I have a table that I have set up for Independent Groups t-test with 6 rows and ~3,000 columns. The last column is the Grouping Column, specifying rows to be “A” or “B” for the t-test - this last column is of type string while all the other columns are of type double. However when I run the t-test I get the error
Execute failed: function values at endpoints do not have different signs, endpoints: [-1, 1], values: [�, �]
I’m not sure what to do to fix this error. If I instead take either half of the table (ie, the first 3 or the last 3 rows) and run them through a single-sample t-test that runs without error.
Can anyone help me troubleshoot this error?
Hello @lparsons42 -
Sorry for the delayed response. I put together a small toy workflow to try to reproduce the problem, but so far I’ve not been successful. Are you able to execute the attached workflow?
Indep_ttest_toy.knwf (22.7 KB)
I understand you might not be able to share your data, but if you can, could you post a workflow demonstrating the problem?
Thank you for reaching out to me on this. I eventually found that there was either a missing value (which will crash the t-test) or something weirdly entered such that there was a string where a numeric value should have been (which of course will also crash the t-test). I had to dig a bit further into my table (Rule-based row filter was helpful here) to find the values that were causing this and remove them. It’s would have been easier to find this if Independent Groups t-test had returned a more specific error pointing at which row (or column) had the troublesome value.
In hindsight the fact that I could run the test on various segments of the table should have been a good hint that there was a troublesome column in there.
Thanks for the feedback - my next suggestion was going to be something along those lines (weird or missing chars in your dataset). Sorry that the error message wasn’t sufficiently specific - maybe that’s something we can work on clarifying.