So pleased to see my paper out in the BU Law Review!
This paper surveys 50+ years of randomized control trials in criminal justice and shows that almost no interventions have lasting benefit -- and the ones that do don't replicate in other settings. 1/
This is one of the most important criminal justice papers of the last decade. Counties send 50% fewer kids to prison if they (rather than states) have to foot the bill. Suggests a substantial portion of incarceration rates are because decision-makers are insulated from costs. 1/
In the US, states pay for incarceration but county-level officials (eg, judges) determine sentences
Incarceration rates fell dramatically in California after 1996, when a reform shifted the costs of incarceration (but not other responsibilities) from the state to the county.
As a researcher, I spend a lot of time alone in my fort (office) engaged in a mental world that most people around me don’t see or understand. Twitter is like having a walkie-talkie that connects you to other people, alone in their forts, who *get* this
I’m teaching a stats intro class to law students and am considering teaching them basics in R, both to give them some exposure to statistical coding as well as to use simulations and graphs to demonstrate concepts. Any recs for a basic R tutorial or other teaching materials?
I got pregnant at 23 with a man who was nowhere near fit for parenthood (nor was I at that age). I was already 6 weeks pregnant when I found out. So grateful that I had the choice to wait until I was ready for a family. My heart goes out to Texan women.
I wouldn't be the person I am, I wouldn't have the grants or the awards or the Princeton professorship or the kids I carefully and lovingly planned for, without the abortion I had many years ago.
How "risky" does someone need to be to justify pretrial detention under current law?
@sandy_mayson
and I lay out the legal & empirical framework to answer this. We show that jail is so harmful that virtually no one is "dangerous" enough to warrant it. 1/
🚨New paper!
There’s a massive system to get defendants to appear in court, including bail and pretrial detention.
Our study shows it’s not defendants who miss court most, it’s other required actors: police officers, lawyers, and civilian witnesses. 1/
I suppose this is as good a time as any to say that I will be starting a new job at UVA Law this summer. It feels funny to share this because, you know, most people are losing jobs right now. I am very lucky. But I will also miss my incredible colleagues at George Mason.
I just noticed that there are no female editors or co-editors at any of the main law & economics journals: JLE, JLEO, JLS, ALER, IRLE, EJLE. 24 total editors listed online and no women. Wuzzup with that? Embarrassing.
I won a prize! The new Ephraim Prize for an early career scholar in law & economics “whose intellectual impact has the potential to reach the legal academy, legal profession, and beyond.” Celebrating by lying in the grass instead of in bed.
#LongCovid
I want to say something about conflict of interest in research; something particularly important in the study of criminal justice. When access to data depends on maintaining warm relationships with the agencies you are studying (eg police) this distorts what you can say! 1/
I love that economists are now interested in race but I am not keen on the recent string of papers using quasi-experimental methods to test for racial bias, using bail as the setting. For one, their model of bail setting behavior is wrong, which makes the inference wrong. 1/
Every once in a while something completely unexpected and nice happens... I won a prize! The Oliver E. Williamson prize for best article published in the Journal of Law, Economics & Organization, selected from all JLEO articles in the last 3 years. :)
I’m very upset by this story. Josh has been a friend to me and many others but that does not excuse this long-standing pattern of terrible behavior. Consent loses meaning when one party wields power over the other. I support these women.
Ok guys, we’ve got to shift norms here. Massive deviations from a pre-registered plan, without acknowledging it in the paper, are NOT OK and grounds for rejection. 1/3
The idea that publication decisions should be based only on design and not on findings is not sensible. Most things in the world don't work. Given this prior, papers that demonstrate effectiveness are all else equal much more informative.
Woohoo! My paper with
@jenniferdoleac
is (finally) forthcoming at AEJ:Policy.
We look at what happens when human judges are given algorithmic risk assessments to help determine who to incarcerate. 1/
I spent the last two days putting together a slide deck to teach about racial bias in risk assessments to a group of California judges. After all that prep I'd love to use the material more than once! Anyone need a lecture on algorithmic fairness in criminal justice? DM me
When you give an algorithmic prediction of future offending to a judge to use in felony sentencing, she may use it in unexpected ways...
Tweet summary of my new paper with
@jenniferdoleac
"Algorithmic Risk Assessment in the Hands of Humans"
1/
I hate Trump but turning to the criminal legal system as the answer for everything is just such an incredibly massively bad habit.
Plus no one even understands the crime he’s accused of, including me, and I teach criminal law.
Unanimous decision by the 5th Circuit Court of Appeals says that it is an unconstitutional conflict of interest for judges to take a cut of bail money for court expenses. Important ruling for New Orleans and beyond.
Below is a new reading list of Black economists writing about crime and criminal justice. For your late-summer reading pleasure. And just in time to update your fall syllabus! Please retweet & share.
#longcovid
folks: I'm trying to get my university to pull IRB approval for a study on high intensity interval training for people with long covid. Can you point me towards sources I can cite that highlight the risks of intensive exercise for people with long covid and/or ME?
Prof.
@MeganTStevenson
has been named a Shannon Center Mid-Career Fellow, which recognizes mid-career faculty who have made significant contributions to their departments and to
@UVA
.
I’ve been finding myself increasingly uninterested in big, sweeping causal questions, eg “how does incarceration affect future employment”. I rarely believe the results of this kind of study. 1/
I agree and would add:
If you are interested in big questions, write descriptive papers. The causal inference tool kit can't provide clean answers to most questions. But facts, time trends and correlations are incredibly valuable in grounding big picture thinking.
Next time I hear the advice that people should pick big questions with imperfect identification over less important but better identified ones, I am going to scream. This is not the game this profession is playing.
If you could not do causal analysis with qualitative methods, people would literally not be able to speak or walk. What we’ve learned about causality from quantitative methods is infinitesimally small compared to what we've learned through observation and experience.
I remember how shocked some people were that you can do causal analysis with qualitative methods, too.
In fact, ethnography *can* provide very robust evidence of causation (particularly because the researcher is often literally RIGHT THERE to observe cause and effect)
What are some fun, publicly available data sets that I can give to my students to practice with in a statistics class? Relevance to law is not necessary but would be a bonus.
When you call the police to “solve” a problem you are in invoking the use of violence. You are creating violence. Arrest is inherently violent, even if no punches are thrown.
This video is a good example of that.
I won an award for a paper that argued that a “evidence-based” criminal justice practice wasn’t actually evidence-based. Epstein presented me the award. In his speech he said he loved my paper because he also hates things that are “evidence based”. That seems blackly comic now.
Wanted to share my 🚨New Essay🚨.
It’s built around a central empirical claim: that most reforms and interventions in the criminal legal space are shown to have little lasting impact when evaluated via gold standard methods. 1/
NBER crime conference, four white male speakers in a row. Sure wish we had some different voices here, especially in the study of crime and criminal justice!
Amazon is out of hand sanitizer. Of all things coronavirus, I find this failure of modern capitalism to provide a basic commodity the most alarming signal yet.
At least when it comes to the type of limited-scope interventions evaluated via RCT, the social world is full of stabilizing forces that resist change.
And causal processes are complex and context dependent, meaning success in one place will be hard to replicate in another. 3/
Call me a grinch, but I am not so keen on the movement to train legions to use causal inference tools. The basic ideas are intuitive and it’s easy to run the commands in Stata/R. But good research requires SOOO much more. The last thing we need is more dabblers.
Some thoughts about the militarized response to recent protests:
I've heard people say things like “What else was administration going to do?” or “You need to enforce the rules consistently.”
I think these arguments are based on a fallacy. 1/
Friends, I’m telling you this because I love you: everyone should use Dragon voice dictation software to write papers and emails. Your hands are delicate birds, don’t send them to work in the coal mines all day long. Plus it’s faster. 1/
I recently listened to a podcast with Daron Acemoglu in which he mentioned using voice recognition software to write his books (and maybe even papers?) Does anyone else do this?
#EconTwitter
Is there an economics literature about resilience, or the idea that people bounce back from difficult experiences? I know there is a large psychology literature on hedonic adaptation, but I’m more interested in research on economic outcomes bouncing back (or not)
I don't want to hear more talk about students right now.
I want to hear clear statements that what’s been happening in Gaza is abhorrent and unacceptable.
The proudly anti-research posture taken in this thread is really concerning. And it is based on some huge misconceptions. First, you don't need to withhold services from clients who would have otherwise gotten them in order to do an RCT. 1/7
THREAD: On RCTs, Bail Funds, and research on people threatened with incarceration: From time to time, we (and other bail funds)are approached by researchers/funders asking us to participate in Randomized Control Trials (RCTs) to "test the impact of money bail". 1/7
#FreeThemAll
I signed this letter. Responding to a teeny tiny quiet protest by calling in riot cops, spraying tear gas, and training sniper rifles on the students is a terrifying escalation. I want answers on how these decisions were made.
a petition from faculty in the college of arts & sciences at the university of virginia has already collected signatures from over 200 professors demanding answers & dropped charges. they also endorse the letter written by the history department.
I have a new paper with
@AurelieOuss
showing a) progressive prosecutors can implement meaningful bail reform and b) a sizeable increase in the fraction of defendants released without cash bail had no detectable adverse consequences in Philly. 1/n
In a new paper, Chris Slobogin and I show that almost 60% of the variation in COMPAS's Violent Recidivism Risk Score can be explained by age -- in other words, people are being labeled "high risk of violent crime" due primarily to their youth. 1/6
Pre-registration does ABSOLUTELY nothing if no one checks and no one cares. If you are a referee, check. If you are an editor, care. If you are an author, make a detailed pre-registration plan and acknowledge any deviations in your paper!! 3/3
Early childhood lead exposure increases crime rates. It’s ridiculous how little funding goes towards environmental solutions compared to what we spend locking people up.
Despite its orgy of carceralism, one of the key tells that reveals how little America actually cares about violent crime is how it refuses to address the problem of lead.
I just nominated
@jenniferdoleac
for APPAM's Kershaw Award. This award is for outstanding contributions to public policy for researchers under 40. The deadline has been extended to June 7, join me in supporting her!
This is what I said.
Those who seek social change have three options.
1. Focus on interventions with direct effects (i.e. feeding people)
2. Continue with systems-conserving change but give up expectations of large effects or the ability to identify effective interventions in advance
4/
Would love some advice about how to interpret the parallel trends robustness test from
@asheshrambachan
and
@jondr44
.
The DD graph below looks pretty good, right? I think so at least. 1/
This is why it's so important to make data and coda available. I hope there is no stigma on the original authors, absent any evidence of negligence or fraud (which I did not see in the thread). Mistakes happen.
Banning the purchase of sex 🚨DOES NOT🚨increase cases of reported rape.
A re-analysis of Ciacci (2024) shows that the paper's headline result comes from an erroneous use of Stata's regression command.
A thread from
@Jopieboy
,
@OlleFolke
, and me 1/11
These days, I don’t mind rejections. Referees are courteous and usually offer great advice. What I don’t like is having that damn paper back on my desk again.
Can someone explain to me a realistic theory that allows us to return to normal life without a) achieving herd immunity or b) finding a cure/vaccine? If not, shouldn't we expect 12-18 months of sheltering in place for everyone who can afford to do so?
Systems-conserving interventions (those that change one factor while leaving the remaining social structure intact) are unlikely to bring about meaningful benefits. 5/
Doesn’t it seem bitterly ironic that the media keeps hyping up the fact that Christian Cooper was a bird watcher from Harvard? Like, they need to associate him with symbols of whiteness and wealth to prevent people from assuming he said something threatening to provoke her.
We are highly unlikely to identify effective interventions via RCT.
We will not be able to pilot test before scaling up.
There is not a set of proven interventions that can be widely adopted. 7/
I had 8 experts in the economics of crime/criminology read the paper and they all said I characterized the RCT literature more or less correctly. Five of them people are “experts" on the Criminal Justice Expert Panel, as am I. 1/
This flawed piece from
@MeganTStevenson
is getting some attention and will likely lead some to wrong conclusions, so I'm going to rebut it below. Dr. Stevenson is an accomplished scholar and this thread challenges this particular article, not her work generally. Read on.
I have a new paper with
@bocowgill
, forthcoming in AEA P&P
"Algorithmic Social Engineering"
We apply classic strategic communication models to "fair machine learning". In a nutshell: nudging people to change behavior by tweaking an algorithm is hard! 1/
(1/9) Our new article on police-related fear (with
@agrahamphd
) shows that Black and White Americans live in different emotional worlds. This thread summarizes our findings.
Data, code, and preprint available
@socarxiv
and at:
Fully agree. It's not a niche position, roughly half our country leans conservative. It's just weird that there are almost none at law schools, or in higher education more generally.
Sandy Mayson & I have a new paper -- to our knowledge, the most comprehensive empirical study of US misdemeanor case processing yet. Despite the scale of the misd. system, and the lasting consequences for those caught in it, there's very little data. 1/
Here, I redacted Wright's complaint to remove personal information of Elyse and Angela. Wright brings a defamation claim for $108MM against them for speaking out re abuse. Note: a reporter told me that additional accusers are coming out later today.
👇Most Black Americans want to maintain or increase police presence, even if crime trends fall and police reform isn’t enacted. This is an important fact to reckon with for the defund movement.
(1/4) New study with PhD student
@LindaBalcar
. Gallup and Pew Research found that most Black Americans want to maintain or increase levels of local police patrol and spending. We tested experimentally if these policy preferences are firm. They are firm.
Can anyone recommend a short (~20p) intro to behavioral economics? I'm looking for an intuitive and accessible introduction to some of the key ideas, eg present bias. No math. Thx!
R in a nutshell: I easily wrote some code to loop through, unzip, edit and rezip a bunch of files. *But* I spent three hours trying to figure out why read.csv gave me an misaligned data frame even though Stata and excel read the csv files perfectly.
This is a fascinating article: an over-sensitive model briefly predicted Republican success in the house and for about an hour US bond yields shot up. Shows a) power of statistical modeling over beliefs and 2) investors see Republican control as risky.
When markets react to fake news
538's faulty election model moved prediction markets night of coverage
Also moved bond yields
Good news: markets process information
Bad news: even if it's
@NateSilver538
's bad data
Namely, it reifies & normalizes the idea that the bail system is designed to sort the dangerous from the not-dangerous, and place the former in jail. This is a legal myth that is racist at its core. 3/
Dear anonymous referees: thanks, whoever you are! So grateful for your thoughtful comments.
Careful refereeing is one of the most beautiful aspects of our profession. It’s a large chunk of valuable time, freely and generously given.
How can you objectively study something when a primary goal is to not piss off the group you are studying? It affects what questions you ask and how you frame the answers you find. It affects other research projects, because you want to maintain a certain sort of "reputation".3/
I gave a talk recently with a combative attendee. It was surprisingly exhilirating, particularly after 8 months of sitting in my office. It made me think of
@AgnesCallard
's excellent article on the fight club aspect of academic talks. Of course, I slayed him, and that helped. 1/
I spent over a year barely able to stand due to long Covid, and I’m still suffering the consequences.
This disease desperately needs funding to match the enormity of its burden.
Nearly 18 million adults and 6 million children suffer from Long COVID.
The time has come to start treating the Long COVID crisis as the public health emergency that it is.
This paper is evidence-based; it is based on hundreds of randomized controlled trials.
But the evidence rejects some of the central tenets of the evidence-based reform movement. 6/
I have access to all of Philadelphia court records since 2006 in digital form. (Thanks to awesome web scraping by a guy who I’m not sure if I should mention publicly by name). Contact me if you want it. For researchers only. Names removed for confidentiality
This fee schedule for the
#Philadelphia
courts is outrageous!
Court records are public information. But a $25,000 fee essentially ensures that the public won't be able to access that information.
Transparency is largely meaningless if it comes with a big price tag.
Yo economists - if its an IV design with a huge effect size it's probably incorrect. The research design might "look good" but there's often a lot of hard-to-detect muck behind the scenes. RCTs rarely find effects, and the ones that do are generally very small.
Update: I relapsed after my previous thread. A total of 5 weeks sick in bed with Covid. On the mend but still weak. Take your paxlovid and don’t push yourself, this disease is weird AF.
Also athletic women in 30s/40s are in a high risk group for long Covid, who knew
I’m quadriple-vaxxed, otherwise healthy and just spent 3 weeks laid out flat in bed with Covid. Like so sick I was barely about to move. Still recovering. Wanted to share the story in case it’s helpful to anyone. 1/
Fellow crime scholars: let’s be careful to use more precise language than “crime” in our work. When we use UCR or police data we are studying “policed crime”. Some of the biggest crimes (e.g. wage theft) are not policed yet shouldn’t get forgotten and omitted in the discussion.
A dude was caught on camera shoplifting from Walgreens; it generated well over 300 stories, in virtually every big outlet.
Walgreens settled a class-action lawsuit for stealing millions of dollars from its own employees through wage theft. One story.
Thanks for the useful suggestions everyone! The PI on the research just got back to me saying that he's already canceled the trial. So phew. I will send these resources onto the IRB department anyway just so they can be better informed for next time.
While this might be disappointing from the perspective of trying to engineer change, I argue it teaches us something important about the structure of the social world. 2/
Newly updated reading list:
Black Economists on Criminal Justice.
Please read, share with your networks, retweet, post on your website, send to journalists, use for literature reviews, and most of all, enjoy. It is a truly excellent body of work.
Did you know professors can avoid the results of a Title IX process being revealed by quitting right before they are terminated? This allows them to go from job to job without accountability. If these women hadn't spoken up, no one would ever have known.
Email from George Mason dean Ken Randall to incoming students. I'm talking with Randall tomorrow morning. If you have ideas for what you'd like to see done differently, reply here. I don't have a stake other than not wanting students and hires to suffer for a decade again.
Dropped my kid off at his first day of elementary school today. Stevie Wonder was blaring on the loudspeaker and all the staff was out on the front step (masked) to welcome the kids. I pretty much cried with gratitude and happiness.
I finally watched the video in which Arbery is shot and am just in tears. The police had the footage for two months and refused to file charges against the killer. Fuck this racist shit.
3 step method for curing phone addiction. 1) Have a baby. 2) Realize how precious life is and stop frittering it away on the internet. JK! Add an extra two hours per day of one handed phone use while breastfeeding, 3) Develop RSI in your thumbs. Phone=pain. Cured.
I love co-authoring because it’s like a little social-intellectual puzzle. You have to figure out what everyone’s comparative advantage is *in that group* and also what people need to thrive. When a coauthoring relationship really takes root it’s a great feeling.
I like and respect the authors of these papers. But I don’t like this direction of economic research and hope that future scholarship will move past thin-concept tests of racial bias and towards a deeper engagement with ameliorating racial injustice. 9/9
So interesting. I'm reading about very early use of predictive algorithms in criminal justice (1930s) and one suggestion was to build "hunch scores" which is basically a crowd-sourced prediction that somebody would reoffend, based on the opinion of inmates who know that person.
Consider a law faculty with, say, 80% white men. I feel like almost all law professors would agree, including many conservatives, that this faculty should try to hire more women and non-white men. I can’t wrap my mind around the argument that this would be illegal, even though
Focusing on bad-apple judges gives a thin sense of racism, one that is solveable by appointing new judges or setting technocratic quotas. But racism in criminal justice is a deep structural aspect of the entire system. 7/
Police officers, victims, other witnesses and private defense attorneys all fail to appear at substantially higher rates than defendants.
Failure-to-appear is not a defendant problem, it’s a systemic problem 2/