c/strangetimes · by idoubtit@lemmy.world · 3y

Do you believe in angels? About 7 in 10 U.S. adults do, a new AP-NORC poll shows

https://apnews.com/article/religion-poll-belief-angels-devil-bee64258d6a47067a046ba7f3c50933a

12 points · 5 comments · view on lemmy.world

5 Comments

NukeminHerttua@sopuli.xyz · 5 pts · 3y (1 reply)

America, ladies and gentlemen (actually I would've thought the number was even higher 😬).

idoubtit@mstdn.social · 1 pts · 3y

[ removed ]

Renevar@lemmy.blahaj.zone · 3 pts · 3y (4 replies)

"The poll of 1,680 adults was conducted May 11-15 using a sample drawn from NORC’s probability-based AmeriSpeak Panel, which is designed to be representative of the U.S. population." (Dont know how to quote on lemmy)

That's a pretty small sample size considering, also they don't give any actual information about what kind of demographic they polled, which seems odd, but I don't know anything about AmeriSpeak so I don't know how that works

crypticthree@lemmy.world · 2 pts · 3y (1 reply)

The analyses on AmeriSpeak Panel survey estimates indicate households recruited through NRFU have a tendency to report more moderate opinions towards policy issues and somewhat more conservative political views than the initially recruited panelists. In other words, probability-based panelists obtained by a less expensive method, compared to the more expensive NRFU method, tend to have stronger (less moderate) opinions and attitudes towards policy issues. The data is not unequivocal and there are exceptions; however, our analyses suggest that probability-based panels with low response rates and/or without a nonresponse follow-up program may overstate political and policy polarization, while the AmeriSpeak NRFU program increases the representation of persons with moderate-to-conservative opinions about policy issues.

idoubtit@lemmy.world · 1 pts · 3y

Good to know! That also makes sense.

(It would be helpful if you could take off the URL link from the entire paragraph, though. It's hard to read this way.)

idoubtit@lemmy.world · 1 pts · 3y

Actually 1000 is a good number for a sample. It narrows your margin of error to around 3%. If it is truly randomly selected people. But that's the hard part. How you sample skews the results. For example, sampling via land line random phone numbers means you get much older people. Sampling on college campuses means younger, etc.

natflow@apollo.town · 1 pts · 3y

[ removed ]