Lightnews — Scholar-powered news

Reposted by Yamil Ricardo Velez

Cambridge University Press Political Science & IR @cambup-polsci.cambridge.org · 1d

#OpenAccess from @polanalysis.bsky.social -

Crowdsourced Adaptive Surveys - https://cup.org/4pNclb0

- @yamilrvelez.bsky.social

An abstract from Political Analysis discussing a Crowd-Sourced Adaptive Survey System (CSAS) for enhancing public opinion surveys with natural language processing and adaptive algorithms.

1 1

Yamil Ricardo Velez @yamilrvelez.bsky.social · 2d

I used to believe that survey aesthetics have minimal effects on completion rates and attentiveness… until I saw that slime-green color scheme

1

Reposted by Yamil Ricardo Velez

Political Analysis @polanalysis.bsky.social · 9d

The latest issue of PA is out now. We have a great collection of papers by @yamilrvelez.bsky.social, @sysilviakim.bsky.social, @mattblackwell.bsky.social, @sophieehill.bsky.social, @dwlee.bsky.social, @melissazrogers.bsky.social, @kaipingchen.bsky.social, @samuelbaltz.bsky.social (1/2)

1 3 6

Reposted by Yamil Ricardo Velez

Julia M. Rohrer @dingdingpeng.the100.ci · Aug 25

Ever stared at a table of regression coefficients & wondered what you're doing with your life?

Very excited to share this gentle introduction to another way of making sense of statistical models (w @vincentab.bsky.social)
Preprint: doi.org/10.31234/osf...
Website: j-rohrer.github.io/marginal-psy...

Models as Prediction Machines: How to Convert Confusing Coefficients into Clear Quantities

Abstract
Psychological researchers usually make sense of regression models by interpreting coefficient estimates directly. This works well enough for simple linear models, but is more challenging for more complex models with, for example, categorical variables, interactions, non-linearities, and hierarchical structures. Here, we introduce an alternative approach to making sense of statistical models. The central idea is to abstract away from the mechanics of estimation, and to treat models as “counterfactual prediction machines,” which are subsequently queried to estimate quantities and conduct tests that matter substantively. This workflow is model-agnostic; it can be applied in a consistent fashion to draw causal or descriptive inference from a wide range of models. We illustrate how to implement this workflow with the marginaleffects package, which supports over 100 different classes of models in R and Python, and present two worked examples. These examples show how the workflow can be applied across designs (e.g., observational study, randomized experiment) to answer different research questions (e.g., associations, causal effects, effect heterogeneity) while facing various challenges (e.g., controlling for confounders in a flexible manner, modelling ordinal outcomes, and interpreting non-linear models).

Figure illustrating model predictions. On the X-axis the predictor, annual gross income in Euro. On the Y-axis the outcome, predicted life satisfaction. A solid line marks the curve of predictions on which individual data points are marked as model-implied outcomes at incomes of interest. Comparing two such predictions gives us a comparison. We can also fit a tangent to the line of predictions, which illustrates the slope at any given point of the curve.

A figure illustrating various ways to include age as a predictor in a model. On the x-axis age (predictor), on the y-axis the outcome (model-implied importance of friends, including confidence intervals).

Illustrated are
1. age as a categorical predictor, resultings in the predictions bouncing around a lot with wide confidence intervals
2. age as a linear predictor, which forces a straight line through the data points that has a very tight confidence band and
3. age splines, which lies somewhere in between as it smoothly follows the data but has more uncertainty than the straight line.

49 280 940

Reposted by Yamil Ricardo Velez

Ethan Porter @ethanvporter.bsky.social · Aug 18

Today, @catiebailard.bsky.social and I take the reins at GW's Institute for Data, Democracy + Politics! We'll work to make IDDP a one-stop shop for research on misinfo, platform manipulation, and threats to democratic ideals. To follow our work + learn about opportunities, visit: bit.ly/45GLggi

Sign up

signup.e2ma.net

6 22

Yamil Ricardo Velez @yamilrvelez.bsky.social · Jul 14

Agreed. Adding links has been my workaround, but that might assume an unrealistically high level of reviewer motivation.

1 2

Yamil Ricardo Velez @yamilrvelez.bsky.social · Jul 7

I had a blast presenting my work on tailored experiments and adaptive surveys at the Summer Institute of Computational Social Science. Nothing better than a room full of engaged and smart people getting into the weeds about algorithms, causal inference, and the messy realities of working with AI.

SICSS-Penn @sicsspenn.bsky.social · Jul 7

@yamilrvelez.bsky.social from @columbiauniversity.bsky.social opened the second week of SICSS-Penn 2025 with his talk, "Generative AI and Respondent-Centered Social Science."

1 1 16

Reposted by Yamil Ricardo Velez

American Political Science Review @apsrjournal.bsky.social · Jun 16

From our new issue: "Confronting Core Issues: A Critical Assessment of Attitude
Polarization Using Tailored Experiments " by YAMIL RICARDO VELEZ (@YamilRVelez) and PATRICK LIU (@patrickpliu) #APSRNewIssue www.cambridge.org/core/journal...

Confronting Core Issues: A Critical Assessment of Attitude Polarization Using Tailored Experiments | American Political Science Review | Cambridge Core

Confronting Core Issues: A Critical Assessment of Attitude Polarization Using Tailored Experiments - Volume 119 Issue 2

www.cambridge.org

4 6

Reposted by Yamil Ricardo Velez

Political Analysis @polanalysis.bsky.social · Jun 4

Currently in FirstView: In “Crowdsourced Adaptive Surveys,” @yamilrvelez.bsky.social introduces a methodology (CSAS) that converts open-ended text from participants into survey items and applies a multi-armed bandit algorithm to determine which questions should be prioritized in the survey.

1 5 10

Yamil Ricardo Velez @yamilrvelez.bsky.social · Apr 18

I played around with Gemma 3 27B for a few hours the other day and was impressed. There are lower quants that should run on your machine. Not sure how it performs with long context.

2

Reposted by Yamil Ricardo Velez

Soubhik Barari @soubhikbarari.bsky.social · Apr 15

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 👇

1 5 8

Reposted by Yamil Ricardo Velez

Patrick Liu @patrickpliu.bsky.social · Apr 2

🧵 Why do facts often change beliefs but not attitudes?

In a new WP with @yamilrvelez.bsky.social and @scottclifford.bsky.social, we caution against interpreting this as rigidity or motivated reasoning. Often, the beliefs *relevant* to people’s attitudes are not what researchers expect.

4 18 40

Reposted by Yamil Ricardo Velez

Tom Costello @tomcostello.bsky.social · Apr 3

New paper!

AI lets us query humanity’s collective knowledge (or a subset of it) – rapidly addressing any given concern about vaccines in detail.

So can information-focused LLM conversations shift vaccination intentions? In an RCT of 1,124 HPV vax-hesitant parents: yes!

David Rand @dgrand.bsky.social · Apr 3

🚨New AI-dialogue WP on vax hesitancy🚨
A conversation with our LLM was >2x more effective than standard CDC messaging for increasing hesitant parents' intent to vax kids against HPV
Many diff reasons for hesitancy = need for tailored responses
PDF: osf.io/preprints/ps...
Bot: healthinfobot.com/hpv

1 4 11

Reposted by Yamil Ricardo Velez

Scott Clifford @scottclifford.bsky.social · Apr 2

New working paper with two great coauthors!

Patrick Liu @patrickpliu.bsky.social · Apr 2

🧵 Why do facts often change beliefs but not attitudes?

In a new WP with @yamilrvelez.bsky.social and @scottclifford.bsky.social, we caution against interpreting this as rigidity or motivated reasoning. Often, the beliefs *relevant* to people’s attitudes are not what researchers expect.

8 25

Yamil Ricardo Velez @yamilrvelez.bsky.social · Apr 2

While there is a growing body of work suggesting information can change beliefs, effects on attitudes tend to be muted. We sketch out why this might be the case, drawing attention to the role of belief relevance. See 🧵 below!

Patrick Liu @patrickpliu.bsky.social · Apr 2

🧵 Why do facts often change beliefs but not attitudes?

In a new WP with @yamilrvelez.bsky.social and @scottclifford.bsky.social, we caution against interpreting this as rigidity or motivated reasoning. Often, the beliefs *relevant* to people’s attitudes are not what researchers expect.

1 6

Yamil Ricardo Velez @yamilrvelez.bsky.social · Feb 12

For those interested in deploying their own CSAS, you can visit tailoredexperiments.com

The code is open source.

Yamil Ricardo Velez - Tailored Experiments using Generative AI

tailoredexperiments.com

1 4

Yamil Ricardo Velez @yamilrvelez.bsky.social · Feb 12

The paper is open access: www.cambridge.org/core/journal...

Crowdsourced Adaptive Surveys | Political Analysis | Cambridge Core

Crowdsourced Adaptive Surveys

www.cambridge.org

1 4

Yamil Ricardo Velez @yamilrvelez.bsky.social · Feb 12

By bridging the gap between researchers and participants, CSAS could ultimately lead to a better understanding of public opinion.

1 1

Yamil Ricardo Velez @yamilrvelez.bsky.social · Feb 12

The method complements traditional surveys. Researchers can dedicate a few slots to CSAS questions while keeping their tried-and-true items. It allows for an exploratory approach to questionnaire construction.

1 1

Yamil Ricardo Velez @yamilrvelez.bsky.social · Feb 12

Key innovation: Instead of relying solely on expert-designed questions, CSAS lets participants shape the survey. I apply CSAS to topics such as issue salience, political beliefs, and local politics, finding that the method surfaces survey items that would likely elude experts.

1 1

Yamil Ricardo Velez @yamilrvelez.bsky.social · Feb 12

Here's how it works: Participants submit responses to open-ended questions that are converted by LLMs into structured survey items. Adaptive algorithms then optimize which participant-generated questions get shown to future participants.

1 1

Yamil Ricardo Velez @yamilrvelez.bsky.social · Feb 12

In this paper, I develop a method called the crowdsourced adaptive survey (CSAS) that allows surveys to evolve with participant input.

1 1

Yamil Ricardo Velez @yamilrvelez.bsky.social · Feb 12

🧵 Knowing which questions to ask is one of the most critical aspects of survey design, but we often rely on guesswork to devise questionnaires. What if we were to rely directly on participants?

Cambridge University Press Political Science & IR @cambup-polsci.cambridge.org · Feb 12

#OpenAccess from @polanalysis.bsky.social -

Crowdsourced Adaptive Surveys - cup.org/4aY2OH4

- @yamilrvelez.bsky.social

"This paper introduces a crowdsourced adaptive survey methodology (CSAS) that unites advances in natural language processing and adaptive algorithms..."

#FirstView

The image features the text "POLITICAL ANALYSIS" in large white letters on a crimson background, with "#OpenAccess" in smaller white letters above a yellow bar below.

2 9 27

Reposted by Yamil Ricardo Velez

Cambridge University Press Political Science & IR @cambup-polsci.cambridge.org · Feb 12

#OpenAccess from @polanalysis.bsky.social -

Crowdsourced Adaptive Surveys - cup.org/4aY2OH4

- @yamilrvelez.bsky.social

"This paper introduces a crowdsourced adaptive survey methodology (CSAS) that unites advances in natural language processing and adaptive algorithms..."

#FirstView

3 5

Reposted by Yamil Ricardo Velez

Political Psychology @ispp-pops.bsky.social · Jan 30

Special Issue Alert!

How do digital & social media fuel alternative identities, extreme narratives & online communities in times of crisis?

Submit your work exploring their role in propaganda, misinformation, and reshaping society!

ispp.org/wp/wp-conten...

@polpsyispp.bsky.social #polisky

Call for papers for a special issue. Special issue title is Responding to socio-political challenges online through radical or extreme narratives and alternative forms of collective identities. Guest Editors: Theofilos Gkinopoulos (Behavior in Crisis Lab, Institute of Psychology,
Jagiellonian University), Malgorzata Kossowska (Behavior in Crisis Lab, Institute of
Psychology, Jagiellonian University), Ana Guinote (University College London), Jesper
Strömbäck (University of Gothenburg)

1 13 27