Alex Coppock Profile picture
Assistant Professor of Political Science, Yale University
May 8, 2023 12 tweets 10 min read
Now out at @ScienceAdvances, @dianagalos and I present a "meta-reanalysis" of audit experiments studying employment discrimination on the basis of gender.

Headline finding is in the title: gender composition predicts gender bias.

doi.org/10.1126/sciadv…

Thread! In our setting, we define (with others) gender bias as the effect on hiring decisions of applying as a woman relative to a man.

A long-standing intuition in this field is that gender bias will be:

(+) in women-dominated employment contexts and
(-) in men-dominated contexts.
Jan 17, 2023 21 tweets 9 min read
Persuasion in Parallel is now out from @UChicagoPress!

In this thread:

- The three key ideas from the book I want to put in people's minds

- Suggestions for how to include it on a syllabus

- Thank yous to the many scholars whose work I replicated or reanalyzed Obligatory photo of books i... Key idea #1: The Persuasion in Parallel pattern is common

Here's a schematic instance of the PiP pattern.

Suppose on average, red triangles oppose a policy, blue circles support it.

When exposed to persuasive information, both groups update their views "in parallel" Schematic visualization of ...
Jan 28, 2019 5 tweets 3 min read
Oliver McClellan and I are very excited to share our "Lucid Validation" paper, out today at @Res_Pol!

Main finding: @lucid_hq is an aggregator of survey respondents that shows promise for researchers doing online survey experiments

Open-access link: journals.sagepub.com/doi/10.1177/20…

1/5
We followed the basic design of the "MTurk validation" paper by @AdamBerinsky, Greg Huber, and @GabeLenz, who replicated a series of original studies on MTurk. We found that the same survey experiments conducted on Lucid got similar answers. 2/5
Nov 5, 2018 7 tweets 2 min read
Brief thread on this "absence of backlash" paper with @andyguess, now out at @BJPolS!

Link to journal: ow.ly/WT5c30mveEs
Ungated: alexandercoppock.com/papers/GC_back… 1/ We would count *oppositely signed CATE estimates for two different groups* as evidence of backlash to a persuasive attempt. This did not occur in 3 experiments (on gun control, minimum wage, and capital punishment) when we split by pre-treatment measures of policy support.
Sep 12, 2018 8 tweets 10 min read
For #DeclareDesign #launchday, here’s a thread about our five #rstats packages for research design and analysis: DeclareDesign, fabricatr, estimatr, randomizr, and DesignLibrary.

5 📦’s in 5 📣’s!

@graemedblair @jasperjcooper @maqartan DeclareDesign is “ggplot for research designs.” You add together design elements – data generating processes, sampling and assignment schemes, and estimators to declare a design. declaredesign.org/r/declaredesig…