I’ve never noticed that note. As I figure it out, this is an attempt to overcome the widely known differences in church attendance between Muslims and Christians: Muslim women are not required to go to Mosque, whereas Christian women go to Church more frequently than Christian men (what about Buddhists?).
So basically it says that there are two different questions asked in ‘Islamic societies’ and the rest of the world, which are stored in a single variable in WVS dataset.
The use of different indicators for measuring the same construct is justified in the reflective measurement logic in which all the indicators are exchangeable, though I doubt it relates to a single-indicator measures.
I checked questionnaire translations in “Islamic societies”, as it seemed to me too strange. What I found is even more surprising. Summing up, the question about praying only Morocco and Malaysia unambiguously asked about praying (though some troubles with options coding in Malaysia). The rest seven translations were either about attending mosques or ambiguous. Find below the detailed report. Thus, in general, we can quietly ignore The Note, just like the questionnaire translators did.
To be serious – when using WVS or some other comparative survey data, every single question should be analyzed in (at least) the way I did it below, especially for your key variables and absolutely necessary for your dependent variables. Google is a handy tool, it gives a crude translation but you get a general idea of what’s in the question. You’ll find plenty of interesting surprises.