Gabriel S. Lenz Profile
Gabriel S. Lenz

@GabeLenz

2,481
Followers
613
Following
75
Media
1,458
Statuses

Prof., UC Berkeley

Joined June 2009
Don't wanna be here? Send us removal request.
Pinned Tweet
@GabeLenz
Gabriel S. Lenz
4 years
Now that the votes are counted, a few plots and observations. It's hard to imagine two candidates more different than Trump and Romney. At the state level, however, states voted for them at very similar rates. It's almost as if voters didn't notice.
Tweet media one
10
35
169
@GabeLenz
Gabriel S. Lenz
2 years
Why did the economist copy the political scientists' book? Because he wanted to follow the leader.
Tweet media one
11
25
347
@GabeLenz
Gabriel S. Lenz
8 months
I'm tired of checking every finding I want to cite for p-hacking. If your p-values are all around 0.05, if you haven't pregistered, if your sample sizes are too small, and especially if you are reporting interactions under these conditions, please, please don't publish.
9
5
89
@GabeLenz
Gabriel S. Lenz
3 years
Now that 2020 income growth data are available, we can draw new inferences about the presidential election. Below, I'll use real disposable income growth, which includes government transfers and taxes. It gives you a sense for people's pocketbooks. The findings are wild. 1/n
6
19
85
@GabeLenz
Gabriel S. Lenz
7 years
CA Governor Brown quoted in the New York Times on his legacy will be?
Tweet media one
1
31
67
@GabeLenz
Gabriel S. Lenz
3 years
Berkeley's searching for an Assistant Professor in Race and Ethnic Politics this fall. The department has a strong preference for candidates whose scholarship focuses on African-American politics. Deadline is September 15. Spread the word!
Tweet media one
2
32
46
@GabeLenz
Gabriel S. Lenz
10 months
The real crisis in academia is not failing to put quotation marks around borrowed phrases. It's that most studies fail to replicate. We're mostly publishing false positives. Average power is below 20% across disciplines. If only billionaires would focus on the real crisis.
0
5
45
@GabeLenz
Gabriel S. Lenz
4 years
Across US history, only two presidents have won reelection with GDP growth below zero. Both are exceptions that prove the rule. From a paper with @ericguntermann and Jeff Myers that examines economic data back to the 1790s.
Tweet media one
1
15
46
@GabeLenz
Gabriel S. Lenz
10 months
Now in print and ungated:
@dbroockman
David Broockman
1 year
NEW w @aaronrkaufman & @GabeLenz : Every political scientist learns in grad school that voters use interest group ratings to help hold elected officials accountable The problem is...we don't actually know this! We find reality is not so encouraging...🧵
Tweet media one
2
52
221
0
13
43
@GabeLenz
Gabriel S. Lenz
4 years
When I wrote Follow the Leader, I really didn’t think people would follow the leader when it meant the deaths of thousands.
@billyscheel
Bill Scheel
4 years
Powerful obituary in today’s AZ Republic. Regular people are starting to boil over
Tweet media one
454
18K
61K
1
12
43
@GabeLenz
Gabriel S. Lenz
8 months
To stir up controversy again, researchers aren't nearly as suspicious as they should be when p-values are near 0.05. Say you are reading a paper with four studies, all with p-values btw 0.04-0.05. What is the chance, with 80% power, that you would see p-values in that range? 1/n
@GabeLenz
Gabriel S. Lenz
8 months
I'm tired of checking every finding I want to cite for p-hacking. If your p-values are all around 0.05, if you haven't pregistered, if your sample sizes are too small, and especially if you are reporting interactions under these conditions, please, please don't publish.
9
5
89
1
10
40
@GabeLenz
Gabriel S. Lenz
4 years
Biden outperforming Congressional Democrats seems consistent with this Bob Erikson paper on anticipatory balancing.
Tweet media one
3
7
38
@GabeLenz
Gabriel S. Lenz
4 years
About 60% of Trump voters support requiring masks in public (only 18% are opposed), and 81% of them think Trump does as well.
@ericguntermann
Eric Guntermann
4 years
I ran a survey of Americans with @GabeLenz and @A_agadjanian . This is what we found: Even With 190,000 Dead, There’s a Lot Voters Don’t Know
4
14
23
2
17
37
@GabeLenz
Gabriel S. Lenz
9 months
I have a recent paper finding that the election year economy has strongly influenced the incumbent president's chance of re-election all the way back to George Washington. This plot from Justin Wolfer's suggests that 2024 could be a good year for Biden.
Tweet media one
1
6
33
@GabeLenz
Gabriel S. Lenz
3 years
Given Trump's spectacular election-year income growth, he should have won the election. As the plot below shows, presidents typically win in landslides with growth that high. Nevertheless, he lost, making him one of the biggest losers in history conditional on income growth.
Tweet media one
1
5
30
@GabeLenz
Gabriel S. Lenz
4 years
State vote shares are even more stable relative to 2016. After four years of Trump, his vote share rose by about one percent and was incredibly stable. Voters either got what they expected from Trump or didn't notice he was president, except for Utah.
Tweet media one
3
6
25
@GabeLenz
Gabriel S. Lenz
10 months
@_adasgupta Definitely a lot of work but science is about replication. We make thousands of decisions in cleaning and merging code. Requiring transparency is huge pain but ultimately good policy. Journals in other fields, like economics, have required this for years.
1
0
28
@GabeLenz
Gabriel S. Lenz
7 years
@polanalysis In the large replication study in Science, low p-value was one of the best predictors of replication.
Tweet media one
2
7
25
@GabeLenz
Gabriel S. Lenz
7 months
@maya_sen John Cochrane thoughts (prompted by the 2020 crunch).
Tweet media one
1
0
25
@GabeLenz
Gabriel S. Lenz
9 years
Top Google searchers related to Hillary Clinton. Check out 2 and 10. http://t.co/JSsJynBmm9
Tweet media one
11
34
20
@GabeLenz
Gabriel S. Lenz
1 year
Aletheia was helpful in ultimately publishing our paper, so I wanted to describe our experience and encourage you to post your papers to it.
@namalhotra
Neil Malhotra
1 year
Another @aletheia_tweets paper by @AliaBraley , @GabeLenz et al. on "The Subversion Dilemma" was a huge success: 4/n
1
2
7
1
10
25
@GabeLenz
Gabriel S. Lenz
3 years
The simplest explanation, however, is that Trump is a dreadful candidate. He bungled the pandemic, needlessly offended many groups including veterans, etc.
2
0
22
@GabeLenz
Gabriel S. Lenz
3 years
IV is a p-hacking machine!
@xuyiqing
Yiqing Xu
3 years
(3) The 2SLS estimates are often much bigger than the OLS estimates.
Tweet media one
1
4
38
4
1
20
@GabeLenz
Gabriel S. Lenz
1 year
Can’t thank these three enough for all they have done for open science. I feel honored just to be able to contribute. The incredible support they are receiving makes me feel better about the world. Please consider donating!
2
5
17
@GabeLenz
Gabriel S. Lenz
2 years
In recorded history, the public has never experienced a fall in their spending power like they have in the last year or so. We're even well below the trendline on spending power, as measured by RDI (income after taxes, transfers, and inflation). 1/3
Tweet media one
1
7
19
@GabeLenz
Gabriel S. Lenz
4 years
Presidents tend not to run and rarely win with election-year GDP growth like Trumps.
Tweet media one
1
2
14
@GabeLenz
Gabriel S. Lenz
3 years
To the extent we believe that income growth matters, these findings raise doubts about Trump's electoral strength in 2024. It’s possible that literally any other Republican presidential candidate would’ve better handled the pandemic and won the election.
1
0
17
@GabeLenz
Gabriel S. Lenz
6 years
Democratic accountability failure. From a t-shirt at a baseball hall. (H/t George Laiolo)
Tweet media one
1
2
16
@GabeLenz
Gabriel S. Lenz
4 years
@dandekadt I grew up in that town. No joke and definitely no dry cleaner.
1
0
16
@GabeLenz
Gabriel S. Lenz
10 months
@JohnHolbein1 “No effect” could mean a precise null effect or could mean an estimate so imprecise that we learn nothing from the paper. Would have been nice for the abstract to be clear. Anyone look at the paper yet and figure out what it found?
2
0
13
@GabeLenz
Gabriel S. Lenz
4 years
In spite of partisanship and echo chambers, Trump is faring worse than any president in the last hundred years except Hoover. Incumbent presidents have had an average national margin of 6.2 points over the last hundred years. Trump's margin is about -11 in national polls.
Tweet media one
0
7
15
@GabeLenz
Gabriel S. Lenz
3 years
Voters, however, don't appear to care much about cumulative growth. Instead, they seem mainly responsive to the election year: they substitute the end for the whole. Achen and Bartels taught us this. Trump did even more spectacularly here.
Tweet media one
Tweet media two
2
1
15
@GabeLenz
Gabriel S. Lenz
6 years
NY Times story about earthquake insurance features our house. Bad sign?
Tweet media one
1
0
15
@GabeLenz
Gabriel S. Lenz
7 years
Nice summary of natural experiments on lead exposure and crime, including a CDC intervention. via @BrookingsInst
0
0
15
@GabeLenz
Gabriel S. Lenz
6 months
@MitsuruIgami Hiding weaknesses seems less than helpful to the accumulation of knowledge.
1
0
11
@GabeLenz
Gabriel S. Lenz
7 months
One of the many interesting findings from this incredible replication effort.
@I4Replication
I4R
7 months
Half of original point estimates significant at the 10% level (but insignificant at the 5% level) become statistically insignificant at the 10% threshold with our robustness checks.
1
2
17
0
0
15
@GabeLenz
Gabriel S. Lenz
10 months
@JohnHolbein1 Thanks. Seems like a precise null.
0
0
13
@GabeLenz
Gabriel S. Lenz
7 years
Nice review of the evidence on following the leader from Brian Resnick. @B_resnick via @voxdotcom
0
5
14
@GabeLenz
Gabriel S. Lenz
9 months
How much does the cliquishness in econ explain its failure to confront its problem with false positives? It's probably no worse than other fields on the low power/false positive front, but there's less angst about it and less attention to it than in psychology or other fields.
1
0
13
@GabeLenz
Gabriel S. Lenz
4 years
@anthlittle Not so worried about the polls this year, but worried on other fronts.
1
0
14
@GabeLenz
Gabriel S. Lenz
10 months
So utterly true.
1
0
12
@GabeLenz
Gabriel S. Lenz
8 months
Nothing worse than waking up to another both-sides-ism headline from the Times. Ugh.
Tweet media one
1
2
12
@GabeLenz
Gabriel S. Lenz
4 years
In a less partisan time, voters did notice differences in presidential candidates.
Tweet media one
0
1
7
@GabeLenz
Gabriel S. Lenz
4 years
Join me for the Citrin Center Postmortem on the election with Lynn Vavreck, Peter Hart, and Nathan Persily. Thursday 11/12 at 1 PM Pacific time. Register here: @vavreck @HartSurveys @persily
Tweet media one
1
2
12
@GabeLenz
Gabriel S. Lenz
8 years
Follow the leader
@GregTSargent
Greg Sargent
8 years
New WaPo poll: Partisanship influences reax to Syrian strikes. But huge swing among Rs: 22% in 2013 to 86% now!
Tweet media one
27
106
103
1
0
12
@GabeLenz
Gabriel S. Lenz
4 years
I’ve a hunch the election forecasters are underestimating voter ignorance about Trump’s response to the pandemic and how much voters will learn this fall. E.g., millions of parents could learn that other countries‘ schools are open. That’s one hell of a campaign fundamental.
0
0
12
@GabeLenz
Gabriel S. Lenz
7 years
How do we explain this distribution of coverage?
Tweet media one
0
5
11
@GabeLenz
Gabriel S. Lenz
4 years
Facing a once-in-a-century crisis, only 40% of survey respondents say Biden is more supportive than Trump on requiring masks in public, and only 41% do the same on closing nonessential businesses. Only 47% say Biden is more supportive of the WHO. Still time to learn.
@ericguntermann
Eric Guntermann
4 years
I ran a survey of Americans with @GabeLenz and @A_agadjanian . This is what we found: Even With 190,000 Dead, There’s a Lot Voters Don’t Know
4
14
23
1
7
11
@GabeLenz
Gabriel S. Lenz
3 years
This spectacular election-year income growth may help explain why Trump made inroads into so many groups between 2016 and 2020, including black voters, hispanic voters, and young voters.
0
3
10
@GabeLenz
Gabriel S. Lenz
7 years
I like this description of Trump: sub-ideological. It also describes most voters --- uncommitted to and unaware of the parties' happenchance bundles of policy positions. via @intelligencer
Tweet media one
0
3
11
@GabeLenz
Gabriel S. Lenz
4 years
Nice Whitman quote about GOP following Trump on his effort to dismantle democracy. A powerful example of following the leader.
Tweet media one
0
1
10
@GabeLenz
Gabriel S. Lenz
2 years
Best to avoid the basic human tendency to see one’s group as better. I’m mostly glad I might’ve been right on the substance and look forward to reading this book. It’s also worth saying that many scholars have contributed to a broad literature on following the leader.
0
0
10
@GabeLenz
Gabriel S. Lenz
1 year
@arthur_spirling And, when the reviews do come in, the review recommendations are drawn at random from urns with mostly black balls.
1
0
10
@GabeLenz
Gabriel S. Lenz
8 months
@AaronChalfin Check out this paper on p-value distributions. It has an important result. You can’t get consistent 0.05 p-values without p-hacking (to simplify a bit).
3
2
10
@GabeLenz
Gabriel S. Lenz
4 years
Most incumbents win by large margins, though less so in polarized periods like the present.
Tweet media one
0
0
7
@GabeLenz
Gabriel S. Lenz
2 years
A few thoughts. Scholars in all fields probably fail to read enough in other fields and to cite what they do read. I wish it were the case but am not sure political scientists are any better and regret implying that here (the joke was too funny to pass up).
2
0
10
@GabeLenz
Gabriel S. Lenz
7 years
Nice summary of research on racism and the economy. via @voxdotcom
0
1
9
@GabeLenz
Gabriel S. Lenz
1 year
@MattGrossmann We have a theory about this: support is declining, not because people don’t like democracy, but because elites have convinced Republicans that the other side is undermining democracy. Younger people seem more persuaded by this rhetoric.
0
2
9
@GabeLenz
Gabriel S. Lenz
4 years
1/n I generally think of voters’ as lacking incentives to pay close attention to politics, and so overly influenced by symbols, identities, and leader cues. Elected officials on the other hand have strong incentives to think about politics and professional staffs to help.
@BethReeseCravey
Beth Reese Cravey 🐘
4 years
Coronavirus ravaged Florida, as Ron DeSantis sidelined scientists and followed Trump - The Washington Post
0
1
1
1
3
9
@GabeLenz
Gabriel S. Lenz
4 years
@cwarshaw Beer was safer than water at the time and often had a low alcohol content. That shows up in the John Snow Cholera accounts. Cholera was less prevalent in and around breweries.
0
0
9
@GabeLenz
Gabriel S. Lenz
3 years
There are many reasons why Trump may have lost despite this income growth. It may be that income growth doesn’t matter as much as we thought. It may be that voters credited Democrats in Congress for the income growth. It may be that the electorate is too polarized.
1
0
9
@GabeLenz
Gabriel S. Lenz
4 years
1/n I understand the incentives to reopen, but we could be much more reopened if politicians listened to experts, as shown by almost every other country. Politicians are much more like voters than I thought.
1
0
9
@GabeLenz
Gabriel S. Lenz
8 years
Nice Uri Simonsohn post on interpreting logit interactions
0
3
7
@GabeLenz
Gabriel S. Lenz
3 years
Income growth under Trump was spectacular, the fourth best in the last 70 years. Wow. Trump poured money into the economy, cut taxes, and benefited from democratic support for pandemic assistance.
Tweet media one
1
0
8
@GabeLenz
Gabriel S. Lenz
4 years
1/n Covid is making a mockery of that view. Republicans governors have followed the leader here into an avoidable calamity in lives and economic devastation. Voters aren’t going forget overflowing morgues and dead friends and relatives.
1
0
8
@GabeLenz
Gabriel S. Lenz
11 months
@RyanDEnos Publish, send Christmas Cards, or perish.
0
0
5
@GabeLenz
Gabriel S. Lenz
8 years
@leedrutman Andy Healy and I have a forthcoming JOP paper that uses related data but reaches the opposite conclusion.
0
2
6
@GabeLenz
Gabriel S. Lenz
8 months
We can run a simple simulation which finds that the probability of an 80% powered study producing a p-value between 0.04 and 0.05 is just 0.027. That's low, really low given how often we see p-values around 0.05. 2/n
1
0
7
@GabeLenz
Gabriel S. Lenz
3 years
I’ve been reading a couple books a month for several years now to understand the causes of the criminal justice nightmare the US is still experiencing. Geoffrey Canada’s memoir is among the most insightful, especially the sections on his childhood.
1
1
6
@GabeLenz
Gabriel S. Lenz
7 years
Article summarizing some of my work with @dougahler , @michaelcdougal , and Jack Citrin
0
5
6
@GabeLenz
Gabriel S. Lenz
7 years
@maya_sen Review need to keep in mind that throw away comments can lead to weeks of work. Mind what you say!
1
0
7
@GabeLenz
Gabriel S. Lenz
11 months
@dylanmatt Breaking Bad was The Wire meets MacGyver.
1
0
5
@GabeLenz
Gabriel S. Lenz
7 years
@matt_blackwell Another reason why serious social scientists should always present the univariate analysis (Shameless self-promotion):
Tweet media one
Tweet media two
0
0
6
@GabeLenz
Gabriel S. Lenz
6 years
UC Berkeley is searching for Asst Profs in Political Science. Application deadline September 24th. Priority areas are Race & Ethnic Politics, Data Science, Law and Politics, IPE, Chinese Politics. Job ad here: #WomenAlsoKnow #POCAlsoKnow
0
6
6
@GabeLenz
Gabriel S. Lenz
10 months
@ryancbriggs I’ll add the self-control not to repeat “please, please, please just buy low-fee index funds.”
1
0
5
@GabeLenz
Gabriel S. Lenz
8 months
@ryancbriggs Sure wish clustering increased my effect sizes. What a depressing results. The pattern seems more consistent with publication bias than p-hacking, though I need to think it.
0
0
6
@GabeLenz
Gabriel S. Lenz
1 year
In the middle of an extraordinary housing crisis in the Bay Area, yet only one crane visible looking south and west from UC Berkeley. Unbelievable. Can you find the only one?
Tweet media one
0
1
6
@GabeLenz
Gabriel S. Lenz
5 years
@dhopkins1776 @mvinaes @PolBehavior @Nate_Cohn GW Bush messaging on Iraq may have increased support for the invasion. @sgadarian has a nice paper on that topic. David Moore has some findings as well.
1
0
6
@GabeLenz
Gabriel S. Lenz
8 months
@jon_mellon @prisonrodeo @carlislerainey I think the social sciences are in a real crisis, probably one that extends to the social sciences as well. Our journals are just jam-packed with false positives. With power as low as 20% on average, it's barely worth taking the time to look at a study.
3
1
6
@GabeLenz
Gabriel S. Lenz
2 years
Last day for nominations for the Emerging Scholar Award (10 years since PhD) from APSA’s Elections, Public Opinion, and Voting Behavior (EPOVB) section. Email nominations to john.holbein @gmail .com, ldd @stanford .edu, and glenz @berkeley .edu or DM me.
0
3
6
@GabeLenz
Gabriel S. Lenz
11 months
The real threat to humanity is humans, not AI. Every decade technology makes it easier for a small group to wipe us out. Will AI improve fast enough to protect us from ourselves?
0
0
6
@GabeLenz
Gabriel S. Lenz
1 year
A third way in which Aletheia assisted us was by providing truly valuable comments from our amazing commentators: @Matt_Graham , @DanBischoh , and @andyguess . The reviewer pool in political science can be hit or miss. @namalhotra will get you hits.
1
0
7
@GabeLenz
Gabriel S. Lenz
2 years
@cwarshaw @ericguntermann E.g., Washington may not have run for a third term, not because he so noble, but because a recession had made him mildly unpopular. We spent a ton of time going through biographies and newspapers from period.
0
0
5
@GabeLenz
Gabriel S. Lenz
10 years
Hands-on dplyr tutorial for faster data manipulation in R http://t.co/zg1OMBtOcY via @rbloggers
0
1
5
@GabeLenz
Gabriel S. Lenz
7 months
@markarian_g @Jopieboy @OlleFolke Nothing comes to mind but weighting by pop is critical in this situation. It’s such a boring question to ask in talks but I usually make myself ask it. Weighting also allows you to generalize to the population instead of precincts/counties.
2
0
2
@GabeLenz
Gabriel S. Lenz
8 years
Fascinating plot. High income Republicans noticed that Trump wasn't a normal Republican. Most others didn't or, if they did, didn't care.
@thomasjwood
Tom Wood
8 years
From today's @monkeycageblog , the striking pattern of income and white pres vote, 1948-2016 , using @electionstudies data.
Tweet media one
9
42
33
1
1
5
@GabeLenz
Gabriel S. Lenz
10 years
@thosjleeper MTurkers are scared of rejection, undergrads aren't
0
1
5
@GabeLenz
Gabriel S. Lenz
5 years
@JohnHolbein1 Agree with others here that focus should be on reciprocity but also on quality, as the average quality of reviews is low.
0
0
4
@GabeLenz
Gabriel S. Lenz
1 year
@ryancbriggs @akoustov Get used to the inanity, unfortunately. Sorry to hear about the desk reject. It’s a great paper on a super important topic. It’ll land somewhere and be appreciated eventually. Collective reasoning in the field seems slightly better than reasoning in the review process.
0
0
5
@GabeLenz
Gabriel S. Lenz
3 years
@BrendanNyhan @ericguntermann Thanks for the plug! Rarely do national candidates take deeply unpopular positions on the most important issue of an election. Trump did precisely that on the pandemic and many voters didn’t know it until the campaign. 1/2 @mattyglesias
1
0
5
@GabeLenz
Gabriel S. Lenz
8 months
@carlislerainey There not everything but, in the many labs replication projects, p-values are the best predictors of replication.
1
0
5
@GabeLenz
Gabriel S. Lenz
6 years
Join us for Terry Moe's talk about his just-released book about power and the politics of education reform after Katrina. Baxter Liberty Initiative Lecture | April 9 | 4:15 p.m. | Sutardja Dai Hall, Banatao Auditorium
Tweet media one
0
2
5
@GabeLenz
Gabriel S. Lenz
8 months
So if you've produced a paper with multiple studies with p-values in the 0.04 to 0.05 range, ask yourself how you could possibly have produced those results. If you see that distribution in another paper, don't believe it. 4/n
1
0
5
@GabeLenz
Gabriel S. Lenz
4 years
@A_agadjanian Use shift=log(2020)-log(2016). It approximates percent change but avoids the additivity and asymmetry problems with percent change.
0
0
5
@GabeLenz
Gabriel S. Lenz
7 months
@Jopieboy @OlleFolke @RiccardoCiacci It's easy to find something in nothing by giving units with small populations equal weight to large populations, especially when studying a rare event. Weighting by population or switching to count models might bring the effect in line with the national outcome of nothing.
2
0
2