iiinova — where your digital self keeps evolving

Soubhik Barari @soubhikbarari.bsky.social · 10/08/2026

Appreciate the summary from @kwcollins.bsky.social and excited to see it in print finally! If you work on the integration of LLMs *into* surveys, there’s something useful in here for you.

020

Reposted by Soubhik Barari

Soubhik Barari @soubhikbarari.bsky.social · 03/06/2026

New Paper: "Measuring Group Alignment in Public Opinion" The standard for issue polling is to report issue-by-issue agreement, yet democratic representation hinges whether coalitions can agree on *issue platforms*. This paper tells you exactly how to do that.

162

Soubhik Barari @soubhikbarari.bsky.social · 03/06/2026

162

Reposted by Soubhik Barari

David Rothschild @davmicrot.bsky.social · 27/05/2026

Preprint "Reply to Westwood: Questioning empirical evidence that AI survey contamination is real and substantial" w/ @soubhikbarari.bsky.social, Trent Buskirk, @andrewgordon.bsky.social, Sunshine Hillygus We respond to claims of "existential threat" of AI on survey research osf.io/preprints/so...

osf.io

OSF

142

Reposted by Soubhik Barari

Stephen Turner @stephenturner.us · 09/03/2026

Peer review reliability is shockingly low. Meta-analyses show reviewer agreement barely above chance, and grant outcomes often depend more on who reviews than what's proposed. Our new preprint with Agnes Urban and Arjun Krishnan @compbiologist.bsky.social : papers.ssrn.com/sol3/papers.... 🧵 1/

5227

Reposted by Soubhik Barari

Social Data Collaboratory @ NORC @sdc.norc.org · 04/03/2026

The instructions are occasionally outrageous, and sometimes the barks sound like human speech. But a fully autonomous, free-thinking, conversational dog park, Moltbook is not.

111

Reposted by Soubhik Barari

Social Data Collaboratory @ NORC @sdc.norc.org · 04/03/2026

Is the "AI takeover" a looming reality or a clever illusion created by human steering? @soubhikbarari.bsky.social, a Senior Research Scientist at @norc.org’s SDC, breaks down why the "Moltbook" phenomenon is less about sentient agents and more about "dogs with walkie-talkies." #Moltbook 🧵

233

Reposted by Soubhik Barari

Kevin Collins @kwcollins.bsky.social · 20/02/2026

Interesting experiment in patient diagnosis using LLMs highlighting the critical differences between LLM output when used optimally, and LLMs as actually used by real people www.nature.com/articles/s41... Key takeaways:

133

Reposted by Soubhik Barari

NORC at the University of Chicago @norc.org · 04/02/2026

Federal agencies often work with data not originally designed for statistical use. NORC researchers developed a framework for how AI, with human oversight, can improve the quality of that data for policy decisions. Find out more. go.norc.org/4aexK5T

go.norc.org

AI Framework Helps Federal Agencies Improve Data Quality & Integration | NORC at the University of Chicago

NORC researchers created a framework for developing AI tools that could allow federal agencies to prepare complex data more efficiently.

011

Soubhik Barari @soubhikbarari.bsky.social · 03/02/2026

*Tap* (is this thing on?) In the event that anyone's still out there, some new writing from me. I looked at country-level AI usage from Anthropic’s Economic Index and asked a simple question: beyond GDP, what cultural values track AI adoption?

120

Reposted by Soubhik Barari

Kevin Collins @kwcollins.bsky.social · 20/01/2026

Thought provoking post from @soubhikbarari.bsky.social on uses of LLMs in social sciences. Screenshotting the section most relevant for surveys. somewhatunlikely.substack.com/p/100-ideas-...

232

Reposted by Soubhik Barari

Kevin Collins @kwcollins.bsky.social · 12/01/2026

Interesting (to me, as native Wisconsinite) post by @soubhikbarari.bsky.social on who identifies as Midwestern somewhatunlikely.substack.com/p/the-midwest

somewhatunlikely.substack.com

Who is Midwestern?

Go (mid)west, young data scientist.

031

Reposted by Soubhik Barari

NORC at the University of Chicago @norc.org · 19/09/2025

College rankings wield enormous influence, but do they actually measure quality? NORC research methodologist @soubhikbarari.bsky.social reveals critical flaws and solutions for improvement to the college ranking system in his latest Expert View. Find out more: go.norc.org/46GIx8l

go.norc.org

College Rankings Are Everywhere, but What Do They Really Tell You? | NORC at the University of Chicago

Today’s college rankings are diverse and influential, but they don’t always measure what they claim or what’s most important to students.

011

Soubhik Barari @soubhikbarari.bsky.social · 22/07/2025

Reality TV is civic education. My review of @eunjikim.bsky.social's excellent new book “The American Mirage: How Reality TV Upholds the Myth of Meritocracy" up now. open.substack.com/pub/somewhat...

open.substack.com

Reality TV is civic education (book review)

Eunji Kim shows how reality TV reinforces meritocracy and shapes American politics.

120

Soubhik Barari @soubhikbarari.bsky.social · 23/06/2025

Fun facts: though everyone knows Zohran is the media darling of the NYC mayor race, Lander's arrest last week skyrocketed his mentions in the national media, more than any other candidate on any day this year. (Also Cuomo has clearly been the more consistently salient candidate in print/broadcast.)

120

Soubhik Barari @soubhikbarari.bsky.social · 28/05/2025

How the public feels about AI. My latest on Substack: a short but sweet round-up of some public opinion studies over the past ~7 years on AI. Some surprising patterns, some not so surprising, and a lot we don't know. Link below 👇

120

Reposted by Soubhik Barari

Soubhik Barari @soubhikbarari.bsky.social · 13/05/2025

This week @aapor.bsky.social, I'm excited for my @norc.org methodology colleagues to present (and debut!) some exciting work at the intersection of #ai and #surveys. A quick round-up 🧵 ⤵️

198

Soubhik Barari @soubhikbarari.bsky.social · 13/05/2025

This week @aapor.bsky.social, I'm excited for my @norc.org methodology colleagues to present (and debut!) some exciting work at the intersection of #ai and #surveys. A quick round-up 🧵 ⤵️

198

Reposted by Soubhik Barari

Soubhik Barari @soubhikbarari.bsky.social · 15/04/2025

Another survey methods talk at 𝗡𝗬𝗔𝗔𝗣𝗢𝗥 coming up next week to put on your calendar: Wed April 23 - 𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝘃𝗲 𝗔𝗜 𝗳𝗼𝗿 𝗦𝘂𝗿𝘃𝗲𝘆 𝗥𝗲𝘀𝗲𝗮𝗿𝗰𝗵 (led by @yamilrvelez.bsky.social , @joshua-lerner.bsky.social , and myself) RSVP below 👇

175

Soubhik Barari @soubhikbarari.bsky.social · 15/04/2025

175

Reposted by Soubhik Barari

NORC at the University of Chicago @norc.org · 01/04/2025

How can we redesign college ranking systems to better serve students and advisors in their college search? Check out this upcoming webinar for insights from NORC's @susanpaddock.bsky.social and @soubhikbarari.bsky.social. Learn more and register: go.norc.org/41S7sn5

go.norc.org

A New Approach to Creating a College Ratings System

NORC has now partnered with Vanderbilt and several other institutions to develop an alternative approach to providing students and their advisors with information to help their college search...

242

Soubhik Barari @soubhikbarari.bsky.social · 31/03/2025

New in 𝗧𝗼𝗯𝗮𝗰𝗰𝗼 𝗖𝗼𝗻𝘁𝗿𝗼𝗹 w/my @sdc.norc.org colleagues @kostygina.bsky.social, Chandler Carter, @andarist.bsky.social, and @sherryemery.bsky.social: on political appeals in nicotine pouch ads. (Which, as "Torches of Freedom", reminds us, is not new). PDF version: soubhikbarari.com/writings/tuc...

033

Reposted by Soubhik Barari

Klint Kanopka @klint.bsky.social · 03/03/2025

Can AI-powered textbots improve survey interviews? Join PRIISM and @soubhikbarari.bsky.social to examine how they can enhance response quality and scalability, while considering challenges in bias and respondent experience. If you're in NYC, join us in person! steinhardt.nyu.edu/events/ai-as...

steinhardt.nyu.edu

PRIISM | AI-Assisted Conversational Interviewing: Effects on Data Quality and User Experience

Can AI-powered textbots improve survey interviews? Join PRIISM and Soubhik Barari to examine how they can enhance response quality and scalability, while considering challenges in bias and respondent ...

121

Reposted by Soubhik Barari

Matt Graham @mattgraham.bsky.social · 24/02/2025

When gauging public reactions to the latest controversy, it's important not to conduct or share misleading polls. Our new paper in Public Opinion Quarterly is a template for how to (and how not to) poll about the effect of political developments. Gift link: academic.oup.com/poq/advance-...

28733

Reposted by Soubhik Barari

NORC at the University of Chicago @norc.org · 18/02/2025

🚨 Upcoming Webinar Alert 🚨 Check out this webinar from NORC's @soubhikbarari.bsky.social and @joshua-lerner.bsky.social on using NLP and generative AI in survey research! RSVP for the webinar on February 26! wapor.org/resources/wa...

Promotional image for a WAPOR webinar titled 'Using NLP and Generative AI in Survey Research'. Features headshots of two speakers, Soubhik Barari from University of Chicago and Joshua Lerner from University of Chicago, with event details including the date as Wednesday, February 26, from 13:00 to 14:30 UTC, and noting online attendance.

053

Reposted by Soubhik Barari

Jae Yeon Kim @jaeyeonkim.bsky.social · 22/12/2024

Some folks asked me how to build a data science career as a social scientist. I published an article on this with other computational social scientists last year: cambridge.org/core/journal...

cambridge.org

Training Computational Social Science PhD Students for Academic and Non-Academic Careers | PS: Political Science & Politics | Cambridge Core

Training Computational Social Science PhD Students for Academic and Non-Academic Careers - Volume 57 Issue 1

11810

Reposted by Soubhik Barari

Jenny Benz @jennybenz.bsky.social · 06/12/2024

Coverage of some results from a population segmentation study we conducted ahead of the election that revealed pervasive cynicism across almost all groups of Americans. Always a joy to collaborate with @trosenstiel.bsky.social, @soubhikbarari.bsky.social, and colleagues.

042

Reposted by Soubhik Barari

NORC at the University of Chicago @norc.org · 22/11/2024

Can AI-powered survey interviews boost response quality? Our study shows conversational LLMs elicit richer, more detailed responses vs standard formats. Read the full brief from NORC staff on enhancing survey methodology with GenAI: go.norc.org/3V2G81D

Comparison graphic between a Standardized Interview and a Conversational AI Interview, both addressing the question 'What do you think is the most important problem facing this country today?' The Standardized Interview features a single-word answer 'inflation.' The Conversational AI Interview includes a follow-up interaction where the AI acknowledges the response and asks if inflation is considered the most important issue, with additional commentary on the impact of inflation on people's everyday lives and quality of life.

165

Reposted by Soubhik Barari

Ariel Edwards-Levy @aedwardslevy.bsky.social · 25/11/2024

this AP-NORC poll on Thanksgiving foods is from last year, but I simply cannot stress enough how much I continue to enjoy the phrase "Casserole unspecified/other" apnorc.org/projects/tur... via apnews.com/article/ap-p...

Table showing least favorite Thanksgiving foods among US adults who celebrate. Topping the list are cranberries/cranberry sauce (18%), turkey (12%) and stuffing/dressing (8%)

4405

Reposted by Soubhik Barari

Tyler Simko @simko.bsky.social · 25/11/2024

Thanks @mirya.bsky.social! See descriptives on when / where nationally salient partisan language appears in local politics (“climate change,” “gun control,” etc.). Identifiably partisan language is (1) common, (2) especially in larger cities, and (3) local D / R “slant” correlates with pres. vote.

Figure shows the relationship between the number of time members of Congress use certain partisan phrases, and how often those phrases are used in local government meetings.

Plot shows city-level distributions of partisan language in local political meetings.

Two figures showing patterns in the appearance of partisan language in local government meetings. Nationally salient terms are used more often in large places. The “slant” of partisan language that leans Democratic or Republican is correlated with local presidential voting patterns.

0254

Soubhik Barari @soubhikbarari.bsky.social · 25/11/2024

🚨New paper out in 𝐏𝐮𝐛𝐥𝐢𝐮𝐬🚨 This one was a labor of love with @simko.bsky.social synthesizing many of our thoughts (and work) on the intersection of computational social science and local political economy over the past few years. Thanks @mirya.bsky.social for the bump!

062

Soubhik Barari @soubhikbarari.bsky.social · 22/11/2024

Excited to share a first look at our AI-in-survey research from this summer! @norc.org

051

Reposted by Soubhik Barari

Soubhik Barari @soubhikbarari.bsky.social · 13/11/2024

Still digging into the election? Tune in to NY@aapor.bsky.social's roundtable on Nov 21. We have a great lineup from the networks (@kabirkhanna.bsky.social /CBS, Arnon Mishkin/Fox, @trussler.bsky.social/NBC), polling (@pollsterpatrick.bsky.social), and academia (@michaelsb.bsky.social). RSVP↓

144

Soubhik Barari @soubhikbarari.bsky.social · 13/11/2024

144

Reposted by Soubhik Barari

Drew Engelhardt @amengel.bsky.social · 13/11/2024

This is exactly @eunjikim.bsky.social's research area. Reality TV shapes views about the American Dream: onlinelibrary.wiley.com/doi/10.1111/... Police procedurals likely influence attitudes about the carceral state (work in progress) Entertainment TV has under-appreciated effects.

47429

Reposted by Soubhik Barari

NORC at the University of Chicago @norc.org · 28/10/2024

Vanderbilt University asked NORC to examine college ranking systems and their methodologies. Find out what we learned and get the details on our work with Cosmic Explorer in the latest NORC Now.

go.norc.org

NORC Conducts Comprehensive Methodological Review of College Ranking Systems

Stories of Insight & Impact

012

Soubhik Barari @soubhikbarari.bsky.social · 18/10/2024

I'm excited to be hosting (NY)@aapor.bsky.social's virtal 🗳️pre-election🗳️ roundtable on October 28th. We're joined by an illustrious group of experts: @ryanenos.bsky.social, @nataliej.bsky.social, @karthick.com, and Ruth Igielnik. Don't miss out! (RSVP 👇)

153

Reposted by Soubhik Barari

Soubhik Barari @soubhikbarari.bsky.social · 11/09/2024

Excited to kick off NYAAPOR's fall series on public opinion/elections/data science. In two weeks, Professor @eunjikim.bsky.social will lead an in-person discussion of her amazing POQ using Yelp reviews to measure anti-Chinese bias during the pandemic. NYC-area folks, RSVP in the reply!

111

Soubhik Barari @soubhikbarari.bsky.social · 11/09/2024

111

Soubhik Barari @soubhikbarari.bsky.social · 07/06/2024

NYC #survey researchers, political scientists, and #polling enthusiasts: Don't miss NYAAPOR's upcoming paper talk/discussion/happy hour this coming Tuesday at 6:30 in Midtown 🍻. Featuring Jason Barabas on the importance of bridging polling and #democracy. Register here: tinyurl.com/nyaapor-poq

000

Reposted by Soubhik Barari

Soubhik Barari @soubhikbarari.bsky.social · 03/06/2024

I’m excited to say my study of political signals from corporate brands on social media is now published in a special issue of the Journal of Quantitative Description! doi.org/10.51685/jqd... I’ll be presenting @ ICWSM on Tuesday at 11:30, so drop by if you’re around ⚡️.

011

Soubhik Barari @soubhikbarari.bsky.social · 03/06/2024

011

Reposted by Soubhik Barari

Andrew Mercer @awmercer.bsky.social · 29/05/2024

I cannot stress this enough, right now if you see a dramatic or alarming headline about young people based on a poll, you should be immediately suspicious. The problem is specifically with online opt-in polls, but these are the ones that most frequently produce these crazy results that go viral.

210849

Soubhik Barari @soubhikbarari.bsky.social · 23/05/2024

Another banger from NYAAPOR for #NYC #survey folks: join us on Tuesday June 4th after work for an interactive workshop on #qualitative methods! Register here: www.eventbrite.com/e/crafting-a...

011

Soubhik Barari @soubhikbarari.bsky.social · 15/05/2024

If you're in town for @aapor.bsky.social #aapor2024 and you're a nerd for complex survey weights, stop by 218 at 2pm on Friday where I'll be presenting some on-going work between @norc.org and #NCHS on combining prob and non-prob surveys 📊

051

Soubhik Barari @soubhikbarari.bsky.social · 14/05/2024

Survey folks: any good resources / advice out there on the design of 'usability' questions (beyond the generic survey satisfaction q)? Much appreciated!

000

Soubhik Barari @soubhikbarari.bsky.social · 08/05/2024

Poli sci/survey/UX folks in the NYC area: NYAAPOR and the Insights Association is hosting a *free* happy hour this Monday ahead of @aapor.bsky.social conf next week. If you're cool, we'll see you there: 😎🥂 tinyurl.com/nyaapor-drinks

tinyurl.com

Spring Fling Happy Hour with Insights Association

A fun-filled evening of research industry networking along with the Insights Association’s Central Atlantic Chapter!

021

Reposted by Soubhik Barari

Brendan Nyhan @brendannyhan.bsky.social · 03/05/2024

"it’s a mistake to argue with a book that contains no arguments."

0182

Reposted by Soubhik Barari

Soubhik Barari @soubhikbarari.bsky.social · 26/03/2024

How many Americans do you think eat pizza every single day? A recent Netflix show committed a statistical felony in answering this question. It's a statistical incrustice, leaving mushroom for improvement. My latest Substack: somewhatunlikely.substack.com/p/alotta-pizza

432