Felony Disenfranchisement and Rights Restoration

What the evidence says

A single randomized mailing that informs recently released felons in a rights-restoration state that they are eligible to vote increases registration by 1.8 percentage points and turnout by 0.9 percentage points against a control-group baseline of 5.9% registered and 3.0% voting (Gerber 2014, RCT, tier A). Among those who actually received the letter, the effect is larger — registration up 3.0 percentage points and voting up 1.5 percentage points — consistent with a mechanism in which the letter removes an informational barrier rather than a legal one, since these recipients were already legally eligible. That informational-barrier mechanism is reinforced by a strong heterogeneity result: released felons who had voted before their incarceration respond far more to the same letter (registration +8.4 points, voting +7.7 points) than eligible non-voters do (registration +1.9 points, voting +0.7 points) — a voting effect about ten times larger for prior voters, though the subgroup estimates rest on small samples.

Incarceration itself is associated with a large drop in participation measured against the same individuals' own pre-incarceration record: registration fell 16.8 percentage points and turnout fell 1.7 percentage points. A single eligibility letter recovers only part of that loss — post-incarceration registration rises from 26% to 34% of its pre-incarceration level, and turnout from 66% to 84% of its pre-incarceration level — meaning even with outreach, participation does not return to its pre-incarceration baseline. This partial-recovery pattern is echoed in Owens (2018): even in Illinois, where the franchise is restored immediately upon release from incarceration rather than after parole or probation, voting participation among people with criminal-justice contact remains suppressed, and material and attitudinal barriers to voting persist among people who are already legally eligible. Owens's data also show that personal connections to civil society organizations do not moderate this suppression — no measured effect on voting despite such connections easing other forms of civic engagement.

At the community level, a 5-percentage-point increase in the felony conviction rate is associated with an approximately 4-percentage-point decrease in expected mayoral-election turnout (Owens 2018, cross-sectional). Owens's synthesis states plainly that declines in voting among people with criminal-justice contact are not a function of disenfranchisement law alone, without quantifying how much of the decline disenfranchisement accounts for versus other factors.

On scope: approximately 93% of formerly incarcerated individuals in the US (about 14 million people, 2011 data) are legally eligible to vote, but the underlying state law landscape restricts many more broadly than that figure suggests. As of 2006, 48 states and DC barred voting during incarceration, 36 states extended that bar through parole or probation, and 11 states imposed lifetime bans; only Maine and Vermont never disenfranchise. Illinois is narrower than the norm, restoring the vote immediately upon release from incarceration. By 2005, disenfranchisement reached 5.3 million citizens (2.5% of the voting-age population) and fell disproportionately on Black Americans — 1.4 million Black men, or 13% of the Black male population.

Two isolated findings describe when disenfranchisement or its reversal can move an election outcome, without generalizing beyond their own case. A counterfactual analysis of the 2000 Florida presidential race estimates that Al Gore would have won the state by more than 30,000 votes had roughly 600,000 disenfranchised people with prior felony convictions voted, against an official 537-vote margin. By contrast, prisoners' actual votes in Ireland's 2007 general election had no measured effect on any constituency outcome, because in every constituency sharing a boundary with a prison, the electoral margin exceeded the number of registered prisoners.

Where the evidence splits

No contradictions in the current sources.

Thresholds and dose-response

A 5-percentage-point increase in the community-level felony conviction rate is associated with an approximately 4-percentage-point decrease in expected voter turnout in the 2015 mayoral election (Owens 2018). No other claim in these sources states a level, rate, or cutoff at which an effect appears, reverses, or changes size.

Open empirical questions

• Whether the eligibility-outreach effect on registration and turnout generalizes to other states, populations, and time periods, or persists with repeated contact over time, is not measured in these sources.
• Whether declines in participation among people with criminal-justice contact are driven by disenfranchisement law itself, unmeasured community characteristics such as disorder or policing intensity, or individual-level factors, is not disentangled here.
• Whether the barriers that persist despite immediate rights restoration in Illinois are material, attitudinal, or administrative in nature is not separately measured.
• Whether the state disenfranchisement-law figures reported here (48 states + DC barring voting during incarceration, 36 restricting parole/probation, 11 lifetime bans, as of 2006; 93% eligibility, 2011 data) still describe current law is not addressed.