While an underscore in variable names is a convention in many applications, it is a bit dangerous in EpiData .rec files because it is possible to create a .qes from a .rec. That .qes will have an underscore that will be interpreted as a new string field of length one. This is the case even now, since the underscore will appear in the 'first word' position on a line.

I suggest reserving underscore for temporary variables that will not be saved.

Jamie

Annemieke wrote:

According to the documentation, an underscore character is allowed in variable names. And, indeed, I have no problem defining new variable names with an underscore and using them in analysis (v2.1.0.151). However, when the data have been saved and one opens the file for further analysis, the variable has been renamed v1 (or v2, etc). In addition, the original variable name containing the underscore is put into the variable label. For example, I created a variable abc_test. After saving the data and opening the file, I find that the variable now has the name v1 and the label "abc_test abc-test".