Megan Stevenson
@meganstevenson.bsky.social
economist & legal scholar studying criminal justice issues.
professor of law
professor of economics (by courtesy)
UVA
https://sites.google.com/view/megan-stevenson/home
professor of law
professor of economics (by courtesy)
UVA
https://sites.google.com/view/megan-stevenson/home
Last words from the inspirational disability activist, Alice Wong.
November 15, 2025 at 1:00 PM
Last words from the inspirational disability activist, Alice Wong.
We look at racial disparities in impact for a Philly bail reform. Even though Black defendants benefited most in an absolute sense, White defendants were disproportionately selected to benefit (had higher BRAI).
Sounds complicated? That’s the point. 13/
Sounds complicated? That’s the point. 13/
October 24, 2025 at 3:31 PM
We look at racial disparities in impact for a Philly bail reform. Even though Black defendants benefited most in an absolute sense, White defendants were disproportionately selected to benefit (had higher BRAI).
Sounds complicated? That’s the point. 13/
Sounds complicated? That’s the point. 13/
In our setting we use it to show how characteristics of beneficiaries (i.e. compliers) differ from the pool of potential beneficiaries (i.e. compliers + never takers). It’s a cool method but probably too much to explain here. Below is a little teaser graph! 12/
October 24, 2025 at 3:31 PM
In our setting we use it to show how characteristics of beneficiaries (i.e. compliers) differ from the pool of potential beneficiaries (i.e. compliers + never takers). It’s a cool method but probably too much to explain here. Below is a little teaser graph! 12/
With a 10 p.p. treatment effect, 100% of the White potential beneficiaries benefited from the reform, whereas only 14% (10/70) of Black potential beneficiaries *actually* benefited. We refer to this as the base-rate adjusted impact measure or BRAI. 8/
October 24, 2025 at 3:31 PM
With a 10 p.p. treatment effect, 100% of the White potential beneficiaries benefited from the reform, whereas only 14% (10/70) of Black potential beneficiaries *actually* benefited. We refer to this as the base-rate adjusted impact measure or BRAI. 8/
In this example, for every 100 Black defendants, 70 of them would have had to pay cash bail absent the reform (i.e. are potential beneficiaries). For White defendants, only 10 of them would have had to pay cash bail absent the reform. 7/
October 24, 2025 at 3:31 PM
In this example, for every 100 Black defendants, 70 of them would have had to pay cash bail absent the reform (i.e. are potential beneficiaries). For White defendants, only 10 of them would have had to pay cash bail absent the reform. 7/
Sometimes, people look at racial disparities in terms of the “percent change” or treatment effect compared to the control group mean for nonmonetary release.
According to this metric, Black defendants gained most: 33% increase versus 11% increase. 5/
According to this metric, Black defendants gained most: 33% increase versus 11% increase. 5/
October 24, 2025 at 3:31 PM
Sometimes, people look at racial disparities in terms of the “percent change” or treatment effect compared to the control group mean for nonmonetary release.
According to this metric, Black defendants gained most: 33% increase versus 11% increase. 5/
According to this metric, Black defendants gained most: 33% increase versus 11% increase. 5/
Typically, when researchers look at racial disparities in impact, they compare treatment effects in absolute terms.
In this example, both Black and White defendants saw a 10 p.p. increase in nonmonetary release as a result of the intervention. 4/
In this example, both Black and White defendants saw a 10 p.p. increase in nonmonetary release as a result of the intervention. 4/
October 24, 2025 at 3:31 PM
Typically, when researchers look at racial disparities in impact, they compare treatment effects in absolute terms.
In this example, both Black and White defendants saw a 10 p.p. increase in nonmonetary release as a result of the intervention. 4/
In this example, both Black and White defendants saw a 10 p.p. increase in nonmonetary release as a result of the intervention. 4/
I’m at an Airbnb and the host is conservative Christian. We talked about trans people a bit and his view was as you would expect. I told him about intersex people and showed him a pic of Alicia Roth Wiegel. Mind BLOWN.
October 4, 2025 at 7:36 PM
I’m at an Airbnb and the host is conservative Christian. We talked about trans people a bit and his view was as you would expect. I told him about intersex people and showed him a pic of Alicia Roth Wiegel. Mind BLOWN.
Paper🧵!
We....
1) develop a framework for identification w/ multiple treatments in a judge IV design
2) find that felony conviction (without incarceration) increases recidivism relative to dismissal
with @johneric.bsky.social Aurelie Ouss @winnievd.bsky.social and Kamelia Stavreva
1/
We....
1) develop a framework for identification w/ multiple treatments in a judge IV design
2) find that felony conviction (without incarceration) increases recidivism relative to dismissal
with @johneric.bsky.social Aurelie Ouss @winnievd.bsky.social and Kamelia Stavreva
1/
March 3, 2025 at 1:19 PM
Paper🧵!
We....
1) develop a framework for identification w/ multiple treatments in a judge IV design
2) find that felony conviction (without incarceration) increases recidivism relative to dismissal
with @johneric.bsky.social Aurelie Ouss @winnievd.bsky.social and Kamelia Stavreva
1/
We....
1) develop a framework for identification w/ multiple treatments in a judge IV design
2) find that felony conviction (without incarceration) increases recidivism relative to dismissal
with @johneric.bsky.social Aurelie Ouss @winnievd.bsky.social and Kamelia Stavreva
1/
Federal judge says jail cells look are packed so tight they look like the holding units of a slave ship from the 18th century
People forced to sleep on the ground 2 feet from the toilet
People forced to sleep on the ground 2 feet from the toilet
January 1, 2025 at 3:59 PM
Federal judge says jail cells look are packed so tight they look like the holding units of a slave ship from the 18th century
People forced to sleep on the ground 2 feet from the toilet
People forced to sleep on the ground 2 feet from the toilet
The primary outcomes listed from the 2021 PAP, linked in fn 9, are here. Very different. 3/
osf.io/prh8j
osf.io/prh8j
November 28, 2024 at 2:23 PM
The primary outcomes listed from the 2021 PAP, linked in fn 9, are here. Very different. 3/
osf.io/prh8j
osf.io/prh8j
The primary outcomes from the paper and the 2023 registration from fn 1 are below. Note that the trainings were completed in early 2021. 2/
www.socialscienceregistry.org/trials/11730
www.socialscienceregistry.org/trials/11730
November 28, 2024 at 2:23 PM
The primary outcomes from the paper and the 2023 registration from fn 1 are below. Note that the trainings were completed in early 2021. 2/
www.socialscienceregistry.org/trials/11730
www.socialscienceregistry.org/trials/11730
This summary of the weak IV literature by Keane and Neal is soooo good
www.annualreviews.org/content/jour...
www.annualreviews.org/content/jour...
November 26, 2024 at 10:29 PM
This summary of the weak IV literature by Keane and Neal is soooo good
www.annualreviews.org/content/jour...
www.annualreviews.org/content/jour...
I’m on a list! 7th most cited scholar in law and economics 🕺💃Saul and I are hanging out in lucky #7. But… they are a year off in my age 🤷♀️
leiterlawschool.typepad.com/leiter/2024/...
leiterlawschool.typepad.com/leiter/2024/...
November 15, 2024 at 8:28 PM
I’m on a list! 7th most cited scholar in law and economics 🕺💃Saul and I are hanging out in lucky #7. But… they are a year off in my age 🤷♀️
leiterlawschool.typepad.com/leiter/2024/...
leiterlawschool.typepad.com/leiter/2024/...
Theoretically, this should have led to lower recidivism rates, since the highest risk people were locked up. This did not happen. We can reject even small declines in recidivism. 8/
January 8, 2024 at 5:54 PM
Theoretically, this should have led to lower recidivism rates, since the highest risk people were locked up. This did not happen. We can reject even small declines in recidivism. 8/
Risk assessment had not impact on racial disparities, likely because judges already sentenced in a racially disparate manner. It led to harsher punishment for young defendants -- but human discretion mitigated the full negative impacts on young people! 6/
January 8, 2024 at 5:53 PM
Risk assessment had not impact on racial disparities, likely because judges already sentenced in a racially disparate manner. It led to harsher punishment for young defendants -- but human discretion mitigated the full negative impacts on young people! 6/
Deviation from the recommendations of the algorithm is systematic: longer sentences for Black defendants and shorter sentences for young defendants. 5/
January 8, 2024 at 5:52 PM
Deviation from the recommendations of the algorithm is systematic: longer sentences for Black defendants and shorter sentences for young defendants. 5/
Below, we compare the *actual* impact of risk assessment in the hands of humans to the *simulated* impact of sentencing by risk assessment alone (no discretion). 4/
January 8, 2024 at 5:52 PM
Below, we compare the *actual* impact of risk assessment in the hands of humans to the *simulated* impact of sentencing by risk assessment alone (no discretion). 4/
Woohoo! My paper with @jenniferdoleac.bsky.social 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/
www.aeaweb.org/articles?id=...
papers.ssrn.com/sol3/papers....
#econsky #polisky
We look at what happens when human judges are given algorithmic risk assessments to help determine who to incarcerate. 1/
www.aeaweb.org/articles?id=...
papers.ssrn.com/sol3/papers....
#econsky #polisky
January 8, 2024 at 5:51 PM
Woohoo! My paper with @jenniferdoleac.bsky.social 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/
www.aeaweb.org/articles?id=...
papers.ssrn.com/sol3/papers....
#econsky #polisky
We look at what happens when human judges are given algorithmic risk assessments to help determine who to incarcerate. 1/
www.aeaweb.org/articles?id=...
papers.ssrn.com/sol3/papers....
#econsky #polisky
So pleased to see my paper out!
It 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/
www.bu.edu/bulawreview/...
#econsky #polisky #metascience #philsci
It 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/
www.bu.edu/bulawreview/...
#econsky #polisky #metascience #philsci
January 2, 2024 at 7:37 PM
So pleased to see my paper out!
It 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/
www.bu.edu/bulawreview/...
#econsky #polisky #metascience #philsci
It 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/
www.bu.edu/bulawreview/...
#econsky #polisky #metascience #philsci
And when police officers and civilian witnesses don’t show up, cases get dropped.
The dismissal rate is twice as high for cases where a witness fails to appear at least once. This strong relationship persists after controlling for charge, criminal record, and demographics. 3/
The dismissal rate is twice as high for cases where a witness fails to appear at least once. This strong relationship persists after controlling for charge, criminal record, and demographics. 3/
December 1, 2023 at 5:58 PM
And when police officers and civilian witnesses don’t show up, cases get dropped.
The dismissal rate is twice as high for cases where a witness fails to appear at least once. This strong relationship persists after controlling for charge, criminal record, and demographics. 3/
The dismissal rate is twice as high for cases where a witness fails to appear at least once. This strong relationship persists after controlling for charge, criminal record, and demographics. 3/
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/
Failure-to-appear is not a defendant problem, it’s a systemic problem 2/
December 1, 2023 at 5:57 PM
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/
Failure-to-appear is not a defendant problem, it’s a systemic problem 2/
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/
papers.ssrn.com/sol3/papers....
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/
papers.ssrn.com/sol3/papers....
December 1, 2023 at 5:56 PM
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/
papers.ssrn.com/sol3/papers....
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/
papers.ssrn.com/sol3/papers....
