Dean Eckles

@eckles
2.2K Followers
2.2K Following
378 Posts
#networks, #contagion, #causality. prof at MIT.
Twitterhttps://twitter.com/deaneckles
Google Scholarhttps://scholar.google.com/citations?hl=en&user=oNQRPLYAAAAJ&view_op=list_works&sortby=pubdate
Websitehttps://deaneckles.com
Dean EcklesAug 22, 2023

Facebook (aka thefacebook) launched at US colleges in 2004 & 2005.

Can this staggered rollout — plus survey data on students' mental health — provide clear evidence on the effects of social media on mental health?

https://statmodeling.stat.columbia.edu/2023/08/22/thefacebook-and-mental-health-trends-harvard-and-suffolk-community-college/

thefacebook and mental health trends: Harvard and Suffolk County Community College | Statistical Modeling, Causal Inference, and Social Science

Show threadDean EcklesApr 4, 2023
I wrote a bit more about this example and the value of plotting more of the data
https://statmodeling.stat.columbia.edu/2023/04/03/when-plotting-all-the-data-can-help-avoid-overinterpretation/
Now with another version of these plots
When plotting all the data can help avoid overinterpretation | Statistical Modeling, Causal Inference, and Social Science

Dean EcklesApr 2, 2023

Seen this plot?

Watch out for onfounding from survey format changes
http://justthesocialfacts.blogspot.com/2023/03/whats-wrong.html

What's wrong?

 A series of questions from a new survey sponsored by the Wall Street Journal is getting a lot of attention.  The figure that's been circula...

Dean EcklesMar 24, 2023
Patron saint of numerical computing
Dean EcklesMar 8, 2023
"New User Experience"
Dean EcklesMar 6, 2023
This is an impressive RCT. One weird thing is that the main analysis seems to avoid using multiple observations per household, despite cluster-randomization. So no estimates of spillovers, but also just discarded data...
https://doi.org/10.1016/S0140-6736(23)00349-5
Dean EcklesFeb 27, 2023

Neat illustration of the bias–variance tradeoff in analysis of a regression discontinuity...

But then the variance turns into bias with file drawer bias
https://vincentbagilet.github.io/causal_exaggeration/summary.html

Causal Exaggeration: Summaries

This page gathers a set of summaries intended for different audiences.

Causal Exaggeration
Show threadDean EcklesFeb 25, 2023

There are analyses like eg http://controversiasbarcelona.com/2019/EvansLacko2017.pdf but with this kind of thing, I often find meta-analyses obscure huge methodological variation.

What are the most credible studies here?

Dean EcklesFeb 20, 2023
When you're reading math articles on Wikipedia and you're not sure whether maybe "abstract nonsense" is a technical term or something
Dean EcklesFeb 9, 2023
Even people with expert training incorrectly associate or mix up confidence intervals and predictive intervals (and also SEs and SDs)
http://jakehofman.com/publication/visualizing-inferential-uncertainty/
@dggoldst
How visualizing inferential uncertainty can mislead readers about treatment effects in scientific results 🏆 | jakehofman.com

When presenting visualizations of experimental results, scientists often choose to display either inferential uncertainty (e.g., uncertainty in the estimate of a population mean) or outcome uncertainty (e.g., variation of outcomes around that mean) about their estimates. How does this choice impact readers' beliefs about the size of treatment effects? We investigate this question in two experiments comparing 95% confidence intervals (means and standard errors) to 95% prediction intervals (means and standard deviations). The first experiment finds that participants are willing to pay more for and overestimate the effect of a treatment when shown confidence intervals relative to prediction intervals. The second experiment evaluates how alternative visualizations compare to standard visualizations for different effect sizes. We find that axis rescaling reduces error, but not as well as prediction intervals or animated hypothetical outcome plots (HOPs), and that depicting inferential uncertainty causes participants to underestimate variability in individual outcomes.

jakehofman.com