Hi Rachael,
Thanks for getting in touch.
On non-numeric values, you would need to systematically replace any string values with numbers in both the xlsform and the dataset, in order for kobo2stata to match them across these two input files. This may or may not be feasible in your case, but given the amount of manual work usually involved in this (and how prone to errors it will be) I cannot recommend it. As a fallback, you can still run your files through kobo2stata to apply at least the variable labels (but not value labels), and then run your categorical string variables through a crude “encode” in Stata for some rudimentary value labels.
On the label-related error message, first please make sure you are using the latest version of kobo2stata (as of today, this is v1.06 - see the version number at the very bottom of the kobo2stata help file). If you have an outdated version, please run Stata’s “adoupdate” command. If you do have the latest version of kobo2stata and the problem persists, you might be using non-standard notation in your label column headers, as a result of which kobo2stata can have troubles identifying the correct label language in a multi-lingual setup. Try removing any label columns for secondary languages (in both the survey and the choices tabs of your xlsform), which might help as a workaround.
