Twitter | Search | |
Michael DeCrescenzo
PhD candidate, political science, UW–Madison. I study elections, voter ID, statistics of a Bayesian flavor. Also: bikes, , GIFs
1,543
Tweets
554
Following
519
Followers
Tweets
Michael DeCrescenzo Dec 13
Replying to @kearneymw
m'employer
Reply Retweet Like
Michael DeCrescenzo Dec 13
Sounds like the math camp lesson that used local coffeeshop data did not stick
Reply Retweet Like
Michael DeCrescenzo Dec 13
Replying to @tjmahr
The decimal pt is important here
Reply Retweet Like
Michael DeCrescenzo retweeted
Ellie Murray Dec 11
Thread on industries trying to avoid of regulation by attempting to dictate the method of used to set regulatory limits.
Reply Retweet Like
Michael DeCrescenzo Dec 12
the purpose of the Microsoft AutoUpdate app is to update the Microsoft AutoUpdate app
Reply Retweet Like
Michael DeCrescenzo Dec 10
torn between loving R Markdown and loathing anything that makes me remember that CSS is a thing
Reply Retweet Like
Michael DeCrescenzo Dec 10
MathJax has some additional fonts, but how do I get them into an R Markdown document? Anybody have luck with this?
Reply Retweet Like
Michael DeCrescenzo retweeted
Michael J. Kane Dec 6
Huh. That's one way to do it.
Reply Retweet Like
Michael DeCrescenzo Dec 5
Replying to @Joshiswrite
Reply Retweet Like
Michael DeCrescenzo retweeted
Don Moynihan Dec 4
You are literally describing a veto.
Reply Retweet Like
Michael DeCrescenzo Dec 4
me: "LaTeX lets you focus on the document's *content* rather than its cosmetics" me: spends two hours trying to get the text font to match the math font
Reply Retweet Like
Michael DeCrescenzo retweeted
David Gerard Nov 28
Not the most glaring issue with , but this plot from Michael Crichton's original 1990 novel looks pretty darn weird. /1
Reply Retweet Like
Michael DeCrescenzo Nov 24
🔸 Researchers have duty to write about what they did, rather than hide what doesn't fit the paper's narrative 🔸Journals that care about honest, accurate work should elevate transparency, deemphasize perfect narrative consistency. There is no framing crisis in social science
Reply Retweet Like
Michael DeCrescenzo Nov 24
🔸 The published record is biased. Replication/reproduction efforts improve on published work & deserve a place in "the literature" 🔸 Tools for transparency and reproducibility facilitate these improvements. Their use is *good for science*—not just a gratuitous technology choice
Reply Retweet Like
Michael DeCrescenzo retweeted
Chris Albon Nov 18
Your population vs. your model
Reply Retweet Like
Michael DeCrescenzo retweeted
Anna Meier Nov 17
"'What we call ‘measurement’ is an estimate,' Dr. Schlamminger said. 'Basically you can estimate what the true value is. The true value, only the universe knows.'"
Reply Retweet Like
Michael DeCrescenzo retweeted
Danielle Navarro Nov 17
Okay. Let's do this. I'm a computational modeller and I *absolutely* support preregistration for confirmatory empirical research. But like many computational modellers I have reservations when empiricists claim that PR is a good tool for our work. Thread.
Reply Retweet Like
Michael DeCrescenzo Nov 17
Is this a place where roll-call scores aren't the appropriate measure? If people care about the Speaker because of how the Speaker structures a policy agenda, ideology estimates from individual members' votes don't seem like the relevant pt of comparison
Reply Retweet Like
Michael DeCrescenzo Nov 16
This figure from 's evaluation of the 's forecasts is something else. Mean error under 1%, standard deviation of 3%, holy moly. And this is before you take credible intervals into account.
Reply Retweet Like
Michael DeCrescenzo retweeted
Jeremy Berg Nov 16
post with statistical analysis of 538 congressional election forecasts compared with election outcomes.
Reply Retweet Like