Limitations on the Ability to Negotiate Justice: Attorney Perspectives on Guilt, Innocence, and Legal Advice in the Current Plea System

Abstract

In the American criminal justice system the vast majority of criminal convictions occur as the result of guilty pleas, often made as a result of plea bargains, rather than jury trials. The incentives offered in exchange for guilty pleas mean that both innocent and guilty defendants plead guilty. We investigate the role of attorneys in this context, through interviews with criminal defense attorneys. We examine defense attorney perspectives on the extent to which innocent defendants are (and should be) pleading guilty in the current legal framework and their views of their own role in this complex system. We also use a hypothetical case to probe the ways in which defense attorneys consider guilt or innocence when providing advice on pleas. Results indicate that attorney advice is influenced by guilt or innocence, but also that attorneys are limited in the extent to which they can negotiate justice for their clients in a system in which uncertainty and large discrepancies between outcomes of guilty pleas and conviction at trial can make it a sensible option to plead guilty even when innocent. Results also suggest conflicting opinions over the role of the attorney in the plea-bargaining process.

In the current American criminal justice system the majority of convictions occur as the result of guilty pleas rather than trial – in 2015, 97.1% of federal cases that were resolved were resolved via a guilty plea by the defendant (United States Sentencing Commission 2014 Sourcebook). The high rate of pleas has occurred largely as a result of the practice of plea bargaining, where a defendant can agree to plead guilty in exchange for a reduced sentence rather than exercise their right to a jury trial (Blume & Helm, 2014; Stuntz, 2006). This practice is well-known and often accepted as an efficient way to dispose of cases. For example, in the case of Missouri v Frye, Justice Kennedy, speaking for a majority of the court stated that “The potential to conserve valuable prosecutorial resources and for defendants to admit their crimes and receive more favorable terms at sentencing means that a plea agreement can benefit both parties” (Missouri v Frye, 2012). This rationale has been used to justify plea bargaining through the idea that a defendant can get a reduced sentence by confessing to something that they have done (Garrett, 2016). However, academics and legal commentators have noted that characteristics of current plea-bargaining practice mean that it is not just operating in a way that allows defendants to get a reduced sentence by confessing to something that they have done, but also to encourage or even coerce innocent defendants into pleading guilty (see, for example, Blume & Helm, 2014; Caldwell, 2011; Dervan & Edkins, 2013; Helm & Reyna, 2017; Zottoli, Daftary-Kapur, Winters, & Hogan, 2016).

Commentators disagree on the way that the current systems functions. Some have argued that it is not a problem that innocent defendants plead guilty because innocent defendants should be able to enjoy the benefits of pleading guilty, including reducing or altogether avoiding custody (Bowers, 2008). Others suggest that prosecutors are encouraging those who are unlikely to be convicted at trial to plead guilty by offering highly coercive plea deals (Bushway & Redlich, 2012; Zottoli et al., 2016). In either case, the decision to plead guilty appears to have become a tactical decision, rather than a moral decision to admit guilt. In this context, the role of the defense attorney is complicated. It will often involve trying to help a client navigate many competing considerations, assessments of risk, and assessments of factual guilt or innocence. It is not clear whether attorneys should be (and are) advising clients on the decision that they should make, based on an assessment of their best interests, or whether the proper role of the attorney is just to provide the relevant information to a client and leave them to make the best decision for themselves. The added consideration of factual guilt or innocence makes this process even more complex. In this paper, we interview criminal defense attorneys about their perspectives on and experiences with the plea system. We examine responses to probe the extent to which innocent defendants are (and should be) pleading guilty in the current legal framework, the way that attorneys view their role in this complex system, and the ways in which defense attorneys consider guilt or innocence when providing advice on pleas.

The Plea Decision

Defendants offered a plea deal must make a decision as to whether they want to exercise their right to trial, or whether they want to plead guilty and receive a certain sentence and charge that is less severe than they would receive if convicted at trial. The terms offered as part of a plea deal can create extreme pressure to plead guilty – in the most extreme case a guilty plea can be the only way to protect against a risk of death (through the imposition of the death penalty) as the result of trial (Blume & Helm, 2014). This means that plea decisions are not as simple as guilty defendants deciding whether or not to admit guilt, but are a tactical decision based on risks and potential consequences. The primary theory examining plea bargaining as a tactical decision is known as “bargaining in the shadow of trial.” According to this theory, parties will forecast the expected sentence after trial, discount it by the probability of acquittal, and offer some proportional outcome in exchange for a guilty plea (Bibas, 2004). Defendants can then consider the options, including potential outcomes at trial, and make a rational decision about which is preferable for them. Typically, a risk-neutral defendant is expected to plead guilty if an offered sentence is less than or equal to her expected value of going to trial.

In terms of deciding which option (plea or trial) is preferable to a defendant, empirical research has confirmed that risk preferences (Bjerk, 2008), probabilities of conviction (Tor, Gazal-Ayal, & Garcia, 2010), and potential sentences (Bordens & Bassett, 1985) influence defendant decisions. However, research also suggests that defendants are influenced by whether they are guilty or innocent (Helm & Reyna, 2017; Tor at al., 2010), pressure from the prosecutor, expediency, and acquiescence (including acquiescing with attorney advice) (Bordens & Bassett, 1985). This makes the plea decision a complex one that is influenced by defendant psychology and attorney practice, as well as by the options involved (Burke, 2007; Covey, 2007; Helm & Reyna, 2017). This also means that guilt or innocence can become a secondary concern in the decision-making process for both defendants and attorneys, and that innocent defendants plead guilty.

Pleading Guilty When Innocent

There is now extensive evidence to suggest that innocent defendants, as well as guilty defendants, plead guilty (see, for e.g., Daftary-Kapur & Zottoli, 2014; Dervan & Edkins, 2013; Tor, Gazal-Ayal, & Garcia, 2010; Zottoli et al., 2016). In a study of guilty pleas in youth and adults who pleaded guilty to felonies in New York City, Zottoli et al, 2016, found that a sizable portion of both youths and adults who had pled guilty claimed to be completely innocent (27% and 19% respectively) or at least not guilty of the crime they were charged with (20% and 41% respectively). In another study, Covey (2013) examined guilty plea rates among defendants in a case that produced a large number of exonerees. He classified participants in the case as innocent, maybe innocent, or guilty, and found that 77% of those classified as innocent pled guilty, compared to 88% of those classified as guilty, and 89% of those classified as maybe innocent (Covey, 2013). This phenomenon has also been confirmed through real wrongful convictions. In 2017, about 18% of exonerations listed in the National Registry of Exonerations were categorized as involving a guilty plea (National Registry of Exonerations, 2017). The true number of wrongful convictions from guilty pleas is likely even higher than this since exoneration is particularly hard for defendants who have pled guilty (Blume & Helm, 2014).

Blume and Helm (2014) identified three specific situations in which innocent defendants appear to be pleading guilty – low level offenses where a guilty plea allows release from jail, cases where defendants have been wrongfully convicted, prevail on appeal, and are then offered a plea bargain that will assure their immediate or imminent release, and cases where defendants are threatened with harsh alternative penalties if they do not plead guilty. In some of these situations the decision to plead guilty seems justifiable and arguably even the only sensible decision. In these situations, there is an added complication for attorneys, who may believe that the best representation is provided by advising the client to plead guilty, despite being (or claiming to be) innocent. This is a complex balance, since it is not desirable for innocent defendants to plead guilty but equally it is not desirable for innocent defendants to risk severe punishments if convicted at trial. This balance can be even more complicated where even the sentence as a result of plea has serious negative consequences for a defendant – for example where it means they will have a serious criminal conviction for the rest of their lives and remain “unexonerated” (Blume & Helm, 2014), or where even a misdemeanor conviction will have severe consequences such as loss of employment or deportation. Here, innocent defendants can end up suffering very serious adverse consequences to avoid the threat of even worse consequences at trial. Especially since the advice of attorneys has been shown to be an important predictor of defendant plea decisions (Viljoen, Klaver, & Roesch, 2005), it is important to understand how they typically advise in these cases and how defense attorney advice might be shaping plea practice.

The Role of Attorneys in the Plea System

As experts on the legal system tasked with advising clients, attorneys have an important role in the plea-bargaining process (Alschuler, 1975, Hessick & Saujani, 2002). This is particularly true since, as noted above, the advice of attorneys has been shown to be an important predictor of defendant plea decisions (Viljoen, Klaver, & Roesch, 2005). However, many questions about attorney advice in this area, particularly the way attorneys consider suspected guilt or innocence when giving advice, remain un-answered.

It is unclear whether attorneys should ever be advising an innocent defendant to plead guilty. Ethical rules typically call on attorneys to refuse to offer evidence that they know to be false and to take reasonable remedial measures where false testimony is offered. However, this is clearly not thought to apply in the case of innocent defendants pleading guilty, especially in the context of Alford pleas where defendants can plead guilty while explicitly claiming to be innocent. Considering the case of the West Memphis Three, where a plea bargain secured the release of the defendants after eighteen years of imprisonment and ensured that none of the defendants would be executed, Blume and Helm (2014) conclude that “Almost any criminal defense lawyer, including the authors of this Essay, would have advised them to take it (and would have cajoled them to take it if they hesitated).” This suggests that attorneys are limited in the extent to which they can really negotiate justice for their clients, and their role becomes protecting a client from the most severe potential consequences. Research suggests that some defense attorneys may be going too far in this regard, having a bias towards plea bargaining in legal cases, persuading clients to enter into plea agreements (Alschuler, 1975; Baldwin & McConville, 1977; Blumberg, 1979), and participating in a “meet and plead” system, in which defendants sometimes meet their attorney for the first time at a court hearing during which a guilty plea is offered, accepted, and entered (Bibas, 2013). This is supported by interviews with defendants reporting infrequent contact with attorneys prior to accepting plea deals, and very short time periods in which to make their decisions (Zottoli et al., 2016).

Attorney obligations in plea cases are being clarified through case law, specifically cases on ineffective assistance of counsel. In a recent case, the Supreme Court ruled in favor of an immigrant whose lawyer falsely told him that pleading guilty to a drug charge would not lead to his deportation (when in fact accepting the plea agreement would certainly lead to deportation) (Jae Lee v. United States). In that case, the defendant’s lawyer urged him to plead guilty in exchange for a lighter sentence and falsely told him that he would not be subject to deportation after doing so. The defendant filed a motion to vacate his conviction after learning that he would be subject to deportation as a result of his plea, and this was approved by the Supreme Court. There are also examples of ineffective assistance of counsel in cases in which an attorney advised a client to go to trial and not plead, such as the well-known cases of Missouri v Frye and Lafler v Cooper. In Frye, it was found that defense counsel was ineffective because of failing to convey a plea offer to a defendant. In Lafler, it was found that defense counsel was ineffective when incompetent advice caused the defendant to reject a guilty plea and proceed to trial. Justice Kennedy delivered the majority opinion which held that the proper test is whether, absent the ineffective counsel, a defendant would have accepted an offered plea that was less severe than his eventual sentence, and the trial court would have accepted the terms of that plea. These cases all provide clear examples of ineffective assistance of counsel (although the Lee case was less clear in terms of whether the defendant was prejudiced by the ineffective assistance), however they also raise interesting questions about exactly what the responsibilities of attorneys are in the context of plea agreements. Is the only responsibility to convey a plea offer and to provide accurate information? Does the responsibility extend to providing sound advice based on the best interests of a client? And should clients claims of guilt or innocence influence attorney advice?

A small number of studies have examined attorney plea recommendations. One study, using an experimental design, showed that probability of conviction, potential sentences, and defendant preference all influenced attorney plea advice (Kramer, Wolbransky, & Heilbrun, 2007). Another study has shown that attorneys can be influenced by the race of a client when giving plea advice – feeling that they could obtain better plea deals with a Caucasian client than with a minority client, even when controlling for perceptions of guilt (Edkins, 2011). However, the relationship between guilt and innocence, attorney advice, and plea decisions has not previously been examined in the literature. In this paper, we utilize attorney interviews to probe the extent to which innocent defendants are pleading guilty in the current system, the way that attorneys view their role in this complex system, and the ways in which defense attorneys consider guilt or innocence when providing advice on pleas.

Method

Participants

Participants were 189 criminal defense attorneys, recruited via defense attorney listservs in New York, Tennessee, Wisconsin, Wyoming, Vermont, Idaho, Iowa, Arizona, and Rhode Island (note that some attorneys chose to leave some questions blank and so we did not have our full sample of 189 attorneys for all questions so the number of attorneys responding to each question is reported). Participants completed the study online and participation took around 10 minutes. Participants were 57.6% male, and 93% white (the remaining attorneys identified as black, Asian, Native America, or of mixed ethnicity). Thirty-three percent of participants classified their political views as very liberal, 31% as somewhat liberal, 23% as moderate, 11% as somewhat conservative, and 2% as very conservative. Our participants had practiced law for 13.52 years (SD = 10.26), and represented the prosecution in criminal cases 2.81% of the time (SD = 11.01), and the defendant in criminal cases 80.43% of the time (SD = 27.45). Participants received monetary compensation for their participation. All participants provided consent to take part in the study, and the project was approved by the Cornell Institutional Review Board.

Participants were first asked whether they had any experience advising clients on whether to accept plea bargains and could select an answer of yes, no, or prefer not to answer. They were then asked a series of interview questions, and finally completed an experimental task involving two hypothetical vignettes.

Interview Questions

Participants were asked to respond to the following questions, in the order that they appear below.

How often do innocent defendants plead guilty?

Participants were asked whether they had ever been involved in a case where a client chose to plead guilty despite maintaining their innocence (and selected yes, no, or prefer not to answer), and whether they had ever advised a client who they believed was innocent to plead guilty (and selected yes, no, or prefer not to answer). They were then asked to provide a rough estimate of what proportion of defendants who plead guilty they think are completely innocent, or guilty of a lesser crime than the one they end up pleading guilty to (note that this question was added to the survey instrument late and therefore only 46 attorneys were asked this question). We asked participants for a written response, and told them that they could just write NA if they felt unable to give a figure.

When should innocent defendants plead guilty?

Next, we asked participants whether there were cases in which they believed that innocent defendants should plead guilty and if so, what the characteristics of these cases were.

The role of attorneys.

Participants were then asked questions probing attorney advice to clients. First, they were asked whether they have ever encouraged a client who wished to take a plea deal to go to trial. Second, they were asked whether they have ever encouraged a client who wished to go to trial to take a plea deal.

Experimental task assessing attorney advice

Participants were given two hypothetical plea vignettes and were asked to decide whether a client should go to trial or plead guilty in the vignettes. One (Case 1) involved a young girl who had been caught with marijuana, and the other (Case 2) involved an adult who was accused of sexual assault (the specific vignettes presented to attorneys are included in the appendix). In both cases, the defendant claims that they are innocent and must choose to plead guilty in exchange for a misdemeanor conviction or go to trial and risk getting convicted of a felony. We randomly assigned half of the participants to see additional facts in the hypotheticals which meant that pleading guilty (even to a misdemeanor) would have particularly harmful consequences for the client - likely rejection of a citizenship application (in the marijuana case) or the loss of a career as a teacher (in the sexual assault case). Participants would either see both cases with these additional consequences, or neither case with these additional consequences. We included this case to mirror scenarios such as that in the Lee case, where even a misdemeanor conviction would have severe consequences, to see whether this would discourage attorneys from advising a client to plead guilty (especially where they suspected they were innocent).

In each case, we asked participants to estimate the percentage likelihood of the defendant being guilty, and the percentage likelihood of the defendant being found guilty at trial. The order of these questions was counterbalanced. We then asked them what they would advise the defendant to do in the hypothetical, and asked them to give a short justification for their response We used these answers to examine the relationships between probability of conviction at trial, guilt and innocence, and plea advice, in a real task.

Following completion of all tasks attorneys completed some short demographics questions asking about gender, political orientation, and years in legal practice.

Results

One hundred and sixty-six participants answered our questions about experiences with the plea-bargaining system. Of these 166 participants, 163 (98.19%, 95% CI [94.82%, 99.38%]) stated that they had experience advising clients on whether to accept plea bargains, one stated that they did not have experience advising clients on whether to accept plea bargains (0.60%, 95% CI [.01%, 3.33%]), and two said that they preferred not to answer (1.20%, 95% CI [.03%, 4.28%]).

Interview Questions

How often do innocent defendants plead guilty?

When asked whether they had ever been involved in a case where a client chose to plead guilty despite maintaining their innocence, 148 (89.16%, 95% CI [83.52%, 93.03%]) of our participants said yes, 14 (8.43%, 95% CI [5.09%, 13.65%]) of our participants said no, and 4 (2.41%, 95% CI [.94%, 6.03%]) of our participants said that they preferred not to answer. Mean comparisons (using t-tests) revealed no significant differences in years of experience (p = .178) or political orientation (p = .353) between those who had been involved in such a case and those who hadn’t, and a chi-square analysis revealed no significant difference between male and female attorneys (p = .994).

Seventy-four (44.58%, 95% CI [37.23%, 52.18%]) of our participants said that they had advised a client who they believed was innocent to plead guilty, 78 (46.99%, 95% CI [39.55% – 54.56%]) said that they had not, and 14 (8.43%, 95% CI [5.09%, 13.65%]) preferred not to answer that question. Mean comparisons (using t-tests) revealed that participants who had advised a client they believed was innocent to plead guilty had significantly more years of experience on average than those who had not (Mhadadvised = 13.93 SD = 10.17 ; Mhadnot = 10.17 SD = 12.08, p=.041), but there was no significant difference in political orientation, and a chi-square analysis revealed no significant difference between male and female attorneys (p = .465).

Thirty attorneys provided an estimate of the proportion of defendants who plead guilty that they believe to be innocent, or guilty of a lesser offense than the one they pled guilty to. Estimates varied widely from less than 5% to 70% (M = 23.82, SD = 19.59) (Figure 1). An examination of correlations between years of attorney experience and estimated proportion of defendants who plead guilty that they believe to be innocent or guilty of a lesser offense revealed that as attorney’s experience increased, the lower their estimates were likely to be (r = −.457, p = .028). A one-way analysis of variance revealed no significant association between political orientation (r = .390, p = .054) and estimates and a t-test revealed no significant difference between estimates of male and female attorneys (p = .418).

Figure 1:

Attorney estimates of the proportion of people who plead guilty who are actually innocent, or guilty of a letter offense than they plead guilty to. Error bars represent 95% confidence intervals.

Are there cases in the current system in which innocent defendants should plead guilty?

We asked attorneys whether they believed there were cases in which innocent defendants should plead guilty. One hundred and fifty-two attorneys gave a yes or no response to this question. Of those attorneys, 119 (78.29%, 95% CI [71.08% – 84.10%]) said there were cases in which innocent defendants should plead guilty given the current system. Thirty-three (21.71%, 95% CI [15.9%, 28.92%]) said that innocent defendants should not ever plead guilty. Mean comparisons (using t-tests) showed that the attorneys that said there were cases in which innocent defendants should plead guilty and those that said that there were not did not differ in mean years of experience (p = .938) or political orientation (p = .745). A chi squared analysis showed that the proportion of attorneys responding that there were cases in which innocent defendants should plead guilty did not significantly differ based on attorney gender (χ² = .704, p=.402).

Attorneys who said that innocent defendants should plead guilty were asked to elaborate on the circumstances in which they thought this was the case. We reviewed responses to identify general (and overlapping) themes that responses would fit into. We identified four broad themes – when possible outcomes at trial were very severe, where a cost-benefit analysis favored taking the plea (e.g. because of strong evidence against a defendant and a good plea offer), where pleading guilty would benefit the client through faster and easier resolution of the case (e.g. getting them out of jail, or avoiding the difficulty and pressure of trial), and avoiding significant repercussions through a plea. Two authors then independently classified each response into one or more of the themes. There was unanimous agreement on the classifications, indicating high reliability. Table 1 shows a summary of the number of attorneys who identified each theme, and some sample responses for each theme. Note that some attorneys identified more than one theme in their response.

Table 1:

Responses when asked whether there are cases in which innocent defendants should plead guilty.

Theme	Number of Attorneys	Sample Responses
Severe outcome at trial	49	Many innocent people take deals when facing draconian mandatory penalties. Faced with decades of prison and offered a year or two, rational people don’t even gamble.
		Even if the risks of losing are extremely small, the consequences of losing can change a client’s life where the consequences of a plea will not.
		If the risks of conviction are great, even if the likelihood is low, it may make sense.
		The state has such incredible leverage and power in many cases, it can be in someone’s best interest to accept a plea to eliminate their risk of catastrophic consequences.
Cost-benefit analysis/ Chance of conviction and sentences involved	76	Where it is unlikely they will be successful at trial AND the plea deal is for a substantially less crime or minor sentence
		When the evidence is overwhelming and the stakes are very high, like mandatory minimum sentences to be avoided.
		Yes – it is a pure costs benefit analysis especially where there are immigration or job related consequences that can be avoided by a plea.
		When the likelihood of conviction and a substantially worse outcome outweighs the likelihood of acquittal or a substantially better outcome.
Faster and easier resolution	17	I have been involved in cases where it would have taken my clients longer than the sentence stimulated in the plea offer to go through trial and litigate the case.
		Clients in pretrial custody with a preexisting criminal record can benefit from pleading guilty to a petty charge for the sole reason that they can get out of custody and resume their lives.
		Unfortunately, clients may have to plead guilty to save money due to incarceration while waiting for trial.
		When a person has been held in jail for several weeks without the means to bond out and the offense is relatively minor, clients often lose their motivation to fight.
A plea offer than avoids any significant	15	Yes, an innocent client might plead guilty if the consequences would not hamper future activity such as job qualifications, or movement.
		A plea that avoids significant repercussions and eliminates the risk of these things is usually worth it.
		An obvious case is where the conviction will be of no practical consequence.

Several attorneys gave examples of cases in which they had likely innocent clients who chose to plead guilty. For example – a case where a likely innocent client was charged with a felony where a loss of life was involved but could plead guilty to a misdemeanor and be released from jail, a case in which a likely innocent client was charged with sexual assaults (and threatened with a 40-year felony sentence) and could pled guilty to disorderly conduct and be sentenced to pay only court costs, and a case in which a likely innocent client was accused of murder and could plead guilty to being an accessory after the fact and be sentenced to time served (8 months in jail) and probation. Many attorneys also noted that the ultimate decision as to whether to plead guilty (when innocent or otherwise) was with the client. Another recurring theme was mandatory sentences (laws in some states that require defendants convicted of certain crimes to serve predefined terms, see Subramanian, & Delaney, 2014). Mandatory sentences were mentioned by 15 attorneys as a reason that innocent defendants pled guilty, due to the harsh outcomes mandated upon conviction at trial, and discrepancies between these outcomes and plea offers.

The role of attorneys.

Examination of responses showed that the number of attorneys who said that they had encouraged a client who wished to go to trial to take a plea deal was significantly higher than the number of attorneys who said that they had encouraged a client who wanted to take a plea deal to go to trial (χ² = 29.37, p<.001). One hundred and fifty-three attorneys answered our question about whether they had ever encouraged a client who wished to take a plea deal to go to trial. Of these attorneys, 97 (63.40%, 95% CI [55.53%, 70.62%]) said that they had encouraged a client who wished to take a plea deal to go to trial, and 56 (36.60%, 95% CI [29.38%, 44.47%]) said that they had not. Mean comparisons (using t-tests) showed that attorneys who said they had encouraged a client who wished to take a plea deal to go to trial had significantly greater average experience (M = 15.47, SD = 11.16) than those who said they had not (M = 10.64, SD = 7.71) (p = .007). The groups did not significantly differ in political orientation (p =.913). A Chi-Square analysis indicated that the proportion of male attorneys who had encouraged a client who wished to take a plea deal to go to trial was greater than the proportion of female attorneys (Male = 68.1%, Female = 52.2%, χ² = 5.378, p=.020).

When asked whether they had ever encouraged a client who wanted to go to trial to accept a plea deal, 154 of our attorneys gave a yes or no answer. Of these attorneys, 138 (89.60%, 95% CI [83.79%, 93.50%]) said that they had encouraged a client who wanted to go to trial to accept a plea deal, and 16 (10.39%, 95% CI [6.50%, 16.21%]) said that they had not. An analysis using t-tests indicated that these groups did not differ significantly in years of experience (p = .469) or political orientation (p = .317). There was also no significant difference in the responses of male and female attorneys (Male = 86.8%, Female = 88.1%, χ² = .306, p=.580). These results suggest that attorneys are advising clients who want to go to trial to plead guilty more often than they are advising clients who want to plead guilty to go to trial.

We also examined attorneys who said they had advised a client who they thought was innocent to plead guilty and attorneys who said they had never advised a client who they thought was innocent to plead guilty separately, to compare the proportion of each group who had encouraged a client who wanted to go to trial to plead guilty. The difference between the two groups was significant – a greater proportion of attorneys who had advised a client who they thought was innocent to plead guilty (94.1%, 95% CI [85.13%, 97.08%]) had encouraged a client who wanted to go to trial to plead guilty than attorneys who had never advised a client who they thought was innocent to plead guilty (71.4%, 95% CI [60.96%, 80.57%]) (χ² = 5.679, p=.017).

We examined the reasons that attorneys gave for encouraging (or not encouraging) a client who wanted to accept a plea deal to go to trial (Table 2) and the reasons that attorneys gave for encouraging (or not encouraging) a client who wanted to go to trial to accept a plea deal (Table 3) (note that not all attorneys gave a reason and some attorneys gave more than one reason). For each set of responses, we identified broad categories into which responses fell and categorized responses into these categories. Coding of each of the two reviewers matched almost exactly with only one attorney response being coded differently, indicating high reliability.

Table 2:

Reasons given by attorneys for encouraging a client to go to trial rather than plead guilty, and for not encouraging a client to go to trial rather than plead guilty.

	Proportion of attorneys in each category giving this reason	95% Confidence Interval
Reason for encouraging a client to go to trial (n=97)
Client doesn’t appreciate the long-term consequences of a plea	.093	.050 – .167
The client is irrational as they are scared	.062	.029 – .129
The client underestimates their chance of success and / or there is a good chance of success at trial	.515	.417 – .613
The plea offer is bad and / or the client has little to lose by going to trial	.340	.254 – .439
Believe that the client is innocent	.124	.072 – .204
Reasons for not encouraging a client to go to trial (n=56)
The unpredictability of jury trial / possibility of jury bias / don’t want to impose risk on the client	.107	.050 – .215
It is the client’s choice what to do	.589	.459 – .708
It is unethical to do so	.036	.010 – .121

Table 3:

Reasons given by attorneys for encouraging a client to plead guilty rather than go to trial, and for not encouraging a client to plead guilty rather than go to trial.

	Proportion of attorneys in each category giving this reason	95% Confidence Interval
Reason for encouraging a client to plead guilty (n=138)
Chances of success at trial are very small	.667	.585 – .740
Potential consequences of trial / risk at trial too great	.674	592 – .747
The unpredictability of trial / minimizing uncertainty	.051	.025 – .101
The client has an unrealistic view of the facts and / or trial	.058	.030 – .110
The costs of trial	.065	.035 – .119
Reasons for not encouraging a client to plead guilty (n=16)
It is the client’s choice what to do	.563	.332 – .769
It is unethical to do so	.063	.011 – .283

Experimental Task Assessing Attorney Advice

Finally, we analyzed the results of our experimental task in which we gave attorneys case facts from two cases (Case 1 involving a school girl accused of marijuana possession, and Case 2 involving an adult accused of sexual assault). One hundred and eighty-nine attorneys answered these questions. Overall, attorneys advised the defendant to plead guilty more in Case 1 (M_case1=.55, SD = .50, M_case2= .37, SD = .49, p<.001 (Fishers Exact Test)). Attorneys also thought that the defendant in Case 1 was more likely to be guilty (M_case1=65.12, SD = 24.58, Range = 0% – 100%; M_case2= 47.00, SD = 20.56, Range = 0% – 100%; t(170)=8.54, p<.001), and that the defendant in Case 1 was more likely to be convicted at trial (Mcase1=77.34, SD=16.12, Range = 10% – 100%; Mcase2 = 62.68, SD =20.77, Range = 9% – 100%; t(170)=8.19, p<.001). Ninety-five attorneys were in our condition where a misdemeanor conviction would have its normal consequences in each case. Ninety-four attorneys were in our condition where a misdemeanor conviction would have an abnormally severe impact (by impacting immigration status in Case 1, or through causing the loss of a career as a teacher in Case 2).

In both cases, we conducted a logistic regression using our misdemeanor impact condition, attorney estimates of the probability of conviction at trial, and attorney estimates of defendant guilt, to predict plea advice. We also controlled for gender which was not the same across our experimental manipulations (in our normal misdemeanor condition, 37% of participants were male, and in our severe misdemeanor condition, 48% of participants were male).¹ The results for Case 1 (marijuana) and Case 2 (sexual assault) are displayed in Table 4.

Table 4:

Logistic regression results using impact of misdemeanor, probability of conviction at trial estimate, probability of guilt estimate, and gender, to predict plea advice (0=Go to trial, 1=Plead guilty).

	B	SE	Wald	OR
Case 1 (Marijuana)
Misdemeanor Impact (0,1)	−1.90	.393	23.30	6.65*
Prob. of conviction estimate	.046	.013	11.68	.955*
Prob. of guilt estimate	.016	.008	3.84	.985*
Gender (0,1)	−.054	.387	.020	.947
Case 2 (Sexual Assault)
Misdemeanor Impact (0,1)	−.419	.430	.950	1.52
Prob. of conviction estimate	.081	.015	31.01	.922*
Prob. of guilt estimate	.025	.013	3.84	.975*
Gender (0,1)	.250	.422	.334	.779

^*p≤.05

In both cases, the probability of conviction estimate and the probability of guilt estimate were significant predictors of plea advice, such that as estimations of the probability of conviction increased and as estimators of the probability of guilt increased, the chance of advising the client to plead guilty increased. In case 1, misdemeanor impact was also a significant predictor, such that when getting a misdemeanor would have a more severe consequence for the defendant (namely influencing immigration status) attorneys were more likely to advise the client to go to trial.

Finally, we looked specifically at attorneys who had stated that there was a less than 50% chance that the defendant was guilty in each case, to specifically examine the decisions of attorneys who believed the defendant was probably innocent.

In Case 1, 29 attorneys thought that there was a less than 50% chance that the defendant was guilty. Of these attorneys, 45% recommended that the defendant plead guilty. A logistic regression showed that in this group attorneys in the condition with a severe impact of a misdemeanor were less likely to recommend pleas (B=−2.873, SE = 1.085, Wald = 7.012, OR =17.70, p=.008), but there was no significant influence of likelihood of conviction at trial, or gender. We examined the reasons why these attorneys, who thought there was a less than 50% chance that the client was guilty, would recommend a guilty plea. Responses referred to the fact that she would be unlikely to win at trial (e.g. She seems unlikely to win at trial) and the more serious consequences of a felony conviction (e.g. The plea would mean that she would not lose her civil rights, would not have a felony conviction on her record…I would want to protect her in case of future police contact. The risks of a felony conviction along with a lifetime of consequences associated with that conviction are too great to accept in this scenario when compared with the far less significant consequences of explaining a misdemeanor charge).

In Case 2, 65 attorneys thought that there was a less than 50% chance that the defendant was guilty. Of these attorneys, 22% recommended that the defendant plead guilty. A logistic regression showed that in this group attorneys who thought there was a higher chance of conviction at trial were more likely to recommend pleading guilty (B=.091, SE = .026, Wald = 11.92, OR =.913, p=.001), but there was no significant influence of our misdemeanor manipulation, or gender. We examined the reasons why these attorneys, who thought there was a less than 50% chance that the client was guilty, would recommend a guilty plea. Responses referred to the fact that the risk of trial would be high (e.g. The risk is too high for a trial, but I would not push the client to make this decision), that juries in child sexual abuse cases often believe the child (e.g. It will be his word against the client and juries are liable to believe the child), and that the likely penalty if convicted would be very serious (e.g. He would face decades in prisons and many other severe repercussions from a felony conviction of this nature. No one could risk the mandatory felony penalties on sex crimes when offered a misdemeanor).

Discussion

This study provides what is, to our knowledge, the first empirical study of attorney perspectives on plea bargaining, and attorney plea recommendations. Although our data is only from a relatively small sample of attorneys, it confirms concerns that have been raised in the legal literature and the psycho-legal literature regarding the extent to which innocent defendants are pleading guilty and the procedures in the current plea system that seem to encourage this practice (see Blume & Helm, 2014; Dervan, 2012; Dervan & Edkins, 2013; Covey, 2013; Hessick & Saujani, 2002). Results provide important insight into how often innocent defendants are (and should be) pleading guilty in the current system, attorney perceptions of their role in the plea system generally and this practice specifically, and the extent to which guilt and innocence might influence attorney advice. This insight allows us to form conclusions about limitations in the current system and to better understand how attorney advice might be shaping plea practice.

Innocent Defendants Do Plead Guilty But It Is Unclear How Often

Scholars supporting plea bargaining have argued that innocent defendants would be unlikely to find plea offers attractive, going as far as to describe the problem of innocent defendants pleading guilty as “barely a perceptible theoretical ripple” when compared with other costs in the plea-bargaining system (Easterbrook, 1992; Schulhofer, 1992). However, accumulating research suggests that innocent defendants might be pleading guilty fairly frequently (for example Dervan & Edkins, 2013; Grisso et al., 2003; Helm & Reyna, 2017; Tor, Gazal-Ayal, & Garcia, 2010). Our results support this accumulating research. They indicate that almost all of our attorneys have experience dealing with clients who claim to be innocent but plead guilty (although we do not have an indication of how often). Importantly, when asked about the proportion of defendants who plead guilty but are really innocent or guilty only of a lesser charge, there was very little consensus among attorneys, with estimates ranging from less than 5% to over 50%, and many choosing not to give a response at all (although note that this question was only given to 46 of our attorneys). This is understandable since it is very difficult for an attorney to know whether a client is factually innocent or guilty. However, importantly, 27 of the 30 attorneys who answered this question believed that of the defendants who plead guilty 5% or more are factually innocent or guilty of a lesser charge. If true, this would make plea bargaining a very important cause of wrongful conviction (although note that wrongful convictions resulting from plea are likely to be convictions for lesser sentences than those resulting from trial), and certainly more of a problem than a barely perceptible theoretical ripple.

The Majority of Attorneys Agree There Are Situations In Which Innocent Defendants Should Plead Guilty And Have Advised Accordingly

Our results indicate that the majority of attorneys do believe that there are situations in the current system in which innocent defendants should plead guilty, and have advised defendants who they believe to be innocent to plead guilty. Specifically, almost half of our attorneys indicated that they had advised clients who they believe to be innocent to plead guilty. Results also provided insight into when attorneys might provide this kind of advice, with the leading two reasons being when a costs benefits analysis promotes this, and when defendants are threatened with severe punishments at trial (even where there is not a high chance of conviction at trial).

The costs benefits analysis scenario fits with arguments justifying the plea system on the basis that innocent defendants are able to, and should be able to, enjoy the benefits of pleading guilty if this is their risk preference (Bowers, 2008; Easterbrook, 1992). According to these arguments, persons at risk of unjust conviction may prefer a certain (but low) punishment in a plea bargain to the risk of conviction and higher punishment after trial (Easterbrook, 1992), and this can lead to less punishment for defendants who are truly innocent (Bowers, 2008). These arguments are based on the idea that defendants will be able to rationally weigh the costs and benefits of a plea offer in a sensible way. However, importantly, our research suggests that from the perspective of attorneys there are other reasons that innocent defendants may find pleading guilty the most sensible option, given the current system. Defendants are threatened with such harsh punishments at trial (something that is exacerbated by mandatory minimum sentences) and offered very lenient plea deals – for example a sentence of time served (8 months) and a misdemeanor conviction when threatened with a murder conviction if convicted at trial. This kind of discrepancy is likely to lead to innocent defendants pleading guilty even when there is only a very small probability of conviction at trial, and the gravity of the threatened punishment does not truly give defendants a fair choice in this situation. This problem can be exacerbated by the practice of vertical “overcharging,” whereby prosecutors include different substantive offences in an initial charge with the intent to dismiss one or more of them (Ross, 1978).

Attorneys Advise Clients Who Want To Go To Trial To Plead Guilty More Than They Advise Clients Who Want To Plead Guilty To Go To Trial

Our results suggest some disagreement among attorneys regarding their role in the plea-bargaining process, with some attorneys providing clients with advice on their decisions based on their knowledge and experience, and other attorneys feeling their job is only to provide information to clients who should then be left to make their own decisions.

Our results suggest that attorneys are more likely to encourage a client who wants to go to trial to enter a plea bargain than to encourage a client who wants to plea bargain to go to trial (although note that this could be because more clients have a bias towards trial, rather than because attorneys have a bias towards pleas). Interestingly, this difference is driven largely by female attorneys, who are more reluctant than male attorneys to encourage a client who wants to plead guilty to go to trial. Responses show that the difference may be due to a reluctance to be responsible for imposing a large risk on a client, especially when consequences of conviction at trial could be severe (even if chances of this outcome are low). Attorneys who had not ever encouraged a client who wanted to take a plea bargain also frequently said that they did not do so because the decision was up to the client, although note that this consideration was important to far fewer attorneys when asked whether they had ever encouraged a client who wanted to go to trial to accept a plea bargain (8 attorneys, compared to 26).

Importantly, the proportion of attorneys noting that they had encouraged a client who wanted to go to trial to plead guilty was greater among our attorneys who said they had encouraged a client they suspected to be innocent to plead guilty. This does suggest that attorneys may be leading to a greater number of innocent clients pleading guilty, although this is not necessarily a bad thing given the current system and the considerations discussed in this paper.

Guilt and Innocence Do Influence Attorney Advice

The results of our experimental task confirm that even in cases in which attorneys think a client is innocent, they may encourage them to plead guilty (45% of attorneys in our first case and 22% of attorneys in our second case who were less than 50% sure the defendant was guilty said they would advise them to plead guilty). Responses suggest that although some attorneys based this advice on a high chance of conviction at trial, attorneys were also influenced by a reluctance to expose clients to the risks of a severe outcome if convicted at trial. Importantly (especially for our Case 2 where there was no impact of our severe misdemeanor consequences condition) some attorneys were advising clients they thought were likely not guilty to plead guilty even where the plea would have severe consequences for them (in Case 2 this was loss of a career in teaching). However, overall results of this task did show that as attorney estimates of guilt increased, the tendency to recommend a guilty plea also increased, even when controlling for the probability of conviction at trial. This suggests that attorneys are more reluctant to encourage defendants who they believe are innocent to plead guilty.

Limitations and Future Directions

The findings of our study should be interpreted in light of some important limitations. First, several attorneys completing our hypothetical plea scenarios noted that not enough information had been provided for them to truly assess the cases. For example, they did not have access to witness statements, were not able to ask questions of the defendant, did not have evidence of the history of the defendant, and did not have expert opinions on the quality of case evidence (e.g. memory evidence in Case 2). This information would allow attorneys to better assess the evidence and their client’s chances at trial, and would thus likely influence plea advice. In addition, we did not include preferences of defendants in our vignettes, and the vignettes were hypothetical and differed in important respects from real legal cases. Another limitation of this work is that it relied on self-report responses by a sub-set of attorneys. Our findings are therefore influenced by which attorneys chose to participate in the study, and what they chose to report to us.

Our findings should also be considered alongside reports of defendants about their interactions with attorneys, which suggest that in some cases the interaction of clients and attorneys may be limited (Zottoli et al., 2016). The amount of time spent with an attorney is likely to influence plea decisions, and is not something that we examined in this study. An attorney who only spends a few minutes with a client prior to a plea hearing may leave a client feeling that they have no option but to plead guilty due to the rushed context in which the discussion takes place which may leave them with little confidence in the attorney. An attorney who spends longer with a defendant is likely to be able to more clearly convey the pros and cons of each decision and leave the client with a clear understanding that they can base their decision on. Results should be interpreted with these limitations in mind, and future research should investigate the findings of this study further when considering real legal cases in which attorneys have advised clients. Future research should also gather more responses from each state, to enable comparisons between attorneys from states with different legal infrastructures (e.g. whether the state allows Alford pleas or not), and data on different types of cases to further compare whether plea recommendations differ depending on the type of case.

Conclusions

In this paper, we examined attorney perspectives on the extent to which innocent defendants are (and should be) pleading guilty in the current system and on their role in this complex system. We also examined the ways in which defense attorneys consider guilt and innocence when advising on pleas, using a hypothetical case. Results show that in the current system innocent defendants are pleading guilty, and attorneys are advising them to do so in certain cases (despite being more reluctant to advise pleading guilty when a client is more likely to be innocent). Most importantly, results suggest that in the current system these attorneys are likely doing the right thing, and protecting their clients from severe consequences at trial.

A theme that has come out of this study is that in the current system there are cases in which, from a practical perspective, innocent defendants should plead guilty, even when the chances they will be convicted at trial are not high. Importantly, a leading reason for this is that there is such a discrepancy between a plea offer and the outcome if convicted at trial that exercising the right to a trial becomes too risky even when the chances of conviction are low. The severe risks that can result from going to trial also appear to be making attorneys reluctant to advise clients to go to trial, since attorneys do not want to be responsible for imposing the risk of a severe conviction on a client when they could have received a much more lenient sentence in exchange for pleading guilty. This undercuts justifications that have been made for innocent defendants pleading guilty, since it takes away the consent of the defendant and the ability of the attorney to prevent the conviction of innocent defendants. When confronted with the chance of a possible murder conviction vs. a misdemeanor, defendants and attorneys often do not truly have a choice even where the chances of conviction are low. Thus, attorney advice may be driving up the number of innocent defendants pleading guilty while also protecting innocent defendants from disastrous consequences at trial. So, perhaps 95% of innocent defendants are accepting pleas when they would have been found innocent at trial, while 5% of innocent defendants are avoiding disastrous consequences that would have occurred at jury trial. Put simply, attorney’s hands are tied. All they can do is to try to protect clients as best they can in the current system, rather than actually negotiating a just outcome.

These results provide further evidence of the need for reforms in the current system, and particularly highlight the need to reduce the discrepancies between sentences and charges given in exchange for pleading guilty and potential outcomes at trial. This could be done either through constraining the discounts offered by prosecutors in exchange for pleas, or by introducing plea-based ceilings so that trial sentences could not exceed plea sentences by more than a modest amount (Covey, 2007). If it is really the case that someone who has committed a certain crime should get a certain sentence, this should be the case for all people who have committed that crime, not just for those who chose to exercise their right to a jury trial. Reducing the discrepancy between outcomes if convicted at trial and outcomes when pleading guilty would allow attorneys to appropriately advise their clients absent the fear of their client receiving a disproportionately high punishment, thus improving their ability to effectively negotiate justice and protect innocent defendants.

Appendix

Case 1 Introduction

Imagine that you are representing a client, Sara. Sara is a senior in college who makes below average grades and has spotty attendance, but she is likely to graduate.

On a recent Saturday evening, a police officer noticed her and a group of people hanging around a school playground. The officer suspected that he smelled marijuana and called for backup. The police quickly surrounded the playground and moved in. Sara tried to run off but was cornered in an alleyway. She was found with a small bag (3 grams or about one-tenth of an ounce) of methamphetamine powder and a list of known drug dealers.

The other 19 people at the scene fled and were not apprehended by police.

Sara claims that she only had the methamphetamine and list because it was handed to her by someone else as they ran away.

The prosecution have offered Sara a plea deal whereby she can plead guilty to possession of methamphetamine (a misdemeanor in the jurisdiction), and other charges against her (including felony trafficking charges) will be dropped.

[Sara is a legal immigrant in the process of applying for US citizenship. If she receives any criminal conviction, it is unlikely that her citizenship application will be successful]

Case 1 Follow-Up Questions

What do you think is the likelihood of Sara being found guilty of a criminal offense at trial?

What do you think is the likelihood that Sara is guilty?

What would you advise Sara to do in this situation?

Please briefly explain why you would give this advice.

Case 2 Introduction

Imagine that you are representing an adult client, James. James is accused of sexually assaulting

10-year-old child. This is based on claims made by the child that while James was babysitting her and she was in the shower, James came into the bathroom and told her to put his penis in her mouth. The child’s mother claims that the child told her what had happened as soon as she arrived home, and was very distressed by the incident.

James was babysitting her on the night in question but strongly denies any wrongdoing. He notes that he did enter the bathroom while the child was in the shower as she cried out needing assistance, but that he did not tell her to put his penis in her mouth or engage in any other sexually suggestive behavior.

The prosecution have offered James a plea deal whereby he can plead guilty to misdemeanor sexual assault against a child, and the felony sexual assault against a child charge will be dropped.

[James is a teacher and if he receives either of these convictions he will lose his current job and will never be able to work as a teacher again.]

Case 2 Follow-Up Questions

What do you think is the likelihood of James being found guilty of a criminal offense at trial?

What do you think is the likelihood that James is guilty?

What would you advise James to do in this situation?

Please briefly explain why you would give this advice.