Voting From Jail WORKING PAPER Anna HarveyOrion Taylor

2025-04-24 0 0 879.82KB 42 页 10玖币

侵权投诉

Voting From Jail∗

WORKING PAPER

Anna Harvey†Orion Taylor‡

October 14, 2022

Abstract

We leverage new data on daily individual-level jail records and exploit the timing of incarceration

to estimate the causal eﬀects of jail incarceration on voting from jail in 2020. We ﬁnd that

registered voters booked into county jails for the full duration of 2020 voting days were on

average 46% less likely to vote in 2020, relative to registered voters booked into the same jails

within 7 – 42 days after Election Day. The estimated negative eﬀect of incarceration on voting

from jail was much larger for Black registered voters, who were 78% less likely to vote in 2020

if booked into county jails for the full duration of 2020 voting days, relative to Black registered

voters booked into the same jails just after Election Day. Placebo tests indicate no eﬀects of 2020

jail incarceration on the 2012 or 2016 turnout of registered voters. We ﬁnd inconsistent eﬀects

of jail incarceration on voter registration in 2020, and eﬀect sizes of comparable magnitude for

turnout unconditional on registration status. Our ﬁndings reveal the pressing need to enable

voting-eligible incarcerated individuals to exercise their constitutional right to vote, and to

address troubling racial disparities in the eﬀect of jail incarceration on the exercise of the right

to vote.

∗We thank Roger Pharr and Andrea Wang for their contributions to data collection and matching. We

thank Ted Enamorado and Ariel White for their helpful insights. This project was supported by Arnold

Ventures and the Chan Zuckerberg Initiative.

†Department of Politics, Public Safety Lab, New York University, anna.harvey@nyu.edu

‡Public Safety Lab, New York University, ojt212@nyu.edu

arXiv:2210.06542v1 [stat.AP] 12 Oct 2022

1 Introduction

In every state, otherwise voting-eligible individuals incarcerated pretrial or to serve a misdemeanor

sentence remain legally entitled to vote while incarcerated (The Sentencing Project,2020,2022).

The vast majority of the approximately 650,000 individuals incarcerated in county jails on any

given day are being held pretrial or to serve misdemeanor sentences (Sawyer and Wagner,2022).

Anecdotal evidence suggests, however, that many of those incarcerated in county jails are not be-

ing given adequate opportunities to exercise their right to vote (The Sentencing Project,2020).

Concerns have also been raised about possible racial disparities in ballot access for those incarcer-

ated in county jails (The Sentencing Project,2022). Yet we lack reliable causal estimates of the

impacts of being incarcerated in a county jail on the exercise of the right to vote, and of any racial

heterogeneity in those impacts.

Some recent papers have sought to estimate the impacts of prison and jail incarceration on indi-

viduals’ post-release voting behavior (Gerber et al.,2017;White,2019b;McDonough, Enamorado

and Mendelberg,2022). White and Nguyen (2022) describe the extent of voting from prison in

Maine and Vermont—the two states permitting individuals serving felony sentences to vote from

prison—but do not estimate the causal impacts of prison incarceration on voting from prison.

We leverage new data on daily individual-level jail records from the Jail Data Initiative and

exploit the timing of incarceration to estimate the causal eﬀects of being incarcerated in a county

jail during 2020 general election voting days on the probability of voting. We probabilistically match

944,985 individual-level booking records from 936 jail rosters with 195,655,326 voter records from 42

corresponding statewide ﬁles for a period of 180 days centered on Election Day (November 3, 2020).

We identify individuals whose periods of jail incarceration began during 2020 voting days in their

state (either mail-in or in-person), or within windows ranging between 7 and 42 days after Election

Day. Our primary analyses are conducted within the sample of jailed individuals who match to

voter records with match probability p > 0.75; we also replicate analyses for the sample with match

probability p > 0.95. We interpret matches as indicating registered voters who were incarcerated

in county jails during our time period of interest. We source information about jailed registered

voters from both jail and voter records. For post-Election Day control group windows ranging

between 7 and 42 days, we conduct a series of balance tests on individual-level and booking-level

characteristics, including age, race, gender, partisanship, and number and kind of booking charges,

to identify the pre-Election Day treatment group windows within which individuals booked into

county jails during 2020 voting days (either mail-in or in-person) were observably indistinguishable

(joint F-test p-value >0.10) from individuals booked into the same jails during post-Election Day

control group windows. For records with match probability p > 0.75, the balanced samples range

in size from 57,821 (28-day control group window) to 103,091 (42-day control group window).

Our identiﬁcation strategy rests on the assumption that individuals booked into the same jail

within a narrow temporal window, who are indistinguishable on observed characteristics, are also

indistinguishable on unobserved characteristics. If this assumption is correct, we can use the 2020

turnout of individuals booked into county jails just after the last 2020 voting day as a valid coun-

terfactual for the turnout we likely would have observed among individuals booked into the same

jails during 2020 voting days, had they not been incarcerated during voting days.

Turnout among registered voters in 2020 is estimated by the U.S. Census Bureau to have been

91.9%.1Registered voters booked into county jails just after the last 2020 voting day had much

lower average turnout in 2020, relative to the general population, underscoring the importance of

identifying an appropriate counterfactual turnout rate for those jailed during 2020 voting days. For

the sample with match probability p > 0.75, turnout among registered voters who were booked

into county jails within 7 – 42 days after 2020 Election Day ranged between 34.9% and 35.7% in

our samples, or between 56% and 62% less than the registered voter turnout observed in the U.S.

population. With a less appropriate control group, we might mistakenly attribute some of the

decreased turnout among those jailed prior to the election to the eﬀects of jail incarceration, rather

than to a lower propensity to vote.

Within our balanced samples, we estimate the impacts of (1) being booked into a county jail

during 2020 voting days (either mail-in or in-person) for any length of time and (2) being booked

into a county jail during 2020 voting days for some proportion of voting days on the probability

that a registered voter voted in the 2020 election. In models that include all individual- and

booking-level covariates along with jail and week ﬁxed eﬀects, we ﬁnd that registered voters who

were booked into county jails during 2020 voting days for any length of time were on average 3.0 –

3.3 percentage points or 8.6% – 9.3% less likely to have voted in 2020, relative to registered voters

who were booked into the same jails within 7 – 42 days after Election Day (avg. eﬀect magnitude

= 9.0%; p <0.01).

The impacts of incarceration on voting from jail in 2020 increased with the duration of in-

carceration. Including all individual- and booking-level covariates along with jail and week ﬁxed

eﬀects, registered voters who were booked into county jails during 2020 voting days for the full

span of voting days in their state were on average 13.7 – 22.9 percentage points or 38.4% – 65.2%

less likely to have voted in 2020, relative to registered voters who were booked into the same jails

within 7 – 42 days after Election Day (avg. eﬀect magnitude = 46.1%; p <0.01). These estimates

imply that registered voters who were booked into county jails during 2020 voting days for the

full span of voting in their state voted from jail at rates ranging between 12.2% and 22.0%. In a

series of placebo tests, we ﬁnd no consistent eﬀects of the timing of individuals’ 2020 bookings on

their probabilities of having voted in the 2012 or 2016 elections. All estimates are of comparable

magnitude in the sample of linked records with match probability p > 0.95.

We also explore racial heterogeneity in the eﬀects of incarceration on voting from jail in 2020,

ﬁnding that the eﬀect of jail incarceration was signiﬁcantly larger for Black registered voters, relative

to white registered voters. White registered voters booked into county jails just after Election Day

2020 had turnout rates ranging between 37.7% and 38.3%, considerably higher than the turnout

rates for Black registered voters booked into county jails just after Election Day 2020, which ranged

between 27.8% and 28.2%. Jail incarceration also had less of a negative eﬀect on turnout for white

1https://www.census.gov/content/dam/Census/library/publications/2022/demo/p20-585.pdf

registered voters. Including all individual- and booking-level covariates along with jail and week

ﬁxed eﬀects, white registered voters who were booked into county jails during 2020 voting days

for the full set of voting days were on average 12.5 – 22.8 percentage points or 32.6% – 60.2% less

likely to have voted in 2020, relative to white registered voters who were booked into the same jails

within 7 – 42 days after Election Day (avg. eﬀect magnitude = 39.8%; p <0.01). These estimates

imply that white registered voters booked into county jails during 2020 voting days for the full set

of voting days voted from jail at rates ranging between 15.1% – 25.8%. The estimated negative

eﬀect of incarceration on voting from jail was an additional 5.0 – 11.8 percentage points or 38.2%

– 55.0% larger for Black registered voters (avg. eﬀect magnitude = 45.9%; p <0.05). The total

estimated eﬀect of jail incarceration for Black registered voters booked into county jails during

2020 voting days for the full span of voting days was a 17.9 – 34.6 percentage point or 61.3% –

121.8% decrease in turnout, relative to Black registered voters who were booked into the same jails

within 7 – 42 days after Election Day (avg. eﬀect magnitude = 77.9%). These estimates imply that

Black registered voters booked into county jails during 2020 voting days for the full set of voting

days voted from jail at rates ranging between 0.0% and 11.3%. Estimates of racially heterogeneous

eﬀects are of comparable magnitude when we restrict the sample to states that report voter race

(eliminating the records for which race is predicted by L2).

We also explore the eﬀects of jail incarceration on voter registration behavior, and on voting

behavior unconditional on registration status. Jail incarceration in 2020 had inconsistent eﬀects

on voter registration. The eﬀects of jail incarceration on turnout in 2020 unconditional on voter

registration are of similar magnitude to the eﬀects estimated from the sample of registered voters

only.

The nature of our data prevent us from exploring the precise mechanisms generating both the

overall decreases in turnout for individuals booked into county jails during 2020 voting days, and

the substantially larger decreases for Black individuals. Nevertheless, at a minimum our ﬁndings

reveal a pressing need for states and counties to take steps to ensure that voting-eligible incarcerated

individuals are given adequate opportunities to exercise their constitutional right to vote, and to

address troubling racial disparities in the exercise of that right.

2 Incarceration and Voting

Whether and by how much being incarcerated in a county jail interferes with an individual’s exercise

of the right to vote, and whether there are any racial disparities in the eﬀects of incarceration on

voting from jail, are empirical questions. We lack systematic evidence on these questions.

Some researchers have sought to estimate the post-release impacts of periods of prison and jail

incarceration on individuals’ exercise of their right to vote (Gerber et al.,2017;White,2019b;Mc-

Donough, Enamorado and Mendelberg,2022). Gerber et al. (2017) used state prison and sentencing

records to investigate the impacts of prior periods of prison incarceration on post-release voting in

the 2012 election, ﬁnding few diﬀerences in turnout between those incarcerated and released from

state prison between 2008 and 2012 and either observably comparable individuals in the Pennsyl-

vania voter ﬁles, or observably comparable convicted defendants who received sentences other than

prison incarceration. Gerber et al. (2017) also found few diﬀerences in post-release turnout between

those convicted defendants sentenced to jail incarceration on their most serious charge and those

receiving non-carceral sentences.

White (2019b) addressed potential bias arising from selection into jail incarceration by estimat-

ing the eﬀects of a sentence to jail incarceration on post-release 2012 turnout using a design based

on the as-if random assignment of ﬁrst-time misdemeanor defendants to judges in Harris County,

Texas. White (2019b) found a 13 percentage point average decrease in post-release 2012 turnout

among marginal Black misdemeanor defendants sentenced between 2008 and 2012, and negligible

eﬀects for marginal white defendants.

McDonough, Enamorado and Mendelberg (2022) used a similar design to estimate impacts

from pretrial jail incarceration between 2008 and 2012 in Philadelphia County on post-release

2012 turnout, leveraging the assignment of defendants to bail judges to attempt to identify causal

eﬀects. McDonough, Enamorado and Mendelberg (2022) found a 39% average reduction in post-

release voter turnout among marginal defendants incarcerated pretrial before the 2012 election, with

eﬀects appearing in the year prior to the election and with larger eﬀects for Black defendants.2

Incarceration may aﬀect turnout not only after a defendant is released, but also, for those de-

fendants incarcerated during voting days, during periods of incarceration (The Sentencing Project,

2020,2022). White and Nguyen (2022) merged state prison records with voter ﬁle records in Maine

and Vermont—the two states in which individuals incarcerated for felony convictions are legally

entitled to vote while incarcerated—ﬁnding that 8% and 6% of those incarcerated in state prison

voted while incarcerated in 2018 in Vermont and Maine, respectively. White and Nguyen (2022)

did not attempt to estimate the causal impacts of incarceration on the incidence of voting from

prison. McDonough, Enamorado and Mendelberg (2022) attempted to distinguish the incapacita-

tion eﬀect of jail incarceration on Election Day 2012 from the post-release eﬀects of prior periods of

jail incarceration, but their design and data did not allow them to cleanly identify an incapacitation

eﬀect.

To date we lack any reliable causal estimates of whether and by how much being incarcerated

in a county jail interferes with an individual’s exercise of the right to vote. Individuals incar-

cerated in county jails are, with few exceptions, not rendered ineligible to vote because of their

jail incarceration.3Anecdotal reports suggest that county jails nonetheless often fail to provide

incarcerated individuals with the ballot access to which they are constitutionally entitled (The

2The assignment of bail judges to cases in Philadelphia County during this period is not independent of case

and defendant characteristics (p <0.001), raising the possibility that non-randomly assigned case and defendant

characteristics may account for the larger estimated reduction in post-release turnout, relative to White (2019b).

3The vast majority of those held in county jails are being held pretrial or for the purpose of serving a sentence

following a misdemeanor conviction, neither of which is disenfranchising in any state. Like individuals not incarcer-

ated in county jails, those incarcerated in county jails may be ineligible to register and vote for reasons unrelated

to their current incarceration, including citizenship status and prior felony convictions. For more information on

enfranchisement status within incarcerated populations, see The Sentencing Project (2020).

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

VotingFromJail*WORKINGPAPERAnnaHarveyOrionTaylorOctober14,2022AbstractWeleveragenewdataondailyindividual-leveljailrecordsandexploitthetimingofincarcerationtoestimatethecausaleectsofjailincarcerationonvotingfromjailin2020.Wendthatregisteredvotersbookedintocountyjailsforthefulldurationof2020voting...

展开>> 收起<<

Voting From Jail WORKING PAPER Anna HarveyOrion Taylor.pdf

共42页,预览5页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

Voting From Jail WORKING PAPER Anna HarveyOrion Taylor

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: