Ariel Edwards-Levy
banner
aedwardslevy.bsky.social
Ariel Edwards-Levy
@aedwardslevy.bsky.social
Polling editor at CNN, keeping (cross)tabs on public opinion and the news. I like puns.
November 21, 2025 at 1:48 AM
"I bet the media won't even cover this" watch, part ♾️
November 20, 2025 at 7:36 PM
unfathomable why anyone would choose to turn on "your followers can reply" when they'd be missing so much truly inexplicable content
November 19, 2025 at 10:45 PM
occasionally I think about this absolutely perfect five-question battery
November 19, 2025 at 8:11 PM
November 19, 2025 at 7:56 PM
November 18, 2025 at 9:05 PM
i think I know what they're casting for

(this may be TOO deep of a '90s edutainment cut, but gamicus.fandom.com/wiki/Alien_T...)
November 18, 2025 at 3:58 PM
190. Pew, 2000: "Do you get a lot of 'spam' or unwanted e-mail messages, or isn't this a problem for you?"
November 13, 2025 at 9:00 PM
189. Gallup, 2003: "Do you happen to know the first name of the First Lady of the United States?"
November 13, 2025 at 8:57 PM
188. SSRS, 2019: "What is your favorite type of Girl Scout cookie?"
November 13, 2025 at 8:56 PM
187. British Institute of Public Opinion, 1946: "Do you think that the Government's popularity has or has not suffered through the recent cut in beer?"
November 13, 2025 at 8:55 PM
186. CBS, 2017: "Which of the following is your favorite summer treat?"
November 13, 2025 at 8:54 PM
185. Penn, Schoen, Berland, 2007 (asked of likely Iowa Democratic caucusgoers -- presumably for Clinton campaign?): "Which would you rather have? A nominee who is strong, a nominee who is inspiring"
November 13, 2025 at 8:53 PM
Another study (using an opt-in sample) found higher rate of reported conspiracy beliefs among low-quality respondents

Also notes that just throwing out all flagged respondents can bring its own problems
November 13, 2025 at 5:53 PM
And from Sunshine Hillygus at Duke:

I love that this includes a *self-reported* measure of whether respondent was just trying to be funny with some responses.

(I don't even know what "Trump Bee Movie" means here but I will certainly google)
November 13, 2025 at 5:48 PM
From Matt Tyler at Rice on measuring support for beliefs in conspiracies: One idea is to look at whether respondents report consistent beliefs over multiple questions in the same survey, with different wording.

(Also: bonus Olympics content!)
November 13, 2025 at 5:39 PM
So how to catch these mischievous responders? Look for variation on traits that *shouldn't* differ between groups.
November 13, 2025 at 5:28 PM
From Joseph Simpian at NYU -- this time looking at an in-person probability survey of teens that asked whether they were adopted. In a more in-depth follow-up, some who said "yes" were misreporting -- and they were much more different from rest of sample than real adoptees

Also some A+ clipart
November 13, 2025 at 5:26 PM
Leib Littman at CloudResearch notes that acquiescence bias is often exacerbated on some opt-in online panels typically used for consumer research, where fraudulent respondents have an incentive to say "yes" and route themselves into more surveys.
November 13, 2025 at 5:14 PM
can I file some sort of appeal
November 13, 2025 at 2:08 PM
i do in fact usually think of myself as a party
November 12, 2025 at 6:02 PM
I mean, I think it's an open question what kind of coverage it gets among the kinds of media outlets disproportionately consumed by people with positive views of Trump available to damage, but I wouldn't say it is not being brought to the public attention
November 12, 2025 at 2:40 PM
🙅‍♀️
November 9, 2025 at 9:47 PM
an attempt was made
November 7, 2025 at 11:41 PM
are women ruining a reasonable duration of daylight

(image via: en.wikipedia.org/wiki/Selene#... )
November 6, 2025 at 10:01 PM