Psychology by BogdanovaThe Encyclopedia of Psychology for the Artificial Era

RESEARCH DOSSIER

Plea Bargaining and Defendant Decision-Making

Abstract. Plea bargaining is a consequential legal decision under uncertainty in which defendants evaluate negotiated case dispositions against the perceived risks, costs, and meanings of trial. This critical narrative research monograph integrates historical legal scholarship, behavioral decision theory, experimental studies of innocent and guilty defendants, survey evidence, observational studies of pretrial detention and actual prosecutorial offers, and recent comparative research. The analysis separates actual guilt from admissions, formal legal voluntariness from experienced pressure, prospective sentence discounts from realized counterfactual outcomes, and statistical association from causal inference. It examines expected-utility accounts, reference dependence, informational inequality, attorney influence, youth and mental-health vulnerability, procedural comprehension, race and socioeconomic context, collateral consequences, institutional incentives, and methodological threats to external validity. Evidence supports the relevance of expected trial outcomes while identifying substantial institutional and psychological determinants that cannot be reduced to one calculation. The monograph closes with concrete standards for transparent evidence collection, experimentally testable safeguards, and cross-jurisdictional research. It is a critical narrative synthesis of existing scholarship; no original defendant data, formal systematic-review protocol, or new pooled meta-analysis is claimed.

Keywords: plea bargaining; guilty pleas; defendant decision-making; legal psychology; false guilty pleas; innocence; risk perception; behavioral decision theory; pretrial detention; defense counsel; voluntariness; coercion; trial penalty; procedural justice; forensic psychology

Introduction: The Psychological Object of Plea Decisions

Plea bargaining is the negotiated resolution of a criminal accusation through an admission or other legally recognized disposition, typically involving concessions concerning charges, sentencing recommendations, or other consequences. The psychological object is the defendant's choice among available legal courses under uncertainty, asymmetrical information, unequal resources, and potentially irreversible effects. A decision can be instrumentally understandable while producing a factually inaccurate admission. The scientific task is to explain how preferences, beliefs, pressures, opportunities, and institutional rules generate decisions, rather than infer mental states from the plea outcome alone. The analysis also treats prosecutorial offers and attorney communication as determinants of the options that reach the defendant. Redlich et al. (2017) Helm (2024)

Four questions organize this monograph. What subjective probabilities and consequences inform the defendant's comparison of trial and settlement? How do actual innocence, cognitive development, health, legal comprehension, social influence, and detention modify the evaluation? How do institutional conditions structure the offer and the opportunity to reject it? Which designs can distinguish causal mechanisms from selection into arrest, detention, representation, bargaining, and conviction? These questions require integration of legal history, decision theory, experimental psychology, administrative case records, and qualitative accounts. Case outcomes establish behavior; valid mechanism claims require independent measurement or credible experimental and quasi-experimental designs. Bibas (2004) Bushway et al. (2014) Redlich et al. (2025)

The geographical evidence base is especially developed for the United States, whose plea practices, sentencing schedules, constitutional protections, and records cannot automatically represent other legal systems. Negotiated dispositions also operate in different institutional forms across jurisdictions. The comparison with Japan's system of cooperation agreements illustrates how incentives to supply information about another person create distinct psychological choices from admitting one's own guilt. The present text is a critical narrative synthesis of identified literature available through November 2026. It reports neither original fieldwork nor a preregistered systematic review or meta-analysis. Hirota & Wakabayashi (2026) Helm (2024)

Definitions: Pleas, Bargains, Admissions, and Decision Outcomes

A guilty plea is a formal procedural action with a particular legal effect; a negotiated bargain is the process or agreement through which conditions for that action may be arranged. Bargaining can occur over charge severity, the number of counts, recommended punishment, collateral arrangements, cooperation, or procedural opportunities. A conviction resulting from a plea is not an independent experimental observation of the defendant's factual culpability. The categories factual guilt, legal liability, admission, conviction, and perceived guilt must therefore be operationalized independently. Experimental manipulations typically define factual status within the study scenario, whereas administrative records often observe only the legal disposition. Redlich et al. (2017) Dervan & Edkins (2013)

The plea discount can be described as a lower offered punishment compared with an anticipated post-trial outcome, but its exact measurement depends on whether one compares an initial offer, the final accepted agreement, an eventual sentence, or a counterfactual sentence never observed. A trial penalty may refer to a larger expected sanction for exercising the trial option, or to the difference between observed outcomes in groups selected into trial and settlement. Those two notions are causally different. New evidence from recorded initial offers in Philadelphia shows why substituting accepted offers for unobserved offers to those who rejected them can reverse conclusions. Abrams et al. (2026) Bibas (2004)

Psychological voluntariness refers to the experienced opportunity to choose without overwhelming pressure, while the legal characterization of a plea as voluntary depends on jurisdiction-specific doctrine and procedural requirements. Comprehension concerns understanding the alternatives and their immediate and delayed consequences; appreciation concerns applying this information to oneself. Competence, informed choice, factual correctness, experienced control, and distributive fairness describe related but nonidentical properties. Their distinctions allow research to test whether a formally valid plea can nevertheless emerge from deficient information or an extremely constrained choice set. Fessinger & Kovera (2023) Redlich et al. (2025) Viljoen et al. (2005)

Historical Emergence and the Institutionalization of Negotiated Justice

Negotiated criminal dispositions have a complex institutional history. Alschuler's historical account documents the growth of plea bargaining and challenges the intuition that its modern predominance merely reflects an unchanged traditional practice. Nineteenth- and twentieth-century expansion of criminal case processing, prosecutorial discretion, professional legal representation, sentencing structures, and judicial workloads helped establish the conditions in which routine bargaining became central. Legal history must separate the existence of occasional negotiated admissions from the later development of system-wide institutional dependence on pleas. Alschuler (1979) Bibas (2004)

In the United States the judicial recognition of negotiated pleas developed alongside attempts to define lawful inducement, representation, and procedural fairness. Later decisions concerning deficient attorney advice and uncommunicated offers, including Lafler v. Cooper and Missouri v. Frye in 2012, highlighted the significance of the bargaining stage for the defendant's substantive interests. These decisions are legal materials, rather than psychological evidence that participants accurately understand or exercise their options. Their significance for behavioral research lies in identifying communication, information, and attorney performance as processes that can change final legal outcomes. Redlich et al. (2017) Helm (2018)

The scientific study of plea choice emerged through formal models and increasingly realistic simulations. Bordens' 1984 mock bargaining experiments manipulated perceived conviction likelihood, threatened punishment, and participant role, finding patterns consistent with strategic consideration of expected outcomes, including willingness by some assigned-innocent participants to accept offers. Later research moved toward actual task-induced innocence or guilt, procedural disclosure experiments, defendant surveys, and datasets recording offers rather than solely convictions. This methodological development gradually exposed problems that conventional outcome statistics could not identify: unrevealed offers, unrecorded threats, mistaken beliefs, and the selection of different cases into trial. Bordens (1984) Dervan & Edkins (2013) Abrams et al. (2026)

Expected Utility and the Shadow of Trial

The canonical decision model expresses the negotiated plea as a comparison between a comparatively certain sanction and an uncertain trial outcome. In its simplest form, a decision-maker accepts a plea carrying sanction S_p if its perceived burden is lower than the probability of conviction p multiplied by the anticipated post-trial sanction S_t. This representation makes the role of evidence, trial risk, and the size of the offer legible. It is a benchmark model under restrictive assumptions: the consequences must be commensurable, preferences approximately linear in the chosen metric, and the defendant must have usable probability estimates. The model concerns subjective expectations and need not imply that the defendant knows the true probability of conviction. Bushway et al. (2014) Bordens (1984)

Bushway, Redlich, and Norris tested a formal shadow-of-trial model using responses from 1,585 prosecutors, defense attorneys, and judges to experimentally varied hypothetical cases. They measured conviction probability assessments, projected trial sentences, and acceptable or offered plea terms, applying structural regressions and corrections for measurement error. The basic model received conditional support for prosecutors and defense attorneys, with individual and jurisdictional characteristics contributing additional explanation. This result establishes genuine explanatory value for trial expectations while indicating that a unified calculation cannot exhaust the bargaining process. The participants were legal professionals evaluating vignettes, and the study did not directly measure the internal utility of actual defendants facing detention. Bushway et al. (2014)

Utility is also broader than months of custody. A negotiated disposition may modify the probability of incarceration, the timing of release, a criminal record, financial liabilities, immigration consequences, reputation, family separation, or access to employment. Preferences vary with the ability to endure uncertainty, the marginal harm of an additional sentence year, and the anticipated social costs of trial. An expected-value model should therefore incorporate multiple consequences or explicitly state what it excludes. What seems irrational under a one-dimensional sentence comparison can become intelligible once the relevant stakes and time horizon are specified. Bibas (2004) Helm (2018)

Prospect Theory, Reference Points, and Loss Framing

Behavioral decision theories predict that alternatives can be evaluated relative to reference points rather than only in terms of final outcomes. A defendant may compare a plea with present liberty, an expected sentence, a prior offer, a lawyer's proposed target, or a worst-case trial scenario. Movement of the reference point alters whether a stated concession feels like a gain or a loss. Descriptions of the same sentencing differential can accordingly produce different emotional reactions, especially when the defendant does not possess calibrated numerical probabilities. The empirical question is whether framing changes actual choice after objective outcomes, understanding, and the sequence of offers are controlled. Helm (2018) Helm (2024)

Labels such as trial penalty and plea discount can be psychologically salient, but salience must not be confused with a universally demonstrated framing effect. In Fessinger and Kovera's experiment, variation in the sentence reduction mattered to willingness to plead and assessments of voluntariness, whereas presenting the offer as a discount or penalty did not change the reported voluntariness measure. This is a particularly useful negative finding. It shows that familiar theoretical expectations require direct tests of the selected outcome, sample, and presentation rather than assuming all framing manipulations shift judgments in the predicted direction. Fessinger & Kovera (2023)

Reference dependence may interact with timing. An initial severe proposal can change the interpretation of a later offer, while an imminent loss of housing or employment can make immediate release the dominant reference. If a study records only the final accepted agreement, it cannot observe these offer-sequence effects. Experiments should vary identical final options after alternative initial anchors, measure the defendant's subjective baseline before exposure, and distinguish changes in perceived punishment from changes in comprehension or trust. Such tests can identify whether apparent negotiation success reflects learning, emotional relief, strategic adaptation, or the structured path of earlier offers. Bibas (2004) Abrams et al. (2026)

Risk, Ambiguity, and Subjective Conviction Probability

Perceived conviction probability is a psychological judgment constructed from available evidence, confidence in counsel, past institutional experience, and beliefs about adjudicators. Defendants frequently lack the information required to estimate this probability objectively. Even lawyers may disagree about evidentiary strength, witness credibility, suppression prospects, or local sentencing practices. When the probability distribution itself is unclear, a person faces ambiguity as well as risk. The distinction matters because responses to an uncertain chance of conviction may differ from responses to a numerical probability that the participant fully understands. Bibas (2004) Bushway et al. (2014)

Bordens varied projected conviction likelihood in an early factorial study and reported an effect on mock defendants' acceptance. This supports the importance of risk information under constructed experimental conditions, while leaving open whether real defendants can identify the inputs needed for the same computation. A probability reported after deciding may also rationalize an earlier preference. Strong studies should measure probability beliefs before the offer, vary independently the quality of evidence and the stated numerical assessment, then record both belief change and final choice. Doing so permits a distinction between persuasion through evidence and compliance driven by a different consequence. Bordens (1984) Helm (2024)

A defendant's subjective risk is also endogenous to the bargaining environment. A very favorable offer might signal that the case is weak, or, conversely, that counsel doubts the trial outcome. Institutional bargaining norms can affect which inference is drawn. Prosecution incentives and attorney recommendations can supply apparently authoritative information about a probability the defendant cannot independently check. Consequently, models that treat p as a stable property of the case miss a feedback mechanism: the offer itself may change beliefs about the likely verdict. Testable models must specify the informational content assigned to offers and the reliability of the defendant's inference. Bibas (2004) Suiter & Metcalfe (2024)

Time Horizons, Discounting, and Immediate Release

Intertemporal choice is central when a defendant must compare relief today with a distant possible cost. Someone detained before trial may value a prompt disposition because the alternative prolongs separation from family, lost wages, health burdens, or uncertainty. An immediate release offer can therefore carry disproportionately high weight even if it entails a permanent conviction. A conventional sentence discount that omits these temporal costs can mistake the practical meaning of the bargain. The psychological mechanism is a combination of present burdens, anticipated future losses, and the accessibility of alternatives. Petersen (2020) Bibas (2004)

Petersen's survival analysis of felony defendants in large urban counties, using historical records from 1990 through 2004, found that pretrial detainees entered guilty pleas 2.86 times faster than released defendants in the fitted comparison. This is a study of timing under observational conditions: detention is associated with faster disposition, but the estimate alone does not identify a universal causal effect of incarceration on individual preference. Case seriousness, risk assessment, available resources, charging practices, and unobserved differences can affect both detention and the pace of bargaining. Nevertheless, its temporal evidence directs attention to an institutional pressure frequently invisible in studies restricted to final sentence lengths. Petersen (2020)

To isolate mechanisms, researchers should record custody status, length and conditions of confinement, access to counsel, anticipated waiting time, and each offered route to release. They should distinguish the value of ending confinement from the belief that trial will end badly. In randomized experimental simulations, these variables can be separated; in real cases, identification may require policy shocks, quasi-random judicial assignment, or detailed pre-offer trajectories. Ethical interpretation should also address how short-term survival incentives interact with delayed legal and social consequences, rather than treating every accepted bargain as a transparent revelation of stable preference. Helm (2024) Redlich et al. (2025)

Preferences, Identity, and the Meaning of an Admission

Legal choices are not exhausted by punishment minimization. An admission of criminal conduct can challenge a person's self-concept, relationships, moral commitments, and perceived standing in the community. Some defendants may refuse an attractive bargain to avoid publicly identifying as guilty; others may accept one to protect their family, minimize publicity, or restore control over an unpredictable proceeding. These goals can coexist with considerations of evidence and sentence. A sound account distinguishes preferences over legal outcomes from the psychological meaning attributed to declaring guilt, even when both are expressed in the same final response. Redlich et al. (2017) Helm (2018)

The innocence literature demonstrates that factual status affects willingness to plead, but does not invariably determine it. If innocence creates a strong nonmonetary cost of admission, the effect of large sentencing incentives becomes theoretically informative: pressure must outweigh or circumvent that cost for an innocent person to accept. Laboratory work inducing an actual task outcome, rather than assigning a fictional defendant role, captures part of this motivational conflict. Yet a university task involving course credit cannot reproduce the layered significance of criminal stigma, immigration loss, imprisonment, or racialized policing. Findings therefore establish a plausible mechanism within defined conditions, not the prevalence of false pleas in a legal jurisdiction. Edkins & Dervan (2013) Dervan & Edkins (2013)

A further challenge is that guilt itself is not a single psychological variable. People may distinguish having performed an act from believing it was blameworthy, agreeing that a statute applies, or thinking the evidence can prove the accusation. Plea decision measures should ask which proposition participants consider true. Where an institution permits particular forms of admission or agreement without an ordinary affirmation of culpability, the psychological meaning of each option must be measured. This prevents the researcher from collapsing factually inaccurate self-identification, strategic procedural acceptance, and differences in legal interpretation into one undifferentiated category. Fessinger & Kovera (2023) Helm (2024)

Innocence, False Pleas, and the Limits of Factual Inference

A false guilty plea occurs when a person admits or legally pleads guilty to conduct for which that person is factually innocent under an appropriate definition of the allegation. It is analytically distinct from an inaccurate confession during interrogation, even though the two may be causally connected. The empirical challenge is that factual innocence is rarely observed independently in administrative case files. Conviction records measure institutional dispositions, and later exonerations capture only a selected set of errors that became discoverable and actionable. Neither source alone can identify the population prevalence of false pleas. The scientific objective is to locate circumstances under which innocent people become willing to accept a false admission and investigate how reliably such circumstances map onto actual prosecution. Redlich et al. (2017) Redlich et al. (2025)

Dervan and Edkins' influential study of a cheating accusation provides an experimental demonstration that innocent participants can accept a negotiated admission in exchange for a known benefit. In the laboratory paradigm, students' involvement or noninvolvement in the prohibited action was experimentally established. A majority of the innocent students accepted the arrangement; the related research report describes an innocent-participant acceptance proportion around 56 percent across the tested conditions. The correct inference is that false admissions can arise through a plea-like incentive under laboratory conditions. The study does not establish that the same fraction of innocent people charged in criminal court plead guilty, because consequences, legal representation, demographics, and the selection into prosecution differ substantially. Dervan & Edkins (2013) Edkins & Dervan (2013)

Recent work complicates any singular causal story. Fessinger and Kovera found that innocent participants were less willing overall to plead guilty than guilty participants but that sentence reduction size affected choices in both groups and influenced how innocent participants rated voluntariness. The distinction matters: innocence can have a protective association while a sufficiently attractive or threatening offer still moves decisions. Lundberg's strategic model, by contrast, derives consequences for wrongful conviction from prosecutor behavior under its specified assumptions; this is theoretical analysis and must not be labeled an experimentally measured real-world false plea rate. Together, experimental and theoretical evidence identifies a genuine problem and a set of hypothesized mechanisms with different levels of empirical support. Fessinger & Kovera (2023) Lundberg (2024)

Experimental Paradigms and Their Ecological Validity

Two common designs study plea decisions. Role-playing experiments assign participants to a vignette in which they imagine being guilty or innocent; behavioral-induction designs place participants in a task where compliance with an instruction can define actual task guilt or innocence, followed by an accusation. The former permits controlled manipulation of multiple legal variables at comparatively low cost. The latter creates a limited first-person conflict between knowledge of innocence and an incentive to admit guilt. Neither automatically reproduces the experiences of defendants facing criminal records, mandatory sentencing rules, public stigma, or incarceration. Ecological validity is a question of the mechanism, stakes, and population for which a result is being generalized. Bordens (1984) Dervan & Edkins (2013)

A factorial experiment can manipulate factual status, expected sentence, conviction probability, attorney advice, and disclosure of consequences independently. Such designs support causal conclusions about the randomized variables within their procedure, although the interaction terms often require substantially larger samples. A study with experimentally assigned guilt status might also induce differences in perceived fairness, anger, and trust, which can mediate decisions. The research report should state whether guilt was assigned as a narrative instruction, induced by observable behavior, or inferred from a respondent's prior legal history. Its conclusion should be framed in that operational language. Fessinger & Kovera (2023) Henderson et al. (2023)

Real-case studies make different gains. They capture actual legal outcomes, institutional variation, time costs, and defendants who face consequential decisions, but they typically lack reliable observations of the evidence presented during bargaining and the offers rejected. Abrams and colleagues' 2026 analysis of recorded initial offers demonstrates the value of collecting the negotiations themselves. The strongest cumulative evidence would triangulate such records with experiments, contemporaneous interviews, and independent assessments of counsel communication, rather than treating the administrative conviction outcome as an adequate proxy for every psychological process. Abrams et al. (2026) Helm (2024)

Mental Health, Symptom Burden, and Decision Vulnerability

Mental-health research concerns heterogeneous clinical states and functional capacities, not an assumption that people diagnosed with a disorder cannot make legal decisions. Relevant mechanisms may include impaired concentration, intense distress, susceptibility to suggestion, hopelessness, acute symptoms, medication effects, or difficulty integrating long-term consequences. Each must be measured at the time of decision and interpreted alongside counsel, confinement, social resources, and the case circumstances. The relation between diagnosis and plea behavior can also arise through differential exposure to arrest, repeated prosecution, or institutional treatment, rather than from a direct cognitive effect of a disorder. Redlich et al. (2010) Redlich et al. (2017)

Redlich, Summers, and Hoover surveyed 1,249 offenders with mental disorders at six sites. Self-reported false guilty pleas varied by site, with reported ranges of 27 to 41 percent; self-reported false confessions also varied. The study documents substantial reported experiences and correlates within a specific sampled population. Because innocence was self-reported and the sample was not a random census of all defendants, the percentages cannot estimate the probability that any accused person with a mental-health diagnosis falsely pleads guilty. Symptom severity, criminal history, and other measured characteristics were associated with reported false admissions; causal pathways remain unresolved. Redlich et al. (2010)

A defensible clinical interface examines decisional abilities relevant to the actual choice: understanding charges, weighing options, communicating with counsel, appreciating consequences, and maintaining a stable expression of choice. It should also distinguish transient crisis from enduring impairment and identify accommodations that restore meaningful participation. Empirical testing of simplified materials, breaks, supported decision-making, and access to qualified assessment can be conducted without treating a psychiatric label as a substitute for functional evidence. Detailed clinical diagnoses, treatment decisions, and formal competence evaluations belong to professional proceedings and neighboring research dossiers. Viljoen et al. (2005) Redlich et al. (2025)

Defense Counsel as an Informational and Social Influence Agent

Defense counsel mediates the flow of information between the legal system and the defendant. Counsel may explain the evidentiary record, estimate trial prospects, identify collateral consequences, describe local practice, and express a recommendation. Because the defendant often cannot independently evaluate these claims, advice functions both as substantive information and as a signal of expertise and trustworthiness. The distinction is crucial: a defendant may accept an offer because a lawyer supplied accurate new information, because the lawyer appeared authoritative, or because rejecting counsel's recommendation felt socially difficult. All three processes can produce the same observed plea. Suiter & Metcalfe (2024) Helm (2024)

Suiter and Metcalfe experimentally varied characteristics of a defense attorney in a national vignette sample. Their abstract reports higher plea acceptance when the lawyer was experienced and empathic, and more favorable perceptions of qualifications under several attorney conditions. The experimental design strengthens inference about manipulated information within the vignette, but respondents did not experience the long-term dependence on counsel characteristic of real criminal cases. The findings place perceived professional competence and empathy within the causal architecture of legal choice, while indicating that these qualities should be distinguished from the objective accuracy of advice. Suiter & Metcalfe (2024)

Henderson, Sutherland, and Wilford tested attorney recommendations in two interactive simulation studies, one involving 106 participants and another 282. The second study found that accept recommendations increased willingness to plead relative to reject recommendations, while recommendations to reject produced a larger change in willingness among those initially inclined to accept. This asymmetry suggests that the direction of advice and its consistency with the defendant's initial preference matter. Future studies should measure initial inclination, confidence in counsel, objective merits, and later understanding separately, so that a measured influence is not automatically interpreted as manipulation or as competent professional assistance. Henderson et al. (2023)

Trust, Communication Quality, and Participation in Negotiations

Attorney-client trust is a multidimensional judgment involving perceived competence, honesty, commitment, empathy, and respect for the defendant's agency. In bargaining it has particular importance because the lawyer may be the only participant positioned to translate complex evidence into a decision-ready account. Trust can facilitate comprehension and productive deliberation, yet uncritical reliance may also allow unsupported probability estimates or neglected collateral consequences to go unchallenged. The optimal research question is not whether trust is always beneficial, but when confidence tracks the informational quality of representation and when it merely increases compliance. Suiter & Metcalfe (2024) Redlich et al. (2017)

Client participation affects the ability to disclose private information that may alter the legal strategy, such as employment obligations, family responsibilities, health needs, immigration risk, or the personal significance of an admission. An agreement developed with little defendant participation can appear efficient to attorneys while poorly representing what the defendant values. Participation itself, however, must be operationalized through behaviors: asking questions, receiving answers, correcting the factual account, comparing alternatives, and retaining a final opportunity to decide. A ceremonial consultation with no decision space does not measure the same construct as collaborative case review. Helm (2024) Redlich et al. (2025)

The experimental literature implies several separable intervention targets. Counsel could provide a written range of possible outcomes, state what evidence remains undisclosed, enumerate immediate and collateral consequences, and invite the defendant to repeat the alternatives in their own terms. Researchers can randomize specific communication components while measuring comprehension, confidence calibration, voluntariness, decisional regret, and plea decisions. These are proposals for testing rather than established universal remedies. Their success must be judged across real representation settings where time pressure, caseloads, institutional incentives, and uneven access to records may constrain implementation. Henderson et al. (2023) Suiter & Metcalfe (2024)

Prosecutorial Discretion and Offer Construction

Before a defendant evaluates an offer, prosecutors and other legal actors determine which offers can be made. Charging decisions, evidentiary screening, sentencing recommendations, cooperation requests, and the order in which proposals are communicated define the defendant's opportunity set. Consequently, a behavioral model that conditions only on the final offer confounds defendant preference with prosecutorial selection. A seemingly strong preference for accepting bargains may partly reflect the absence of reasonable alternatives, while an observed refusal may reflect an unworkable initial proposal rather than an enduring preference for trial. The process must be studied jointly from offer generation through negotiation and disposition. Bibas (2004) Helm (2024)

Votruba and Tisdale developed an experimental case-file paradigm in which community participants acted in a prosecutorial role and reviewed manipulated case details. The materials varied the alleged offense severity, defendant criminal history, and stated motive. The abstract reports that severity and relevant criminal history increased the length of plea offers, whereas manipulated motive did not produce the hypothesized effect. These findings concern an experimentally structured analog of prosecutorial reasoning, not actual prosecutorial behavior in contested courts. Nonetheless the design shows how variables that influence offers can be separated and tested before defendants encounter them. Votruba & Tisdale (2021)

Lawyers' and prosecutors' forecasts need not coincide with defendants' beliefs. In Bushway and colleagues' professional sample, the trial-shadow account explained some decision patterns, but covariates associated with institutional actors and jurisdictions added explanatory information. In Lundberg's theoretical strategic model, prosecutorial choices shape the relationship between plea systems and wrongful conviction under assumed incentives. Both sources support a process-level approach in which defendants respond to offers deliberately selected by actors with different information and objectives. Stronger empirical designs would record all rejected and revised offers and match them to each participant's predicted trial outcome at the time of negotiation. Bushway et al. (2014) Lundberg (2024)

The Empirical Architecture of the Trial Penalty

A trial penalty is often treated as the difference between a negotiated sentence and a harsher outcome following conviction at trial. That comparison is ambiguous when defendants reaching trial differ from defendants settling, when initial offers are changed, and when unobserved case strength affects both trial selection and sentencing. The observed sentence differential may reflect bargaining incentives, charging decisions, judicial practice, case selection, or several mechanisms together. Causal interpretation requires specifying the counterfactual: which sentence would the very same defendant have received under the alternative procedural choice, holding legally relevant circumstances constant? Bibas (2004) Abrams et al. (2026)

Abrams, Nguyen, Ouss, and Reinhold analyzed initial offers across approximately 23,000 felony cases in Philadelphia during 2012–2017. Their 2026 article reports that rejected initial offers were, on average, longer than eventual sentences, including outcomes after trial, contrary to a simple implication that rejecting an initial plea necessarily leads to a longer actual sentence. They also report longer initial offers for Black defendants after adjustment for detailed observed case factors, especially where defendants were detained pretrial. These findings are from a particular jurisdiction and years, and their observational design does not establish that every rejection is advantageous. They demonstrate why selection and offer history are essential for a valid empirical trial-penalty estimate. Abrams et al. (2026)

A psychologically useful analysis separates objective initial offers, the defendant's perception of expected post-trial punishment, perceived certainty of a stated recommendation, and the actual sentence later imposed. The behaviorally effective penalty may be a feared possibility even if the realized aggregate comparison shows no corresponding average increase. Conversely, a favorable average sentence after rejection could result from systematic case selection. Future datasets should preserve the sequence of proposals, dates, counsel, detention status, the evidence provided at each stage, and a standardized measure of defendants' beliefs. This design would connect institutional outcomes with the decision mechanism actually operating at the moment of choice. Abrams et al. (2026) Redlich et al. (2025)

Pretrial Detention, Resource Inequality, and Negotiating Leverage

Pretrial detention changes the bargaining environment in ways that extend beyond formal sentencing rules. It can impair access to employment, reduce opportunities to gather records or consult family, increase reliance on counsel, limit privacy, and make an otherwise delayed resolution feel urgent. The defendant's willingness to reject an unfavorable offer is therefore partly a function of available time and material resources. A person who cannot tolerate another week of confinement is not evaluating the same effective alternatives as someone with stable housing and funds for extended litigation. Psychological analyses need to measure these circumstances rather than categorize all defendants as facing comparable choice menus. Petersen (2020) Bibas (2004)

The association between detention and faster pleas reported by Petersen is consistent with strong temporal incentives but may also reflect differences in case trajectories. A defendant detained in a case that proceeds rapidly may not have had an equivalent opportunity to review evidence or compare offers with counsel. Researchers should examine hazard rates of acceptance alongside the sequence of detention, disclosure, defense access, and amendments to offers. An isolated coefficient for custody status cannot tell us whether delay, fear, objectively weaker trial prospects, or limited legal resources accounts for the association. Designs that exploit plausibly exogenous detention variation can strengthen causal inference, although their assumptions must be openly tested. Petersen (2020) Helm (2024)

Resource inequality is a structural determinant of apparent preferences. Differences in legal representation, available expert review, bail access, translation, transportation, or capacity to miss work may influence both the terms offered and how trial risk is experienced. Rehavi and Starr's federal sentencing analysis illustrates how prosecutorial charging decisions can contribute substantially to measured racial punishment disparities, although their study is not a direct randomized investigation of plea choice. Combined with recent initial-offer data, this points toward models in which defendant decisions emerge from a stratified institutional environment, rather than from free-standing risk preferences alone. Rehavi & Starr (2014) Abrams et al. (2026)

Race, Socioeconomic Position, and Unequal Decision Environments

Disparities in negotiated justice can enter at different causal points: surveillance and arrest, initial charge selection, detention, legal representation, disclosure, offer severity, and acceptance. An analysis of plea acceptance by race without the opportunity set will attribute to defendants' preferences differences that may have been produced earlier in the process. A useful decomposition examines which charges were initially filed, which offers were communicated, which were rejected, and what consequences followed. Socioeconomic conditions also shape the ability to bear prolonged proceedings, obtain expert assistance, maintain employment, and accept financial obligations. Statistical adjustment must reflect these pathways rather than indiscriminately control for variables that may themselves be consequences of disparity. Rehavi & Starr (2014) Abrams et al. (2026)

Rehavi and Starr's federal analysis found that much of the observed racial disparity in sentence length was associated with initial charging, including charges carrying mandatory minimums. Their reported findings belong to a specific period and federal setting and are not a causal experiment on defendants' bargaining psychology. Abrams and colleagues' Philadelphia offer dataset later found that initial offers were longer for Black defendants even after detailed observed controls, particularly among people detained pretrial. The two studies examine different stages, jurisdictions, and outcomes, but together demonstrate that bargaining conditions are shaped upstream of final admissions. The proper research conclusion is to examine institutional allocation and defendant responses jointly. Rehavi & Starr (2014) Abrams et al. (2026)

Potential mechanisms include differences in perceived institutional legitimacy, counsel quality, expected adjudication bias, community experience, and the costs of conviction. Each requires direct measurement. A correlation between demographic identity and plea choice cannot by itself distinguish rational responses to unequal exposure from altered subjective risk or social influence. Prospective studies should measure evidence quality using blinded procedures, capture the defendant's expectation of treatment, and preserve initial and revised offers. They should also evaluate whether interventions improve decisional fairness across groups rather than increasing aggregate comprehension while leaving serious disparities unchanged. Helm (2024) Redlich et al. (2025)

Probation, Collateral Consequences, and Multidimensional Loss

Many negotiated outcomes replace custody with supervision or other conditions. Such options may appear lenient when compared with imprisonment, yet supervision can entail restrictive travel rules, financial obligations, mandatory reporting, housing vulnerability, and exposure to sanction for violations. The actual choice therefore concerns a bundle of future states rather than a single sentence length. If research describes probation only as a brief or favorable punishment, participants may underestimate its obligations. Complete experimental descriptions and actual counseling records should include both the duration of the sanction and relevant requirements. Yan et al. (2024) Helm (2018)

The randomized online experiment by Yan, Wilford, and Ferreira included 906 respondents and manipulated guilt status, probation duration, and detailed disclosure. Their findings indicate that information and sanction length can shape willingness to plead, though effects differed across comparisons. The study illustrates how ostensibly minor choice architecture can matter when a noncustodial bargain is treated as an attractive alternative. Its evidence does not quantify real-world probation violations or predict which individuals will experience a particular legal outcome. Further research should examine recall of conditions, correction of misunderstandings, and downstream consequences among individuals entering community supervision. Yan et al. (2024)

Collateral consequences can also outlast the sentence itself. Criminal-record visibility, occupational licensing, housing, education, family life, or migration status may be more consequential to particular defendants than small changes in custodial duration. What is regarded as collateral in a legal taxonomy may be central to the psychological utility calculation. Surveys should elicit personalized concern before counsel advice, and experimental manipulations should present realistic consequences without exaggeration. Interactions between immediate freedom and delayed burdens are especially relevant where defendants are detained and an offer appears to end a crisis. Bibas (2004) Petersen (2020)

International Variation and the Structure of Negotiated Cooperation

Plea systems vary in the nature of admissions, charging discretion, judicial participation, discount rules, and whether bargaining concerns the accused's own culpability or cooperation against others. Consequently, the phrase plea bargaining does not denote a uniform psychological treatment across countries. Comparative research should first construct a procedural inventory specifying what is exchanged, who can make and withdraw an offer, whether sentencing concessions are predictable, and what disclosure and review rights exist. Only then can the researcher compare subjective probability, autonomy, trust, and incentives across institutional systems. Helm (2024) Redlich et al. (2025)

Hirota and Wakabayashi's 2026 Japanese research investigated decisions in a cooperation-based arrangement introduced in Japan in 2018. Their studies addressed the effects of perceived guilt, punishment severity, conviction probability, and closeness to the accused person on decisions involving information about another person's crime. These findings bear on false statements, interpersonal loyalty, and risk-bearing incentives, but the choice structure differs from an American defendant's admission to their own alleged offense. It would be methodologically unsound to merge raw acceptance rates as if both studies administered the same psychological test. Hirota & Wakabayashi (2026)

Cross-jurisdictional designs should therefore compare latent constructs through validated instruments while treating procedural categories as potentially non-equivalent. A defendant's statement of voluntariness may have different practical meaning when the judge negotiates directly, when counsel is required earlier, or when sentence discounts are bounded by statute. Researchers should record procedural culture and legal institutions alongside individual characteristics. Multilingual cognitive interviewing can identify whether terms such as bargain, guilt, concession, and consent are understood consistently. The Artificial Era adds new documentary and decision-support infrastructures, making it particularly important to distinguish information accessibility from legal authority. Helm (2024) Redlich et al. (2025)

Measurement, Sampling, and Research Design

Research measurement begins by defining the dependent variable precisely. Acceptance of a specific offer, willingness to accept, confidence in counsel, perceived voluntariness, understanding of a waiver, expressed factual guilt, and later regret measure different psychological outcomes. A study should record the actual offer and alternatives before collecting global attitudes, then measure intermediate beliefs such as conviction probability and expected sentence separately. Psychometric evaluation requires reliability, convergent and discriminant validity, comprehension testing, and invariance checks across relevant groups. Retrospective explanations of decisions are useful narrative data but may be influenced by memory reconstruction and the current legal outcome. Redlich et al. (2017) Helm (2024)

The experimental hierarchy includes between-person vignettes, behavioral paradigms with induced allegations, interactive attorney simulations, and controlled variations in the disclosure or timing of offers. Internal validity improves through random assignment, preregistered hypotheses, transparent exclusion rules, and sensitivity analyses for small interaction cells. External validity requires realistic stakes, actual defendants, differences in counsel availability, and attention to legal context. Observational data add temporal and institutional realism but bring confounding and missingness. Rejected offers and undocumented informal proposals are missing precisely where causal inference may depend on them. Henderson et al. (2023) Abrams et al. (2026)

An integrated design would follow cases from the first charge and initial offer through consultation, successive negotiations, release or detention, disposition, and subsequent outcomes. It would preserve exposure timestamps, offer terms, counsel contact, known evidence, the defendant's subjective trial forecast, and a direct comprehension test. Researchers could estimate time-to-decision models, multilevel relationships within courthouses, and appropriately identified causal effects of exogenous reforms. Sensitivity analyses should examine unobserved case strength, missing negotiations, changes in charging, and nonrandom access to counsel. Open deidentified datasets with carefully controlled privacy access could enable independent replication of key associations. Petersen (2020) Redlich et al. (2025)

Evidence Quality, Replication, and Inferential Boundaries

The evidence base combines controlled studies with relatively modest or convenience-based samples, professional surveys, defendant self-reports, theoretical models, and jurisdiction-specific administrative records. These sources answer complementary questions but cannot substitute for one another. Experiments directly test responses to manipulated offers; surveys document perceptions or experiences; strategic models expose implications of assumptions; administrative records reveal actual practices while facing selection and measurement limits. Apparent agreement across these traditions is most informative when operational definitions align and alternative mechanisms have been addressed. Bushway et al. (2014) Redlich et al. (2017)

Three recurrent inference failures deserve attention. First, a high false-plea rate within a laboratory scenario should not be presented as an estimate of the real criminal court population. Second, a statistical association between detention and speed of pleading should not be described as a universal individual causal effect without a credible identification strategy. Third, observed differences between trial and plea sentences should not be treated as the causal effect of declining an initial offer when cases self-select into trial and the rejected offer itself is unobserved. The Philadelphia 2026 analysis provides direct empirical motivation for the third warning. Dervan & Edkins (2013) Petersen (2020) Abrams et al. (2026)

Replication should extend beyond repeating a vignette with similar participants. It should vary jurisdictions, case seriousness, counsel assignment, ability to communicate in a preferred language, race and class contexts, the salience of prior detention, and different bargaining formats. Where prior studies are preregistered or supply open data, independent reanalysis can test robustness and alternative specifications. Researchers should report both null and positive findings, uncertainty intervals, and the actual content of each manipulation. In the available sources, an absent effect of sentence severity in one cheating paradigm and an absent framing effect in a later study are substantive constraints on any simple theory of uniformly increasing incentives. Edkins & Dervan (2013) Fessinger & Kovera (2023)

Transparency, Disclosure, and Testable Procedural Safeguards

The most direct institutional safeguard is visibility into the decision that is actually offered. A defendant can compare alternatives meaningfully only when the relevant accusations, likely legal outcomes, outstanding evidence, and obligations attached to the proposed agreement are sufficiently accessible. Research should evaluate disclosure before the decision, not merely the fact that a written agreement existed afterward. Redlich and colleagues' 2025 expert synthesis identifies transparency of discovery and negotiated practices among urgent unresolved issues. Its recommendations are a research agenda informed by scholarship and practice, rather than randomized evidence that any one legal reform will work in every court. Redlich et al. (2025)

A practical evaluation could randomize an accessible comparison sheet against usual explanation, where ethically and legally permissible, with equal factual content presented in alternative forms. Outcomes should include comprehension, correct recall of collateral consequences, calibration of conviction beliefs, decisions about offered terms, experienced pressure, and later perceptions of fairness. A separate arm could test whether a cooling-off interval improves deliberation after a high-pressure proposal. Recording offers and revisions would permit independent assessment of whether the intervention changes the bargain itself, the defendant's understanding, or merely the style in which an unchanged decision is recorded. Yan et al. (2024) Helm (2024)

Safeguards must also be evaluated for unintended consequences. More disclosure could produce cognitive overload if delivered too late or without counsel assistance; mandatory delay could prolong pretrial detention; a rigid form might obscure individualized consequences. Outcomes need to be stratified by detention status, developmental stage, language, educational access, and actual opportunity to consult legal counsel. A defensible policy evaluation compares benefits in decision quality against these predictable burdens and reports results separately for innocent and guilty experimental conditions where factual status can be established. Redlich et al. (2025) Fessinger & Kovera (2023)

Critical Synthesis: A Multilevel Model of Defendant Decision-Making

The literature supports a multilevel explanatory organization. At the individual level, the defendant has beliefs about conviction, preferred outcomes, risk tolerance, comprehension, temporal priorities, and the moral meaning of admission. At the interpersonal level, counsel communication, perceived expertise, recommendations, and opportunities to ask questions influence both beliefs and available strategies. At the institutional level, prosecutors construct offers while detention policy, charging rules, sentencing structures, and court workloads shape the effective alternatives. At the societal level, resource inequality, differential exposure to enforcement, and long-term consequences modify what each choice costs. These levels interact dynamically across time rather than remaining fixed attributes of a defendant. Bibas (2004) Redlich et al. (2017) Helm (2024)

Several results can be stated with different evidentiary force. Controlled experiments establish that plea-like incentives can produce false admissions under certain conditions; controlled vignettes also show that counsel attributes, recommendations, and detailed probation terms can shift choices or perceived voluntariness. Surveys reveal substantial self-reported false plea experiences among selected populations with mental disorders, without producing population innocence prevalence estimates. Observational timing analyses associate pretrial detention with earlier pleas. Finally, datasets containing rejected initial offers demonstrate the danger of inferring a causal trial penalty from outcomes alone. A comprehensive explanation must preserve these differences in design and inference. Dervan & Edkins (2013) Suiter & Metcalfe (2024) Redlich et al. (2010) Petersen (2020) Abrams et al. (2026)

The strongest synthesis is conditional. Expected punishment and trial risk matter, but the calculation depends on how people understand, value, and receive those quantities. Factual innocence generally influences willingness to plead but does not guarantee refusal. Representation can enhance informed choice or amplify social influence, depending on the information and interaction. The threat associated with trial can be experienced even when later observable sentencing contrasts fail to reveal a uniform realized penalty. The resulting scientific model predicts that altering objective outcomes and altering the presentation, timing, credibility, or accessibility of information may produce different changes in behavior. This is an integrative interpretation of established studies rather than a newly validated quantitative theory. Bushway et al. (2014) Fessinger & Kovera (2023) Henderson et al. (2023)

Open Questions and a Replicable Research Agenda

The first major unanswered question is how accurately defendants perceive trial probabilities and possible sentences in actual pending cases. It requires repeated belief measures taken before and after specific evidence disclosures and offers, accompanied by independent assessments of case strength and documentation of what counsel actually communicated. A second question concerns the causal magnitude of detention, delayed trial, expiring offers, and immediate-release incentives, ideally examined through ethically acceptable policy evaluations or strong quasi-experimental designs. A third concerns the conditions under which innocent defendants will refuse beneficial offers when stigma, evidence, and legal representation are varied. Helm (2024) Redlich et al. (2025)

The field also needs standardized measurement of meaningful choice. Such measurement should separate objective understanding, application of information to oneself, subjective voluntariness, perceived ability to refuse, trust in counsel, and postdecision outcomes. Instruments require validation in populations with varied languages, disabilities, ages, experiences of detention, and access to counsel. The same instrument should be retested across time to distinguish durable comprehension from a courtroom response produced by immediate social expectations. Multi-site research can assess whether a safeguard operates consistently where the size and structure of sentence concessions differ. Viljoen et al. (2005) Fessinger & Kovera (2023)

The final research priority is transparency about what was offered, what was refused, and how negotiations changed. Longitudinal data linking initial charges, detention, factual disclosures, offer histories, court events, and outcomes would make competing causal interpretations more discriminating. Where sensitive records are involved, privacy-protective access and independent oversight are essential. Preregistered replications should report effect sizes, uncertainty intervals, missingness, subgroup sample sizes, and negative results. Cross-jurisdictional comparison should first align procedural definitions before pooling effects. Advances in data collection, including carefully audited digital systems, will matter most when they reveal previously invisible decision mechanisms and improve the possibility of accountable review. Abrams et al. (2026) Redlich et al. (2025)

Conclusion

Plea bargaining and defendant decision-making form a distinct psychological research field at the intersection of cognition, social influence, development, clinical vulnerability, and institutional criminal procedure. The negotiated admission is an observable procedural result of a choice shaped by subjective trial expectations, immediate and delayed consequences, counsel advice, prosecutorial selection, detention, information access, and social conditions. These determinants operate through identifiable psychological constructs whose measurement requires more than recording the final conviction. Research therefore gains explanatory precision when it follows the full sequence from the first accusation and offer through deliberation, acceptance or rejection, and subsequent consequences. Redlich et al. (2017) Helm (2024)

The available empirical record establishes consequential facts while imposing boundaries on generalization. Innocent participants can accept false admissions in controlled environments; advice, offer size, probation terms, and custody conditions matter in different designs; professional survey evidence supports parts of the trial-shadow model; and new administrative evidence reveals why initial offers and rejected proposals are indispensable for measuring the meaning of a supposed trial penalty. None of these findings alone supplies a universal causal estimate for every court system. Scientific integrity requires distinguishing hypothetical choices, induced experimental behavior, self-reported experience, strategic models, and actual legal records. Dervan & Edkins (2013) Bushway et al. (2014) Abrams et al. (2026)

The central conclusion is that meaningful defendant choice depends simultaneously on the quality of the available options and the person's capacity to understand and evaluate them. A psychologically informed legal system can examine factual accuracy, comprehension, opportunity to refuse, information equality, and procedural fairness as separate measurable objectives. The most promising path forward is cumulative evidence obtained through transparent offer records, robust causal designs, validated measures, international comparison, and carefully tested procedural safeguards. That research program provides a foundation for evaluating negotiated justice with reference to both human decision processes and the institutions that construct their alternatives. Redlich et al. (2025) Helm (2024)

References

  • Abrams, D. S., Nguyen, V., Ouss, A., & Reinhold, J. (2026). A (plea) offer you can refuse. The Journal of Legal Studies, 55(2), 277–316. https://doi.org/10.1086/742147
  • Alschuler, A. W. (1979). Plea bargaining and its history. Columbia Law Review, 79(1), 1–43. https://chicagounbound.uchicago.edu/journal_articles/1006/
  • Bibas, S. (2004). Plea bargaining outside the shadow of trial. Harvard Law Review, 117(8), 2463–2547. https://papers.ssrn.com/sol3/papers.cfm?abstract_id=464880
  • Bordens, K. S. (1984). The effects of likelihood of conviction, threatened punishment, and assumed role on mock plea bargaining decisions. Basic and Applied Social Psychology, 5(1), 59–74. https://doi.org/10.1207/s15324834basp0501_4
  • Bushway, S. D., Redlich, A. D., & Norris, R. J. (2014). An explicit test of plea bargaining in the shadow of the trial. Criminology, 52(4), 723–754. https://doi.org/10.1111/1745-9125.12054
  • Dervan, L. E., & Edkins, V. A. (2013). The innocent defendant's dilemma: An innovative empirical study of plea bargaining's innocence problem. Journal of Criminal Law and Criminology, 103(1), 1–48. https://scholarlycommons.law.northwestern.edu/jclc/vol103/iss1/1/
  • Edkins, V. A., & Dervan, L. E. (2013). Pleading innocents: Laboratory evidence of plea bargaining's innocence problem. Current Research in Social Psychology, 21, 14–22. https://repository.belmont.edu/lawfaculty/50/
  • Fessinger, M. B., & Kovera, M. B. (2023). An offer you cannot refuse: Plea offer size affects innocent but not guilty defendants' perceptions of voluntariness. Law and Human Behavior, 47(6), 619–633. https://doi.org/10.1037/lhb0000548
  • Helm, R. K. (2018). Cognitive theory and plea-bargaining. Policy Insights from the Behavioral and Brain Sciences, 5(2), 195–201. https://doi.org/10.1177/2372732218786974
  • Helm, R. K. (2024). The psychology of guilty plea decisions. Annual Review of Law and Social Science, 20, 183–199. https://doi.org/10.1146/annurev-lawsocsci-041822-032654
  • Henderson, K. S., Sutherland, K. T., & Wilford, M. M. (2023). Reject the offer: The asymmetric impact of defense attorneys' plea recommendations. Criminal Justice and Behavior, 50(9), 1321–1340. https://doi.org/10.1177/00938548231172515
  • Hirota, T., & Wakabayashi, K. (2026). Decision-making in Japanese plea bargaining: Impact of guilt, punishment severity, probability of conviction, and intimacy with the accused. Japanese Journal of Psychology, 97(2), 59–69. https://doi.org/10.4992/jjpsy.97.23046
  • Lundberg, A. (2024). Do prosecutors induce the innocent to plead guilty? Economic Inquiry, 62(2), 650–674. https://doi.org/10.1111/ecin.13203
  • Petersen, N. (2020). Do detainees plead guilty faster? A survival analysis of pretrial detention and the timing of guilty pleas. Criminal Justice Policy Review, 31(7), 1015–1035. https://doi.org/10.1177/0887403419838020
  • Redlich, A. D., Summers, A., & Hoover, S. (2010). Self-reported false confessions and false guilty pleas among offenders with mental illness. Law and Human Behavior, 34(1), 79–90. https://doi.org/10.1007/s10979-009-9194-8
  • Redlich, A. D., & Shteynberg, R. V. (2016). To plead or not to plead: A comparison of juvenile and adult true and false plea decisions. Law and Human Behavior, 40(6), 611–625. https://doi.org/10.1037/lhb0000205
  • Redlich, A. D., Bibas, S., Edkins, V. A., & Madon, S. (2017). The psychology of defendant plea decision making. American Psychologist, 72(4), 339–352. https://doi.org/10.1037/a0040436
  • Redlich, A. D., Becker, L., Dervan, L. E., Donnelly, M. P., Frazier, A., Gallen, M., Gazal-Ayal, O., Helm, R. K., Johnson, B. D., MacLean, E., Reimer, N. L., Wilford, M. M., & Zottoli, T. M. (2025). Urgent issues and prospects in guilty plea research and practice. Legal and Criminological Psychology, 30(2), 193–211. https://doi.org/10.1111/lcrp.12306
  • Rehavi, M. M., & Starr, S. B. (2014). Racial disparity in federal criminal sentences. Journal of Political Economy, 122(6), 1320–1354. https://doi.org/10.1086/677255
  • Suiter, E., & Metcalfe, C. (2024). The impact of the defense attorney on plea decision-making: An experimental analysis. Criminal Justice and Behavior, 51(2), 307–328. https://doi.org/10.1177/00938548231208201
  • Viljoen, J. L., Klaver, J., & Roesch, R. (2005). Legal decisions of preadolescent and adolescent defendants: Predictors of confessions, pleas, communication with attorneys, and appeals. Law and Human Behavior, 29(3), 253–277. https://doi.org/10.1007/s10979-005-3613-2
  • Votruba, A. M., & Tisdale, C. N. (2021). Examining prosecutorial decision-making in plea bargaining: An experimental paradigm in a community sample. Analyses of Social Issues and Public Policy, 21, 595–620. https://doi.org/10.1111/asap.12234
  • Yan, S., Wilford, M. M., & Ferreira, P. A. (2024). Terms and conditions apply: The effect of probation length and obligation disclosure on true and false guilty pleas. Journal of Experimental Criminology, 20, 457–483. https://doi.org/10.1007/s11292-022-09543-9