ***@***.***> wrote:
@jmobrien <https://github.com/jmobrien> do you mind taking a look at
this, to see what you think? I don't think I've done much with
variableNames in my real work.
—
Reply to this email directly, view it on GitHub
<#343 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/ABWEMIXHT2AKTGSWYGAY6DDZBLD2NAVCNFSM6AAAAABHN2IU7CVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCMBRG42DEMRZGU>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
There was a problem hiding this comment.
Thank you so much for this PR @Haunfelder, and for your patience! 🙌
It would be nice to get this behavior under testing, but we are looking at totally revamping our testing strategy as outlined in #294 (comment) so I don't think it's a high priority for now.
|
Awesome, thanks @juliasilge! I appreciate you all taking the time to understand the issue I was seeing. The qualtRics package has been very useful for me! |
If variableNames are non-null, use them for factor levels, otherwise use the choiceText. It may be more appropriate to add an explicit flag to the function to not cause any issues with people relying on choiceText. From my testing, all cases where variableNames do not match choiceText causes an error in fetch_survey as explained in the issue.
Fixes #342