Lightnews — Scholar-powered news

Ariel Edwards-Levy

@aedwardslevy.bsky.social

38K followers 610 following 4.5K posts

Polling editor at CNN, keeping (cross)tabs on public opinion and the news. I like puns.

Posts Media Videos Starter Packs

Pinned

Ariel Edwards-Levy @aedwardslevy.bsky.social · Mar 24

Excited to share some tips for writing about polls for anyone coming in cold to the subject!

(Pictured: a thing on which pollsters are probably not calling you these days).

Ryan Teague Beckwith @ryanbeckwith.bsky.social · Mar 24

How do you read a poll? CNN polling editor Ariel Edwards-Levy walks you through it yourfirstbyline.substack.com/p/how-to-rea...

How to Read a Poll

CNN polling editor Ariel Edwards-Levy walks you through everything you need to know.

yourfirstbyline.substack.com

11 40 230

Ariel Edwards-Levy @aedwardslevy.bsky.social · 8h

ah, an instrument survey instrument

Danny Dougherty @dannydougherty.bsky.social · 8h

Me, sending out questionnaires to the nerds I make play music with me

A Google drive survey form with open ended questions stating "I have other thoughts about band business, which I have written below" then "I have other thoughts about Danny's survey design, which I have written below"

3 16

Ariel Edwards-Levy @aedwardslevy.bsky.social · 8h

please let her know she's very welcome at the survey research conference

1 5

Ariel Edwards-Levy @aedwardslevy.bsky.social · 9h

can't believe you didn't save any injera to share with the rest of the ̷c̷l̷a̷s̷s̷ social media site

1 27

Ariel Edwards-Levy @aedwardslevy.bsky.social · 9h

once again asking all surveys to include an "anything else you would like to tell us" at the end because I guarantee you I will have helpful* thoughts on your questionnaire design

*YMMV, but then again YMM be wrong, particularly if all your fave questions have "never heard of" but not "no opinion"

7 4 36

Reposted by Ariel Edwards-Levy

Michael McDonald @electproject.bsky.social · 15h

I served on the National Academy of Sciences advisory committee to the Census Bureau on the privacy formula and the biggest concern was it resulted in undercounting communities of color — and it was the Biden administration that implemented it for the 2021 data release

Renee DiResta @noupside.bsky.social · 16h

“Who was president in 2020?” is the question of our time

Senator Jim Banks V
@SenatorBanks
Follow g..
The 2020 Census was a fraud. The Biden admin used a shady "privacy" formula that scrambled the data and miscounted 14 states.
It included illegal immigrants and handed
Democrats extra seats. Americans deserve a fair count and l'm fighting to fix it.
Zinited Statcs SScnate
MASHNORON DEYNN
October 6, 2025
an. Thow arewutn
Socretary
U.S. Department of Commerce Washington, DXC 20230
Dear Secretary Lutaick:
I urge you to investigate and correct erroes from the 2020 Census that handed disproportionntd political power to Democrats and illogal alicns. The Census Burcau adoptod a now and opaque
voting districts. As preparod by the Biden administration, the 2020 Censas roports miscountod the population of fourteen states, ' wrongly allocating six congressional scats aod Electoral
xx conwuu tmora uy, toc taroro tu mot đoo tnn Cunu lc pung t imber of votine districts. And the reoorts de aniovety inchaded ilecnl nhens widhout track those aliens' citiz enship status, If left uncorrected, these erroes will coetinue diluting the political power of American citizens.
Consus data plays a crucial role in allocating political roprescntation and goverment funding
Under the Constitution, cach state gets Congressional represcotutives and Electoeal College votes- based on "the whole number of persons" within the state." The number of persoes is in
tum clabaabod by a consttubocalty-mandalcd dcccacaal ccavas. Pcual agcco nely on census data to allocate billions of dollars in federal funding, much of which hinges ee opulation. And states use consus data to deaw congrossicnal and voting districts, »hich the
The it eiy of tie soaren alie n tor dur at ce hor inc ae oam toy ee n
nght wilbout statsica ly signilicant ciroo.
Diffcrcntial privacy is opaque and liable to mistakcn count totals. Sure cnough, the 2020 Ccnsus ovarocunted the noculation in cicht states and undercounted it in six." Tihe most extrcmi andercount was Arkansas, at 5.04%, and the larg…

1 16 49

Ariel Edwards-Levy @aedwardslevy.bsky.social · 15h

when you wanted AmeriSpeak but were thwarted by an errant autocorrect :(

Ariel Edwards-Levy @aedwardslevy.bsky.social · 16h

oh I have lost significant amounts of time to this particular genre of rabbit hole

Ariel Edwards-Levy @aedwardslevy.bsky.social · 16h

the lackadaisical approach to writing about polls is unfortunately shared by plenty of other outlets/writers (many of whom pride themselves on being savvy and should know better!)

Ariel Edwards-Levy @aedwardslevy.bsky.social · 16h

How To Write About Polls 101
yourfirstbyline.substack.com/p/how-to-rea...

What are some things you should always include in your write-up?

When I’m writing about a survey, I always want to include at a minimum: the name of the pollster, the identity of any sponsors, the dates during which it was fielded, the sample size, the margin of error, and some basic information about how it was conducted.

I also want to be consistent when describing which group the poll is intended to represent: “all US adults” is a different group of people than “registered voters,” or “voters who are likely to turn out in the next election.” A great recent example of why this last one matters: people are more likely to tune in for a State of the Union or other presidential address if they’re already supporters of the president. That means “speech watchers” tend to be a much friendlier group than the full American public – which is extremely important context for why such speeches tend to get such positive ratings.

And finally, a link back to the data you’re writing about is always a good thing!

1 15

Ariel Edwards-Levy @aedwardslevy.bsky.social · 16h

afaik

Ariel Edwards-Levy @aedwardslevy.bsky.social · 16h

Ariel Edwards-Levy @aedwardslevy.bsky.social · 7d

The buff vs. crying dog meme

Buff dog is labeled "A good methodology statement for a survey conducted online" and contains a partial excerpt from Pew's ATP methodology statement.

Sad dog is labeled "A bad methodology statement for a survey conducted online" and reads simply "The survey was conducted online." (You would think this is an exaggeration of how oversimplified bad methodology statements are, but it is not.)

2 3 46

Ariel Edwards-Levy @aedwardslevy.bsky.social · 16h

I am once again asking people who encounter a poll to ask whether they have the necessary information to fully understand at a minimum a) who was surveyed and b) what they were asked

3 13 78

Ariel Edwards-Levy @aedwardslevy.bsky.social · 16h

poll:
is sponsored by company with a vested interest in results and which is clearly using survey as a promotional tool

writeup:
does not link to toplines
no info on question wording
no methods info
confuses "percentage" and "percentage point"

Herb Scribner @herbscribner.bsky.social · 17h

Hack — Sleep on vacation.

Americans Are Using PTO to Sleep, Not for Vacation—Report

An Amerisleep.com survey revealed that 37 percent of Americans used vacation days in the past year just to rest.

www.newsweek.com

4 13 94

Reposted by Ariel Edwards-Levy

Dan Berman @dhberman.bsky.social · 18h

Here are some of the questions Bondi has refused to answer
www.cnn.com/politics/liv...

Live Updates: Attorney General Pam Bondi testifies before Senate Judiciary Committee | CNN Politics

Attorney General Pam Bondi faces questions from lawmakers on James Comey, Jeffrey Epstein, Jack Smith, immigration and drone strikes. Follow here for the latest.

www.cnn.com

1 4 8

Ariel Edwards-Levy @aedwardslevy.bsky.social · 19h

never once in my life have I wanted a space after paragraph

6 20

Ariel Edwards-Levy @aedwardslevy.bsky.social · 19h

i'm disappointed in all of you

15 33 380

Ariel Edwards-Levy @aedwardslevy.bsky.social · 19h

tremendous content, etc

Ariel Edwards-Levy @aedwardslevy.bsky.social · 19h

Anyway, as is often the case, the first step toward thinking about public opinion here without sending yourself down a rabbit hole is accepting that there isn't a "true" measurement where every person is on a binary of either supporting or rejecting violence and you can measure which side.

Ariel Edwards-Levy @aedwardslevy.bsky.social · 19h

(Also I do not know what happened with my attempts at tagging people there!)

1 1

Ariel Edwards-Levy @aedwardslevy.bsky.social · 19h

This topic takes a lot of the edge cases where gauging public opinion is particularly tricky -- views held by very small but consequential shares of the public, topics subject to partisan cheerleading, measuring support for hypothetical action -- and basically wraps them up with a bow.

1 2

Ariel Edwards-Levy @aedwardslevy.bsky.social · 19h

Some thoughtful recent work on measuring support for political violence from @gelliottmorris.com, Nate Cohn, @lilymasonphd.bsky.social and @nathankalmoe.bsky.social

www.gelliottmorris.com/p/most-polls...
www.nytimes.com/2025/09/20/u...
www.dropbox.com/scl/fi/9blbj...

2 3 8

Ariel Edwards-Levy @aedwardslevy.bsky.social · 1d

"in the news OR in advertisements" really doing a lot of work there

1 2

Ariel Edwards-Levy @aedwardslevy.bsky.social · 1d

oh look, it's my least favorite question structure

2 4

Ariel Edwards-Levy @aedwardslevy.bsky.social · 1d

have just encountered possibly the bravest individual in the universe: a person whose email signature contains the phrase "I will return all calls & emails within 24 hours"

7 5 72

Ariel Edwards-Levy @aedwardslevy.bsky.social · 1d

such is the lot of the Swift vote veteran for truth

2 1 10