SEARCH

SEARCH BY CITATION

Keywords:

  • coercion;
  • confession;
  • discounting;
  • inconsistencies;
  • juror decisions

Abstract

  1. Top of page
  2. Abstract
  3. Background
  4. EXPERIMENT 1
  5. Method
  6. Results and discussion
  7. EXPERIMENT 2
  8. Method
  9. Results and discussion
  10. GENERAL DISCUSSION
  11. Acknowledgements
  12. References
  13. Supporting Information

Purpose

Although inconsistencies undermine the credibility of evidence from a witness or victim, anecdotal evidence from many court cases suggests that they do not reduce the impact of confession evidence. This research provides the first empirical test of this idea by experimentally manipulating the consistency of confession evidence. Drawing on principles from attribution theory, we hypothesized that inconsistencies would undermine the credibility of confession evidence only when there was a salient, plausible alternative explanation (other than guilt) for why the defendant confessed.

Methods

In two experiments (total = 245), participants were presented with information about a crime, including a confession statement, and asked to act as jurors in a courtroom case. As well as manipulating whether the confession was consistent or inconsistent with verifiable facts of the crime, we manipulated whether there was a salient alternative explanation for the confession: specifically, the presence of coercion ('EXPERIMENT 1') or the desire to protect another suspect ('EXPERIMENT 2').

Results

Inconsistencies influenced participants' verdicts regardless of whether an alternative explanation was made salient, such that inconsistent confessions resulted in fewer guilty verdicts than consistent confessions. Additional mediation analysis of the data from 'EXPERIMENT 2' suggested that these effects occurred, in part, because the presence of inconsistencies prompted participants to generate alternative explanations for why the defendant confessed (regardless of whether such explanations were salient in the available evidence).

Conclusions

Contrary to the existing literature, these results indicate that inconsistencies can undermine the credibility of confession evidence.


Background

  1. Top of page
  2. Abstract
  3. Background
  4. EXPERIMENT 1
  5. Method
  6. Results and discussion
  7. EXPERIMENT 2
  8. Method
  9. Results and discussion
  10. GENERAL DISCUSSION
  11. Acknowledgements
  12. References
  13. Supporting Information

In 1983, 22-year-old Earl Washington was arrested for alleged burglary and malicious wounding. After being held in custody and questioned for 2 days, Washington – who had an IQ of 69 – confessed to five different crimes. Four confessions were dismissed after the victims failed to identify Washington as the perpetrator. However, Washington was charged with the fifth crime, a rape and murder. His fourth attempt at a rehearsed confession was accepted by police; this was the only evidence connecting him to the crime. Washington did not know the race of his victim, the location of the crime, or that she had been raped. His confession also contained numerous statements that were inconsistent with the facts of the crime. He testified that the victim was short (she was 5′8″), that no one else was present at the time of the crime (her two children were in the apartment), and that he had stabbed her two or three times (she had 38 stab wounds). Following a 50-min deliberation, a jury convicted Washington, who was sentenced to death. Washington was eventually exonerated on the basis of DNA evidence in 2001, having spent 17 years in prison and coming within 9 days of execution (Innocence Project, 2012).

Earl Washington's is one of numerous cases in which an innocent defendant has been convicted largely on the basis of a confession that later proved to be false (Dwyer, Neufeld, & Scheck, 2000; Kassin, Bogart, & Kerner, 2012; Leo & Ofshe, 1998). Prior research has addressed many questions relating to false confessions, such as why innocent people confess to crimes they did not commit (Kassin & Kiechel, 1996; Madon, Guyll, Scherr, Greathouse, & Wells, 2012), what procedural factors increase the risk of false confessions (Madon, Yang, Smalarz, Guyll, & Scherr, 2013; Narchet, Meissner, & Russano, 2011; Perillo & Kassin, 2011), and why jurors' verdicts are swayed strongly by confessions even when circumstances suggest that the confession may be false (Kassin & Sukel, 1997; Kassin & Wrightsman, 1980).

This research examined the effect of inconsistencies in confessions on juror judgements of guilt. Anecdotal and case study evidence from individual trials suggests that inconsistencies in confessions have little effect on jurors' judgements (Malloy & Lamb, 2010). That is, although inconsistencies undermine the credibility of evidence from a witness or victim, they do not undermine the credibility of evidence from a defendant. This study provides an empirical test of whether inconsistencies in confession evidence affect juror judgements of likely guilt, and tests whether the presence of inconsistencies in a confession change the way that jurors process trial evidence. Drawing on principles from the attribution theory framework (Kelley, 1973), we hypothesized that inconsistencies would prompt jurors to engage in a greater level of attributional thinking, whereby they would be more likely to consider alternative explanations – other than guilt – for why a suspect confessed (Fein, McCloskey, & Tomlinson, 1997).

Testimonial inconsistencies: Different effects for eyewitness evidence versus confessions?

Inconsistencies in evidence can take numerous forms. Information provided in a police interview may contradict information given by the same person in a previous interview. A person may change the nature of their evidence entirely, such as when a suspect confesses to a crime then recants their confession. Inconsistencies can also take the form of errors; that is, the information provided in testimony might be inconsistent with verifiable facts of the crime (as in Earl Washington's case).

When an eyewitness provides testimony, inconsistencies and errors in their evidence typically undermine the credibility of the witness and reduce the impact of the evidence on jurors' verdicts (e.g., Berman & Cutler, 1996; Berman, Narby, & Cutler, 1995; Brewer & Hupfeld, 2004; Semmler & Brewer, 2002; but cf. Brewer & Burke, 2002, for an example of null results). In contrast, anecdotal evidence from numerous case studies suggests that inconsistency does not have the same undermining effect on the credibility of confession evidence (Malloy & Lamb, 2010). That is, inconsistencies in a confession do not seem to detract from the perceived credibility of the confession or reduce the likelihood of guilty verdicts from jurors. Several empirical studies have provided evidence consistent with this view (Najdowski, Bottoms, & Vargas, 2009; Redlich, Ghetti, & Quas, 2008). In these studies, a child provided evidence about a misdemeanour and immediately retracted it (a strong form of inconsistency). The key manipulation was the role of the child, who was either a witness to the event or a suspect confessing to the act. Mock jurors tended to accept the retraction if the evidence was provided by a witness but disregard the retraction if the evidence was provided as a confession, suggesting that inconsistency did not undermine the credibility of the confession.

Importantly, however, these studies did not include a condition in which the evidence was consistent (not retracted). Because no prior study has experimentally manipulated the consistency of confession evidence, it remains an empirical question as to whether inconsistencies in confession evidence affect juror judgements.

A theoretical framework: Discounting and augmentation

We investigated whether jurors disregard inconsistencies in confessions when rendering a verdict and, if so, why this occurs. As a framework, we drew on the principles of augmentation and discounting (defined below) from attribution theory (Kelley, 1973). Observers often seek to explain another's behaviour by attributing it to a likely cause. A classic example involves observing a student who is observed stopping to assist a professor (Reeder, Vonk, Ronk, Ham, & Lawrence, 2004). In the absence of an alternative explanation, the observer might attribute the student's helpful behaviour to their nature (being a helpful person). However, additional contextual information can alter such attributions via augmentation and discounting. Augmentation refers to a process whereby attributions are strengthened when the observer is aware of contextual factors that would normally inhibit the observed behaviour. For example, if the observer knows that the student was on their way to a lecture but still stopped to help the professor, the observer would be especially convinced that the student is a helpful person. Discounting refers to the opposite process: attributions are weakened by the presence of factors that provide an alternative explanation for the observed behaviour. If the observer knows that the student has been nominated for an award to be judged by the professor, the student's helpful behaviour might be attributed to a desire to win the award. Although the majority of work on augmentation and discounting has dealt with judgements about dispositional versus situational attributions for observed behaviour, these principles also apply to other aspects of judgements, such as the perceived validity of information (Brandt, Vonk, & van Knippenberg, 2011).

Inconsistent testimony

The principles of augmentation and discounting offer a potential explanation for why inconsistencies might undermine eyewitness evidence but not confession evidence. This explanation rests on the availability of a plausible reason as to why the inconsistent evidence was provided. A juror hearing evidence may initially assume that the person giving evidence is trying to provide an accurate account of events, given that people generally tend to assume that others are telling the truth (Bond & DePaulo, 2006). But inconsistent evidence may prompt a juror to become suspicious that the person providing the evidence has an ulterior motive; in turn, the juror may consider alternative explanations for why the evidence is being provided. For example, if the testimony comprises incriminating evidence from a witness or victim, the juror may think that the person giving evidence is attempting to help the prosecution obtain a conviction (Fein et al., 1997). If such an explanation is generated and deemed plausible, the juror will likely discount the relevant evidence when formulating to their verdict.

However, when inconsistent evidence is provided in a confession, it is less likely that jurors will be able to generate a plausible alternative explanation for why the evidence was provided (other than guilt), because people find it extremely difficult to imagine circumstances that would lead them – or another person – to confess to a crime they did not commit (Henkel, Coffman, & Dailey, 2008; Kassin et al., 2010). Indeed, because the confession was provided despite the strong inhibiting effect of the ensuing consequences, it is likely that the process of augmentation would further push jurors towards concluding that the person who confessed really must be guilty.

This account of why jurors' verdicts are not influenced by inconsistencies in confessions generates testable predictions. According to this explanation, inconsistencies in confession evidence fail to undermine the credibility of the evidence due to a lack of plausible alternative explanations for why the confession was made (other than guilt). However, when a plausible alternative explanation is made salient, inconsistencies in confession evidence should undermine the credibility of that evidence. As a result, the confession should have less bearing on jurors' verdicts.

We tested this idea in two experiments. In each, participants were asked to act as jurors and were presented with information about a fictional court case. The information included some verifiable details of a crime, and a copy of a signed confession provided by the defendant. The consistency of the confession evidence was manipulated by varying several details in the confession so that they matched or contradicted the verifiable facts of the crime. We also manipulated the salience of a plausible ulterior motive for the defendant to confess. In 'EXPERIMENT 1', we varied the degree to which the confession was coerced, rather than provided voluntarily. Participants read either that the defendant confessed spontaneously at the start of a police interview, or that they were coerced to confess after hours of suggestive questioning. In 'EXPERIMENT 2', we manipulated motive by providing some participants with information implying that the defendant had confessed in order to protect another suspect from investigation.

EXPERIMENT 1

  1. Top of page
  2. Abstract
  3. Background
  4. EXPERIMENT 1
  5. Method
  6. Results and discussion
  7. EXPERIMENT 2
  8. Method
  9. Results and discussion
  10. GENERAL DISCUSSION
  11. Acknowledgements
  12. References
  13. Supporting Information

'EXPERIMENT 1' aimed to (1) test the effects of inconsistencies in confessions on jurors' verdicts, and (2) test whether any effect of inconsistencies on verdicts is moderated by the presence of a plausible, salient explanation for the confession (other than guilt). We expected that inconsistencies would have no effect on verdicts when no explanation was salient. However, when an explanation was made salient (via information suggesting that the confession was coerced), we expected that inconsistent confession evidence would produce fewer guilty verdicts than consistent confession evidence.

Method

  1. Top of page
  2. Abstract
  3. Background
  4. EXPERIMENT 1
  5. Method
  6. Results and discussion
  7. EXPERIMENT 2
  8. Method
  9. Results and discussion
  10. GENERAL DISCUSSION
  11. Acknowledgements
  12. References
  13. Supporting Information

Participants and design

Participants (= 129; 84 female; aged 18–54 years, = 23.56 years, SD = 7.04), comprised 50 undergraduate students (who received course credit or $10) and 79 members of the broader community (who received $10). We excluded three additional participants who were ineligible for jury duty (aged 17 years) and one who did not provide a verdict. This had a negligible effect on the results. Participants were randomly allocated to one cell of a 2 (Coercion) × 2 (Consistency) between groups design.

Materials

Stimulus materials comprised a single page information sheet describing a crime, and a copy of a signed confession made by the defendant (see Supporting Information online for examples).

The information sheet included a description of the crime, a list of facts relating to the crime that had been verified by police, and some additional information about the case. The crime was an armed robbery and homicide. The alleged perpetrator entered a pharmacy, demanded money, and shot the pharmacist before leaving with the money. The facts of the crime included information about the time of the crime, the number of gunshots received by the victim, a description of a duffel bag found near the scene, and the results of DNA testing of blood found at the scene. The additional information included a brief description of the police interview in which the defendant had confessed, based on an audio recording of the interview obtained by the defence attorney. Four versions of the information sheet were constructed, corresponding to the four experimental conditions described in the following section.

The confession statement was written in first person and described the defendant, Mr. Lucas, entering the pharmacy and walking over to the perfume display. Mr. Lucas then approaches the pharmacist with a gun and demands money be put into a duffel bag. The pharmacist refuses and Mr. Lucas shoots him before fleeing the pharmacy. The confession was designed to look like a photocopy of an actual statement obtained during a police interview (i.e., with appropriate letterhead and signatures from the confessor and the attending officers). A single version of the confession statement was used across all experimental conditions. Participants received no information about whether the confession was recanted.

Experimental manipulations

The presence of inconsistencies was manipulated by varying three facts of the crime (described in the information sheet) so that they matched or did not match the information in the confession statement. The confession stated that the crime had occurred at 4:30 p.m., a green duffel bag was used, and the victim was shot five times. In the consistent condition, all facts described in the information sheet matched those described in the confession. In the inconsistent condition, the information sheet stated that the crime had occurred in the morning (the panic alarm had been triggered at 10:18 a.m.), a red backpack was found near the crime scene, and the victim had been shot once.

The presence of coercion in the police interview was manipulated via the description of the police interview given in the information sheet. In the no coercion condition, this information stated: ‘At the beginning of the interview, police asked Mr. Lucas if he was involved in the crime and Mr. Lucas confessed’. In the coercion condition, the information stated: ‘At the beginning of the interview, police asked Mr. Lucas if he was involved in the crime and Mr. Lucas denied involvement in the crime. During the 12-hr interview that followed police continued to tell Mr. Lucas that he was in fact guilty of committing the crime, despite Mr. Lucas denying involvement in the crime a total of 32 times. At the end of the 12-hr interview, Mr. Lucas confessed'. The magnitude of these inconsistencies might appear large at first glance. However, inconsistencies in actual cases can also be large (e.g., Earl Washington claimed to have stabbed his victim two or three times, when in fact she was stabbed 38 times).

Procedure

Participants completed the study individually. Participants read one version of the information sheet describing the crime, followed by the signed confession statement. Participants provided a verdict (guilty or not guilty) and rated their confidence in their verdict on a scale from 0 to 100% with 10% increments. Participants also rated the probability that the defendant was actually guilty of committing the crime (also on a 0–100% scale with 10% increments). The latter measure was included because people sometimes render verdicts that do not match their actual opinion about guilt (e.g., Kassin & Sukel, 1997). It should be noted that no specific charge was mentioned in the instructions. The fact that no participants expressed surprise or asked for clarification on this matter suggests that this was not problematic.

Participants handed their stimulus materials and responses to the experimenter before completing a final questionnaire which asked them to describe any inconsistencies in the evidence they read earlier, to rate the extent to which the confession was consistent with the facts of the crime (on a scale from 1 – not at all consistent to 7 – very consistent), and report whether they believed the confession was given voluntarily (yes/no).

Verdict measures

We conducted analyses using three measures of verdict: (1) dichotomous verdict judgements; (2) a verdict preference score calculated by combining verdict and verdict confidence ratings (see Tenney, MacCoun, Spellman, & Hastie, 2007); and (3) probability of guilt judgements. These analyses produced virtually identical results. For simplicity, we only present results of the verdict data here. Details of the analyses of additional verdict measures in 'EXPERIMENT 1' and 2, and the method for calculating verdict preference scores, are available in the Supporting Material online (see Section B and Table S1).

Results and discussion

  1. Top of page
  2. Abstract
  3. Background
  4. EXPERIMENT 1
  5. Method
  6. Results and discussion
  7. EXPERIMENT 2
  8. Method
  9. Results and discussion
  10. GENERAL DISCUSSION
  11. Acknowledgements
  12. References
  13. Supporting Information

An alpha level of .05 was used. Effect sizes were estimated using Cohen's d (for differences between means) and Cohen's w (for differences between proportions). Small, medium, and large effects – respectively – are suggested by cut-offs of 0.2, 0.5, and 0.8 for d and .1, .3, and .5 for w.

Manipulation checks

Both manipulations had the desired effect. A lower proportion of participants reported that the confession was given voluntarily in the coercion condition (.35) than the no coercion condition (.65), χ2(1, = 129) = 15.50, < .001, = .34. The number of inconsistencies recalled was greater in the inconsistent condition (= 1.79, SD = 1.00) than the consistent condition (= 0.05, SD = 0.21), t(67.18) = −13.56, < .001, d = 2.43. Likewise, ratings of the degree of consistency between the confession and the facts of the crime were lower for participants in the inconsistent condition (= 2.72, SD = 1.77) than the consistent condition (= 5.59, SD = 1.30), t(109.58) = 10.38, < .001, d = 1.86. (Two participants did not provide a rating of perceived consistency.)

Verdict data

The consistency manipulation affected participants' verdicts, but the pattern of results differed from expectations (see Table 1). A 2 (Consistency) × 2 (Coercion) × 2 (Verdict) hierarchical log-linear analysis yielded a significant Coercion × Consistency interaction on verdicts, χ2(1, = 129) = 4.38, = .036. Follow-up chi-square tests indicated that, when coercion was present, the proportion of guilty verdicts was higher in the consistent condition that the inconsistent condition, χ2(1, = 64) = 5.16, = .023. Likewise, with no coercion present, there was a higher proportion of guilty verdicts in the consistent condition that the inconsistent condition, χ2(1, = 65) = 22.30, < .001. But, contrary to our hypothesis, the effect of consistency on verdicts was – if anything – greater when there was no coercion (= .59) than coercion (= .28; Table 1 shows 95% CIs around these differences in proportions).

Table 1. The effect of inconsistencies in confession evidence on the proportion of ‘guilty’ verdicts [with 95% CIs] rendered by mock jurors in 'EXPERIMENT 1' and 2
 ConsistentInconsistentDifference in proportions
'EXPERIMENT 1'
No coercion.91 [.76, .97].34 [.20, .52].57 [.34, .72]
30 of 3311 of 32 
Coercion.61 [.44, .75].32 [.19, .50].28 [.04, .48]
20 of 3310 of 31 
'EXPERIMENT 2'
No motive.96 [.82, .99].40 [.25, .58].56 [.34, .72]
27 of 2812 of 30 
Motive.80 [.63, .91].18 [.08, .36].62 [.37, .77]
24 of 305 of 28 

These results do not support the notion that jurors will discount inconsistencies in a confession unless there is a salient explanation for them. When an explanation (coercion) was made salient, the effect of inconsistencies on verdicts was actually smaller than when no explanation was made salient. This unexpected pattern of results may indicate a floor effect, whereby the effect of inconsistencies is constrained if a proportion of participants convict regardless of how problematic a confession is. From a broader perspective, the fact that inconsistencies significantly reduced the proportion of guilty verdicts in both coercion conditions runs contrary to the idea that jurors will render a guilty verdict despite the presence of inconsistencies in a confession. These data provide evidence that, in at least some circumstances, jurors' verdicts are influenced by inconsistencies between a confession and the facts of the crime.

We further explored the effect of confession inconsistencies on participants' verdicts by testing whether the effect was mediated by the number of inconsistencies detected by participants (using the INDIRECT macro for SPSS; Preacher & Hayes, 2008). This would indicate that the relationship operates in a graded manner, as opposed to an all-or-none manner whereby the magnitude of the effect does not vary with the number of inconsistencies detected.

The effect of the consistency manipulation on verdicts was fully mediated by the number of inconsistencies detected by participants. The consistency manipulation significantly affected the number of inconsistencies detected, with more inconsistencies detected in the inconsistent condition than the consistent condition, = 1.75, SE = .13, = 13.85, < .001. In turn, the number of inconsistencies detected was related to participants' verdicts, such that more detected inconsistencies were associated with fewer guilty verdicts, = −1.06, SE = .31, = −3.38, < .001. Bootstrap confidence intervals (5,000 samples) indicated a significant indirect effect of the consistency manipulation on verdicts via the number of inconsistencies detected, ab = −1.86, 95% CI [−3.26, −0.83], SE = .63. When the mediator was controlled for, the effect of the consistency manipulation on verdicts became trivial, = −0.13, SE = .63, = .20, = .839. These results suggest that the effect of inconsistencies on verdicts operates in a graded manner.

EXPERIMENT 2

  1. Top of page
  2. Abstract
  3. Background
  4. EXPERIMENT 1
  5. Method
  6. Results and discussion
  7. EXPERIMENT 2
  8. Method
  9. Results and discussion
  10. GENERAL DISCUSSION
  11. Acknowledgements
  12. References
  13. Supporting Information

'EXPERIMENT 2' was a replication of 'EXPERIMENT 1'. Again, the aim was to test whether inconsistencies in confession evidence affect jurors' verdicts, and whether such effects vary depending on the presence of a salient explanation for the confession. Here, a different type of explanation was used. Specifically, we manipulated whether the defendant had an ulterior motive for confessing to the crime: a desire to take the blame to protect another suspect in the investigation. Prior research suggests that this type of motivation accounts for approximately half of self-reported false confessions made during police interviews (Gudjonsson & Sigurdsson, 1994).

Method

  1. Top of page
  2. Abstract
  3. Background
  4. EXPERIMENT 1
  5. Method
  6. Results and discussion
  7. EXPERIMENT 2
  8. Method
  9. Results and discussion
  10. GENERAL DISCUSSION
  11. Acknowledgements
  12. References
  13. Supporting Information

Participants and design

Participants were 116 undergraduate students (72 female; aged 18–58 years, = 24.68 years, SD = 9.15 years) who had not participated in 'EXPERIMENT 1'. Participants received either course credit or $10 and were randomly allocated to one cell of a 2 (Motive presence) × 2 (Consistency) between groups design.

Materials and procedure

The stimulus materials and measures were identical to those used in 'EXPERIMENT 1', save for the differences relating to the manipulation of motive.

Experimental manipulations

The manipulation of inconsistencies between the confession and facts of the crime was identical to 'EXPERIMENT 1'. Motive was manipulated via a statement in the information sheet that described the police investigation. In the motive condition, this statement read: ‘Police initially focused their investigation upon Mr. Lucas' 17-year-old son as the prime suspect in this case, but this line of investigation was closed when Mr. Lucas came forward and provided the confession'. In the no motive condition, the statement read: ‘Police initially focused their investigation upon one other individual as a potential suspect, but this line of investigation was closed when Mr. Lucas came forward and provided the confession'. The effectiveness of this manipulation was assessed by asking participants whether they thought the defendant had an ulterior motive for providing the confession (after all other dependent measures were collected).

Results and discussion

  1. Top of page
  2. Abstract
  3. Background
  4. EXPERIMENT 1
  5. Method
  6. Results and discussion
  7. EXPERIMENT 2
  8. Method
  9. Results and discussion
  10. GENERAL DISCUSSION
  11. Acknowledgements
  12. References
  13. Supporting Information

Manipulation checks

The experimental manipulations were successful. The confession evidence was rated as more consistent with the facts of the crime in the consistent condition (= 5.72, SD = 1.14) than the inconsistent condition (= 2.31, SD = 1.34), t(114) = 14.80, < .001, = 2.74. The proportion of participants who reported that the defendant has an ulterior motive for confessing was higher in the motive condition (.76) than the no motive condition (.42), χ2(1, = 115) = 13.56 < .001, = .34. (One participant who did not provide an answer to the relevant question was omitted from this analysis.)

Verdict data

A 2 (Consistency) × 2 (Motive) × 2 (Verdict) hierarchical log-linear analysis produced a significant main effect of consistency, with the proportion of guilty verdicts higher in the consistent condition (.88) than the inconsistent condition (.29), χ2(1, = 116) = 48.04, < .001, = .59. Contrary to our hypothesis, the Motive × Verdict interaction was non-significant, χ2 < 1. As shown in Table 1, the strength of the effect of inconsistencies on verdicts was very similar in the motive condition, χ2(1, = 58) = 22.37, < .001, = .62, and the no motive condition, χ2(1, = 58) = 20.94, < .001, = .60.

These results converge with those of 'EXPERIMENT 1' to indicate that inconsistencies between confession evidence and the facts of the crime affect jurors' verdicts. Contrary to our expectations, they offer no support for the idea that jurors will discount inconsistent confession evidence only when provided with a salient alternative explanation for the presence of the inconsistencies (i.e., the defendant was innocent and had an ulterior motive for confessing).

Additional exploratory analyses

Prompted by these unanticipated results, we considered an alternative hypothesis: that inconsistency in confession evidence might prompt participants to consider alternative explanations for the confession (other than guilt) regardless of whether a plausible alternative reason was made salient (i.e., regardless of whether the stimulus materials suggested an ulterior motive). If this were the case, then the data should show an effect of the consistency manipulation on participants' perceptions that the defendant had an ulterior motive to confess (i.e., participant's responses to the motive manipulation check), and – importantly – this effect should not vary depending on whether an ulterior motive was provided. Consistent with this notion, a hierarchical log-linear analysis showed that participants were more likely to report that the defendant had an ulterior motive for confessing when the evidence was inconsistent (.72) rather than consistent (.46), χ2(1, = 115) = 11.17, = .001, = .27. This effect was not moderated by the motive manipulation, χ2 < 1.

To further explore this account, we tested whether the effect of inconsistency on verdicts were mediated by perceptions that the defendant had an ulterior motive to confess. For this analysis, we used versions of Baron and Kenny's (1986) method and the Sobel test that have been modified for use with categorical mediator and outcome variables (Mackinnon & Dwyer, 1993). Logistic regression analysis confirmed that the perceived presence of an ulterior motive was associated with fewer guilty verdicts, = −1.55, SE = .53, = .003, and that consistency accounted for a significant proportion of variance in verdict, Nagelkerke R2 = .43, = −2.87, SE = .50, < .001. When the perceived presence of an ulterior motive was controlled for, consistency remained a significant predictor of verdict but accounted for less variance, Nagelkerke R2 = .31, = −2.74, SE = .52, < .001. A Sobel test showed that the indirect effect of consistency on verdict via perceived presence of an ulterior motive was significant, ab = −2.06, SE = .86, = .039. Thus, the effect of the consistency manipulation on verdicts was partially mediated by the perceived presence of an ulterior motive for confessing. This suggests that confession inconsistencies prompted some participants to generate and consider alternative explanations for why the defendant confessed (other than being guilty of the crime), even when such explanations were not made salient to them in information about the case. In turn, the perceived presence of an ulterior motive led to the confession being discounted when participants make their verdict judgement.

GENERAL DISCUSSION

  1. Top of page
  2. Abstract
  3. Background
  4. EXPERIMENT 1
  5. Method
  6. Results and discussion
  7. EXPERIMENT 2
  8. Method
  9. Results and discussion
  10. GENERAL DISCUSSION
  11. Acknowledgements
  12. References
  13. Supporting Information

Contrary to the results of previous studies (Malloy & Lamb, 2010; Najdowski et al., 2009; Redlich et al., 2008) this research demonstrates that inconsistencies in confession evidence can indeed undermine the credibility of that evidence. In two experiments, participants who read an inconsistent confession were far less likely to render a guilty verdict than those who read a consistent confession. Additional mediation analyses of the data from 'EXPERIMENT 1' indicated that the effect of inconsistencies on verdicts was fully mediated by the reported number of inconsistencies detected by participants, suggesting that the effect of inconsistencies on verdicts operates in a graded (rather than all-or-none) manner.

One unexpected finding was that inconsistencies did not undermine the credibility of confession evidence only when a plausible alternative explanation for why the defendant confessed (other than guilt) was made salient. Instead, the data from 'EXPERIMENT 2' suggest that inconsistencies prompted participants to generate and consider their own alternative explanations for why the confession was rendered. In turn, the perceived presence of a plausible alternative explanation led to fewer guilty verdicts being rendered. This result points to a potential explanation for the discrepancy between our results (which show that inconsistencies can undermine confession evidence) and previous research (which suggests that they typically do not do so). Specifically, it could be that certain features of a case (e.g., the nature of the crime; individual characteristics of the defendant) make it more or less likely that jurors will generate plausible reasons – other than guilt – for why a confession was given. In turn, the extent to which such reasons can be generated may dictate the extent to which inconsistencies undermine the credibility of confession evidence. This explanation warrants further testing.

Although our results do not align with those of previous studies of inconsistencies in confessions, three caveats must be emphasized. First, there may be conditions under which the pattern of results hypothesized in this research does emerge. For example, one anonymous reviewer suggested that some participants in the no motive condition of 'EXPERIMENT 2' may have assumed that the other suspect being investigated was known to the defendant (no details of the identity of the other suspect were provided to participants), and may have concluded that the defendant had confessed to protect someone they knew, as in the motive condition. This possibility warrants further investigation as it could potentially explain the unexpected effect of inconsistencies on verdicts in the no motive condition.

Second, our research examined the effects of one particular type of inconsistency: between information in the confession and verifiable facts of the crime. Prior research on eyewitness evidence has shown that the effects of inconsistency on credibility can vary depending on the type of inconsistency involved. For example, credibility is undermined to a greater extent by contradictions within a statement of evidence given in court than by the provision of new information in court that was absent from an earlier statement of evidence given to police (Berman & Cutler, 1996). Thus, the effects of inconsistencies on juror judgements may vary for different types of inconsistencies, such as recantations (Najdowski et al., 2009; Redlich et al., 2008).

Third, our experiments used simple trial stimuli. Although this helps isolate the effects of inconsistencies from the effects of other incidental variables, a more complicated relationship between confession evidence and verdicts may emerge with more realistic stimuli. For instance, because the presence of a confession can bias evaluations of other types of evidence (e.g., eyewitness testimony and the results of forensic science tests; Hasel & Kassin, 2009; Kassin et al., 2012), the consistency of confession evidence may moderate the effects of other types of evidence on juror judgements.

Despite these caveats, this research provides initial, experimental evidence that inconsistencies can undermine the credibility of confession evidence, and they do so in part by prompting jurors to consider alternative explanations (other than guilt) for why a defendant confessed.

References

  1. Top of page
  2. Abstract
  3. Background
  4. EXPERIMENT 1
  5. Method
  6. Results and discussion
  7. EXPERIMENT 2
  8. Method
  9. Results and discussion
  10. GENERAL DISCUSSION
  11. Acknowledgements
  12. References
  13. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Background
  4. EXPERIMENT 1
  5. Method
  6. Results and discussion
  7. EXPERIMENT 2
  8. Method
  9. Results and discussion
  10. GENERAL DISCUSSION
  11. Acknowledgements
  12. References
  13. Supporting Information
FilenameFormatSizeDescription
lcrp12048-sup-0001-SuppInfo.docWord document494K

Data S1. Stimulus materials.

Data S2. Analyses of additional verdict measures.

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.