Thread by @RachelBitecofer

1. Survey update: To all who contributed to my research study on extremism in the GOP I wanted to updated you on where I'm at. Right now, I'm haggling over some options trying to max the # of questions I can include w the # of what are called "demographics" bc I really need both

2. to make best use of the data for analysis. Goal is most questions w most demos w best methodology possible given the MASSIVE budget constraint we have. Again, bc I only need data collection & sample & can guerillo some of the other aspects, we're gonna pull this sucker off on

3. about 1/3 of the normal budget. Which is CRAZY! And I'll never do it again. Until the next time I decide to do it again. Right now, my general design idea is a GA statewide with an oversample of respondents from MTG's district w enbough completes to be able to run cross tabs.

4. Anytime you see someone "interpreting" crosstab data from a survey of n=500 or 600 completes they should not be doing that bc the margins of error (as well as other issues are very high at low n), no matter who this person is & how fancy their job title is. This means NO TALK

5. about anything other than what we call topline data Candidate A 47% to Candidate B's 53% w MOE +-3.5%. If someone since "Democrats," or "women," or "Independents" or "college edu" they're talking crosstabs & if the survey MOE is 3.5% the crosstab MOE is about double (roughly)

6. You might be thinking, "but Dr. B I see doing this every day on TV & everywhere when n is not above say, 700. Are you saying they are violating basic good statistical practices & telling me shit that may, or may, be actually accurate?" Yes. Yes, that's exactly what I'm telling

7. you. So n is really, really imp IF you care about crosstabs & science (hopefully at least some of you really appreciate the deep, deep irony here). So I am trying to max n, given # of questions, w the best methods I get get "live calls vs online poll." Lost 4 days to being

8. bedridden last week, which delayed me on all fronts. Should have fielding finalized in the coming week & will be able to show y'all the questionnaire. What I have envisioned in my brain thingy is cool if I can pull it off at Walmart prices. Goal again, measure "extremism."