Luckily, this can be simply "solved" from a statistical approach by just deleting the option "Christian" from the survey.
https://twitter.com/ryanburge/status/1480547479770869763
When you have an option that is confusing respondents, there's a solution! Delete it! Force them to choose between the empirically meaningful options.
Also, the actual best way to survey religion is to ask people, "Think about the place you most often worship. What is its name?" and then to do a follow up, "Do you happen to know what religion this place of worship is associated with?" then give religion options.
Then once you've ID'd a religion, ask, "Do you happen to know what denomination this place of worship is associated with?" then give major denominational families.
The problem is we usually do this in a weird order.
The problem is we usually do this in a weird order.
We ask people "What is your religious preference?" or something absurd like that, and we give like "major families" of religions and it's kind of a survey of religion and kind of a vocabulary test.
When what we should do is ask people an extremely verifiable fact they probably know: what is the name of the place you usually attend religious services? If they don't know it, you can try to classify based on follow-ups, but odds are they're de facto nonreligious anyways.
What's nice about "What's the name of the place you usually go for religious services?" is that it directly queries the actual social fact we are interested in and then gives the researcher really nice, granular microdata to recode as they see fit. With follow-ups, even moreso.
The reasons we don't do this kind of question format in most surveys are: 1) inertia. Survey-taking involves a huge amount of question- and concept-borrowing. 2) Data processing burden. You have to recode and classify tons of custom responses. 3) Privacy. People may not want...
... to give the name of their congregation.
Now, high-quality surveys like GSS actually already do something kinda like this for their congregation follow-up survey. But even they start with religious preference and work down.
Now, high-quality surveys like GSS actually already do something kinda like this for their congregation follow-up survey. But even they start with religious preference and work down.
For my surveys, I don't ask about specific religious ID because my research is strongly focused on other fertility-related questions and I don't want to squander my "respondent trust budget." And that concern impacts a whole lot of surveys!
• • •
Missing some Tweet in this thread? You can try to
force a refresh
