ARGHHH it's so frustrating. And there's a better way, too! We even wrote a follow-up on the Trump indictments specifically to try to make it even easier to avoid using change questions:
alexandercoppock.com/barari_etal_...
(forthcoming at POQ!)
0 replies
1 reposts
3 likes
This post resonates with my view that all we ever really learn about is the ITT and we should be grateful we ever even learn about the ITT in the first place
0 replies
2 reposts
7 likes
Tim Gill has moved on from trolling academics to trolling... Phish?
2 replies
0 reposts
4 likes
Thank you Eric, thank you Margit and Josh for this huge honor. I'm grateful to the committee and to everyone who helped me make PiP - esp. those whose work I replicated or reanalyzed.
See you in Philly for APSA2024 -- here's a throwback to APSA2023 for the author-meets-critics alternative for PiP
1 replies
0 reposts
11 likes
That would be awesome, thank you!!
1 replies
0 reposts
1 likes
+1 to the main takeaway from this post that pilots are great for reasoning about your eventual standard error, less great for learning about the treatment effect.
Complements this chapter nicely: book.declaredesign.org/lifecycle/pl...
And I couldn't resist rewriting these loops in DeclareDesign:
1 replies
4 reposts
10 likes
Reposted by Alex Coppock
Make me king for a day and I swear to all that is holy, I'll require that anyone attempting to field a "support more/support as much/support less" question will be required to spend fifteen goddamn minutes googling "response substitution." Such a huge whiff.
0 replies
2 reposts
3 likes
When I'm doing meta-analysis work, I ask a lot of people for their replication datasets.
Good news is that norms appear to have really shifted -- for one project:
2 "let me get back to you"
2 "data unavailable"
8 data shared
4 left on read
50% is progress!
{please post your data, TY!}
0 replies
0 reposts
5 likes
Reposted by Alex Coppock
"... people are not good at articulating the reasons for their choices. So they may say – and honestly feel! – that an issue is important. But when they say that, they are not revealing that this issue will affect how they vote."
0 replies
3 reposts
5 likes
Not to be glib -- I 100% agree that academic incentives push scholars away from correcting others' work -- but I do think @steveharoz.com's efforts here could legitimately be a CV line under "disciplinary service"
0 replies
0 reposts
3 likes
Here's a nice new paper from the fully blueskyless author team of Philip Moniz, Rodrigo Ramirez-Perez, Erin Hartman, and Stephen Jessee, showing that survey exp. effects are similar for "eager" and "reluctant" respondents, with positive implications for generalizability
doi.org/10.1017/pan....
0 replies
11 reposts
22 likes
dark
1 replies
1 reposts
18 likes
Reposted by Alex Coppock
Today's @mattyglesias.bsky.social newsletter focuses on my book, The Invented State: Policy Misperceptions in the American Public. www.slowboring.com/p/the-misinf...
0 replies
3 reposts
12 likes
Reposted by Alex Coppock
My experiments seminar at @dartmouthartsci.bsky.social in the student newspaper - amazed, as always, at how wonderful my students are and how much we can do together www.thedartmouth.com/article/2024...
0 replies
4 reposts
30 likes
This op-ed by @ryanenos.bsky.social makes the case that universities should practice "forbearance" when it comes to responding to student protesters who are breaking some university policies when protesting peacefully.
www.thecrimson.com/article/2024...
3 replies
3 reposts
17 likes
"Destination: Various Classrooms"
should be the title of the Michelson biopic
0 replies
0 reposts
1 likes
Q is such a missed opportunity!
0 replies
0 reposts
6 likes
I missed this piece last month -- political science experimentalists especially should give it a read!
0 replies
0 reposts
0 likes
Those editor comments are crazy bad!
Obvi, the false positive rate (fraction of experiments that are significant, conditional on the null being true) is controlled by alpha and does not go *up* when sample sizes increase. 🙄
0 replies
0 reposts
3 likes
yes!
a good template is something like
using [research design] we find that the average effect of [treatment] on [outcome] is [estimate]([std.error])
1 replies
0 reposts
1 likes
Reposted by Alex Coppock
The childcare reimbursement form for @APSAtweets 2024 conference in Philly is now open! It is due by August 2nd. It covers $500 of expenses, which can include a Philly based provider OR travel costs for a non-guardian caregiver connect.apsanet.org/apsa2024/chi...
1 replies
20 reposts
25 likes
do it!
"is associated with" isn't a get our of jail free card and
"these findings are descriptive" isn't a legitimate dodge and
"while not causal, these findings are suggestive" is just innuendo
1 replies
0 reposts
1 likes
Reposted by Alex Coppock
I’m also coming down more and more on the side of “no, causal over-claiming cannot be blamed on journalists.” Sometimes, they may be the source of misinterpretations. But often, the articles themselves are carefully crafted to imply causality while maintaining plausible deniability.
1 replies
2 reposts
16 likes
We thank the commenter for their kind words: "This is actually very timely advice." In the revised draft, we are sure to insist on \textit{timeliness} as an important feature of unsolicited advice (p. 12).
0 replies
0 reposts
1 likes
unsolicited advice: do not group reviewer comments together then provide a summary response -- go through each reviewer's points point-by-point. Otherwise it's too easy to mischaracterize what a reviewer was getting at or worse, to skip responding to some comments altogether.
1 replies
0 reposts
8 likes
Reposted by Alex Coppock
So excited this is finally out and that a UCD colleague and I could contribute!
For anyone interested, here is a link to our replication’s discussion paper, published last June:
ideas.repec.org/p/zbw/i4rdps...
0 replies
2 reposts
3 likes
My contribution to this exciting initiative was an "almost entirely successful" robustness replication of López-Moctezuma, Wantchekon, Rubenson, @thomasfujiwara.bsky.social, and Pe Lero (AJPS 2022).
Thanks to @i4replication.bsky.social for their incredible efforts!
0 replies
3 reposts
5 likes
oof, hell yeah -- was looking for my soundtrack, this is it
0 replies
0 reposts
1 likes
😠
0 replies
0 reposts
0 likes
for sure :)
I'm just grumpy -- why present a mediation analysis at all when there's simply no design-based reason to think it's correct. We already thought it was a possibility before doing any data analysis, so this isn't even "suggestive"
1 replies
0 reposts
1 likes
would like to also include author's caveat that the mediation analysis could be wrong
1 replies
0 reposts
3 likes
Browsing Josh Knobe's website to discover a section on responses to his work... Don't believe I've seen a political scientist's website do anything like this! @xphilosopher.bsky.social
0 replies
4 reposts
12 likes
Reposted by Alex Coppock
Hi Folks: Firing up this account for the first time to note that today marks one year since the Russian government detained my colleague Evan Gershkovic in an act of supreme cynicism and cowardice. WSJ’s response is this stunning front page. #FreeEvan
16 replies
370 reposts
1127 likes
oh man. NA, NA, NA, NA, NA, NA, NA
0 replies
0 reposts
3 likes
ooh that's interesting maybe if you don't know the show its just "that doofus"
0 replies
0 reposts
1 likes
yes, the negative correlation with subject partisanship really shows up for her... funny how it's there even for those who don't know the show!
1 replies
0 reposts
1 likes
YES so well put
1 replies
0 reposts
0 likes
Furiously making appendix figures for #MPSA2024 over here.
...for a study with @ethanvporter.bsky.social and Don Green, we did a warmup exercise in which we asked subjects to guess the partisanship of characters from The Office.
could not resisting posting
5 replies
12 reposts
29 likes
Reposted by Alex Coppock
The econ paper that found a 42-64% increase in rapes after prostitution banned (in Sweden)? Turns out result from coding error + Stata prioritization rues + seasonality.
* always plot raw data
* always sniff test your estimates
* open science works!
On X or here:
drive.google.com/file/d/1oGP3...
3 replies
63 reposts
178 likes
Would love thoughts from survey researchers and REP scholars about the measurement properties of this new question.
To my mind it addresses the persistent MENA mismeasurement and has consistent race/ethnicity then national origin structure across groups
Should we update AmPol surveys to this Q?
0 replies
0 reposts
2 likes
ooh even prettier!
0 replies
0 reposts
1 likes