Citation
Evidence Interaction

Material Information

Title:
Evidence Interaction A More Complete Story of Evidence Contamination in Criminal Trial
Creator:
Kienzle, Megan R
Place of Publication:
[Gainesville, Fla.]
Florida
Publisher:
University of Florida
Publication Date:
Language:
english
Physical Description:
1 online resource (141 p.)

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Criminology, Law, and Society
Sociology and Criminology & Law
Committee Chair:
LEVETT,LORA M
Committee Co-Chair:
LANZA KADUCE,LONN M
Committee Members:
HASEL,LISA E
WEST,ROBIN L
Graduation Date:
8/9/2014

Subjects

Subjects / Keywords:
Confidence interval ( jstor )
Defendants ( jstor )
Experimentation ( jstor )
Guilty verdicts ( jstor )
Interrogations ( jstor )
Jurors ( jstor )
Legal evidence ( jstor )
Police ( jstor )
Trials ( jstor )
Verdicts ( jstor )
Sociology and Criminology & Law -- Dissertations, Academic -- UF
decision-making -- evidence -- juror
City of Gainesville ( local )
Genre:
bibliography ( marcgt )
theses ( marcgt )
government publication (state, provincial, terriorial, dependent) ( marcgt )
born-digital ( sobekcm )
Electronic Thesis or Dissertation
Criminology, Law, and Society thesis, Ph.D.

Notes

Abstract:
Recent research has shown that when confession evidence is presented to investigators, witnesses, or other legal decision makers, the knowledge of that evidence can corrupt and alter the decisions they make about other evidence. However, generally people assume each piece of evidence in a trial to be independent of the others; thus it is not known whether jurors are able to identify this potential for contamination in the evidence presented during trial. The current study examined how jurors weigh evidence that was potentially contaminated by other evidence. Community members read a trial stimulus containing two pieces of evidence. I varied the quality of both pieces (high v. low quality confession evidence, and high v. low quality eyewitness evidence) and counterbalanced the order of the presentation of the evidence. Finally, I varied the potential for contamination (present v. absent) in some conditions by having the attorney in the stimulus point out that the eyewitness likely overheard about the confession from the suspect. I additionally included 2 control conditions with no eyewitness evidence presented to have comparisons for the high and low quality confession conditions. Results indicated that jurors were able to identify differences in confession evidence quality levels and change their verdict and probability of guilt accordingly. Participants were able to do the same for eyewitness quality levels, though not consistently. Finally, though contamination was noticed by participants in the manipulation check questions, it was not a significant factor in their verdict, probability of guilt decisions, or their scaled responses. Therefore, while jurors are able to appropriately weigh evidence of varying quality, they are not able to identify problem with the evidence such as contamination. Given this, investigators need to be particularly cautious to keep all evidence separate throughout the investigation to ensure the evidence is properly weighed at trial. ( en )
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis:
Thesis (Ph.D.)--University of Florida, 2014.
Local:
Adviser: LEVETT,LORA M.
Local:
Co-adviser: LANZA KADUCE,LONN M.
Electronic Access:
RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2016-08-31
Statement of Responsibility:
by Megan R Kienzle.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Embargo Date:
8/31/2016
Classification:
LD1780 2014 ( lcc )

Downloads

This item has the following downloads:


Full Text

PAGE 1

EVIDENCE INTERACTION: A MORE COMPLETE STORY OF EVIDENCE CONTAMINATION IN CRIMINAL TRIAL By MEGAN R. KIENZLE A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIR EMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2014

PAGE 2

© 2014 Megan R. Kienzle

PAGE 3

To my parents, Carol and Doug, and my sister Emily

PAGE 4

4 ACKNOWLEDGMENTS First , I have to thank Dr. Lora Lev ett for all the times she has kept me going over the last 6 years . I cannot say enough how much I appreciate her kind words and always constructive criticism especially her complement cookie. Further, I appreciate all the input and advice Dr. Lonn Lanz a Kaduce, Dr. Lisa Hasel, and Dr. Robin West have provided while working though each stage of this research project. Their comments and feedback made the project what it is, and I am grateful for their support in graduate school as well. Further, I am ver y thankful of the American Psychology in Aid which helped to fund this project. After receiving this award for my thesis and now otherwise would remain uncompleted. I am proud to be a member of such an amazing organization. Last, I want to thank my family for always supporting me and helping get me to where I am today. They are truly great people and I love them all dearly and a special thanks to my dog, Nessie for spending the numerous hours working on this project lying by my side for support !

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGURES ................................ ................................ ................................ .......... 9 ABSTRACT ................................ ................................ ................................ ................... 10 CHAPTER 1 LITERATURE REVIEW ................................ ................................ .......................... 12 Introduction ................................ ................................ ................................ ............. 12 Evidence Contamination Research ................................ ................................ ......... 18 ................................ ............. 18 Contamination within Evidence Type ................................ ............................... 20 Contamination of Physical Evidence ................................ ................................ 22 Why Does Evidence Contamination O ccur? ................................ ........................... 24 Evidence Contamination in the Courtroom ................................ ............................. 27 Evidence Q uality. ................................ ................................ ............................. 29 We ighing Complex E vidence. ................................ ................................ ........... 31 2 THE CURRENT STUDY ................................ ................................ ......................... 32 3 METHOD ................................ ................................ ................................ ................ 38 Part icipants ................................ ................................ ................................ ............. 38 Design ................................ ................................ ................................ ..................... 39 Independent Variables ................................ ................................ ..................... 39 Quality of Evidence ................................ ................................ .................... 39 Evidence Contamination ................................ ................................ ............ 46 Order of Evidence ................................ ................................ ...................... 47 Control Condition s ................................ ................................ ..................... 48 Dependent Measures ................................ ................................ ....................... 49 Verdict, Confidence Ratings, and Explanation of V erdict. .......................... 49 Evaluations of E vidence. ................................ ................................ ............ 49 Procedure ................................ ................................ ................................ ......... 52 4 PILOT TESTING ................................ ................................ ................................ ..... 54 5 RESULTS ................................ ................................ ................................ ............... 58 Data Cleaning ................................ ................................ ................................ ......... 58 Manipulation C hecks ................................ ................................ ............................... 60

PAGE 6

6 S caling Dependent M easures ................................ ................................ ................. 63 Control Group Comparisons ................................ ................................ ................... 64 Does Eyewitness Quality, Confession Quality and Contamination of Evidenc e ................................ ................................ 73 Does the Confession Quality, Eyewitness Quality, or Contamination P resence vidence? ................................ ... 7 9 Testing for Covariates. ................................ ................................ ............................ 88 6 DISCUSSION ................................ ................................ ................................ ......... 90 Quality of Confession Evidence ................................ ................................ .............. 94 Quality of Eyewitness Evidence ................................ ................................ .............. 96 Contamination of Evidence ................................ ................................ ..................... 99 Recomm endations for Future Research ................................ ............................... 101 Policy Recommendations ................................ ................................ ..................... 106 Limitations ................................ ................................ ................................ ............. 108 Conclusions ................................ ................................ ................................ .......... 110 APPENDIX A INFORMED CONSENT ................................ ................................ ........................ 112 B VOIR DIRE QUESTIONNAIRE ................................ ................................ ............. 113 C MASTER TRIAL STIMULUS ................................ ................................ ................. 115 D POST TRIAL QUESTIONNAIRE FOR PILOT STUDY ................................ ......... 126 E POST TRIAL QUESTIONNAIRE FOR THE CURRENT STUDY .......................... 128 F PILOT STUDY DEBRIEFING ................................ ................................ ............... 134 G CURRENT STUDY DEBRIEFING ................................ ................................ ........ 135 LIST OF REFERENCES ................................ ................................ ............................. 136 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 140

PAGE 7

7 LIST OF TABLES Table page 5 1 Number of Participants per Condition in the 692 Person Sample. ...................... 61 5 2 Number of Participants per Condition in the 438 Person Sample. ...................... 63 5 3 Means and Confidence Intervals for Planned Comparisons to High Quality Confession Control Group on Probability of Guilt Measure. ............................... 66 5 4 Means and Confidence Intervals for Planned Comparisons to Low Quality Confession Control Group on Probability of Guilt Measure. ............................... 68 5 5 Means and Confidence Intervals for Planned Comparisons to High Quality Confession Control Group on Co ntinuous Verdict Measure. .............................. 69 5 6 Means and Confidence Intervals for Planned Comparisons to Low Quality Confession Control Group on Continuous Verdict Measure. .............................. 70 5 7 Means and Confidence Intervals for Planned Comparisons to High and Low Quality Confession Control Groups on Confession Strength Scale. ................... 71 5 8 Means and Confi dence Intervals for Planned Comparisons to High and Low Quality Confession Control Groups on General Weighing of Confession Evidence During Trial Scale. ................................ ................................ .............. 72 5 9 Logistic regression examining evi dence quality and contamination on dichotomous verdict selection (Guilty/Not Guilty) with n=692 ............................. 74 5 10 Logistic regression examining evidence quality and contamination on dichotomous verdict s election (Guilty/Not Guilty) with n=438 ............................. 74 5 11 intervals) of confession strength, general weighing of confession ev idence during trial, eyewitness strength, and specific weighing of eyewitness evidence in this case with n=692 . ................................ ................................ ....... 80 5 12 nce intervals) of confession strength, general weighing of confession evidence during trial, eyewitness strength, and specific weighing of eyewitness evidence in this case with n=692 . ................................ ................................ ....... 81 5 13 M intervals) of confession strength, general weighing of confession evidence during trial, eyewitness strength, and specific weighing of eyewitness evidence in this case with n=438 . ................................ ................................ ....... 82

PAGE 8

8 5 14 intervals) of confession strength, general weighing of confession evidence during trial, eyewitness strength , and specific weighing of eyewitness evidence in this case. with n=438 ................................ ................................ ........ 83 5 15 Mean Values (Confidence Intervals) for Main Effect of Contamination Presence, and Interaction Effects between 1) C ontamination Presence and Quality of Confession Evidence, and 2) Contamination Presence, Quality of Decisions ................................ ................................ ................................ ............ 85 5 16 Mean Val ues and Confidence Intervals for Main Effect of Contamination Presence, Two way Interaction between Contamination Presence and Quality of Confession Evidence, and Three way Interaction between Contamination Presence, Quality of Confession Evidence, and Qua lity of Eyewitness Evidence on Scaled Dependent Measures. ................................ ..... 86

PAGE 9

9 LIST OF FIGURES Figure page 2 1 Graphical representation of hypotheses for A) Contamina tion Absent, and B) Contamination Present conditions. ................................ ................................ ..... 36

PAGE 10

10 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy EVIDENCE INTERACTION: A MORE COMPLETE STORY OF EVIDENCE CONTAMINATION IN CRIMINAL TRIAL By Megan R. Kienzle August 2014 Chair: Lora Levett Major: Criminology, Law and Society Recent research has shown that when confession evidence is presented to investigators, witnesses, or other legal decision makers, the knowledge of that evidence can corrupt and alter the decisions they make about other evidence . However, generally people assume each piece of evidence in a trial to be independ ent of the others; thus it is not known whether jurors are able to identify this potential for contamination in the evidence presented during trial. The current study examined how jurors weigh evidence that was potentially contaminated by other evidence. C ommunity members read a trial stimulus containing two pieces of evidence. I varied the quality of both pieces (high v. low quality confession evidence, and high v. low quality eyewitness evidence) and counterbalanced the order of the presentation of the e vidence. Finally, I varied the potential for contamination (present v. absent) in some conditions by having the attorney in the stimulus point out that the eyewitness likely overheard about the confession from the suspect. I additionally included 2 contr ol conditions with no eyewitness evidence presented to have comparisons for the high and low quality confession conditions.

PAGE 11

11 Results indicated that jurors were able to identify differences in confession evidence quality levels and change their verdict and p robability of guilt accordingly. Participants were able to do the same for eyewitness quality levels, though not consistently. Finally, though contamination was noticed by participants in the manipulation check questions, it was not a significant factor in their verdict, probability of guilt decisions, or their scaled responses. Therefore, while jurors are able to appropriately weigh evidence of varying quality, they are not able to identify problem with the evidence such as contamination. Given this, i nvestigators need to be particularly cautious to keep all evidence separate throughout the investigation to ensure the evidence is properly weighed at trial.

PAGE 12

12 CHAPTER 1 LITERATURE REVIEW Introduction Evidence contamination is defined as how one piece of evidence can influence or change the collection or interpretation of another piece of evidence and has received increased interest from those studying psychology and law in recent years. If two pieces of evidence interact in this manner, one piece of evid ence can influence the way that the second piece (or all subsequent pieces) are collected. We know that confessions can influence other pieces of evidence such as eyewitnesses, forensic decisions, and alibi corroborators (discussed in detail bel ow, see Hasel & Kassin, 2009; Dror & Charlton, 2006; Kienzle & Levett, manuscript under review, for some examples). Further we know other types of evidence can serve as this contamination starting point as well, such as an eyewitness influencing another e yewitness (Levett, 2012). Given that evidence contamination occurs, it is necessary to understand how triers of fact perceive contaminated evidence. Therefore , this project tested how jurors weigh evidence that was likely influenced by another piece of ev idence (compared to independent evidence), while also accounting for a second the question of whether jurors are able to identify and correctly weigh potentially conta minated pieces of evidence. One of the major problems with potentially contaminated evidence is that this type of evidence is complex to consider and laypeople are generally not able to correctly weigh complex pieces of evidence (Horowitz et. al, 1996; He uer & Penrod, 1994). Whether evidence is contaminated could be considered an indicator of evidence

PAGE 13

13 quality. Generally, research shows that jurors may not take into consideration whether the evidence presented to them was of higher quality or lower quality in assigning weight to evidence ( Kovera & McAuliff, 2000 ). Thus, it is possible that jurors may perceive high quality, contaminated evidence similar to high quality, uncontaminated evidence. Further, both of these may be perceived as the same as low qual ity evidence. In this study, I assessed whether jurors could identify potentially selections, and if the quality of this type of evidence interacted with these effects. Un fortunately, given that the evidence in a trial is legally presumed to be independent, researchers, investigators, and jurors/judges alike have generally treated any piece of evidence in a criminal trial as a single, independent factor in the case against the defendant, though this is likely inaccurate (Hasel, 2012). Most cases of wrongful conviction have multiple pieces of incriminating evidence in the case; in over 50% of the exonerations reported by the Innocence Project, more than one factor is listed as contributing to that conviction ( Kassin, 20 12; www. innocenceproject.org). Similarly, a recent archival analysis of several Innocence Project case files revealed that 55% of case files had listed multiple causes of the wrongful conviction. This study sh owed that the presence of multiple factors leading to an erroneous conviction is a potential problem and that errors are likely made with more than one piece of evidence in many cases (Kassin, Bogart, & Kerner, 2012). As a case builds, it is possible that the evidence, and more likely indicating guilt over innocence. This belief may influence how subsequent pieces of evidence are obtained, interpreted, and made par t of the criminal

PAGE 14

14 investigation and trial proceedings. When this occurs, pieces of evidence in a case are not independent. Thus, it is possible that some form of evidence contamination may have occurred in a non negligible portion of cases of wrongful con viction. Further, in these cases, the jury rendered an incorrect decision based on that evidence. That is, the jury was likely unable to distinguish between accurate and inaccurate evidence. It is also possible that the jury may not have noticed the eviden ce contamination in these cases, and thus rendered a verdict assuming the evidence was independently obtained and sound. One may think that this could be corrected on appeal; however the initial cases have been tried while assuming the evidence is indepen dent. Harmless error analysis is completed by appeals court judges to determine if the confession was allowed in to the original trial erroneously (Sweeny, 1995). If they determine that the confession should have been deemed inadmissible, then the judges need to decide if this error was harmful to the trial outcome meaning that the verdict would have been different had the confession not been present in the original trial. If so, it is considered a prejudicial error and the case would likely be remande d back to the lower court to be retried. If not, it is considered a harmless error; the verdict would have been the same anyways. The re are a few ma jor problem s with this analysis; first is that this type of analysis considers pieces of evidence to be ind ependent. Here, judges may have believed that he or she could simply mentally delete the influence of the confession and determine how the case would have turned out had the problematic information not been present in the original trial.

PAGE 15

15 Research sugges ts this assumption is likely erroneous. A study by Kassin and Sukel (1997) showed that jurors who read trial transcripts that contained an inadmissible confession were more likely to render guilty verdicts than jurors who read identical trial transcripts w ithout mention of the confession. This occurred even if the confession was explicitly identified as being elicited under coercion and even if the juror reported they would not consider it. In a second study, Wallace and Kassin (2012) sent trial stimuli t o several judges to determine verdict, some with confession evidence that was elicited under high or low pressure. While the results were not as clear cut as the study examining jurors, the researchers found that if judges were presented with a high press ure confession and little other corroborating evidence, 38% of judges still maintained that the suspect was guilty, even if the judges agreed that the confession was legally coerced. Thus, it seems like both jurors and judges have trouble properly weighin g confession evidence. Therefore appellate court judges may think they can determine what the outcome of a trial would have been by mentally removing a piece of evidence, but the research shows this is likely not the case. Second , even if we assume that ju dges can mentally discount a piece of evidence, mentally removing the problematic piece of evidence does not account for how that evidence may have influenced other pieces of evidence in the case. Removing the single piece of evidence mentally (and after t he fact) does not adequately reduce the weight given to pieces of evidence that were collected with knowledge of the confession (and thus, possibly contaminated). That is, removing the problematic evidence does not account for the idea that the other piece s of evidence were somewhat dependent on that piece of problematic evidence.

PAGE 16

16 For example, a case could be made up of a confession elicited with coercion, an eyewitness, and some kind of forensic analysis. On appeal, the confession was determined to be a n error in the original trial, but a harmless one because the judge noticed that there was also an eyewitness and forensics present. Thus, given the other, strong evidence in the case, the judge believed the jury would have voted guilty. The problem with evidence inter dependence is that the eyewitness could have been unintentionally influenced to choose a particular person out of the lineup because the investigators already had a suspect in mind after his confession. Or, the eyewitness could have been told that the suspect confessed which could have increased the likelihood the eyewitness would choose the suspect from the lineup and the investigators may have made inadvertent commen ts when requesting the forensic testing that indicated to the analyst who the suspect is or whether the goal of the propensity to find a match (Thompson, 1997). None of thi s would be taken into account when an appellate judge determined the confession was a harmless error, so we cannot safely assume that the potential for contamination will be corrected after the original case is over through the appellate process. Last, the harmless error analysis is based on the assumption of a legal issue with the evidence, not a psychological one. The analysis would be completed if the confession evidence is determined to have been legally coerced which is a higher level, more stringent definition of coercion compared to the psychological understanding of coercion in confessions , likely involving extreme and harsh measures to elicit the

PAGE 17

17 confession. In the psychological literature, the tactics and techniques used to elicit a confession ca n be termed coercive earlier than they would be considered legally coercive. These tactics may involve measures that involve more nuanced kinds of influence that appear to the legal world as non coercive but are shown in psychological research to be coerci ve . For example, using minimization or maximization techniques are not considered legally coercive, but according to psychological research investigating the effects of using those techniques in eliciting a false confession, they would be considered psych ologically coercive. Therefore, there may be many more cases that constitute a coerced confession according to the psychological research (and are thus susceptible to the host of issues discussed above) than would be termed legally coerced. These confessi ons would not rise to the legal attention of the court for the chance to complete a harmless error analysis, which further demonstrates that this is not a protection against problematic evidence . Most likely, this is where the bulk of the problematic case s such as those discussed in this paper would fall . A great deal of the evidence presented at trial has some amount of human decision making involved in it. As such, the evidence will be subjected to many knowledge of other evidence) that could evidence (Kassin, 2012). As the case builds, every new piece of evidence has the potential to contaminate any subsequently collecte d piece of evidence; once information that followed in a different way framed by that preconceived opinion. If the person who provided, collected, or interpreted the secon d piece of evidence had

PAGE 18

18 prior to collecting the second piece of evidence, that knowledge may have affected how the person interpreted or presented the other pieces of e vidence (Hasel, 2012). Evidence Contamination Research Of the few studies that have experimentally tested the possibility of evidence contamination during an investigation, we have seen that person evidence (m eaning evidence used at trial that requires the testimony of a human being, e.g., eyewitness testimony and alibi corroboration) can be influenced by external information or knowledge about the investigation or evidence that the witness has acquired. The f irst of these studies examined if confession information, which is known to be highly 2009). Sixty percent of eyewitnesses who made an initial lineup selection from a target absent lineup and were later presented with information that another person from the lineup confessed changed their selection to be consistent with confession evidence. Further, 50% of eyewitnesses who initially made a correct choice by rejecting the lineup , who then heard another lineup member confessed, altered their decision away from the correct choice and toward the lineup member who had confessed. Thus, eyewitnesses were influenced by another piece of evidence (the confession). Another stud y looked at a similar effect, but instead of eyewitnesses, this study examined the possibility of evidence contamination in potential alibi corroborators (Kienzle & Levett, manuscript under review). These researchers tested whether the knowledge of a confe

PAGE 19

19 alibi; most alibi corroborators who heard the suspect confessed withdrew their initial corroboration. This effect was similar across alibi corroborator suspect relationship types; friends and strangers were influenced similarly by the presence of a confession. Therefore, evidence contamination has the potential to put more incriminating evidence in court through the presence of a non independent eyewitness. Further, evidence contamination can also keep exonerating evidence out of court through the withdrawal of a non inde pendent alibi corroborator. Given this, it seems as though many parts of confession during the investigation (Kassin, 2012). One thing these two studies have in common is th ey both further demonstrate the striking influence that a confession can have on a case, whether the confession is truthful or not (Kassin, Drizin, Grisso, Gudjonsson, Leo, & Redlich, 2010; Kassin, 2012). An archival analysis by Kassin and colleagues exam ined how confessions exert influence in the cases with defendants who have been exonerated with the assistance of the Innocence Project (Kassin, Bogart, & Kerner, 2012). These researchers found that cases with a confession present were more likely to have multiple evidence errors listed as contributing to the wrongful conviction than cases without confession evidence. Additionally, in those cases that contained multiple errors, the confession was most often the first piece of evidence collected (though it is possible that other evidence could also have contaminated a confession as well; Kassin, 2012). This led to the conclusions that 1) when multiple pieces of evidence were present in a case, there is a reasonable chance of evidence contamination if inves tigators were not careful to keep

PAGE 20

20 the evidence separate until trial, and 2) confessions could (and likely do) contaminate other pieces of evidence even in real world cases. Contamination within Evidence Type Evidence contamination literature has not confin ed itself to examining also how other types of evidence (like the presentation of false information or on subsequent evidence. For example, a common tactic used during a suspect interrogation is to present false evidence against that suspect to encourage the person to confess (Inbau, Reid, Buckley, & Jayne, 2 013 ). However, studies have shown that present ing false incriminating evidence led to increased rates of confessions and greater internalized feelings of guilt for the crime, even if the suspect did not actually commit the crime (Kassin & Kiechel, 1996; Perillo & Kassin, 2011). In presenting false in criminating evidence, it is likely that the investigator might have used a (potentially non existent) eyewitness or DNA evidence that places the suspect at the scene of the crime in order to elicit a confession. Framed this way, the increased rates of con fessions with this interrogation tactic of presenting knowledge of other evidence could demonstrate evidence contamination as well. Specifically, these studies show that some pieces of evidence (that may be false or at least weaker than implied) could pos sibly have original (correct) decision of not confessing, to the contaminated (incorrect) version of falsely confessing to a crime. Here, the confession is the piece of evidence contaminated by pieces of false or low quality evidence.

PAGE 21

21 C o witness influences could also be conceptualized as evidence contamination because an eye reported memory or identification decision has the potential to influence another witnes Gabbert, Healey, & Lenton, 2007; Gabbert, Memon, & Allan, 2003). Any knowledge that one witness had about the crime or identification could be transferred to the second eyewitness, thus conta Kemp, & Ng, 2011). For example, one study looked at the relationship between co witnesses and found that, all pairs who conversed with each other after witnessing the crime reported less accurate information than pairs who did not converse, showing that this influence occurs even if the co witnesses are not close friends (Hope, Ost, Gabbert, Healey, & Lenton, 2007). Another study had co witnesses watch different views of the same crime (each view with unique items). The researchers found that those co witnesses who were participan ts who did not discuss with the co witness after they watched the crime (Gabbert, Memon, & Allan, 2003). This becomes especially problematic when any information that is relayed between co witnesses is inaccurate because it becomes a part of the other witn Further, this contamination effect is not simply limited to self reported memory of the crime details, but it extends into the eyewitness identification made by the witnesses as well; eyewi tnesses who knew that another witness chose from a lineup were more likely to select someone from the lineup compared to those who did not hear such

PAGE 22

22 information, (Levett, 2012). Based on these studies, we can conclude that contaminating influence could occ ur within multiple versions of the same kind of other eyewitness. Contamination of Physical Evidence Additionally, studies have also shown these contamination effects within physical evidence (evidence that can be documented through physical means rather than relying footage). For example, some forensic science studies have looked at how know ledge of contextual information, like a confession or pictures from the crime scene, altered series of studies in which participants were asked to make match decisions ei ther with or without contextual information such as emotional crime scene photos or knowledge of a confession in the case (Dror, Peron, Hind, & Charlton, 2005; Dror, Charlton, & Peron, 2006; Dror & Charlton, 2006 ). One study used fingerprint experts as pa rticipants and presented those experts with fingerprint pairs that they had previously determined as matches by the experts themselves (in the previous 5 years); the earlier match/no sion. The researchers found that the experts were influenced by the contextual information and made their new match decisions in line with that information 16.6% of the time, meaning ing that he was in custody at the time of the crime, the experts were more likely to say the prints did were more likely to

PAGE 23

23 Even though the sample sizes were small with only 6 examiners reassessing 48 print match decisions, studies like this show that person evidence (e.g., eyewitnesses and alibi corroborators) is not the only type of evidence that can be contaminated; forensic evidence is also susceptible to evidence contamination. Further, polygraph examiners have also been shown to be subject to these types of conta minating external influences (such as knowledge of who the suspect is) when making their decision of whether a person is lying (Elaad, Ginton, & Ben Shakhar, 1994 panels, the researchers found that having a prior expectation about the outcome of the panel (for example, knowing the suspect eventually confessed) led participants to determine evidence of deception within the panel when the charts were somewhat ambiguous. Here, we see that physical evidence (like a fingerprint) as well as testimonial based evidence (like polygraph results) can be subjected to a change based on the influential knowledge of confession information. Contamination of forensic science is not quite as big of a problem as contamination of person evidence primarily because the original evidence is typically preserved and could be reexamined at a later time if needed. However, with person ry) from its initial state forever so there is no chance to retest it later. Because of this, it is crucial that ideally all evidence, but particularly person evidence, is recorded in its original, unaltered state. This likely means video recording the q uestioning of potential alibis or eyewitness identification procedures, a protection that has been suggested elsewhere

PAGE 24

24 (Kienzle & Levett, manuscript under review; Wells, Small, Penrod, Malpass, Fulero, & Brimacombe, 1998; Kassin, 1998). This also means tha t investigators have a great deal of pressure to get all information from the witness as soon as possible, in a non leading, non contaminating manner before any unavoidable contamination has occurred. Physical evidence should also be kept in a safe environ ment, and every effort should be made to keep any and all evidence separated until it is presented at trial. Whether evidence contamination occurred is something that jurors should consider when determining how much weight to give to a piece of evidence b ecause evidence that is contaminated should not hold as much weight as evidence that is independent of other collected evidence. Understanding why evidence contamination occurs in the first place may help us to better understand how jurors may react to the evidence. Why does evidence contamination occur? Confirmation biases may be to blame in cases with contamination of evidence . Confirmation biases occur when people interpret and seek new information in a way that matches their preexisting beliefs, regar dless of whether the original information is correct (Boring, 1930; Leeper, 1935). Thus, old beliefs or information biases the interpretation of later details in the case. That is, confirmation biases occur when individuals tend to listen only to informati on that is in line with their original idea, also showing the problem with how individuals in the legal world are evaluating information (Kassin, Dror & Kukucka, 2013; Charman, 2013). For example, as previously discussed, confessions have been most often shown to contaminate other pieces of evidence, or produce a contamination effect (though other types of evidence may also contaminate evidence, too). The contamination effect

PAGE 25

25 refers to the legal actor (be it an investigator, eyewitness, alibi corroborator, etc.) who a fingerprint match decision, selecting from a lineup, or serving a s an alibi corroborator) instead of their original, untainted memory of their interaction with the suspect or blind examination of the evidence (Dror & Charlton, 2006; Elaad, Ginton & Shakhar, 1994 ; Hasel & Kassin, 2009; Kienzle & Levett, manuscript under review). Though evidence contamination studies have not explicitly tested confirmation biases, it is likely that the process that produces the contamination effect is similar. That is, because people want to be correct in their decisions, especially as t hey relate to the legal system, it is likely that people allow the lens through which they view subsequent evidence to be contaminated by the beliefs created by kno wledge of prior evidence. Confi rmation biases are also an innate human bias, affecting all human decision making. The legal realm is no different as much of evidence gathering takes place in a social context, so it is likely that we search for information and interpret new information in a case to be consistent with our prior beliefs. More spec ific to the legal world, fingerprint analysts who had contextual information when making their match/not match decision (see the studies by Dror and colleagues) could be showing confirmation biases. These individuals had prior onfession which may have given them the pre existing belief which led them to make a decision in line with that confession when deciding if the prints matched. Knowing that the print was from the suspect might have created a guilt presumptive belief that t

PAGE 26

26 evaluating all the information that comes forward on its own merits, confirmation biases lead people to evaluate information in light of their prior knowledge. This is problematic because the initia l beliefs may be incorrect, thus biasing all of the subsequent information . In other words, hearing about some piece of incriminating evidence likely leads people to have preconceived notions that the suspect is guilty, resulting in a confirmation bias. They then encode and process all other information in a way that fits with that belief. Confirmation biases are likely the processes that 1) account for the information the legal actor is paying attention to, 2) explain how they are evaluating the informat ion, 3) determine if they seek out additional information to support their initial hypotheses, and 4) ultimately decide whether they believe the information or disregard it. If the subsequent information lines up with their original idea that the suspect i s guilty, then people will listen to and retain the new piece of information without question (Ask, Rebellius & Granhag, 2008 that demonstrates crime), that actor may be more critical of this information and interpret it in a biased way. Here the investigator will likely make excuses for why the DNA does not fit with Nickerson, 1998 ). This reliance on human decision making in our justice system leaves open the possibility that evidence in a case could be at least partially a function of the p

PAGE 27

27 Evidence Contamination in the Courtroom C onfirmat ion biases that explain why evidence contamination occurs may also explain why jurors may n ot be good at identifying potentially contaminated pieces of evidence at trial . For example, confirmation biases may be at work when jurors weigh alibi evidence differently in the context of a courtroom than they do in the context of an investigation (Somm ers and Douglass, 2007). In this case, it is possible that jurors give less weight to alibi evidence that is presented at trial because they believe the case would not have proceeded to trial if the alibi was strong. That is, they gain a preexisting belie f by looking at the context (that the case went to trial) and use it to interpret the strength they should give to the alibi evidence. Thus, they put more weight on alibi evidence in an investigation context because they do not have a preexisting belief a bout the guilt of the suspect, however once the evidence gets to the courtroom they interpret it in terms of their preexisting belief that the defendant is guilty (because the case made it to trial) and thus put less emphasis on the potentially exculpatory evidence of the alibi. Confirmation biases could also be at fault when jurors cannot identify contaminated evidence. It is possible that jurors latch onto the story presented by one of the attorneys, making it difficult to negate or subtract evidence fro m their guilt judgments. For example, a prosecutor may first present confession evidence to the jury. Once a confession has been presented, we know that jurors are likely influenced by this confession, even if a defense attorney tries to weaken that evid ence by showing that the confession was coerced or trying to show enough evidence of innocence to negate the confession ( Leo & Liu, 2009; Kassin & Sukel, 1997 ). Jurors have made a belief about that initial story and evidence (in this case, a confession) a nd may have trouble breaking away from that story even when given conflicting information; they

PAGE 28

28 have likely heard the information but decided to disregard it because of the confirmation bias to be more critical of information that conflicts with their init ial idea. Given that jurors are unable to appropriately judge coerced confessions, I believe they will be similarly unable to judge potentially contaminated evidence. So, do people perceive contaminated evidence similarly to non contaminated evidence? Fur ther, if they do not, are people aware that they do not weigh these two types of evidence differently? Jurors may not possess the skills necessary to identify potentially problematic pieces of evidence and know to discount that information when rendering their verdict. Even if jurors are able to spot contaminated evidence when it is presented at trial, it seems unlikely that jurors will adequately negate the influence of that information in their final verdict choice, much like they can identify coerced c onfessions, but still give guilty verdicts more often in cases with coerced confessions compared to cases with no confession evidence even though the jurors say they are not considering that evidence (Kassin & Sukel, 1997). I believe potentially contamina ted evidence may be similar. Much like with the confession evidence, it is possible that even when jurors are told about potentially contaminating influences on a piece of evidence (assuming they recognize these influences as problematic in the first plac e), they will be unsuccessful in negating the impact of that piece of evidence entirely when rendering their verdict. One study, to date, has examined specifically how mock jurors weigh potentially contaminated evidence (Hasel, Vogler, & Lusizzo, 2013). I n this study, participants watched a videotaped trial stimulus depicting either 1) one piece of independent evidence, 2) two pieces of independent evidence, or 3) two pieces of evidence (one of

PAGE 29

29 which was likely contaminated by the other). After the video, participants completed dependent measures assessing their verdict choice, confidence, and their perceptions of the reliability of evidence. Here, the researchers found that participants were able to identify potentially contaminated evidence; the guilt a nd confidence ratings were higher in cases with two independent pieces of evidence than only one piece or the two pieces with one piece contaminated. However, the study used criminology students, who likely are more knowledgeable than the community juror population about the effects of evidence contamination. So, the current study expanded on these findings by testing the hypotheses using community members and added the variation of examining the quality of the evidence . Evidence quality Another considera tion in how jurors may weigh potentially contaminated (or contaminating) evidence is the quality of that evidence. One hopes that jurors will be able to identify evidence that is collected under optimal (or dismal) conditions and assign weight accordingly . In fact, evidence that is high in quality that contaminates other evidence may not be as much of a concern at trial as evidence that is low in quality that contaminates other evidence. Ideally, all evidence should be independently collected and free from contamination. However, if that is not the case, then the quality of evidence may determine the depth of the contamination problem. That is, If jurors are considering the totality of the evidence and the initial evidence is of high quality, then it follow s that the secondary evidence should not matter as much as if the initial evidence is of low quality, even if that secondary evidence is contaminated. If the initial evidence is low quality, then the secondary evidence is likely more important in the tota lity of the evidence; if that evidence was contaminated by the initial, low quality evidence, this is

PAGE 30

30 more concerning than if the evidence was either not contaminated or if the secondary evidence was contaminated by high quality evidence. However, research suggests jurors may have trouble distinguishing between high quality and low quality evidence. We have seen that both jurors (laypeople) and judges (presumably more knowledgeable about evidence quality than laypeople) are unable to identify variations in the quality of expert testimony without assistance (Kovera & McAuliff, 2000; Kovera, McAuliff & Hebert, 1999; Levett & Kovera, 2008; Levett & Kovera, 2009). As described in detail next, jurors are not good at understanding complex scientific evidence rega rdless of how sound that evidence may be (Helgeson & Ursic, 1993). One type of evidence that is often misunderstood by jurors is expert testimony describing scientific studies or experiments (Bornstein, 2004; Cooper, Bennett and Sukel, 1996). If we loo k at the investigation process as an experiment and we know that jurors have trouble picking out problems and confounds within experiments (Kovera and McAuliff, 2000), we might expect to find that jurors will have similar issues picking out problems and co nfounds within the investigation process as well (such as contamination of evidence).Thus, I concluded that jurors will not be able to identify problematic evidence gathered during the investigation and presented at trial either. Therefore, in this study, I manipulated the quality of evidence by changing the way it was collected in low and high quality conditions. I also manipulated the presence perceptions of evidence. I predic ted that if contamination did occur in the investigation process, then it is something that could be pointed out during the trial. Quality of evidence is important in that consideration because it is possible that jurors will assign a

PAGE 31

31 high weight to evide nce collected under both good and bad circumstances when they should be discounting the information gained from low quality evidence. They should also be discounting the information gained through contamination of high quality evidence because of the chan ce for error. Further, they should be discounting the information gained from evidence that was subject to evidence contamination because that evidence is likely being presented in trial as stronger than it actually is; pointing out the contamination shou ld help them weigh evidence more accurately. Weighing complex evidence In general, jurors are usually considered fairly good decision makers, and for most pieces of evidence they assess the weight or importance of that evidence appropriately (Levett, Dani elson, Kovera & Cutler, 2005). There are several notable exceptions to this statement; one is discussed above in that jurors may not distinguish between high quality and low quality evidence appropriately. Another notable exception is how jurors evaluate complex evidence. Generally, complex trial evidence is difficult for jurors to understand (Helgeson & Ursic, 1993). Overall, as the amount of evidence (and the cognitive difficulty of that information) increases, jurors understand less of the evidence an d are likely to make more mistakes in processing that information (Heuer & Penrod, 1994). It is possible, that contaminated evidence could be defined as complex evidence in that if jurors are going to weigh the evidence, properly, it requires considering the influence of one or more pieces of evidence on other evidence all at one time. It stands to reason then, that since jurors are not generally adept at comprehending and weighing complex evidence, they may not be able to understand (or even identify) th e complex process of evidence interdependence in a court case.

PAGE 32

32 CHAPTER 2 THE CURRENT STUDY In this study, I planned to add to the literature on evidence contamination by examining whether potential jurors could properly weigh contaminated versus independ ent (non contaminated) evidence. Given we know evidence contamination occurs in actual cases, the next logical step was to examine if jurors were able to identify when a piece of evidence presented in court was potentially biased by other evidence. Previ ous literature has shown that criminology students were able to distinguish between contaminated and uncontaminated evidence (Hasel et al., 2012), however, other literature examining how jurors evaluate complex evidence suggests that jurors might not be as skilled (Kovera & McAuliff, 2000; Levett & Kovera, 2009). I tested if jurors could differentiate between good, unbiased information and those pieces of evidence that may have been interpreted or collected differently because of evidence contamination. I a lso examined how jurors weighed evidence by manipulating the quality of the evidence presented to mock jurors as well as the potential for contamination between evidence. In this study, I presented jurors with a trial containing two pieces of incriminatin g evidence: an eyewitness, and a confession. These two pieces of evidence were selected because they have been shown to be significant contributors to wrongful conviction, and have been explicitly tested in the evidence contamination literature, showing t hat it is possible for contamination to occur when an eyewitness hears about whether a lineup member confessed. Additionally, I varied the quality of the evidence by describing during the trial stimulus how the potentially contaminating evidence was coll ected. The conditions with

PAGE 33

33 high quality evidence show how the evidence was collected using procedures and protections that have been shown to be best practices; conditions with low quality evidence depicted evidence that was collected using practices used by police in the real world, but are not best practices. These manipulations will be described in greater detail in the method section. In pertinent conditions, I also indicated that the secondary evidence (the eyewitness) was contaminated. If jurors a re able to differentiate between contaminated versus independent evidence, and also account for the quality of evidence in that determination ¸ then the problem of contamination of evidence is likely not as serious of a problem as the literature has portray ed it thus far. The contamination may be problematic during the investigation, but could be corrected when the case gets to trial as jurors will see that the evidence was of low quality and/or contaminated, thus necessitating a not guilty verdict. The bes t case scenario is that the evidence is of high quality, independent, and successfully leads to a guilty verdict. In all other scenarios, jurors who weigh the evidence properly should be more likely to render not guilty verdicts; however, this may vary bas ed on the quality of the evidence and whether the two pieces of evidence were independent. To test these things, I created a study in which community members acting as mock jurors read a trial transcript that presented both pieces of evidence. I manipulat ed 1) the evidence quality by describing the manner in which the evidence was collected, 2) the potential for contamination by describing the information that the legal actor had knowledge of prior to making their decision and 3) the order of evidence, cou nterbalanced to ensure no primacy or recency effects in which piece of evidence

PAGE 34

34 jurors adhered to more readily. All participants heard about an eyewitness (high or low cond heard the eyewitness evidence followed by the confession evidence, and the other half heard the confession evidence followed by the eyewitness. In this way, I was able to de termine if participants could differentiate between the quality levels of different pieces of evidence as well as how much weight they put on any one piece of evidence to determine their verdict option. Further, I asked several questions of the participan t after they made their verdict selection to determine how each piece of evidence was evaluated and assessed their confidence in their decision (all described in more detail in the method). Prior research suggested that criminology undergraduate students c ould identify contamination (Hasel et al., 2012); however other research showed how jurors are generally not able to properly weigh certain pieces of evidence (Kovera, McAuliff & Hebert, 1999; Levett & Kovera, 2009). Thus, there are two manners in which t he data could be informative ; either jurors weigh evidence properly or they do not. One is consistent with null hypotheses and the other is consistent with traditional research hypotheses. If jurors weigh evidence properly, then I expected to see H1: a m ain effect of contamination, meaning that verdict and confidence ratings (along with scaled responses to additional dependent measures) are more indicative of guilt when contamination is absent than when it is present. Further, I predicted H2: a main effe ct of quality of confession evidence, meaning that dependent measures are more indicative of guilt in high quality evidence conditions than in low quality evidence

PAGE 35

35 conditions (but only for the confession evidence). I also expected to see H3: a three way i nteraction between quality of confession evidence, quality of eyewitness evidence and contamination on verdict and probability of guilt ratings , depicted in graphical form in Figure 2 1 below . More specifically, beginning with the figure on the left ( conta mination absent ) , I expected conditions with high quality confession evidence and high quality eyewitness evidence (the blue diamond in the upper left corner) to be the strongest case for a guilty verdict; I expected that switching the eyewitness to low qu ality (the red square directly below) would lessen guilt ratings, though only slightly because there would still have been a high quality, uncontaminated confession to use (i.e., the confession had not been contaminated). Again, in conditions with contami nation absent, if the confession was low quality and the eyewitness high quality (the blue diam ond in the upper right corner ) , I expected the guilt ratings to be high middle range because the confession evidence was weak, but there was still eyewitness evi dence to base a verdict on. Finally, in contamination absent conditions with low quality confession and eyewitness evidence (the red square in the bottom right corner), I expected the guilt ratings to be lowest because both pieces of evidence were collect ed poorly and of low quality. In conditions in which contamination was present (the figure on the right of Figure 2 1) , I expected the guilt ratings to be lowest in the condition with low quality eyewitness and confession evidence (the red square in the bo ttom right) because both pieces were weak evidence , and the eyewitness evidence was fu r ther compromised because of the contamination . In the condition with high quality eyewitness evidence (the blue diamond on the right side) , I expected guilt ratings to move up slightly since there was a

PAGE 36

36 strong eyewitness (though it was still very low since the witness was contaminated). In conditions with high quality confession evidence (the red square in the top left corner, the blue diamond cannot be seen) , I expecte d relatively high guilt ratings for either eyewitness quality level because in both conditions there was contamination on the eyewitness (weakening it from either high or low quality) however the juror still had a high quality confession to base a guilty v erdict on , thus the guilt ratings remain high with essentially only one piece of evidence to rely on the confession . Here the quality of the eyewitness does not matter because it was weakened regardless. Figure 2 1: Graphical representation of hypot heses for A) Contamination Absent, and B) Contamination Present conditions. Conversely, jurors without this prior experience of criminology undergraduate courses may not be able to properly weigh evidence or account for contamination effects. If this was true, then I predicted a set of alternative hypotheses. I predicted no main effects for either H1: contamination, or H2: quality of evidence, and no H3: three way interaction effect. I expected verdict and confidence ratings as well as additional 0 1 2 3 4 5 High quality confession Low quality confession Probability of Guilt Contamination Absent High quality eyewitness Low quality eyewitness 0 1 2 3 4 5 High quality confession Low quality confession Probability of Guilt Contamination Present High quality eyewitness Low quality eyewitness

PAGE 37

37 dependen t measures to be equal in conditions with contamination present and absent, and in both high and low quality evidence conditions (for both the confession and the eyewitness evidence). This would be a problematic outcome in that jurors would be weighing hig h and low quality evidence and contaminated and uncontaminated evidence the same even though there is actually basis for different verdict outcomes based on either factor. For example, a case could have an otherwise high quality piece of eyewitness eviden ce (in that it was collected using science based, best practices) but this piece of evidence could have still been contaminated by other evidence (for example, if the eyewitness t the jurors to be able to identify that contamination potential and reduce the weight jurors place on the contaminated evidence accordingly. Further, jurors ideally should be able to differentiate between evidence of high and low quality (for example, eye witness identifications conducted using either a sequential or simultaneous lineup). However, I also postulated that it is unlikely jurors will be able to do so, which means that jurors would still put the same amount of weight on pieces of evidence that were substantively of different importance. If this is true, then even low quality or contaminated evidence may still lead to guilty outcomes in court.

PAGE 38

38 CHAPTER 3 METHOD Participants I recruited a final sample of 692 community members through the use of greatest amount of jury eligible citizens in the country. All recruited participants were 18 years or older (birth years ranged from 1937 1996 with a modal year of 1989) . Several participants entered a two digit number instead of a four digit year, so it is possible participants may have mistakenly entered their age here instead of their year of birth. Thus, some participants were not included in the ranges because I wa s unsure what their value meant (i.e. if the participant entered 48, he might mean he was 48 years old, or that he was born in 1948). Further, participants received a small amount of monetary compensation ($3) for their time. The amount of compensation is consistent the current minimum wage. The sample was fairly evenly split males (53.2%) and females (46.8%). The sample was mostly White, non hispanic (73.7%) with 5.9% Bl ack, non hispanic, 13.3% Asian, 5.1% Hispanic, 0 .1% Cuban and 1.9% indicating Other for their racial identification. Most were single (55.3%), followed by married (34.5%), divorced (7.8%), and widowed (1.6%), with 0 .7% leaving their response blank for the question about their current marital status. The majority of participants (86.3%) had not had prior jury service experience, and 13.3% had previously served on a jury, with 0 .4% leaving this response option blank. Finally, 24.6% of the participants indic ated that they do attend religious services at least once a month, with 75.0% indicating they do not, and 0 .4% leaving the response options blank. Participants were

PAGE 39

39 automatically removed from the final analysis if they failed the attention test question hi dden within the survey, or had reasonable suspicions as to the true purpose of the study. Some participants were also removed depending on the amount of time he or she spent attending to the stimulus (described below in the data cleaning section) . Finall y, some analyses are reported after removing those who failed manipulation check questions (this strategy is described in more detail below in the results section, and the percentages for demographics did not differ in the two separate analyses). Design A 2 (confession evidence quality: high v. low) x 2 (eyewitness evidence quality: high v. low) x 2 ( evidence contamination: present v. absent ) x 2 ( order of evidence : confession followed by eyewitness evidence v. eyewitness followed by confession evidence ) + 2 (control with no eyewitness identification, contamination absent: high quality confession v. low quality confession) between s ubjects factorial design was used . Prior to collecting the data, a power analysis was conducted using the G*power program asses sing the required sample size needed to observed a small to medium effect size with a .95 power level and = .05 in our 18 group design. Results for an ANOVA to assess the effect of our IVs on any scaled items that may be created required a total sample size of n = 480 (assuming a small to medium effect size of f = .1649, half way between the standard small effect size of .1 and medium effect size of .25). Independent Variables Quality of E vidence Both pieces of evidence that I varied (the confession and the eyewitness

PAGE 40

40 piec es of evidence also had three or four manipulation check questions to help distinguish whether participants noticed that the evidence was high or low quality (attached in the appendices and explained in the dependent measures section next). In the high qu ality c onfession condition, the confession was described as being elicited under optimal interrogation tactics : 1) there was no coercion or pressure by the interrogators , 2) the interrogation was conducted for a short amount of time , quantified as about tw o hours , 3) the interrogators included Blume (the detective who arrested our Smi , and 4) the confession was hand w ritten by the suspect himself. The section of the trial stimulus direct testimony depicting this information was stated as: The detective stated that the police officers brought in several suspects for questio Michael Smith. Detective Blume stated that he asked another officer, uninvolved in the case prior to the interrogation, to question Mr. Smith to drive any of the interrogation questioning. Blume reiterated that while both Blume and the other officer were present for the entire interrogation, Blume remained in Blume did no t participate in the interrogation can be seen in the interrogation video submitted into evidence. Blume reported that Smith waived his Miranda rights freely, knowing he was free to leave at any point during the interrogation. The officer interrogated Sm ith for approximately 2 hours, stating that the interrogation was mainly conversational for the entire 2 hours with only a few times when the interrogator needed to really push Mr. Smith for additional information. During the first hour of the interrogati on Mr. Smith denied any involvement in the crime, stating that he was out with a friend until about 9pm on the 18 th then went home alone for the remainder of the night and morning of the 19 th . Blume reported that the interrogator used techniques that are standard for use in the Interrogation Manual used by the Brevard Police Department and others around the country, such as open ended questioning and encouraging the suspect to describe as much as they can with little interruption by the officers. Blume re did kill Ms. Bennet during the early morning hours of September 19 th . The

PAGE 41

41 officer requested that Smith write a confession, including as many details about the murder as he could, and Smith complied with the request, including details such as the extension cord used to strangle Ms. Bennet. After Mr. Smith signed the handwritten confession, he was escorted from the room. In the low quality c onfession condition, the confession was elicited under cond itions that are typically more likely to elicit a false confession : 1) the interrogators used minimization (making the crime seem less serious than it is, thus not problematic to confess to) and maximization (making the crime seem more serious than it is, thus the suspect will need to cooperate with the police to receive any leniency) techniques to intimidate the suspect, 2) the interrogation lasted for ten hours, 3) the interrogators included the same officer that arrested the suspect (Detective Blume) , an d 4) the confession was written out by the investigators and only initialed by the suspect. For those participants who received the low quality confession evidence in their version of the trial stimulus, they read the following paragraph with the confessi on collection details modified to demonstrate low quality evidence: The detective stated that the police officers brought in several suspects for Michael Smith. Detective Blume stated that Smith waived his Miranda rights after a bit of coaxing, knowing that he was free to leave at any point during the interrogation. Blume and another officer interrogated Smith for a total of approximately 10 hours, stating that quite frequently during the questioning, Blume needed to raise his voice to encourage the suspect to tell him the truth. During the first 5 hours of the interrogation Mr. Smith denied any involvement in the crime, stating that he was out with a friend until about 9pm on the 18 th then went home alone for the remainder of the night and morning of the 19 th . However, Detective Blume reported that did kill Ms. Bennet during the early morning hours of Septemb er 19 th . Blume encouraged Smith to initial throughout the report and to sign at the bottom of the page, whic h Smith did. During the interrogation, Blume reported that he used techniques that, while they may be controversial due to the potential for coercion, are standard for use in the Interrogation

PAGE 42

42 Manual used by the Brevard Police Department and many others t hroughout the country. These techniques included questions designed to make the crime seem less important than it is and thus, no big deal to confess to. Additionally, there were questions that make the crime seem more important and a much bigger deal th an it really is. This later question indicates the suspect should cooperate and confess so the police can help him as much as possible. A video recording of the interrogation was submitted into evidence, showing that both Blume and the other officer enga ged in the interrogation of Smith. In addition to this main section describing the details of the confession, we also repeated the details of how the confession was elicited in the opening and closing statements to ensure that participants noticed and enco ded the details of the investigation. The high quality confession information was described in the The confession that was elicited by our police detectives was a high quality confession one that you should pay close attention to and Mr. Smith freely volunteered this information when talking with police. An innocent man would not do this. In addition, in the closing statement, the prosecutor said Keep in mind these details were given by Mr. Smith during a two hour discussion with police. There was little influence on Mr. Smith to confess; little influence other than his own guilt for the crime of course. The low quality confession information was opening and closing s Further, the amount of pressure used ten hours of questioning, my client still did not write out a confession, but merely initialed a piece of paper t hrown in his face by the same people who arrested him. This pressure that is caused by having to sit in a little room with police officers drilling questions at you one after another for ten hours. That is enough

PAGE 43

43 pressure fo r many people to say something, anything to get out of the situation. I know it would be enough pressure for me e closing statement. In appropriate conditions, participants also heard testimony from an eyewitness indicating the eyewitness identified the suspect and again answered four manipulation check questions to ensure they noticed the quality of the eyewitness evidence. In the low quality eyewitness conditions, the eyewitness completed the task under poor conditions. Specifically, participants heard that 1) the eyewitness heard biased lineup instructions that failed to indicate the actual perpetrator may not be present, 2) the eyewitness made an identification from a show up with a mug shot photo in jail clothes rather than plain clothes, 3) the re was a two week time delay between seeing the crime occur and making their identification, 4) there was a few blocks distance when witnessing the suspect at the crime scene, 5) there was little to no light, and therefore the viewing conditions were poor, 6) the witness was not paying close attention to the suspect, 7) the witness only watched the suspect for a few minutes, and 8) the eyewitness heard positive post identification feedback after the identification in the form Nearly two weeks later, on October 4 th , Mr. Griffin was asked to report to the police station for an identification. Mr. Griffin reported that he sat at a table in a small room and the detectives explain ed how the identification procedure would work; they would show him one photo and he could take as long as he needed to look at it. The police stressed how important it was for him to make an accurate identification, because they did not have any physical evidence tying the perpetrator to the crime scene and were relying heavily on his identification. He said the detectives presented him with the mug shot of a person wearing an orange jumpsuit who fit his description, and Mr. Griffin reported that he said it was the same person he had seen outside his house. The police asked if he was certain and to look at the picture again. Mr. Griffin reported that he spent several minutes looking at the picture again, and confirmed once again that this was the person he had seen. The police detectives congratulated him and

PAGE 44

44 a computer to record his identification and certainty. On cross examination, the defense attorney asked how Mr. Griffin c ould be certain of his identification, especially since he had seen the person in the early morning hours, in the dark. Mr. Griffin reported that yes, it had been dark out especially since their street light had burned out a few weeks before. The attorne y asked how long Mr. Griffin had been watching the person for, and Mr. Griffin replied only a few minutes and that he was not watching his actions very closely but he was still pretty sure that they had found the right guy. The attorney further asked if h e was stressed at all during the identification procedures. He reported that yes, he was nervous, but felt that he had done a good service when the police said he had picked the right guy. In the high quality eyewitness conditions, the eyewitness : 1) iden tified the suspect from a photo with the suspect wearing plain clothes rather than a jailhouse jumpsuit, 2) completed the task less than two days after witnessing the crime, 3) saw the suspect from a short distance of a few feet, 4) under well lit conditio ns, 5) while paying close attention to the suspect, 6) and watching him for nearly 20 minutes, 7) heard unbiased instructions warning that the actual perpetrator may not be present, and 8) received no feedback after the identification. These conditions re ad as follows: The next day, September 21 st , Mr. Griffin was asked to report to the police station for an identification. Mr. Griffin reported that he sat at a table in a small room and the detectives explained how the identification procedure would work; they would show him one photo and he could take as long as he needed to look at it. They further warned him that the actual perpetrator may not be in the photo and to be certain before making an identification. The police stressed how important it was f or him to make an accurate identification because they did not have any physical evidence tying the perpetrator to the crime scene, but that they would continue the investigation whether he made the identification or not. He said the detectives showed him a photo of a man in jeans and a t shirt fitting his description, so he stared at the photo for a while then told the officer that was the man he had seen near his house at the time of the crime. The police asked if he was certain and Mr. Griffin spent an other minute looking at the photo then said he was sure that was the man he had seen. The police detectives thanked him and escorted Mr. Griffin to another room with a computer to record his identification and certainty,

PAGE 45

45 being sure not to say anything to confidence level. On cross examination, the defense attorney asked how Mr. Griffin could be certain of his identification, especially since he had seen the person in the early morning hours, in the dark. Mr. Griffin rep orted that yes, it had been dark out, but there were street lamps along the road with bright lights and the man had gone under them a few times while he was pacing. Mr. Griffin reported that he watched the man for about 20 minutes, paying close attention to where the man was walking to ensure his new car in the driveway was not stolen. The attorney further asked if he was stressed at all during the identification procedures. He reported that yes, he was nervous, but felt that he had done a good service. A gain, the eyewitness identification collection details were reiterated in both the opening and closing statements to ensure that participants remembered the way the opening statements include d , The witness you will hear from today was given unbiased instructions prior to making his identification, and our police officers followed all the procedures that have been shown to elicit accurate identifications, like ensurin g the identification photo was taken in plain clothes, not a jailhouse jumpsuit. Further, the perpetrator was right in the front yard only a few feet from the window our witness was looking out of, watching his actions for nearly 20 minutes, and he made h is identification only 2 days after the crime. and the closing statements said , Keep in mind, this witness identified our man Smith only 2 days after the crime, after watching him closely for 20 minutes practically in his front yard under a street light, a nd with the warning to only make an identification if he was sure. Further, our officers followed every procedure by the book, making sure not to give any hints before the identification, using method photo with the suspect in street clothes rather than a potentially biasing jailhouse jumpsuit and taking down his confidence level right away after then identification. The low quality eyewitness information was opening statements, saying,

PAGE 46

46 Further, the witness was asked to identify someone over 2 weeks after the crime occurred. Can you remember who you might have seen over 2 weeks ago, never mind identifying a person you saw from nearly a block away and for only a few minutes? Further, researchers have learned that many of t he procedures that protect against mistaken identifications were not followed here. In particular, the officers had our witness identify the suspect using a show up (using only 1 picture) rather than a photo array, tion of a photo with the suspect wearing an orange jumpsuit, and even encouraged the witness to pick someone, all of which have been shown to lead to mistakenly selecting the wrong person. And in We have already e stablished the whole host of issues that surround the crime and the identification, the procedures employed by the officers of using a mug shot instead of a photo with everyday clo thing, even the things the officers said before, during, and after the identification all have been shown to make the chances of that witness being able to accurately identify the perpetrator slim to none. To make matters worse, this man made his identifi cation after seeing the person from a block away, in the but sometimes they get it wrong. And he was clearly wrong here today. Evidence Contamination In conditions in which contami nation was present, the participant heard during testimony that there was potential for the confession to have influenced the conditions heard that the eyewitness overhea confession while waiting to complete the identification procedure. This information was presented through cross examination questioning by the defense attorney across both high and low quality conditions when conta mination was present. It was omitted from the conditions where the pieces of evidence are considered independent (when contamination is absent). In the contamination present conditions, participants read: The defense attorney asked if Mr. Griffin heard a nything while waiting to begin his identification and Mr. Griffin reported that he had overheard a

PAGE 47

47 couple of police officers talking at their desks about how great it was that the suspect had confessed that morning and how this case was shaping up nicely. They had wished that all cases were so easy. The defense attorney asked if he thought that hearing this right before making an identification may have affected his decision any, and Mr. Griffin reported that he had based his identification decision on hi s memory from the night he saw the man pacing in the road, not what the police officers had said. Further, the closing statements also mentioned the potential for contamination by ession. In witness already knew the suspect had confessed even prior to making his identification. How can the officers expect him to have a clear state of mind if e ven the witness himself I did not include this manipulation in the opening statement as I had with the other manipulations because the issue was brought up through cro ss examination and might not have been known to the defense attorney before the trial started. Given this, it would have seemed unrealistic to discuss this in the opening statements. All participants were also asked three manipulation check questions abou t the possibility of contamination or influence of the confession on the eyewitness to ensure they noticed the manipulation. Specifically, they were asked to agree or disagree to the statements that 1) the eyewitness overheard people talking about the con fession prior to their identification, 2) they were sheltered from any information about the case before completing their identification, and 3) their decision could have been influenced by their knowledge of the confession. Order of Evidence The stimulus sections discussing the confession and eyewitness evidence were rearranged to depict one piece before the other, according to the condition; no

PAGE 48

48 substantive information was altered based on this variable. In conditions where confession evidence was present ed first , participants were presented first with the confession information, showing that the suspect confessed to the crime when he was questioned earlier in the investigation. They then read about the eyewitness information. In the eyewitness informat ion presented first conditions, the first piece of evidence presented in the trial was the eyewitness information, followed by the confession information. The order the evidence was presented in was counterbalanced to ensure that the weight assigned to an y piece of evidence was not due to whether it was heard first versus last or suffered from any primacy or recency effects. Control Conditions I also included two control conditions to allow for more complete comparisons among our variables. In these condi tions, (denoted by removing the green text in Appendix C below) the same high quality and low quality confession information was described, however there was no mention of an eyewitness or potential for contamination. By including these two conditions, I was conviction rates were reaching ceiling effects based solely on the confession information, thus masking the effect of contamination. I expected conviction rates to be highest when there were two pieces of evidence pointing to wards guilt and somewhat lower when there was only one piece of evidence pointing to guilt. I also thought that if guilt ratings were equal in conditions with just confession evidence compared to both pieces of evidence, then I could say that the confessi on evidence was possibly masking

PAGE 49

49 Dependent Measures A full copy of dependent measures as they were shown to participants is included in Appendix D (for the pilot study ) and E ( for the current study) . Verdict , confidence ratings, and explanation of verdict There were two versions of the verdict choice: 1) scaled dichotomously as guilty v. not guilty, and 2) scaled continuo Likert type scale (strongly disagree (1) to strongly agree (6)). Participants were also asked to rate their conf idence in their verdict choice on a 1 100% confidence scale and to complete a probability of guilt rating (again on a 1 100% scale). Following their confidence response, they were asked to explain their reasoning for the verdict they rendered and how they weighed the evidence to come to that verdict. Evaluations of evidence Additionally, there were perceptions of each piece of evidence and how that evidence contributed to their ultimate verdict decision . All ite ms were rated on a 1 to 6 Likert type scale with 1 indicating strong disagreement and 6 indicating strong agreement. All reverse coded items are indicated by a (R) below, and were recoded so that higher ratings on each item indicated greater strength of e vidence. Out of these several items, I created four scales based on the factor analysis results reported below. First, the items generally assessing the strength of the confession evidence held the confession is the strongest piece of evidence in this case; 2) If I were the poli ce

PAGE 50

50 investigator, I would have collected the confession information differently (R); 3) I believe the confession was elicited under optimal conditions; 4) I thought the detective hat the suspect confessed because he is truly guilty; 6) I believe that jurors should consider the confession in this case is a weak piece of evidence (R). A second constr uct assessing the strength of the eyewitness evidence also had following items: 1) I believe the eyewitness identification is the strongest piece of evidence in this c ase; 2) The eyewitness identification evidence was not collected correctly(R); 3) I believe the identification was made under the best possible conditions; 4) I thought the witness seemed untrustworthy(R); 5) I am certain that the person selected from the showup is truly guilty; 6) I believe the eyewitness identification should believe the eyewitness evidence was a high quality piece of evidence. A third construct, Gener al Weighing of Confession Evidence During Trial , assessed the level to which individuals would rely on the confession evidence for their unreliable (R); and 2) I would not rely on confessions if I were deciding on verdict in a real criminal trial (R). The fourth and final construct also containing two items assessed the extent to This sc ale, Specific Weighing of Eyewitness Evidence In This Case , included: 1) The

PAGE 51

51 details about the showup were important in deciding whether to believe the evidence or not and 2) When choosing a verdict, I used my knowledge about the showup. Finally, the last section of the questionnaire contained all of our manipulation of the following questions to ensure they could accurately remember the details about each piece of eviden ce: 1) the confession was handwritten by the suspect; 2) The confession was collected without coercion or force from the investigator; 3) The confession was given after a short interrogation (less than 3 hours); 4) The investigators who heard the confessio n did not know about the guilt of the suspect (they were not the same officers who had completed the earlier stages of investigation); 5) The eyewitness was told that the actual perpetrator may or may not be present in the showup; 6) the eyewit ness complet ed the showup task two days after seeing the crime; 7) the police identification (R); 8) during the identification, the eyewitness saw only 1 picture (this question was mistake nly not fixed before collecting the data and thus, was not used for the eyewitness manipulation check scale); 9) before making the eyewitness identification, the eyewitness overheard people talking about the person who confessed (R); 10) the eyewitness did not hear any information about the case prior to completing been influenced by the confession (R). The last section of questions gathered some demographic information (gender, race/ethnicity, age, educational background, and whether they consider themselves to be extremely knowledgeable about Psychology/Law issues). Finally, I asked if they had reasonable suspicions as to the

PAGE 52

52 true intent of the study procedures by ask ing them if they believed the information that was given to them in the case and if they thought they were given any false information in the case description (and explain what they did not believe for each of those questions). Procedure Participants were recruited from the online website MTurk (a large, diverse group of community members from around the country). They were directed to a Qualtrics website that contained the entirety of our study. P articipants read an informed consent document and agree d t o partic ipate in the study. I also included an attention question in the questionnaire telling them to answer a question a certain way; this gave me the opportunity to assess which participants were really paying detailed attention to the materials and fo llowing the instructions. The first screen after the informed consent asked several demographic questions to simulate voir dire questioning by attorneys. The next screen told participants to act as if they were jurors hearing this case, stressing the impo rtance of paying attention to the details and all information presented as if they were in a real courtroom with real attorneys and a real judge. Then participants read the entire trial stimulus (the master stimulus containing all possible conditions is a ttached in Appendix C with condition changes noted), s etting up the crime scenario and depicting each piece of evidence. This written trial stimulus followed standard trial procedures; therefore I included opening judicial instructions, opening statements from both prosecution and defense, presentation of evidence (prosecution, then defense with some cross examination questioning), closing statements from prosecution and defense, and finally closing judicial instructions and the charge to the jury. After r eading the stimulus, participants

PAGE 53

53 were presented with the dependent measures (as described above). Each of the screens was timed to ensure that participants were paying adequate attention to the stimulus and questions before moving on to the next section. After completing all the dependent measures and manipulation check questions, participants read a full debriefing to explain the purpose of the study, and were thanked for their participation. Additionally, I gave participants the principle investigator should they have any questions about their participation in the study. There was also a procedure for participants to contact the researcher through the MTurk website as well.

PAGE 54

54 CHAPTER 4 PILOT TESTING Prior to running the cur rent study, I conducted pilot testing to ensure that our stimulus was calibrated the way I intended. That is, to fully see the effect of any manipulations, I needed to ensure variation in verdict. To do this I conducted a study mirroring the procedure of the actual study with 60 students, asking them to provide guilt ratings/confidence of the defendant given the evidence presented to them and to answer manipulation check questions (for a copy of the quest ionnaire, see Appendix D ). Thirty students read thr ough the strongest stimulus with non contaminated evidence (confession evidence first, followed by eyewitness evidence; both pieces were high quality), and thirty students read the weakest stimulus with contaminated, low quality evidence (again, confession evidence presented first followed by eyewitness evidence, both pieces being of low quality). Students were randomly assigned to one of these two conditions. Participants then rendered a verdict to ensure that the evidence presented leads to statistically different verdict distributions in the strongest and weakest cases. square test revealed that our verdic ts were not significantly , p = .22. Of the 30 participants hearing the high quality, uncontaminated evidence, 30% rated the defendant guilty ( SD = .47) whereas only 17% of those who heard the low quality, contaminated evidence rated the defendant as guilty ( SD = .38). I tested the belief in believe the defendant should be found guilty of the charges brought again 6 Likert type scale (1 indicating strongly disagree to 6 indicating strongly agree); the results of an independent subjects t test revealed the participants in the stronger

PAGE 55

55 condition ( M = 3.00, SD = 1.53) were more likely to agree with the s tatement than those in the weaker condition ( M = 2.27, SD = 1.34); t (58) = 1.98, p = .05. With these results, I do see some difference in perceived guilt with the continuous scale and full ranges in participant responses to each of the above items (0 and 1 for the dichotomous verdict choice and 1 to 6 for the continuous verdict scale). The pilot testing was completed using undergraduate criminology students who might be very skeptical of criminal evidence as a result of their education. I used a communit y sample in the actual study, under the assumption that the community member sample would be less skeptical of criminal evidence than the sample in our pilot study. In addition, given that I did find variation on our dependent measures, I believed the mate rials were strong enough to use in testing for the effects in our larger community sample. A second goal of the pilot study was to ensure that participants were able to identify our manipulations, meaning that they noticed when the confession was high or l ow quality, when the eyewitness was high or low quality, and if there was the potential for contamination of the eyewitness by the confession or not. I created three summed scales for the manipulation check questions assessing 1) confession quality ( four questions), 2) eyewitness quality ( four questions), and 3) contamination presence ( three questions). Some questions were reverse coded so that higher scores indicated higher quality evidence (or absence of contamination). A series of independent samples t tests revealed that participants in the high quality confession conditions ( M = 2.80) scored significantly higher on the confession manipulation check scale than those in the low quality confession conditions ( M = 0.50), indicating that they correctly ag reed with

PAGE 56

56 the manipulation check questions than those in the low quality confession condition; t (56) = 12.90, p < .001. With the confession manipulation checks, 5 people agreed with all 4 questions, 15 agreed with 3 questions, 9 agreed with 2 questions, and 1 agreed with 1 question, meaning that 20 of 30 (67%) endorsed 3 or more of the eyewitness manipulation questions). For the confession manipulation questions, 15 people correctly rejected all 4 questions, 12 people agreed with 1 question, 1 person ag reed with 2 questions, and none agreed with 3 or all 4 questions (meaning that 27 of 28 (96.42%) correctly rejected at least 3 of the 4 confession manipulation questions). The same pattern held for the eyewitness quality, meaning that those in the high qu ality eyewitness conditions ( M = 3.55) scored significantly higher on the eyewitness manipulation check scale than those in the low eyewitness quality conditions ( M = 0.93); t (56) = 12.27, p < .001. Overall, 19 of those in the stronger condition (with hi gh quality, uncontaminated evidence) agreed with all 4 eyewitness questions, 8 agreed with 3 questions, 1 agreed with 2 questions, and 1 agreed with only 1 eyewitness manipulation question (meaning 27 of 29 (93%) endorsed 3 or more of the eyewitness manipu lation questions). Of those in the weaker condition (meaning they should reject the manipulation check questions), 10 correctly rejected all 4 eyewitness questions, 13 agreed with 1 question, 4 agreed with 2 questions, 2 people agreed with 3 questions, and none agreed with all 4, meaning that 23 of 29 (79.31%) correctly rejected at least 3 of the 4 eyewitness manipulation check questions. Finally, those who heard about the potential for contamination ( M = 1.93) scored higher on the contamination scale (mean ing less presence of contamination) than those

PAGE 57

57 who did not hear about the contamination section ( M = 0.52) in our stimulus; t (57) = 6.74, p < .001. Finally, 8 people agreed with 3 of the 3 contamination questions, 12 agreed with 2 questions, and 10 agree d with 1 question, meaning 20 of 30 (67%) endorsed 2 or more of the contamination manipulation questions). For the contamination manipulation check questions, 19 people correctly rejected all contamination questions, 6 people agreed with 1 question, 3 peop le agreed with 2 questions, 1 person agreed with all 3 contamination questions, meaning that 25 of 29 people (86.21%) correctly rejected at least 2 of the 3 contamination manipulation check questions. Thus, I believed that the manipulations were noticed by participants in the pilot study, however given the number of people who did not pass all manipulation checks and after consultation with my committee members, I decided to boost the manipulation slightly for the community member sample so that I was able to effectively test the hypotheses in the actual study. This is when I added in the extra reinforcement of the manipulations in the opening and closing statements and additional details about the quality of evidence for both the confession and the eyewitn ess .

PAGE 58

58 CHAPTER 5 RESULTS Data Cleaning As the data for the current project was collected online through the Qualtrics website, there was no need to worry about inaccuracies in data entry. After downloading the results in SPSS format, I cleaned the data. This cleaning included recoding and summing the timed results indicating how long the participants spent attending to the stimulus (these results were per page initially) as well as how long the participants spent completing the entire study overall. The first part of cleaning the data if they did not pay adequate attention to the stimulus. First, I removed those who failed the attention question embedded in the dependent measures questions. This question asked pa rticipants to give a particular response to a question to indicate they were paying attention and reading all instructions carefully. Of the 950 people who completed the study, 166 failed to correctly respond to the attention question, therefore their res ponses were eliminated. The reasoning to do so was that if they did not read instructions clearly enough to answer this question as indicated, then they likely were not paying enough attention to the rest of the materials and may have been an extra source of noise in the data. After determining the average amount of time spent to complete the study, any participants who took too long or too short to finish the study were removed for the actual analyses. Again, the purpose of removing participants for takin g either too long or too short to complete the study was to ensure participants paid attention to the study. That is, those who completed the study too quickly may have gone through the study too quickly to really pay attention. And, those who took to o lon g to complete the study

PAGE 59

59 may have left the computer while completing the study or may not have completed the study all in one sitting. T aking t oo long to complete the study was quantified as more than three standard deviations above the mean. T aking t oo sho rt to complete the study was quantified as more than one standard deviation below the mean (because of skewed timing data, the standard deviations used on e ither end could not be equal). I completed this separately for those in the 16 experimental conditio ns compared to the two control conditions as the control groups did not have any mention of an eyewitness in the trial stimulus, so their average length of completing the study should be shorter and they might have been mistakenly viewed as outside of the parameters when in fact they were not. For the 16 factorial conditions, I began by removing six outliers whose time was well above the normal distribution (over 297 minutes to complete the study) . The average amount of time those in the factorial conditi ons spent to complete the study was 39.64 minutes (standard deviation: 20.45 minutes). Therefore, 60 people were removed for spending less than 19.19 minutes and 15 people were removed for spending more than 100.98 minutes on the study. For the two contr ol conditions, I first removed one outlier who spent 402.33 minutes completing the study (and was thus well above the normal distribution). T he average amount of time spent to complete the study was 33.49 minutes (standard deviation: 18.43 minutes). In t hese two conditions, I removed 9 people who spent less than 15.06 minutes and 1 person who spent more than 88.79 minutes. Finally, I checked for any reasonable suspicions as to the true purpose of the study (none were reported that lead me to believe par ticipants were aware of what I was testing). Then I recoded any of the questions phrased in the reverse direction to ensure

PAGE 60

60 that results with higher numbers indicated greater confidence in or strength of the evidence. The final sample consisted of 692 peo ple and is described in the participants section. Manipulation checks Analyses for manipulation checks were conducted in two ways: first, I completed three independent samples t tests to ensure there were significant differences between the high and low qu ality groups for scales measuring perceptions of 1) the confession evidence, 2) the eyewitness evidence, and 3) the present and absent contamination conditions. After recoding any reverse coded questions to ensure higher responses indicated higher qualit y evidence, I created a summed scale out of the four confession quality manipulation check questions ; these items were scored as 0 indicating the participant recalled low quality evidence and 1 indicating the participant recalled high quality evidence . Th ose who read a vignette depicting high quality confession evidence scored higher ( M = 2.92) on the confession scale than those who read a vignette with the low quality confession evidence ( M = .56), t (685) = 31.53, p < .001. Similarly, I created another s ummed scale of the three eyewitness quality manipulation check questions ; again, these items were scored as 0 indicating the participant recalled low quality evidence and 1 indicating the participant recalled high quality evidence . Again, those who read a vignette depicting high quality eyewitness evidence scored higher ( M = 2.69) on the eyewitness scale than those who read a vignette with the low quality eyewitness evidence ( M = .56), t (611) = 37.14, p < .001. Finally, I created a third summed scale out of the three contamination manipulation check questions, here higher scores indicated absence of the

PAGE 61

61 contamination ; again, items were scored such that 0 indicated contaminated evidence and 1 indicated uncontaminated evidence . Those who read a vignette wit h no mention of contamination of evidence scored higher ( M = 2.91) on the contamination scale than identification ( M = .58), t (686) = 20.84, p < .001. Given that there were sign ificant differences between groups for all the manipulation check questions, I could safely conclude that on average, participants noticed the differences between manipulated conditions. This sample left an adequate amount of participants in each condition ; the number of participants per cell can be seen in Table 5 1. Thus, the first set of analyses reported will include all 692 participants. Table 5 1. Number of Participants per Condition in the 692 Person Sample. Controls: no eyewitness, no contamination High Quality Confession: n= 35 Low Quality Confession: n= 40 Evidence Order Confession before Eyewitness Eyewitness before Confession High Quality Confession Low Quality Confession High Quality Confession Low Quality Confession Contamination Present H igh Quality Eyewitness n = 40 n = 38 n = 37 n = 34 Low Quality Eyewitness n = 44 n = 33 n = 45 n = 39 Contamination Absent High Quality Eyewitness n = 41 n = 36 n = 38 n = 38 Low Quality Eyewitness n = 35 n = 37 n = 43 n = 39 However, given that even though as a g roup, participants noticed the differences between conditions, I also used manipulation check questions to complete a second set of analyses. That is, there is still the possibility that participants would be included in the analyses who did not notice th e differences between conditions. So, to remove the possible noise from those who may have failed manipulation check questions, I reran all

PAGE 62

62 analyses after removing those who answered more than one manipulation check question inaccurately in each of the thr ee groups. For example , there were four manipulation questions to confirm that participants noticed and remembered the quality of the confession evidence, so anyone who missed more than one of the manipulation questions on the confession scale was removed . This same pattern was repeated for the three eyewitness manipulation questions and the three contamination manipulation questions, meaning anyone who was inaccurate on more than one question was removed . Overall, 254 participants were removed due to fai led manipulation checks. Some participants failed more than one manipulation; therefore, the numbers indicated below will total more than the 254 who were ultimately removed. For example, a participant may all three of the manipulations: the confession m anipulation, the eyewitness manipulation, and the contamination manipulation (and be included in each of those counts), but would only count as one of the 254 ultimately removed. There were 1 49 people who were removed because they failed the criteria for the confession manipulations, 95 participants were removed for failing the eyewitness manipulation, and 173 participants were removed for failing the contamination manipulation c heck questions . This left a final sample of 438 people, and this group is the basis for the second set of analyses. This set of analyses may be problematic due to a lack of power in that some conditions had a relatively small number of participants compared to others , but the ANOVA test is relatively robust even with uneven cell s izes, and I used ANOVA with the majority of analyses. The sample breakdown per cell can be seen in Table 5 2.

PAGE 63

63 Table 5 2. Number of Participants per Condition in the 438 Person Sample. Controls: no eyewitness, no contamination High Quality Confession: n=2 9 Low Quality Confession: n=33 Evidence Order Confession before Eyewitness Eyewitness before Confession High Quality Confession Low Quality Confession High Quality Confession Low Quality Confession Contamination Present High Quality Eyewitness n =25 n =29 n =17 n =17 Low Quality Eyewitness n = 28 n = 18 n =27 n = 25 Contamination Absent High Quality Eyewitness n =23 n =27 n =26 n =28 Low Quality Eyewitness n =18 n =25 n =20 n =23 Scaling dependent measures Finally, before running the analyses, I c reated several scales out of the many different questions described above in the dependent measures section. I ran a principle components factor analysis with varimax rotation with all items to assess which items would hold together and generally indicate the same underlying construct. Any items that loaded above a rounded .70 on the factor were kept in the scale (i.e., the lowest value of any included item was .65) . This particular number was selected as that it is the standard teaching in our methods c ourses and commonly applied in criminological or psychological studies . Each scale was tested to ensure adequate reliability of at least any items that were reverse coded are indicated by an (R) after the statement. Scales are reported in the method section. Results were reported using a 95% confidence interval; this is st andard practice for most psychology studies. Additionally, any findings that were significant at the 0.10

PAGE 64

64 confidence level were reported as marginally significant and indicated as such with the noted p value . Control Group Comparisons To determine the ove rall effects of the various levels of evidence on the continuous verdict measures (probability of guilt and the continuous verdict rating), I included two control conditions that did not contain eyewitness testimony (one with low quality confession evidenc e and one with high quality confession evidence). The idea behind including the control conditions was that it would allow a more thorough examination of whether jurors were able to fully discount the effects of contamination (i.e., whether they would rend er guilty verdicts in contaminated conditions at similar rates compared to conditions with no eyewitness evidence). That is, if the eyewitness evidence was dependent on (or contaminated by) the confession evidence, theoretically, jurors should discount the eyewitness evidence; this allowed me to test whether they would discount the contaminated eyewitness evidence to the same levels as if it were not included in trial. Further, as an exploratory analysis, I compared the other conditions within the high qua lity confession manipulation to the control group (high quality confession with no eyewitness evidence), and the other conditions within the low quality confession manipulation to the control group (low quality confession with no eyewitness evidence). Beca use these comparisons could not be fully crossed within the factorial design (they were dangling controls), I used one way ANOVAs to analyze the data. The first analyses are conducted using the full sample of 692 participants and the second round of analys es with those who failed more than one manipulation check question removed are mentioned briefly as these results mirrored the first. I ran a one -

PAGE 65

65 way ANOVA with eight planned comparisons to simultaneously assess the differences between the experimental co nditions and the control conditions on the probability of guilt measure. For the experimental conditions, I collapsed conditions across order of evidence given that the comparison control group did not include variation in presentation of evidence (becaus e of the lack of eyewitness evidence in the control conditions). Thus, there were eight experimental groups (varying the quality of the confession (low, high), the quality of the eyewitness (low, high) and evidence contamination (present, absent) and two c ontrol groups (high quality confession evidence with no eyewitness evidence and low quality confession evidence with no eyewitness evidence). The eight planned comparisons in the ANOVA compared each of the experimental conditions to the appropriate contr ol condition; high quality confession evidence conditions were always compared to the high quality confession evidence control, and low quality confession evidence conditions were always compared to the low quality confession evidence control. Results show the experimental conditions and the control groups, F (9, 688) = 4.11, p < .01. First, I will report the comparisons within the high quality confession conditions. Participants in conditi ons with a high quality confession, a low quality eyewitness, and contamination rated the probability of guilt marginally lower than those participants in the control comparison (with only a high quality confession and no mention of the eyewitness), p < .1 0. Participants in conditions with 1) a high quality confession, a high quality eyewitness, and no contamination, ( p = .15) 2) a high quality confession, a high quality eyewitness, and contamination ( p = .20) and 3) a high quality confession, a low qualit y

PAGE 66

66 eyewitness, and no contamination ( p = .13) were not significantly different from participants in the high quality confession control condition on this measu re. All means are reported in T able 5 3 . Thus, it appears that if the confession evidence was hi gh quality, the presence and quality of the eyewitness does not necessarily matter. In fact, the presence of a low quality, contaminated eyewitness actually caused participants to rate the probability of guilt lower than if there was no eyewitness (althoug h this effect was only marginally significant). That is, it seems as though having a low quality contaminated eyewitness is hurting the prosecution even more so than not presenting the eyewitness at all if there is high quality confession evidence to rely on. Here, however, it is unclear if contamination or the quality of the eyewitness was driving the difference in decisions, especially given this was the only (marginally) significant finding. It could be the totality of the quality of the eyewitness evid ence (i.e., the fact that it was both low quality and contaminated) resulted in the (marginally) significant difference. Results for the second round of analyses with the manipulation check failures removed for the high quality comparisons showed no signi ficant differences (all p thus are not reported here to save on repetition. Table 5 3 . Means and Confidence Intervals for Planned Comparisons to High Quality Confession Control Group on Probability of Guilt Measure. Conditions Mean Confidence Interval High quality confession, high quality eyewitness, no contamination 47.03 39.74 54.31 High quality confession, high quality eyewitness, contamination 47.97 40.60 55.35 High quality confession, Low quality eyewitness, no contamination 46.46 3 8.47 54.45 High quality confession, Low quality eyewitness, contamination 45.44 a 38.79 52.09 Control: high quality confession, no eyewitness, no contamination 56.20 a 44.73 67.67 Subscript a indicates significance at the .10 level

PAGE 67

67 ss the results for the planned comparisons in the ANOVA comparing the experimental conditions to the control in low quality confession low quality confession control gr oup and the experimental conditions. Specifically, participants in the conditions with 1) a low quality confession, a high quality eyewitness, and no contamination, ( p = .01), and 2) a low quality confession, a high quality eyewitness, and contamination, (p = .05) were both likely to rate the probability of guilt higher than participants in the control group with only the low quality confession. Participants in the conditions with 1) a low quality confession, a low quality eyewitness, and no contamination ( p =.33), and 2) a low quality confession, a low quality eyewitness and contamination ( p = .33), were not significantly different from participants in the low quality confession control group on this measure. The high quality eyewitness evidence (even wh en it is contaminated) is able to increase probability of guilt assignments even with low quality confession eviden ce. All means are reported in T able 5 4 . So here, it whe ther the eyewitness is contaminated. That is, if the quality of the confession evidence is low, then the quality of the eyewitness evidence may be more likely to affect contaminatio n does not seem to be factoring into the jurors decisions appropriately. Results for the second round of analyses with the manipulation check failures removed mirrored these results for the low quality comparisons, and thus are not reported here to save o n repetition.

PAGE 68

68 Table 5 4 . Means and Confidence Intervals for Planned Comparisons to Low Quality Confession Control Group on Probability of Guilt Measure. Conditions Mean Confidence Interval Low quality confession, high quality eyewitness, no contamination 42.18 a 34.74 49.61 Low quality confession, high quality eyewitness, contamination 39.22 b 31.95 46.49 Low quality confession, low quality eyewitness, no contamination 32.83 26.01 39.65 Low quality confession, low quality eyewitness, contamination 3 2.96 26.71 39.21 Control: Low quality confession, no eyewitness, no contamination 26.90 a b 18.71 35.09 Subscripts a b indicates significance at the .05 level To test the same comparisons for the continuous verdict measure, I also ran an ANOVA comparing the eight experimental groups to the two control groups using the same planned comparisons (always comparing the low quality confession experimental conditions to the low quality control group and the high quality confession experimental conditions to the high quality control group) using the continuous verdict measure as the dependent variable and found somewhat similar results, F (9, 690) = 3.97, p < .01. Again, first I will report results for the high quality confession planned comparisons. Participan ts in conditions with 1) a high quality confession, a low quality eyewitness, and no contamination ( p < .10) rated the probability of guilt marginally lower than the control comparison (with only a high quality confession and no mention of the eyewitness). Participants in conditions with 1) a high quality confession, a high quality eyewitness, and no contamination, ( p = .25), 2) a high quality confession, a high quality eyewitness, and contamination ( p = .36), and 3) conditions with a high quality confessi on, a low quality eyewitness, and contamination were not significantly different from the control condition with only a high quality confession ( p = .15) did not significantly differ from participants in the control condition (depicting only a high quality

PAGE 69

69 confession and no mention of the eyewitness) on this meas ure. All means are reported in T able 5 5 . Thus, similar to the above analysis, it appears as though the presence of than not hearing mention of the eyewitness at all. Again, contamination of the eyewitness seems to not be a consideration in verdict ratings (or at least not used appropriately). Table 5 5 . Means and Confidence Intervals for Planned Comparisons to High Q uality Confession Control Group on Continuous Verdict Measure. Conditions Mean Confidence Interval High quality confession, high quality eyewitness, no contamination 2.96 2.59 3.33 High quality confession, high quality eyewitness, contamination 3.04 2. 64 3.44 High quality confession, Low quality eyewitness, no contamination 2.71 a 2.34 3.09 High quality confession, Low quality eyewitness, contamination 2.88 2.52 3.23 Control: high quality confession, no eyewitness, no contamination 3.34 a 2.72 3.97 Subscript a indicates significance at the .10 level In examining the planned comparisons for the low quality confession group, participants in conditions with a low quality confession, a high quality eyewitness, and contamination, ( p < .05) rated th e probability of guilt higher than participants in the control group with only the low quality confession. There were not significant differences conditions with a low quality confession, a high quality eyewitness, and no contamination, ( p = .22), 2) the condition with a low quality confession, a low quality eyewitness, and no contamination ( p =.75), and 3) a low quality confession, a low quality eyewitness and contamination ( p = . 70). All means are reported in T able 5 6 . Again, similar to the above analysis, the high quality eyewitness seemed to make a

PAGE 70

70 contaminated eyewitness. Again, this indi cates that participants are not correcting for contamination. Results for the second analysis with people who failed the manipulation check questions removed showed no significant results for comparisons to either the high or low quality confession contro l group (all p Table 5 6 . Means and Confidence Intervals for Planned Comparisons to Low Quality Confession Control Group on Continuous Verdict Measure. Conditions Mean Confidence Interval Low quality confession, high quality eyewitness, no conta mination 2.54 2.15 2.93 Low quality confession, high quality eyewitness, contamination 2.82 a 2.41 3.22 Low quality confession, Low quality eyewitness, no contamination 2.25 1.94 2.56 Low quality confession, Low quality eyewitness, contamination 2. 03 1.72 2.34 Control: Low quality confession, no eyewitness, no contamination 2.15 a 1.68 2.62 Subscript a indicates significance at the .05 level I also conducted a series of ANOVAs to test these planned comparisons comparing the experimental condit ions to the control group for two of the four scales created from rest of the dependent measures: the confession strength and general weighing of confession evidence during trial scales. Given that the participants in the control conditions did not read ab out an eyewitness, we did not analyze the data from First, for the confession strength scale, results again showed significant differences between our factorial conditions and the control g roups, F (9, 686) = 6.03, p < .01. For those planned comparisons comparing the high quality confession evidence experimental conditions to the high quality confession evidence control condition, participants in conditions with 1) a high quality confession , a low quality eyewitness, and no contamination ( p < .05), 2) a high quality confession, a low quality eyewitness, and

PAGE 71

71 contamination ( p < .10) rated the confession strength lower than the control comparison. Participants in conditions with 1) a high quali ty confession, a high quality eyewitness, and no contamination, ( p = .11) 2) a high quality confession, a high quality eyewitness, and contamination ( p > .10) were not significantly different from the control conditions (depicting only a high quality confe ssion and no mention of the eyewitness). There were no significant difference between the factorial conditions and control conditions for the low quality confession groups (all p 16). All means are reported in T able 5 7 . As with the previous results, it appears as though presenting a low quality eyewitness is this additional (p roblematic) evidence is also lowering the evaluations of the other evidence (even when that evidence is actually high quality). Again, participants do not seem to be correcting for the contamination potential either. Results for the second analyses witho ut people who failed with manipulation checks again showed no significant results (all p Table 5 7 . Means and Confidence Intervals for Planned Comparisons to High and Low Quality Confession Control Group s on Confession Strength Scale. Condition s Mean Confidence Interval High quality confession, high quality eyewitness, no contamination 3.59 3.31 3.87 High quality confession, high quality eyewitness, contamination 3.58 3.29 3.88 High quality confession, Low quality eyewitness, no contamina tion 3.38 b 3.07 3.69 High quality confession, Low quality eyewitness, contamination 3.56 a 3.32 3.81 Control: high quality confession, no eyewitness, no contamination 3.99 ab 3.59 4.39 Low quality confession, high quality eyewitness, no contaminatio n 3.04 2.75 3.33

PAGE 72

72 Table 5 7 . Continued Conditions Mean Confidence Interval Low quality confession, high quality eyewitness, contamination 3.10 2.79 3.41 Low quality confession, low quality eyewitness, no contamination 2.83 2.61 3.06 Low quality confession, low quality eyewitness, contamination 2.99 2.71 3.27 Control: low quality confession, no eyewitness, no contamination 2.77 2.51 3.03 Subscript a indicates significance at the .10 level, Subscript b indicates significance at the .05 level Results from the second ANOVA testing the effect of the planned comparisons weighing of the confession evidence did not show any significant differences between our factorial conditions and the control groups on the general weighing of confession evidence during trial scale, p = .38. Means and confidence intervals can be seen in Table 5 8 below. Results from the second round of analyses with the manipulation check failures removed mirrored these results (all p .30). Ta ble 5 8 . Means and Confidence Intervals for Planned Comparisons to High and Low Quality Confession Control Group s on General Weighing of Confession Evidence During Trial Scale. Conditions Mean Confidence Interval High quality confession, high quality ey ewitness, no contamination 3.79 3.54 4.04 High quality confession, high quality eyewitness, contamination 3.80 3.53 4.08 High quality confession, Low quality eyewitness, no contamination 4.03 3.79 4.27 High quality confession, Low quality eyewitne ss, contamination 3.99 3.75 4.23 Control: high quality confession, no eyewitness, no contamination 3.94 3.54 4.35 Low quality confession, high quality eyewitness, no contamination 4.04 3.78 4.30 Low quality confession, high quality eyewitness, con tamination 3.93 3.68 4.18 Low quality confession, low quality eyewitness, no contamination 3.68 3.46 3.90

PAGE 73

73 Table 5 8 . Continued Conditions Mean Confidence Interval Low quality confession, low quality eyewitness, contamination 4.01 3.79 4.24 Con trol: low quality confession, no eyewitness, no contamination 3.70 3.42 3.98 Does Eyewitness Quality, Confession Quality and Contamination of Evidence For the next set of analyses, I excluded the control conditio ns to allow an exploration of the hypothesized interaction effects. In addition, I ran two sets of analyses. The first set of analyses examined the full set of participants (based on the manipulation check result that groups significantly differed in perce iving the various manipulations). The second set of analyses excluded those who failed more than one manipulation check per manipulation (as described above). a backwards, ste pwise logistic regression . Because there were competing hypotheses for this study, I chose to use this type of regression to ensure the best possible fitting model of the data. Given the competing hypotheses, the study could be considered more exploratory in nature. Thus, the backwards, stepwise procedure is likely the best type of analysis to ensure the final model is the best fitting model for the data. C onfession quality, eyewitness quality, contamination presence, and evidence order were included as in dependent variables in the first step. All possible interactions were included in the second step. The low quality and contamination absent conditions were coded as comparison groups. 2 ( 2 5) = 7 . 8 9, p < .05, R 2 = .0 2 , and included one significant main effect and one marginally significant

PAGE 74

74 selection and a marginally significant main effect of ey ewitness quality on participant s verdict sele ction (see Table 5 9 ); participants who read a trial vignette containing a high quality confession were more likely to render a guilty verdict (28%) than those who read a vignette depicting a low quality confession (21.1%) . Similarly, participants who rea d about a high quality eyewitness (28%) were marginally more likely to render a guilty verdict than those who read about a low quality eyewitness (21.1%) . All other main effects and interactions were not significant. Table 5 9 . Logistic regression examin ing evidence quality and contamination on dichotomous verdict selection (Guilty/Not Guilty) with n=692. Variable St. Error Sig. Confession Quality . 40 ** .1 9 4.33 .0 37 1. 48 Eyewitness Quality .36* .19 3. 58 .05 9 1.43 ** indicates significance (.05) , * indicates marginal significance (.10) For the second set of analyses with manipulation check fai lures removed, I ran 2 ( 1 ) = 27.15 , p < .01, R 2 = .1 1 , and it included one main effect. There was a main effect n (see T able 5 10 ); participants who read a trial vignette containing a high quality confession were more likely to render a guilty verdict (34.0%) than those who read a vignette depicting a low quality confession (10.7%) . There was no difference found be tween eyewitness quality levels, presence or absence of the contamination, nor the order evidence was presented. Table 5 10 . Logistic regression examining evidence quality and contamination on dichotomous verdict selection (Guilty/Not Guilty) with n=438. V ariable St. Error Sig. Confession Quality 1. 34 * .2 7 24.48 .000 3.83 * indicates significance (.05)

PAGE 75

75 each person to respond on a 1 6 Likert type scale (ra nging from strongly disagree to independent variables (confession quality, eyewitness quality, contamination presence, and evidence order) as well as their interactions on this dependent continuous verdict question, I conducted an ANOVA. The same comparison groups were used for this analysis as well: the low quality evidence conditions and the absent contamination conditions. Again, two main effects were found. First, confession quality significantly affected F (1, 6 16 ) = 13.59 , p < .01, 2 = .0 2 . In addition, eyewitness quality o n participants responses to the continuous verdict item, F (1,6 16 ) = 8.28 , p <.01, 2 = .01. Participants who read a trial vignette depicting a high quality confession ( M = 2. 8 9 , standard error = .09 , CI = 2.71 3.06 ) were more likely to agree with the st atement than participants who read about a low quality confession ( M = 2. 41 , standard error = .09 , CI = 2.22 2.59 ). Participants who read a trial vignette depicting a high quality eyewitness ( M = 2.83, standard error = .09 , CI = 2.65 3.02 ) were more l ikely to agree with the statement than participants who read about a low quality eyewitness ( M = 2.46, standard error = .09 , CI = 2.28 2.64 ). There was also a marginally significant interaction between confession quality, eyewitness quality, and eviden ce order, F (1,6 16 ) = 3.07 , p <.1 0 , 2 = .01 . Here, participants who read about a high quality confession and a high quality eyewitness (in M = 3.23, standard error = .18, CI = 2.87

PAGE 76

76 3.58) than those who re ad about a high quality eyewitness and a high quality confession (in that order; M = 2.76, standard error = .19, CI = 2.87 3.58). All other comparisons within this interaction were not significant. There were no differences between contamination present or absent conditions or other interactions , all > .10. The second set of analyses with manipulation check failures removed also showed similar results. One main effect was found for confession quality on participants responses to the continuous verdict item, F (1, 375 ) = 37.08 , p < .01, 2 = . 09 . Participants who read a trial vignette depicting a high quality confession ( M = 3.0 5 , standard error = .1 2, CI = 2.81 3.28 ) were more likely to agree with the statement than participants who read about a low quality confession ( M = 2.04, standard error = .1 2, CI = 1.82 2.27 ). There were no differences between eyewitness quality level s , contamination present or absent conditions or evidence presentation order, all > .10. While verdict is an important co nsideration for trial vignette studies, there are other concepts that may be able to show differences in the way participants think about the evidence that should be examined also. To that end, I included another question to ty of guilt perceptions that the offender had actually committed the crime. For the probability of guilt question, I conducted a second ANOVA to assess the impact of the quality of confession and eyewitness evidence, the presence or absence of contaminat ion, and th e evidence order on participant s s probability of guilt. During the post trial questionnaire, participants were presented with a slider bar to indicate the percentage (0% is the probability that you believe the defendant is actually guilty of the charges brought

PAGE 77

77 as the comparison groups. One significant main effect and one marginally signifi cant main effect was found. Confession quality significantly impacted probability of guilt assigned to the defendant, F (1, 6 14 ) = 14.72 , p < .01, 2 = .0 2 . Participants who read a trial vignette depicting a high quality confession assigned a higher probability of guilt ( M = 46.48 , standard error = 1. 7 6 , CI = 43.02 49.94 ) than participants who read about a low quality confession ( M = 36.71 , standar d error = 1. 84, CI = 33.10 40.33 ). Eyewitness quality marginally impacted the probability of guilt assigned to the defendant, F (1, 6 14 ) = 3. 48 , p < .10, 2 = .01. Participants who read a trial vignette depicting a high quality eyewitness were marginal ly more likely to assign a higher probability of guilt ( M = 43.97, standard error = 1.80 , CI = 40.41 47.53 ) than participants who read about a low quality eyewitness ( M = 39.22, standard error = 1.78 , CI = 35.71 42.73 ). There was one marginally signif icant interaction between the eyewitness quality and the evidence order, F (1, 6 14 ) = 3. 80 , p < .10, 2 = .01 . Participants who read about a low quality eyewitness rated the defendants probability of guilt higher in the eyewitness then confession condition ( M = 41.26, standard error = 2.47, CI = 36.41 46.10) than in the confession then eyewitness condi tion ( M = 37.18, standard error = 2.59, CI = 32.10 42.27). The opposite pattern held for the high quality eyewitness conditions: those who heard about the confession then the eyewitness ( M = 46.90, standard error 2.53, CI = 41.93 51.86) rated the prob ability of guilt higher than those who heard about the eyewitness then the confession ( M = 41. 05, standard error = 2.60, CI = 35.95 46.14).

PAGE 78

78 There were no differences between contamination present or absent conditions nor other interactions , all > .10. The second set of analyses with the manipulation check failures removed showed similar results for the probability of guilt analysis. One significant main effect and one marginally significant main effect w ere found. Confession quality significantly impacted the probability of guilt assigned to the defendant, F (1, 375 ) = 29.28 , p < .01, 2 = . 0 8 . Participants who read a trial vignette depicting a high quality confession assigned a higher probability of guilt ( M = 50. 1 4, standard error = 2. 33, CI = 45.56 54.73 ) than participants who read about a low quality confession ( M = 32.52 , standar d error = 2. 28, CI = 28.05 37.00). Evidence presentation order marginally impacted the probability of guilt assigned to the defendant, F (1, 375 ) = 3.3 1 , p < .10, 2 = .01. Participants who read a trial vignette depicting the confession before the eyewi tness were more likely to assign a higher probability of guilt ( M = 44.29, standard error = 2.2 7, CI = 39.84 48.75 ) than participants who read about an eyewitness and then a confession ( M = 38.37, standard error = 2.3 4, CI = 33.77 42.98). There was one significant interaction between the eyewitness quality and the evidence order, F (1, 375 ) = 4.38 , p < .0 5 , 2 = .01 . Participants who read about a high quality eyewitness conditions rated the probability of guilt higher in the confession then the eyewitness conditions ( M = 50.09, standard error 3.05 , CI = 44.09 56.09) compared to those who heard about the ey ewitness then the confession ( M = 37.36, standard error = 3.39, CI = 30.68 44.03). There was no such difference in the low

PAGE 79

79 quality eyewitness group . There were no differences between eyewitness quality levels or contamination present or absent conditio ns, all > .10. Does the confession quality, eyewitness qual ity, or contamination presence affect p articipan vidence? Beyond verdict differences, the differences in the independent variables could also have impacted partic weight during trial and strength. Therefore, I ran a MANOVA to test the effect of confession quality, eyewitness quality, contamination presence, evidence order, and all possible to the four scales developed and described in the data cleaning section : confession strength, eyewitness strength, general weighing of confession evidence during trial , and specific weighing of eyewitness evidence in this case . These scales were all inclu ded in the same analyses because they assess similar concepts about the relative value of the evidence and how each piece should be weighed in making court decisions. Two main effects were found. There was an overall significant effect of confession quali ty on the scales, = . 92 , F (4, 585 ) = 13.66 , p < .01, 2 = . 09 . At the scalar level, I found a main effect for confession quality on t wo of the four scales (Table 5 11 for means and confidence intervals). Overall, there was an effect of confession quality on participant F (1, 6 04 ) = 42.87 , p < .01, 2 = .0 5 . Post hoc tests using LSD showed participants who read a high quality confession rated the confession strength higher than participants who read a low quality confession. evaluations of specific weighing of eyewitness evidence in this case , F (1, 6 04 ) = 8.02 , p < .01, 2 = .01. Post hoc tests using LSD showed participants who read a high quality

PAGE 80

80 confession rated the weight during trial of the eyewitness as lower than those who read about a low quality confession. The effects of confession quality on the other two scales were not significant. Table 5 11 . Main effects of confession quality on parti intervals) of confession strength, general weighing of confession evidence during trial , eyewitness strength, and specific weighing of eyewitness evidence in this case with n=692 . Scale High Quality Confession Low Quality Confession Confession Strength 3.5 1 (3. 38 3. 65 ) * 2.9 8 (2.8 4 3. 12 ) * General Weighing of Confession Evidence During Trial 3.91 (3.78 4.03) 3.91 (3.79 4.04) Eyewitness Strength 2.83 (2.71 0 2.94) 2.80 (2.68 2.91) Specific Weighing of Eyewitness Evidence In this Case 4. 43 (4. 30 4. 55 ) * 4. 66 (4. 53 4. 79 ) * * indicates significance (.05) There was also an overall significant effect of eyewitness quality on the scales, = .8 6 , F (4, 585 ) = 24.51 , p < .01, 2 = .1 4 . At the scalar level, I found a main effect for eyewitness quality on t wo of the four scales (Table 5 1 2 for means and confidence intervals). Overall, there was an overall main effect of eyewitness quality o n F (1, 6 04 ) = 63.09 , p < .01, 2 = . 10 . Post hoc tests using LSD showed participants who read a high quality eyewitness rated the strength of the eyewitness as higher than those who read about a low qualit y eyewitness . specific weighing of eyewitness evidence in this case , F (1, 6 04 ) = 8.96 , p < .01, 2 = .01. Post hoc tests using LSD showed participants who read a high qua lity eyewitness rated the weight during trial of the eyewitness as lower than those who read about a low quality eyewitness. There was no effect of eyewitness quality on the other two scales.

PAGE 81

81 Table 5 1 2 . Main effects of eyewitness quality on mean ratings (confidence intervals) of confession strength, general weighing of confession evidence during trial , eyewitness strength, and specific weighing of eyewitness evidence in this case with n 692 . Scale High Quality Eyewitness Low Quality Eyewitne ss Confession Strength 3.31 (3.17 3.45) 3.18 (3.05 3.32) General Weighing of Confession Evidence During Trial 3.88 (3.76 4.01) 3.94 (3.82 4.06) Eyewitness Strength 3.14 (3.0 2 3.2 5 ) * 2.49 (2.37 2.6 0 ) * Specific Weighing of Eyewitness Evidence In this Case 4.42 (4.29 4.55) * 4.67 (4.54 4.79) * * indicates significance (.05) There were no significant effects found for contamination presence, the order of evidence presentation, or the interaction of the independent variables on the scaled res ponses, all > .10. I additionally ran each scale through an ANOVA as well to ensure that my MANOVA results were correct in case the four scales might not be tapping the same general concept of evidence evaluation. The four ANOVA results mirrored the overall MANOVA results, indicating they are representative of the overall evidence evaluation concept, so they are not reported here to avoid repetition. In the second set of analyses with those who failed the manipulation check questions removed, there we re again two main effects found. There was an overall significant effect of confession quality on the scales, = .82, F (4, 352) = 20.02, p < .01, 2 = .19. At the scalar level, I found a main effect for confession quality on three of the four scales (T able 5 1 3 for means and confidence intervals). Overall, there was an F (1, 371) = 83.72, p < .01, 2 = .14. Post hoc tests using LSD showed participants who read a hi gh quality confession rated the confession strength higher than participants who read a low quality confession. There was a overall main effect of confession

PAGE 82

82 F (1, 371) = 6.84, p < .05, 2 = .02 . Post hoc tests using LSD showed participants who read a high quality confession rated the general weighin g of confession evidence during trial scale higher than participants who read a low quality confession. There was an overall main effect of confess specific weighing of eyewitness evidence in this case , F (1, 371) = 8.07, p < .05, 2 = .02. Post hoc tests using LSD showed participants who read a high quality confession rated the weight during trial of the eyewitness as lower than those who read about a low quality confession. The effect of confession quality on the specific weighi ng of eyewitness evidence in this case scale was not significant. Table 5 1 3 intervals) of confession strength, general weighin g of confession evidence during trial , eyewitness s trength, and specific weighing of eyewitness evidence in this case with n=438 . Scale High Quality Confession Low Quality Confession Confession Strength 3.75 (3.6358 3.93)* 2.79 (2.61 2.96)* General Weighing of Confession Evidence During Trial 3.99 (3 .83 4.15)* 3.89 (3.73 4.04)* Eyewitness Strength 2.85 (2.71 2.99)* 2.60 (2.46 2.73* Specific Weighing of Eyewitness Evidence In this Case 4. 52 (4.36 4.68) 4.85 (4.69 5.00) * indicates significance at the .05 level. There was also an overall significant effect of eyewitness quality on the scales, = .8379, F (4, 352) = 22.79, p < .01, 2 = .21. At the scalar level, I found a main effect for eyewitness quality on two of the four scales (Table 5 1 4 for means and confidence intervals). Overall, there was an overall main effect of eyewitness quality on F (1, 371) = 43.46, p < .01, 2 = .12. Post hoc tests using LSD showed participants who read a high quality eyewitness rated

PAGE 83

83 the strength of the eyewitness as higher than those who read about a low qual ity weight during trial , F (1, 371) = 7.00, p < .05, 2 = .02. Post hoc tests using LSD showed participants who read a high quality eyewitness rate d the weight during trial of the eyewitness as lower than those who read about a low quality eyewitness. There was no effect of eyewitness quality on the other two scales. Table 5 1 4 dence intervals) of confession strength, general weighing of confession evidence during trial , eyewitness strength, and specific weighing of eyewitness evidence in this case with n=438 . Scale High Quality Eyewitness Low Quality Eyewitness Confession Stren gth 3.26 (3.08 3.43) 3.28 (3.11 3.46) General Weighing of Confession Evidence During Trial 3.88 (3.73 4.04) 4.00 (3.84 4.16) Eyewitness Strength 3.07 (2.94 3.22)* 2.38 (2.23 2.52)* Specific Weighing of Eyewitness Evidence In this Case 4.55 ( 4.39 4.70)* 4.83 (4.67 4.98)* * indicates significance at the .05 level. Again, there were no significant effects found for contamination presence, the order of evidence presentation, or the interaction of the independent variables on the scaled resp onses, all > .10. Given the particular interest in testing whether jurors are able to identify and appropriately discount contaminated or interdependent evidence, I created two tables to fully untangle all levels of effects for the contamination variab le. More specifically, Tables 5 15 and 5 16 show means and confidence intervals for all the dependent measures discussed in this section: dichotomous verdict, continuous verdict, and probability of guilt are depicted in Table 5 15 for both the 692 sample as well as the 438 sample groups (as labeled). Further, the four scales that were created from additional

PAGE 84

84 dependent measures, Confession Strength, Eyewitness Strength, General Weighing of Confession Evidence During Trial, and Specific Weighing of Eyewitne ss Evidence in this Case , are depicted in Table 5 16 (again with both the 692 sample and the 438 sample). For both tables, t he Contamination and No Contamination rows depict main effects o f only contamination presence. T he High and Low Quality Confession rows depict two way interactions between the contamination presence and the quality of the confe ssion. T he High and Low Quality Eyewitness rows depict a three way interaction between contamination presence, the quality of the confession and the quality o f the eyewitness. As shown, none of the main effects or interactions were significant for any of the dependent measures s > .10).

PAGE 85

85 Table 5 1 5 . Mean Values ( Confidence Intervals ) for Main Effect of Contamination Presence, and Interaction Effects betw een 1) Contamination Presence and Quality of Confession Evidence, and 2) Contamination Presence, Quality of Confession Evidence, and Quality of Eyewitness Evidence on Dependent Measure Condition Verdict: G or NG N = 692 Verdict: G or N G N = 438 Verdict 1 6 N = 692 Verdict 1 6 N = 438 Probability of Guilt N = 692 Probability of Guilt N = 438 Contamination (all) 24.8 (20.0 29.7) 20.9 (14.9 26.8) 2.68 (2.50 2.86) 2.54 (2.31 2.77) 41.25 (37.73 44.78) 39.80 (35.22 44.37) * Hi gh Quality Confession (all) 28.2 (21.6 34.8) 32.3 (24.1 40.6) 2.95 (2.70 3.20) 3.12 (2.80 3.44) 46.58 (41.77 51.38) 48.62 (42.31 54.93) ** High Quality Eyewitness 29.5 (19.8 39.1) 31.9 (19.4 44.4) 3.02 (2.66 3.39) 3.20 (2.71 3.68) 47.71 (40.67 54.75) 49.60 (40.02 59.18) ** Low Quality Eyewitness 27.0 (18.0 36.0) 32.7 (22.0 43.5) 2.88 (2.54 3.21) 3.04 (2.62 3.45) 45.45 (38.90 51.99) 47.63 (39.41 55.85) * Low Quality Confession (all) 21.4 (14.4 28.5) 9.4 (0.8 18.1) 2.42 (2.15 2.68) 1.96 (1.62 2.29) 35.92 (30.77 41.08) 30.98 (24.36 37.60) ** High Quality Eyewitness 27.9 (17.9 37.9) 12.1 ( 0.1 24.2) 2.82 (2.44 3.19) 2.22 (1.75 2.69) 39.16 (31.87 46.45) 33.01 (23.70 42.32) ** Low Quality Eyewitness 15.0 (5.0 25.1) 6.8 ( 5.5 19.1) 2.01 (1.64 2.39) 1.69 (1.22 2.17) 32.69 (25.39 39.99) 28.95 (19.53 38.37) No Contamination (all) 24.2 (19.4 29.1) 23.5 (17.7 29.4) 2.61 (2.43 2.79) 2.55 (2.32 2.78) 41.94 (38.40 45.49) 42.87 (38.39 47.35) * High Quality Confession (all) 27.8 (21.0 34.6) 34.7 (26.1 43.4) 2.82 (2.57 3.08) 2.98 (2.64 3.31) 46.38 (41.41 51.35) 51.67 (45.03 58.32) ** High Quality Eyewitness 29.0 (19.5 38.6) 32.5 (21.1 43.9) 2.96 (2.60 3.32) 3.03 (2.59 3.48) 46.86 (39.91 53.81) 54.12 (45.40 62.84) ** Low Quality Eyewitness 26.6 (16.8 36.3) 36.9 (24.0 49.9) 2.69 (2.32 3.05) 2.92 (2.41 3.42) 45.90 (38.80 53.01) 49.22 (39.20 59.25) * Low Quality Confession (all) 20.7 (13.8 27.6) 12.3 (4.5 20.2 ) 2.40 (2.14 2.65) 2.13 (1.82 2.43) 37.50 (32.44 42.56) 34.07 (28.04 40.09)

PAGE 86

86 Table 5 15 . Continued Condition Verdict: G or NG N = 692 Verdict: G or NG N = 438 Verdict 1 6 N = 692 Verdict 1 6 N = 438 Probability of Guilt N = 692 Probability of Gu ilt N = 438 ** High Quality Eyewitness 25.6 (15.7 35.4) 16.3 (5.6 27.1) 2.54 (2.17 2.90) 2.20 (1.78 2.61) 42.16 (34.98 49.34) 38.16 (29.94 46.38) ** Low Quality Eyewitness 15.8 (6.1 25.5) 8.3 ( 3.2 19.8) 2.23 (1.89 2.62) 2.06 (1.61 2.5 0) 32.84 (25.72 39.97) 29.97 (21.17 38.78) Table 5 1 6 . Mean Values and Confidence Intervals for Main Effect of Contamination Presence, Two way Interaction between Contamination Presence and Quality of Confession Evidence, and Three way Interaction be tween Contamination Presence, Quality of Confession Evidence, and Quality of Eyewitness Evidence on Scaled Dependent Measures. Dependent Measure Confession Strength Eyewitness Strength General Weighing of Confession Evidence During Trial Specific Weighi ng of Eyewitness Evidence In this Case n=692 n=438 n=692 n=438 n=692 n=438 n=692 n=438 Contamination (all) 3.29 (3.15 3.43) 3.26 (3.08 3.43) 2.82 (2.70 2.93) 2.68 (2.54 2.82) 3.94 (3.82 4.06) 3.94 (3.78 4.10) 4.55 (4.42 4.67) 4.73 (4.57 4.89) * High Quality Confession (all) 3.56 (3.37 3.75) 3.76 (3.51 4.01) 2.85 (2.70 3.01) 2.84 (2.65 3.04) 3.91 (3.75 4.08) 3.98 (3.76 4.20) 4.42 (4.25 4.59) 4.58 (4.36 4.80) ** High Quality Eyewitness 3.56 (3.29 3.84) 3.75 (3.38 4.13) 3 .11 (2.88 3.34) 3.16 (2.86 3.46) 3.84 (3.59 4.08) 3.93 (3.59 4.27) 4.32 (4.07 4.57) 4.65 (4.31 4.99) ** Low Quality Eyewitness 3.56 (3.30 3.81) 3.76 (3.45 4.08) 2.60 (2.39 2.81) 2.53 (2.28 2.78) 3.99 (3.76 4.21) 4.03 (3.74 4.31) 4.5 2 (4.29 4.75) 4.51 (4.23 7.80) * Low Quality Confession (all) 3.02 (2.82 3.22) 2.75 (2.50 3.01) 2.78 (2.61 2.95) 2.51 (2.31 2.71) 3.97 (3.79 4.15) 3.91 (3.68 4.14) 4.67 (4.48 4.85) 4.88 (4.65 5.11)

PAGE 87

87 Table 5 16 . Continued Confession Str ength Eyewitness Strength General Weighing of Confession Evidence During Trial Specific Weighing of Eyewitness Evidence In this Case n=692 n=438 n=692 n=438 n=692 n=438 n=692 n=438 ** High Quality Eyewitness 3.05 (2.76 3.34) 2.67 (2.31 3.02) 3.10 (2. 86 3.34) 2.76 (2.47 3.04) 3.89 (3.64 4.14) 3.80 (3.48 4.12) 4.47 (4.21 4.73) 4.68 (4.36 5.00) ** Low Quality Eyewitness 2.99 (2.70 3.27) 2.84 (2.48 3.20) 2.46 (2.22 2.70) 2.26 (1.98 2.55) 4.05 (3.80 4.30) 4.02 (3.69 4.34) 4.86 (4.60 5.12) 5.07 (4.75 5.40) No Contamination (all) 3.20 (3.07 3.34) 3.29 (3.11 3.46) 2.81 (2.69 2.92) 2.77 (2.64 2.91) 3.88 (3.76 4.00) 3.94 (3.78 4.09) 4.54 (4.42 4.67) 4.64 (4.49 4.80) *High Quality Confession (all) 3.47 (3.28 3.66) 3 .75 (3.50 4.01) 2.80 (2.64 2.96) 2.87 (2.67 3.07) 3.90 (3.73 4.07) 4.01 (3.78 4.23) 4.43 (4.26 4.61) 4.47 (4.24 4.70) **High Quality Eyewitness 3.58 (3.31 3.86) 3.80 (3.47 4.14) 3.18 (2.95 3.40) 3.32 (3.06 3.59) 3.78 (3.54 4.02) 3. 89 (3.59 4.19) 4.29 (4.05 4.54) 4.24 (3.94 4.54) **Low Quality Eyewitness 3.36 (3.08 3.63) 3.70 (3.31 4.09) 2.43 (2.19 2.66) 2.41 (2.11 2.71) 4.02 (3.78 4.27) 4.12 (3.77 4.46) 4.57 (4.32 4.82) 4.70 (4.35 5.05) *Low Quality Confessio n (all) 2.94 (2.74 3.14) 2.82 (2.59 3.05) 2.82 (2.65 2.98) 2.68 (2.50 2.86) 3.86 (3.68 4.03) 3.87 (3.66 4.08) 4.65 (4.47 4.83) 4.82 (4.61 5.03) **High Quality Eyewitness 3.04 (2.76 3.32) 2.80 (2.48 3.13) 3.16 (2.93 3.40) 3.06 (2.80 3.31) 4.02 (3.77 4.27) 3.91 (3.62 4.20) 4.59 (4.34 4.85) 4.61 (4.32 4.90) **Low Quality Eyewitness 2.84 (2.56 3.11) 2.84 (2.50 3.17) 2.47 (2.24 2.69) 2.30 (2.04 2.57) 3.69 (3.45 3.93) 3.83 (3.53 4.14) 4.71 (4.46 4.95) 5.02 (4.72 5 .33)

PAGE 88

88 Testing for Covariates Finally, after running the hypothesized analyses reported above, I re ran all analyses a third time while including several covariates, specifically race and gender of the participant. In this way, I could ensure the random assignment to condition was completed correctly, and that there were no results or effects that were dependent on the group of participants I had in this study. There was a main effect of the race of the participant significantly in the MANOVA analysis re ported above, = .95, F (4, 580) = 7.77, p < .01, 2 evaluations of the general weighing of confession evidence during trial , F (1, 602) = 12.60, p < .01, 2 = .02. Those who identified as As ian were likely to rate the weight during trial as lower than any other racial group (Table 5 17 for means and confidence intervals). There were no significant differences between any other racial comparisons. The race of the participant also evaluation of the specific weighing of eyewitness evidence in this case , F (1, 602) = 5.21, p < .05, 2 = .01. Again, those who identify as Asian were likely to rate the weight during trial as lower than any other racial group; additionally Blacks were marginally weight during trial as lower than Whites. Finally, the race of the participant also affected the participants evaluations of the eyewitness strength, F (1, 602) = 5.51, p < .05, 2 = .01. Again, those who identify as Asian were likely to rate the strength of the eyewitness as lower than any other racial group . Importantly, race did not interact with any of the independent variables, which indicates that the effects of race were consistent across manipulated variables. Therefore, the main effects of race reported are not a huge concern in understanding the res ults of the current study. Gender did not affect responses to the scaled items.

PAGE 89

89 Table 5 1 7 . Means and Confidence Intervals for Race of Participants Responses to Scaled Items Race of Participant Asian White Black Hispanic Other General Weighing of Confe ssion Evidence During Trial 3.28 abcd (3.05 3.51) 3.98 a (3.88 4.07) 3.91 b (3.53 4.30) 4.22 c (3.83 4.61) 4.18 d (3.58 4.77) Specific Weighing of Eyewitness Evidence in this Case 4.33 abcd (4.08 4.58) 4.80 ae (4.70 4.90) 4.53 be (4.11 4.94) 4.89 c (4.48 5.31) 4.98 d (4.33 5.62) Specific Weighing of Eyewitness Evidence in this Case scale indicates a marginally significant effect at the .10 level.

PAGE 90

90 CHAPT ER 6 DISCUSSION In this study, community members read one of 18 different trial stimuli depicting information based on a real criminal court case. In the trial vignette, all participants read about a confession that was elicited from a suspect; the confes sion was later recanted. All participants except those in the two control groups read about an eyewitness who viewed the crime occurring and positively identified a suspect from a showup procedure. In some vignettes the confession was depicted as a high q uality confession, meaning it was elicited under optimal conditions without bias; in other vignettes the confession was considered a low quality confession, meaning it was collected using procedures that may introduce bias or coercion. Thus, jurors should weigh the high quality confession more heavily than the low quality confession in determining guilt of the defendant. The same type of manipulation occurred for the eyewitness evidence; in some vignettes the eyewitness was depicted using high quality pro cedures for conducting the identification (termed system variables in the eyewitness literature; Wiliford & Wells, 2013; Rodriguez & Berry, 2010; Fulero, 2009) and conditions under which the eyewitness viewed the criminal (termed estimator variables in the eyewitness literature; Wiliford & Wells, 2013; Rodriguez & Berry, 2010; Fulero, 2009). In other conditions the eyewitness was of low quality by depicting biased and poor identification procedures and viewing conditions. In our two control conditions, th ere was no mention of an eyewitness at all. Finally, half of our vignettes mentioned the possibility of evidence contamination by having the eyewitness overhear about the confession prior to making his identification whereas the other half did not hear su ch information.

PAGE 91

91 I predicted twofold hypothesis; simply put, either jurors are able to properly weigh evidence or they are not. Prior literature supported the possibility of both predictions: jurors have trouble with complex evidence and sometimes weigh ev idence improperly (Helgeson & Ursic, 1993), however prior evidence interdependence research showed that college students were able to identify trial vignettes that showed potential for contamination and render less guilty verdicts than vignettes that did n ot depict interdependence (Hasel, Vogler & Lisuzzo, 2013). Results from the current study indicated that confession quality was a consistent predictor of a range of the dependent measures. More specifically, confession quality affected verdict selection ( both dichotomously and continuously), probability of guilt, and scores on the four scales that were created. Participants who read about a confession of high quality were more likely to render guilty verdicts, and assign a higher probability of guilt. Ho was somewhat inconsistent sometimes high quality confessions resulted in higher ratings of quality or reliability, sometimes lower. Further, the two control groups that were includ ed allowed us to look for ceiling effects in the confession evidence and get a better picture about how well jurors were able to correct for evidence contamination (if at all). Results from the control comparisons seem to indicate that ceiling effects wer e not a problem. Specifically, several findings demonstrated differences between verdict options in the high quality confession factorial conditions and the high quality confession control group. It seems that participants are able to differentiate betwe en quality levels of confessions. However, the comparisons of the experimental conditions with the control conditions

PAGE 92

92 showed that sometimes the quality level of the eyewitness did not significantly influence the verdict ratings when the confession was high quality, but sometimes it did; that is, participants are able to differentiate between quality levels of the confession, but even a high quality confession may be subject to the other evidence presented. Specifically, having a low quality eyewitness pres ented in the case was more detrimental to the case (even with a high quality confession) than not mentioning an eyewitness at all. Further, it seems that this is driven by the quality of the eyewitness, not that it was contaminated. That participants in t his study were able to distinguish between quality levels of confession evidence is surprising, as past research has shown this evidence to be so potent that any mention of it (coerced or weak as it may be) automatically raised dependent measures. So, jur ors may not be perfect at evaluating evidence, but in this study, the results are promising that jurors are somewhat able to effectively evaluate the factors that are important in determining when to trust evidence and when not to: specifically the factors about the collection of that evidence which should be the basis for deciding whether to trust it or not. as they were by the confession quality manipulation. While eyewitness quality was a significant predictor in the expected direction in most of the analyses with the larger sample of participants included (dichotomous verdict, continuous verdict, and marginally predicting probability of guilt), these effects seems to disappea r when participants who failed all but one of the manipulation checks per variable were removed, while the confession effects remain. Again, the impact was somewhat inconsistent in that reading about a high quality eyewitness sometimes resulted in higher r atings of confession or

PAGE 93

93 eyewitness strength or weighing the evidence during trial scales . Sometimes reading about a high quality eyewitness resulted in lower ratings of confession or eyewitness strength or weighing the evidence during trial scales . Final ly, results indicated that the evidence contamination manipulation had no effect on any of our dependent measures, despite the fact that when contamination was present, it was pointed out in the trial. Therefore, the possibility that the eyewitness knew a verdict, probability of guilt, confidence, or any of our scaled item responses. The analyses examining the planned comparisons between the experimental and control comparisons fu rther supported this conclusion because of the lack of differences between contamination present and contamination absent results. Even though contamination should be a part of the quality that participants are assessing, this particular part of quality m ay not be well known or identifiable to community members currently. Given these results, it seems that our two tailed hypotheses were partially confirmed for different variables. In regards to quality, results indicate that the participants in this study were able to assess procedural factors that are significant and then properly weigh the evidence. However this was only consistently true of confession evidence; even though participants correctly identified several of the factors crucial to determining eyewitness quality, they did not properly weigh the evidence in terms of weight during trial or strength scales that should have been associated with the manipulated variations in eyewitness quality. Further, it appears that potential jurors are

PAGE 94

94 also not considering the possibility of evidence contamination in a way that results in a weaker assessment of the evidence or in not guilty verdicts. Quality of Confession Evidence If participants did notice the difference between high and low quality confessions as these results show, what might this mean for the proposed confirmation bias theoretical explanation that was discussed earlier? Given that confessions are generally viewed as a very potent piece of evidence that jurors latch onto regardless of th e fact ors about it (Leo & Drizin, 2010 ), past research might have said confirmation biases were at work. More specifically, the fact that jurors were not typically differentiating between high or low quality confession evidence suggests that they may be process ing the confession evidence as simply indicative of guilt regardless of the details inherent in the confession (such as how it was collected). Thus, jurors would seek out this evidence because it is one of the best things that can support their preexistin g belief that the defendant is guilty. This belief is likely because the case has gone to trial charges; both of these are factors that may create a preexisting inclination of gu ilt in the suggest that they not only will seek out evidence to support that, but they may not listen to evidence that suggests otherwise or find a reason to discount it. Past research shows that this is likely to occur; confession evidence can become more important than any other piece of evidence in the trial, even above and beyond demonstrations of innocence (Kassin & Sukel, 1997; Kassin, 1997). Therefore, it seems as though past research could say that confirmation biases were likely at work in the way that jurors rated and evaluated confession evidence in criminal court cases.

PAGE 95

95 Instead, the participants in this study were able to differentiate between quality level s of the confession which suggests they are not blindly following their preexisting about how a confession is collected, distinguish between high and low quality confessi ons, and render verdicts and evaluate evidence in line with that differentiation. It seems as though the confirmation biases that could be present leading participants to simply rate any confession as a reason to vote guilty are not working that way in th is study. Another possible explanation is that the past research suggesting that jurors do not differentiate between confession quality levels is now outdated. Given the more commonplace discussions of wrongful convictions and issues with legal evidence t hat occur in everyday life (through television shows or news broadcasts), laypersons may be more knowledgeable about the issues relating to confessions and how they are elicited. Several recent mainstream movies have depicted or discussed the tactics used to garner confessions from suspects. It is possible that the flaws with this evidence are starting to influence potential jurors to help them better evaluate confession details. They may be more readily able to identify potential for coercion and false confession factors, and alter their decision making according to these factors. Past research shows that even those who say they recognize that a confession is problematic and will not consider it render more guilty verdicts than when the confession is no t presented at all (Kassin & Sukel, 1997; Kassin, Meissner, & Norwick, 2005), meaning the influence is potentially an unconscious one. This may be changing in that jurors are not evaluating high quality and low quality evidence the same, thus they identif y differences

PAGE 96

96 consciously and render different verdicts and evaluations of evidence in line with those distinctions. Another important factor to consider is that evidence order was almost never a significant predictor of verdict, probability of guilt, or s caled item responses. If confirmation biases were playing a role in the way that jurors weigh confession evidence, I would expect that those conditions with confession evidence presented first followed by eyewitness evidence would render more guilty verdi cts, higher probabilities of guilt, and higher scaled ratings. However, with the exception of two marginally significant interactions on the continuous verdict measure and the probability of guilt measure, evidence order was not significant. Further, the interactions did not depict a consistent finding regarding evidence order; sometimes the confession before eyewitness conditions lead to higher probability of guilt ratings, sometimes the eyewitness before confession conditions lead to higher probability of guilt ratings. Therefore, the lack of consistent, significant evidence order findings further suggests to me that confirmation biases may not be playing out as expected in regards to the quality of evidence. Quality of Eyewitness Evidence Eyewitness ev idence results were somewhat similar, though to a lesser degree. Participants in the current study sometimes noticed the quality differences between high and low quality eyewitness evidence and sometimes did not distinguish between the two. Given that th is type of evidence is generally viewed as less potent that confessions (at least in past research), this may be unsurprising. Past research has shown that confessions are able to contaminate eyewitness evidence (Hasel & Kassin,

PAGE 97

97 200 9 ), yet no studies to d confession. One idea to note is that the confession quality main effects remained significant when removing those participants who failed manipulation checks (the second round of analyses), how ever most often the eyewitness quality main effects either dropped out completely or switched to marginal significance. This further supports the idea that the eyewitness evidence in general is somewhat weaker than confession evidence. It also seems to i ndicate that confirmation biases may play more of a role with the eyewitness quality assessments than with the confession quality assessments. Here, participants might be relying on their preexisting notion that the defendant is guilty and thus not differ entiating consistently between an eyewitness identification elicited under optimal conditions and one elicited under poor conditions. Instead, they simply look at the eyewitness as another piece of evidence that likely points toward their guilty verdict d ecision and do not bother to evaluate it on the merits of the evidence in this case compared to what they already know about an eyewitness. However, this may not be the case because our control comparisons seemed to suggest that in some instances, having a weak eyewitness mentioned was worse for the case than having no eyewitness mentioned at all. These results seem to suggest that jurors are able to differentiate (though maybe not consciously) the difference between high and low quality eyewitness evidenc e. Thus, confirmation biases may not be working in the traditional manner as I have described. Instead, the participants seemed to be narrowing in on the low quality collection skills used to gather the eyewitness evidence, and overcompensate by lowering the probability of guilt more than they would

PAGE 98

98 have if the evidence was absent in the first place. Given all this with the eyewitness quality results from this study, I am still uncertain exactly why jurors are weighing this evidence as such. Another poss were unsure how to evaluate the eyewitness. The current study depicted a showup procedure instead of a lineup in order to ensure that our contamination manipulation would be clear. In other words, I wanted the participants to fully understand that same person selected during the subsequent identification, not simply another unknown person. However, many laypersons are likely onl y used to hearing about lineups or perhaps, when they do watch a showup procedure on TV, they are not aware that is what it is or of the downfalls associated with it. Further, they may not be aware of the procedures that should be associated with a showup or be able to fully evaluate a high(er) quality showup compared to a low quality one. Therefore participants were unable to appropriately weigh the evidence because it was not in the format they were used to seeing. For example, in the control compariso ns, when the confession evidence was high quality, the eyewitness quality level did not seem to matter in determining probability of guilt, the continuous verdict measure, or on the general weighing of confession evidence during trial scaled responses. So , it is possible that participants did not differentiate the quality of eyewitness evidence when they had a high quality confession to still rely on this one piece of evidence was enough that they may not have bothered to put forth the effort to properly untangle the eyewitness evidence.

PAGE 99

99 Further, participants might have focused on the showup as a factor of a low quality confession and evaluated the evidence as lower based on that one factor rather than the other good qualities like close distance, clear l ighting, short time from witness to identification, and others. Therefore it is possible that even what was described as a high quality eyewitness in this study may have seemed like a lower quality eyewitness to the participants, despite the manipulation. It may even be acceptable for participants to judge the showup this way as research indicates that showups likely do results in poorer quality identifications than lineup procedures (Yarmey, Yarmey, & Yarmey, 1996). Therefore, the factors depicted to de monstrate system and estimator information indicative of a high quality eyewitness were negated by the overall framing of a showup instead of the more commonly understood (by laypeople at least) lineup. Future research should determine if jurors are weigh ing these factors together to come to an overall evaluation of that evidence, or if they pick and choose those that are most salient to them. Contamination of Evidence The current study mimicked a real world case of wrongful conviction (Damon Thibodeaux, 1 997). However, the potential for evidence interdependence was also brought to the forefront based on the evidence presented in the case stimulus to further test jurors understanding of the potential for contamination of evidence during an investigation. This allowed us to experimentally examine whether evidence contamination is a factor that mock jurors understand and are able to properly weigh. In this study, there was no evidence that showed jurors were able to identify cases of potential contamination and appropriately discount the weight associated with the second piece of evidence , ( s ee Tables 5 15 and 5 16 f or means and confidence

PAGE 100

100 intervals ). This finding was in contrast to the prior research showing that students were able to identify and discount c ontaminated evidence (Hasel, Vogler & Lisuzzo, 2013). The growing body of research on the subject is showing that 1) these situations do impact decisions for eyewitnesses, alibi corroborators, forensic analysts and others (Hasel & Kassin, 2009; Dror & Cha rlton, 2006; Kassin, 2009; Dror, Charlton, & Peron, 2006) and 2) the current research shows that community member mock jurors are unable to differentiate between cases of interdependent evidence and cases of independent evidence. Therefore our criminal ju stice system has a problem in the way cases with multiple pieces of evidence are processed and tried. Given that most cases contain multiple pieces of evidence (Kassin, Bogart, & Kerner, 2012), it is a problem that may be bigger than we are currently awar e. In addition, the attorney in this case attempted to educate the jurors about evidence contamination in pertinent conditions, so it is even more troubling that jurors did not use contamination in their decision making If evidence contamination is taking place in criminal court cases and investigations every day, it is crucial that we understand more about this phenomenon. One of the big steps in this study was to determine if the jury is able to identify this problem when presented with pieces of informa tion like they are in a real trial. In terms of the confirmation bias literature discussed earlier, the lack of significant findings for contamination may support the idea that confirmation biases are at work. Specifically, we may have had two sets of pr eexisting beliefs that governed the way participants thought of the evidence: beliefs about confessions and beliefs about lineups. It is possible that, given these beliefs that the participant developed early in the trial, the

PAGE 101

101 n of the contaminated evidence may have be one example of jurors discounting or underemphasizing the information that is contrary to their initial ideas. Since the discussion of the contaminated evidence was in opposition to their guilty preexisting belie f, they may have negated that information in favor of maintaining their guilt mentality. This is supported by the results which show conditions with contamination discussed we re no different than conditions where contamination was not discussed; participa nts are not able to identify this weakness in their evidence evaluation skills. It seems as though, even if they have identified a problem with the evidence, they are not able to do so based on the current study results. This is especially important in th e context of high quality evidence. Since jurors were unable to identify and appropriately correct for potential contamination, it is possible that jurors are considering this high quality evidence strong enough to justify a guilty verdict when, in fact, it only seems like strong evidence because of the influence that contamination played. Contamination on its own should be enough to determine a piece of evidence low quality, and therefore remove any or most of the weight assigned to it. However, this is not occurring. We have the potential for high quality evidence to be contaminated during the investigation, and we see that jurors still consider it high quality when rendering their verdicts. Therefore, future research should focus on how to solve this problem or prevent it from occurring. Recommendations for Future Research Several recommendations for future research have emerged during this research. First, as mentioned above, it may be useful to try to untangle contamination results without using co nfession evidence. While confession contamination is a well demonstrated part of the contamination world, the effects are generally too strong in an

PAGE 102

102 experiment setting to effectively test other variables. I would like to run a very similar study with an eyewitness and another piece of evidence to see if that can replicate the results found here. For example, the above mentioned possibility of participants only relying on one piece of evidence to render their guilty verdicts, thus allowing the possibility of mentally deleting the eyewitness influence and still seeing the same result, is an interesting possibility for future research. That is, are these problematic pieces of evidence such as those with low quality being treated as weak evidence to consider with low weight? If this is the case, then the relative weight of other pieces of evidence would likely be more influential on verdict or the overall balance of evidence. Or are they being negated completely and not factoring into the verdict determinati on at all? It is possible that the evidence is being treated as if it does not exist when it may be better to consider all evidence and then more effectively train jurors on how to balance and weigh each and every piece of evidence, however weak they think it may be. Further, a somewhat altered study would be able to better test for this overwhelming influence that the confession has in many cases. Therefore, I would like to further test to see if the finding where participants seem to only rely on one pie ce of evidence is dependent on the type of evidence present. That is, do we see the same results where participants rely on one piece of evidence if the evidence grouping does not contain a confession? Perhaps testing to see an eyewitness and an alibi to gether could be interesting to see inculpatory versus exculpatory evidence pitted against each other. In this situation, is it that jurors disregard the low quality evidence altogether, or does a weighing process evaluating both pieces of evidence determi ne verdict? Further, it would be interesting to see how the so

PAGE 103

103 (DNA) factors into the contamination potential. Prior research shows the possibility that this occurs in real world cases (Dror & Charleton, 200 6 ), but we d o not yet know how jurors who rely on DNA evidence very heavily are able to comprehend the problems of quality or contamination. Another line of future research mentioned above is the examination of the qualitative data component of this research. Coding and analysis of the qualitative questions is ongoing to further untangle this result. In particular, one question asked n will help to shed light on the motivations and evaluations within each trial vignette context that led to the selected detail as possible, how you weighed the evide nce to come to the above marked would be very valuable in better understanding the role that quality of evidence plays in weight assigned to evidence. Further, see ing how participants weighed the evidence in this case, depending on the condition they read, could provide further support for one aspect of our two tailed hypothesis. Examining the qualitative data using different models of juror decision making may be able to provide a current test of which models seem to represent the way jurors think and process through the evidence presented to them at trial. For example, mathematical models (Davis, 1973) may not be useful with our current findings, but perhaps the story model of juror decision making (Pennington & Hastie, 1993) might be. Given the type of information gathered and the ability to gain insight into the mock

PAGE 104

104 of the methods could be done with this data. This would be helpful for attorneys, trial consultants, as well as other researchers interested in jury research in the future. Additionally, a better experimental test of how showup procedures are interpreted in court could be informativ e. Given how common they are, do we see jurors weighing this evidence similarly to an eyewitness who had selected someone from a lineup? Further, it would be interesting to vary different factors about the quality of the showup compared to a high or low q uality lineup selection. For example, I could test to see how lineup. If I see differences in choosing rates based on this one factor, it would be problematic to fail t o see different influences when that evidence makes it to court. Another useful form of education for jurors could be found in training on how to evaluate the quality of any piece of evidence. However, we are currently unaware of whether jurors are assess ing just one or two factors that are salient to them in their quality determination, or if they are taking an overall assessment of the information provided about the evidence. For example, I encoding of the crime occurrence as well as their identification procedures. Which, or how many, of those numerous factors was the determination in qualifying the evidence eyewitness evidence it may have been the fact that it was a showup that lead participants to categorize both the low and high quality eyewitness into a low quality condition, and thus, why I did not see consistent influences of the eyewitness on verdict. That may also explain w hy the contamination results were negative; it could be a floor effect. If participants are already considering the showup to be low quality, the

PAGE 105

105 contamination simply makes the evidence further low quality. If future research shows this to be the case, j uror education would be useful to try to show that the quality of a piece of evidence should be determined by an overall evaluation of myriad factors about the collection of that evidence. While this might be difficult as it introduces a new weighing proc ess in determining how much weigh to assign to each factor, it would at least encourage jurors to make a holistic evaluation of the evidence rather than focusing in on something they picked up from CSI or Law and Order. Another study that could be interest ing to do would try to tease out the influence of the confession a bit more. If a study showed that a confession contaminated an eyewitness, then later recanted and was not presented at trial (just simply mentioned in passing), however the contamination p otential was pointed out as it was in this study, would we still see the effect? In that case, we could see the presence or absence of the contamination, but may not have the same impact that confessions tend to have in other studies. The judge could iss ue a warning to the mock jurors mid trial that the confession was inadmissible and to not consider it, and also include a control group where the confession is not mentioned. We may see that jurors are better able to pick out the contamination problem bec ause they are explicitly told it is problematic, though given past research it is possible this judicial warning may not be strong enough to Further, other forms of contamination discussion in the trial context should be investigated. In the current study, the contamination problem was mentioned during ways that it could be discussed to make the issue more salient at trial, and thus

PAGE 106

106 hopef ully more weight would be associated with it in verdict determinations. Perhaps the judge discussing the problem would give more weight to the issue, or the defense attorney could make it part of his own presentation of evidence and devote more time to en sure jurors are able to understand the influence this could play rather than just a mention of it in closing or during cross examination. It may even involve an expert Policy R ecommendations As detailed above, there are many more studies that could be conducted branching out from this research. Many of these will be needed before any conclusive policy recommendations are likely to be enforced, however this research alone helps to justify some recommendations to be aware of. First, ensuring that our officers keep any and all evidence independent during the investigation is the best way to stop the contamination problem before it gets to the jury. If proper procedures are follow ed and officers do not discuss the investigation or any piece of evidence in places where they may be overheard by those who are not part of the investigation, we can be certain that this problem does not become one of the courts. Further, ensuring that d ouble blind procedures are used when discussing any aspect of the investigation with members of the public, such as witnesses, alibi corroborators, or other members of the investigation such as the forensic team, can eliminate the chances that a small comm ent might be influencing or contaminating any decision. This is likely easier to recommend than it is to actually impose, but every effort must be made to stop the contamination from occurring; this is especially true because the jury seems to be an inade quate check for this type of contamination to be corrected.

PAGE 107

107 Given the potential problems mentioned above with the use of the showup in this study, one recommendation I would make is to help jurors better understand the problems associated with showup proce dures and the proper procedures to follow when using this piece of evidence. While many departments still use showups on a daily basis in conducting their police work, there are many problems inherent to showing a witness only one person and asking for a positive identification. Although one information, better education on how to evaluate the eyewitness evidence, particularly if it is a showup, would not hurt. Anot her policy recommendation is to better educate attorneys about how to spot potentially contaminated evidence. Given that stopping contamination from occurring in the investigation stage may be unlikely, it would be useful for attorneys to be able to ident ify cases where this might have occurred and be able to either develop a trial strategy for dealing with it during trial, or effectively argue that the evidence should be deemed inadmissible before trial begins. One way this goal could be reached is throu gh trial consulting. Given that consultants are research based practitioners, they may be better able to understand the potential problem this poses for the court case, and effectively communicate that to the attorney. They would also be able to test pot ential trial strategies to see which strategy is the most effective way of mitigating the influence the contaminated evidence has on the jury. Finally educating the jurors about the problem is another possible remedy for the contamination effect. It is po ssible that a warning with some amount of information and explanation by the judge could be one way of downplaying the impact that contaminated

PAGE 108

108 evidence has on jurors. Another option is to have the education be part of the jury selection process to ensure that all potential jurors are able to understand this kind of complex explanation and then properly weigh evidence. At any rate, ensuring that attorneys are able to identify the problem means they can attempt something to help their jurors recognize the bias. Limitations As with any study, there are a few limitations with the study that need to be mentioned. First, while I was able to add to the literature by testing the influence of contamination on community members in the entire region of interest, th is study was done online. The benefits of conducting the study online are that it allowed me to reach a sample of community members who may be more representative of those who actually serve on juries compared to criminology students. In addition, conduct ing the study online allowed swift and efficient data collection. However, online studies are not without drawbacks. For example, there is potential for participants to not pay as much attention as I would hope for the study, or to treat the study as they would a real life jury situation. I attempted to ensure this did not occur by pilot testing the materials several times to gage a response time adequate enough to study and encode all the information. The community member sample generally did fall withi n this time range of 20 45 minutes depending on condition, and those who took inordinately long to complete the study were thrown out for one set of analyses. However the lower end of the group may not have been adequately filtered because there were stil l many individuals who took less than 20 minutes to complete the study. In any case, a study that takes between 20 and 45 minutes to complete does not effectively replicate a day or several days long trial. Additionally, participants were

PAGE 109

109 paid $3 for thei r time, which may not have been enough of an incentive to encourage participants to really dedicate their attention to the work. Mturk also has a group of people who have been classified as Masters, meaning they have completed a number of projects with suc cessful results, and have been nominated as such by an individual requesting their work. I initially planned to collect from only this group hoping that they would be more detailed in their work, however the data collection trailed off rather quickly and I worried that I had used up their entire group of Masters given the large sample that I needed for this study. While I did find that these participants were much more likely than non Masters to pass the attention question, there was no difference in resu lts when examining only those with the Masters classification compared to the current results depicting both groups. Another potential problem is in the data collected for this study. Those in the control groups were given an option to indicate that they had not read about an eyewitness (and similarly for a confession even though all conditions included that evidence). However, some participants still filled in answers for the dependent measures questions even though they were not supposed to. Some also indicated in emails to me that they were confused about the mention of the friend at the end of the trial compared to what they thought an eyewitness should be. I did not analyze these data for participants who did not hear about an eyewitness, but it was confusing for participants and should be avoided in future research. Another potential downfall of the current study is the use of the showup as our eyewitness manipulation. Given that it seems like participants did not fully understand the evidence or did not properly weigh the factors to determine it of high versus low

PAGE 110

110 quality, it is possible this may explain the lack of significant eyewitness findings in our results and the tendency for significant eyewitness effects to turn marginally significant in the second round of analyses. Future research should examine ways to test contamination in the context of a lineup; perhaps indicating in the overheard statement something distinctive about the suspect or which number in the lineup he was. This would be able to convey both that the suspect confessed as well as who the suspect is to the person making their identification. If the same results occur, I can be more certain that jurors are unable to identify contamination as a problem in the trial, not that i t could be a result of the evidence choices I made for the study. Finally, the mention of contamination was relatively weak (only a sentence or two) so it is possible that it may not have been strong enough to remain salient through the rest of the trial s timulus and dependent measures. It was intentionally kept out of the opening paragraphs where the quality information was reiterated because it would not have made sense as it was brought up during cross examination. Future research should try to make th e discussion of contamination more salient to ensure an adequate test of the manipulation. However, given that participants noticed the differences between conditions in which contamination was a problem and in which it was not, this limitation is not of h uge concern in the current study. Conclusions This study was able to contribute to the knowledge base about several concepts. First, jurors are likely not able to properly weigh contaminated evidence compared to independent evidence. As such, it is not likely that a harmless error type analysis on appeal is enough to solve the problem of contaminating evidence; we need to account for this influence earlier, either at the investigation stage to ensure evidence is not

PAGE 111

111 contaminated in the first place, or th rough one of the trial actors such as the jurors in those cases where contamination does occur by helping them to be more aware and more critical of potentially contaminated evidence. This would also better untangle the problem seen with the different dis tinctions between legal and psychological coercion definitions. It is also possible that the deliberation process is able to serve as a safeguard for this complex information as well. Maybe discussing the information with fellow jurors helps to clarify an y problems with the evidence and determine what weight should be attached, however future research would be needed before this could be considered a successful safeguard against the influence of contamination. Additionally, the current study was able to ad d to the field of knowledge on evidence contamination by accounting for the quality of evidence. One of the particularly problematic parts of the evidence contamination issue is that this contamination effect occurs regardless of how the evidence is gathe red. This means that even poorly gathered, and thus, low quality, evidence contaminated other pieces of evidence, such as those that are gathered in the appropriate manner and thus of high quality. Pieces of evidence that ideally should be trusted and ke pt in as the case goes to trial get deleted and replaced with other evidence that should be questioned from the start, not put in the forefront as a criminal case develops. Ideally, participants should be able to differentiate between high and low quality evidence so that there is less chance of evidence contamination occurring at all, and particularly less chance of obscuring high quality evidence with low quality evidence. However, this was not found with the current research, indicating a greater need to stop contamination before it gets to trial.

PAGE 112

112 APPENDIX A INFORMED CONSENT Protocol Title: Juror Decisions in a Criminal Trial Please read this consent document carefully before you decide to participate in this study. Purpose of the research study: The p urpose of this study is to examine how jurors make decisions when presented with evidence in a criminal court case. What you will be asked to do in the study: You will be asked to complete a voir dire questionnaire to simulate the jury selection process o f a trial. After completing that questionnaire, you will read through a trial stimulus, depicting opening statements, presentation of evidence, closing statements, and judicial instructions for a jury trial. Then you will complete a series of questions a bout your verdict selection and thoughts on the evidence. Time required: 30 minutes Risks and Benefits: There is minimal risk in participating; participating is similar to everyday activities like surfing the internet or reading the newspaper. As a benefi t, you will learn about the research process. Compensation: You will be awarded credit through the Department of Criminology, Law and Society Participant Pool for participating in the study. Confidentiality: Your identity will be kept confidential to the extent provided by law. Any information you provide will not be linked to your personal information. Voluntary participation: Your participation in this study is completely voluntary. There is no penalty for not participating . In addition, you do not hav e to answer any question you do not wish to answer. Right to withdraw from the study: You have the right to withdraw from the study at any time without consequence. Whom to contact if you have questions about the study: Megan Kienzle, M.A. , Doctoral St udent, Department of Sociology and Criminology & Law, PO Box 117330, Gainesville, FL 32611, mkienzle@ufl.edu. Lora Levett, Ph.D., Assistant Professor, Department of Sociology and Criminology & Law, PO Box 117330, Gainesville, FL 32611, llevett@ufl.edu. W hom to contact about your rights as a research participant in the study: IRB02 Office, Box 112250, University of Florida, Gainesville, FL 32611 2250; phone 392 0433. Agreement: I have read the procedure described above. I voluntarily agree to participat e in the procedure and I have received a copy of this description. Participant: ___________________________________________ Date: _________________ Principal Investigator: ___________________________________ Date: _________________

PAGE 113

113 APPENDIX B VOIR DIR E QUESTIONNAIRE Instructions: Before rendering your judgments in the cases, we would like to ask you a few questions about yourself. All answers to this questionnaire are confidential and anonymous, your name will not be linked to any of the information t hat you provide today. 1. Gender (circle one) FEMALE MALE 2. Year of birth _______________ 3. Are you a U.S. Citizen? YES NO 4. Are you registered to vote? YES NO YES NO 6. What is your occupation? ___________ ___________________ 7. How many children do you have, if any? __________________ How many of your children are under age 18? __________ 8. Which of the following statements best describes your highest educational achievement? ______ SOME HIGH SCHOOL _ _____ HIGH SCHOOL GRADUATE ______ TRADE SCHOOL ______ SOME COLLEGE ______ COLLEGE GRADUATE ______ SOME GRADUATE SCHOOL ______ GRADUATE DEGREE 9. What is your racial/ethnic background? (circle one) White, Non Hispanic Black, Non Hispanic Asian His panic Cuban Other ____________ 10. What is your current marital status? SINGLE MARRIED DIVORCED WIDOWED 11. Have you ever served on a jury before? YES NO If yes, was the trial: CIVIL CRIMINAL Were you the foreperson on the jury? YES NO 12. Generally speaking, do you usually think of yourself as a Republican, a Democrat, or an Independent? 1 2 3 4 5 6 7 Strong Republica n Not very strong Republica n Independen t leaning Republican Independen t Independen t leaning Democrat Not very strong De mocra t Strong Democra t 14. Which of these opinions best represents your views? 1 2 3 4 5 6 7 Extremel y liberal Libera l Slightl y liberal Moderate/middl e of the road Slightly conservativ e Conservativ e Extremely conservativ e

PAGE 114

114 15. Do you attend reli gious services at least once a month? YES NO

PAGE 115

115 APPENDIX C MASTER TRIAL STIMULUS IN THE BREVARD COUNTY SUPERIOR COURT FOR THE STATE OF FLORIDA State of Florida ) v. ) ) Michael Smith ) SUMMARY OF Defendant ) TRIAL PROCEEDINGS FROM ______________________________________________________________________ ________ Opening Judicial Instructions L adies and gentlemen of the jury, y ou have been selected and sworn as the jury to try the case of State of Florida v. Michael Smith. This is a cri minal case. Michael Smith is charged with Second Degree Murder. The definition of the elements of Second Degree Murder will be explained to you later. It is your solemn responsibility to determine if the State has proved its accusation beyond a reasonable doubt against Mr. Smith. Your verdict must be based solely on the evidence, or lack of evidence, and the law. The fact that Mr. Smith was indicted for this crime is not evidence and is not to be considere d by you as any proof of guilt. It is the judge's r esponsibility to decide which laws apply to this case and to explain those laws to you. It is your responsibility to decide what the facts of this case may be, and to apply the law to those facts. Thus, the province of the jury and the province of the cour t are well defined, and they do not overlap. This is one of the fundamental princ iples of our system of justice. Before proceeding further, it will be helpful if you understand how a trial is conducted. At the beginning of the trial the attorneys will hav e an opportunity, if they wish, to make an opening statement. The opening statement gives the attorneys a chance to tell you what evidence they believe will be presented during the trial. What the lawyers say is not evidence, and you are not to consider it as such. Following the opening statements, witnesses will be called to testify under oath. They will be examined and cross exam ined by the attorneys. After the evidence has been presented, the attorneys will have the opportunity to make their final argume nt. Following the arguments by the attorneys, the court will instruct you on the law applicable to the case. You should not form any definite or fixed opinion on the merits of the case until you have heard all the evidence, the argument of the lawyers and the instru ctions on the law by the judge. The case must be tried by you only on the evidence presented during th e trial in your presence and in the presence of the defenda nt, the attorneys and the judge. In every criminal proceeding a defendant has the abs olute right to remain silent. At no time is it the duty of a defendant to prove his innocence. From the exercise of a defendant's right to remain silent, a jury is not permitted to draw any inference of guilt, and the fact that a defendant did not take the witness stand must not influence your verdict in any manner whatsoever. We will now begin with opening statements. Opening Statement from Prosecution Attorney

PAGE 116

116 Ladies and gentlemen of the jury, as this case is presented, we will show you exactly how Mr. S mith murdered Ms. Bennet, in cold blood, on the early morning of September 19 th , 2000. Mr. Smith himself even confessed to doing so when he was questioned by police officers after the killing. His confession is filled with details of the crime that only the extension cord hanging from the tree above the crime scene to strangle Ms. Bennet that morning thereby ending her life. The officers will testify that Mr. Smith was calm and cool w hile reciting the intimate details of his crime, with the detached nature of a ruthless murderer. (only in high quality confession conditions)The confession that was elicited by our police detectives was a high quality confession one that you should pay close attention to and Mr. Smith freely volunteered this information when talking with police. An innocent man would not do this. (blue text is removed in control conditions) Further, we have an eyewitness, a man who lives near the crime scene, who pla ces Mr. (only in high quality eyewitness conditions) The witness you will hear from today was given unbiased instructions prior to making his identification, and our poli ce officers followed all the procedures that have been shown to elicit accurate identifications, like ensuring the identification photo was taken in plain clothes, not a jailhouse jumpsuit. Further, the perpetrator was right in the front yard only a few f eet from the window our witness was looking out of, watching his actions for nearly 20 minutes, and he made his identification only 2 days after the crime. pacing by the road that morning fully depicts the man sitting in this courtroom today. The witness was even able to make this identification confidently . Ms. Bennet and her family deserve justice for her murder and Mr. Smith deserves punishment for this brutal crime. Opening Statement from Defense Attorney Ladies and gentlemen of the jury, Mr. Smith did not murder his cousin. While the murder of Ms. Bennet is a horrendous tragedy, the evidence that will be presented w ho murdered Ms. Bennet and there is absolutely no physical evidence to support the was murdering his cousin. He was not out committing any crime, let alone one of this mag nitude. And his clean rap sheet will show just how unlikely that would be. Further, the prosecution mentioned that my client was detached during the police interrogation. We do not argue this fact; Mr. Smith was detached during police questioning, just as any person would be when suffering the news that they have been framed for a murder and are in danger of going to prison. (only in low quality confession conditions) Further, the amount of pressure used during my client s interrogation by police was un necessary, even inhumane. After 10 hours of questioning, my client still did not write out a confession, but merely initial ed a piece of paper thrown in his face by the same people who arrested him. This kind of pressure can lead people to act unlike them situation? (only in low quality eyewitness conditions and removed for control conditions ) Further, the witness was asked to identify someone over 2 weeks after the crime occurr ed. Can you remember who you might have seen over 2 weeks ago, nevermind identifying a person you saw from nearly a block away and for only a few minutes?

PAGE 117

117 Further, researchers have learned that many of the procedures that protect against mistaken identific ations were not followed here. In particular, the officers had our witness identify the suspect using a show up (using only 1 picture) rather than a photo orange ju mpsuit, and even encouraged the witness to pick someone, all of which have been shown to lead to mistakenly selecting the wrong person . case depends on circumstantial evidence that fails to conclusively prove my client is the perso n responsible for this cruel and unthinkable tragedy that Ms. Bennet endured. The defense will show there is not proof beyond a reasonable doubt that Mr. Smith killed her. Thank you for your time and service in this case. ORDER OF PROSECUTION EVIDENCE WILL BE FLIPPED DEPENDING ON CONDITION (CONFESSION FIRST, OR EYEWITNESS FIRST) Testimony of Detective Blume During direct examination, Detective Blume testified that he was the Chief Detective of the Brevard County Police Department. He further stated that he was in charge of the was filed with the police department. He testified that he put together a county wide search party and after 13 hours of se arching, he was notified of the discovery of a body near the Indian River Bridge. Detective Blume testified as to the state of the body when that her face and upper tors o had been beaten and badly bruised. In addition, the The detective and his officers searched the area surrounding the body, looking for any evidence or information that may contribute to finding the responsible party. ( Removed for control conditions) D etective Blume stated that there were two houses within a short distance of the bridge, and when officers talked with residents of those houses, an older male who lives at the closer of the two houses stated he saw a man pacing near the area in the middle of the night. The prosecuting attorney asked if this m an was Mr. Griffin, and the detective confirmed and said that Mr. Griffin was brought to the police station 2 days later. Blume stated that Mr. Griffin identified Mr. Smith from a showup. When the prosecutor asked Detective Blume to elaborate, Blume expl ained that showups were common practice for many police departments. He had presented the witness with a photo of Mr. Smith and was able to ascertain his confidence after making the positive identification . The detective went on to state that no other ph ysical evidence was found at the crime scene, and that he ordered forensic tests to check for rape and semen from the body, however the tests indicated there was no evidence of ( show either good or bad confession paragraph , depending on condition) BAD CONFESSION CONDITION: The detective stated that the police officers brought cousin, Michael Smi th. Detective Blume stated that Smith waived his Miranda rights after a bit of coaxing, knowing that he was free to leave at any point during the interrogation. Blume and another officer interrogated Smith for a total of approximately

PAGE 118

118 10 hours, stating t hat quite frequently during the questioning, Blume needed to raise his voice to encourage the suspect to tell him the truth. During the first 5 hours of the interrogation Mr. Smith denied any involvement in the crime, stating that he was out with a friend until about 9pm on the 18 th then went home alone for the remainder of the night and morning of the 19 th . However, Detective Blume reported that after another 3 hours ly morning hours of September 19 th including details about the murder, such as the extension cord wrapped around Ms. t he bottom of the page, which Smith did. During the interrogation, Blume reported that he used techniques that, while they may be controversial due to the potential for coercion, are standard for use in the Interrogation Manual used by the Brevard Police D epartment and many others throughout the country. These techniques included questions designed to make the crime seem less important than it is and thus, no big deal to confess to. Additionally, there were questions that make the crime seem more importan t and a much bigger deal than it really is. This later question indicates the suspect should cooperate and confess so the police can help him as much as possible. A video recording of the interrogation was submitted into evidence, showing that both Blume and the other officer engaged in the interrogation of Smith. GOOD CONFESSION CONDITION: The detective stated that the police officers brought in several suspects for questioning at the station. One of these men was the tective Blume stated that he asked another officer, uninvolved in the case prior to the interrogation, to question Mr. Smith to ensure Blume reiterated that while both Blu me and the other officer were present for the entire Smith. The fact that Blume did not participate in the interrogation can be seen in the interrogation video submitte d into evidence. Blume reported that Smith waived his Miranda rights freely, knowing he was free to leave at any point during the interrogation. The officer interrogated Smith for approximately 2 hours, stating that the interrogation was mainly conversat ional for the entire 2 hours with only a few times when the interrogator needed to really push Mr. Smith for additional information. During the first hour of the interrogation Mr. Smith denied any involvement in the crime, stating that he was out with a f riend until about 9pm on the 18 th then went home alone for the remainder of the night and morning of the 19 th . Blume reported that the interrogator used techniques that are standard for use in the Interrogation Manual used by the Brevard Police Department and others around the country, such as open ended questioning and encouraging the suspect to describe as much as they can with little stating he did kill Ms. Bennet dur ing the early morning hours of September 19 th . The officer requested that Smith write a confession, including as many details about the murder as he could, and Smith complied with the request, including details such as the extension cord used to strangle Ms. Bennet. After Mr. Smith signed the handwritten confession, he was escorted from the room. investigated for this murder. The detective reported that there had been incr eased

PAGE 119

119 reports of property theft and vandalism in the area in the weeks prior, but they had not been able to identify a suspect for those crimes yet. The defense attorney further asked how frequently the detectives are unable to find any sort of physical ev idence at the crime scene that could tie a person to a murder. The detective reported that quite frequently they are unable to gather physical evidence, either because it was not left by the perpetrator, or because the sample has degraded to the point tha t it is no longer available by the time the police are called to the body. The detective stated that the department has to frequently rely on eyewitness evidence and their skills at eliciting a confession from the perpetrators in order to prove the person The defense attorney also asked if the techniques such as minimization and maximization could be felt as pressure to confess to the suspect, especially since he had since recanted this confession. The detective stated that it was possible that Mr. Smith felt pressured into confessing, but that these tactics are commonly used by many police departments around the country and have been useful to the Brevard County Police Department in eliciting confessions from truly guilty suspe cts. Testimony of Mr. Griffin After Mr. Griffin was called to the stand, the prosecuting attorney asked what his body had been found at the bridge. He stated that he had lived in the house for about 15 years now, and often saw people coming to the river, even at night to go swimming. He stated that a few years back a few teenagers had hung a rope from a tree branch to fin thought it was dangerous so he would occasionally cut the rope down. He went on to say he would inevitably find another form of swing put up not long after. Mr. Griffin stated that he had heard about Ms. Bennet and the search party, but did not know t he status of it until officers knocked on his door on the morning of September 20 th . He stated that officers told him they had found a body by the bridge (bad eyewitness) about a block down the road (good eyewitness) across the street and asked if he had seen or heard anything suspicious in the previous days. Mr. Griffin reported to police officers that he had woken up at about 1:30am the morning of the 19 th and went to get a drink of water from the kitchen. He noticed a person pacing back and forth (b ad eyewitness) a little ways down the road (good eyewitness) across the street , near the bridge. He stated that, as he said earlier, it was not uncommon for young adults to go out to the bridge at night to go swimming or skinny dipping, and that he did no t think much of it at the time. He said the police asked if he could describe what the person looked like, or if he remembered any other details about that early morning. Mr. Griffin said that he gave police a brief description, saying that he saw a man, about average height, with short brown hair wearing a dark shirt and jeans. He further stated that he did not hear any sort of commotion or noise that seemed out of the ordinary. (show either good or bad eyewitness (2 paragraphs), depending on conditio n) GOOD EYEWITNESS CONDITION: The next day, September 21 st , Mr. Griffin was asked to report to the police station for an identification. Mr. Griffin reported that he sat at a table in a small room and the detectives explained how the identification proced ure would work; they would show him one photo and he could take as long as he needed to look at it. They further warned him that the actual perpetrator may not be in the photo and to be certain before making an identification. The police stressed how imp ortant it

PAGE 120

120 was for him to make an accurate identification because they did not have any physical evidence tying the perpetrator to the crime scene, but that they would continue the investigation whether he made the identification or not. He said the detect ives showed him a photo of a man in jeans and a t shirt fitting his description, so he stared at the photo for a while then told the officer that was the man he had seen near his house at the time of the crime. The police asked if he was certain and Mr. G riffin spent another minute looking at the photo then said he was sure that was the man he had seen. The police detectives thanked him and escorted Mr. Griffin to another room with a computer to record his identification and certainty, being sure not to s ay anything to influence Mr. On cross examination, the defense attorney asked how Mr. Griffin could be certain of his identification, especially since he had seen the person in the early morning hours, in the dark. Mr. Griffin reported that yes, it had been dark out, but there were street lamps along the road with bright lights and the man had gone under them a few times while he was pacing. Mr. Griffin reported that he watched the man for about 20 minutes, paying c lose attention to where the man was walking to ensure his new car in the driveway was not stolen. The attorney further asked if he was stressed at all during the identification procedures. He reported that yes, he was nervous, but felt that he had done a good service . The defense attorney asked if Mr. Griffin heard anything while waiting to begin his identification and Mr. Griffin reported that he had overheard a couple of police officers talking at their desks about how great it was that the suspect had confessed that morning and how this case was shaping up nicely. They had wished that all cases were so easy. The defense attorney asked if he thought that hearing this right before making an identification may have affected his decision any, and Mr. Grif fin reported that he had based his identification decision on his memory from the morning he saw the man pacing in the road, not on what the police officers had said. (RED TEXT IS ONLY INCLUDED IN THE CONTAMINATION PRESENT CONDITION). BAD EYEWITNESS CONDIT ION: Nearly two weeks later, on October 4 th , Mr. Griffin was asked to report to the police station for an identification. Mr. Griffin reported that he sat at a table in a small room and the detectives explained how the identification procedure would work; they would show him one photo and he could take as long as he needed to look at it. The police stressed how important it was for him to make an accurate identification, because they did not have any physical evidence tying the perpetrator to the crime sc ene and were relying heavily on his identification. He said the detectives presented him with the mugshot of a person wearing an orange jumpsuit who fit his description, and Mr. Griffin reported that he said it was the same person he had seen outside his house. The police asked if he was certain and to look at the picture again. Mr. Griffin reported that he spent several minutes looking at the picture again, and confirmed once again that this was the person he had seen. The police detectives congratulat to another room with a computer to record his identification and certainty. On cross examination, the defense attorney asked how Mr. Griffin could be certain of his identification, especiall y since he had seen the person in the early morning hours, in the dark. Mr. Griffin reported that yes, it had been dark out especially since their street light had burned out a few weeks before. The attorney asked how long Mr. Griffin had been watching t he person for, and Mr. Griffin replied only a few minutes and that he was

PAGE 121

121 not watching his actions very closely but he was still pretty sure that they had found the right guy. The attorney further asked if he was stressed at all during the identification procedures. He reported that yes, he was nervous, but felt that he had done a good service when the police said he had picked the right guy. The defense attorney asked if Mr. Griffin heard anything while waiting to begin his identification and Mr. Griffin reported that he had overheard a couple of police officers talking at their desks about how great it was that the suspect had confessed that morning and how this case was shaping up nicely. They had wished that all cases were so easy. The defense attorn ey asked if he thought that hearing this right before making an identification may have affected his decision any, and Mr. Griffin reported that he had based his identification decision on his memory from the morning he saw the man pacing in the road, not what the police officers had said. (RED TEXT IS ONLY INCLUDED IN THE CONTAMINATION PRESENT CONDITION). After this testimony, the prosecution rested. Testimony of Michael Smith When Michael Smith took the stand, his defense attorney ask ed about his relationship with the victim, Ms. Bennet. Mr. Smith stated that they were cousins, but did not keep in close contact. He stated that he had not seen her in about 3 months when he found out that she had been killed. The attorney asked where he was on the morning of the murder, and Mr. Smith reported that he had been out with his friend Ben until about 9pm on the night of September 18 th , then went home alone and went to bed. Mr. Smith elaborated, saying that he and the friend had dinner toget her, then went to a movie which ended just before 9pm. He reported that he went straight home after leaving the movie then went to sleep until getting up for work the next morning at 5am, arriving at the office just after 6am on September 19 th . The defens e attorney asked Mr. Smith about the interrogation by the police officer, and Mr. Smith reported that he only confessed to the murder because of the pressure he felt by the interrogator. He stated that he was distraught over breaking up with his girlfrien d just days prior, and was not in a good place mentally to withstand that kind of intense questioning. The defense attorney asked if he had ever met Mr. Griffin, and Mr. Smith replied that he had not met him, but that he did go out to the bridge area by t he lake occasionally to swim, so he stated that he must have confused when he saw him. Testimony of Ben Thomas The defense attorney brought Ben Thomas, friend and coworker of Michael Smith to the stand. Ben testified about spending time with Michael Smith the night before Ms. Bennet was reported missing. He confirmed that the two had gone to dinner just before 6pm at their favorite restaurant before seeing a movie and each going home alone. The defense attorney asked Mr. Thomas to expand on their night a bit more, so Mr. Thomas explained that they had gone to a pizzeria around the corner after getting off work. He said both he and Michael were working on a new project to try and impress their boss in the hopes of getting promotions, so they were putting in long hours each day. Mr. Thomas stated that they go to that particular restaurant quite frequently, and nothing seemed out of the ordinary that night. After dinner, they caught the 7:10pm show of a movie that had just come out and once it was over, th ey each headed back to their own apartments since they were planning to be to work early again the next day.

PAGE 122

122 On cross examination, the prosecutor asked Mr. Thomas if they had talked at all after separating for the night. Mr. Thomas replied that he could n ot remember for sure, but that they usually did not talk to each other on the phone so he was fairly certain they had not spoken again that night after the movie. The prosecutor asked if Mr. Smith made it in to work at 6 the next morning (the 19 th ) as he had planned, and Mr. Thomas replied that he was closer to 6:30am, but that their job was pretty flexible and they could make their own hours, so it was not unusual. After this testimony, the defense rested. Closing Statement from Prosecuting Attorney Ladie s and gentlemen of the jury, thank you for your time in hearing the evidence presented in this case. As I stated before the trial even began, we have clear and convincing evidence that Mr. Smith killed his young cousin on the early morning of September 19 th , 2000. (removed in control conditions) Y ou heard testimony from a reliable witness that places the defendant at the scene of the crime during the time in question, and pacing as if worried about something, no less! (only in high quality eyewitness con ditions) Keep in mind, this witness identified our man Smith only 2 days after the crime, after watching him closely for 20 minutes practically in his front yard under a street light, and with the warning to only make an identification if he was sure. Fur ther, our officers followed every procedure by the book, making sure not to give any hints before the identification, using method photo with the suspect in street clothes rather than a potentially biasing jailhouse jumpsuit and taking down his confidence level right away after then identification. Further , you heard the detective in charge of the case testify that Mr. Smith admitted to his crime during the interrogation and give details of the murder that only the true killer would know. (only in high qua lity confession conditions) Keep in mind these details were given by Mr. Smith during a 2 hour discussion with police. There was little influence on Mr. Smith to confess ; little influence other than his own guilt for the crime of course . How much more ev idence is needed to Bennet justice today when you deliberate and make your verdict selection. Please do what is right for this county by taking a killer off the streets and find Mr. Smith guilty. Closing Statement from Defense Attorney As my colleague has pointed out, this case boils down to 2 main pieces of evidence: an eyewitness , and a confession. Throughout the trial, we have shown problems with both , which is more than enough to demonstrate reasonable doubt. Reasonable doubt is some level of uncertainty in the guilt of my client, some amount of question as to his involvement in this crime. Reasonable doubt is the pressure put on a man during an interrogation that has b een shown to lead people to confess to crimes they did not commit. (only in low quality confession conditions) Th is pressure that is caused by having to sit in a little room with police officers drilling questions at you one after another for ten hours . That is enough pressure for many people to say something, anything to get out of the situation. I know it would be enough pressure for me. Reasonable doubt is how a man looking out his window in the darkness of the early morning and seeing another person w alking down the road. How is that man able to remember exact facial features of that person later on? (only in low quality eyewitness conditions) We have my client . The extended time delay between the crime and the identif i cation, the

PAGE 123

123 procedures employed by the officers of using a mug shot instead of a photo with everyday clothing, even the things the officers said before, during, and after the identification all h ave been shown to make the chances of that witness being able to accurately identify the perpetrator slim to none. To make matters worse, this man made his identification after seeing the person from a block away, in the dark, and for only a few minutes! And he was clearly wrong here today. (only in contamination present conditions) Reasonable doubt is the fact that the witness already knew the suspect had confessed even prior to ma king his identification. How can the officers expect him to have a clear state of mind if even the witness himself reports that he overheard about the confession right before making his identification? The prosecution is grasping for anything they can fin you all know it too. Thank you. Closing Judicial Instructions Members of the jury, I thank you for your attention during this trial. Please pay attention to the inst ructions I am about to give you. Michael Smith , the defendant in this case, has been accused of Murder in the Second Degree . If you find that Ms. Bennet was killed by Michael Smith , you will then consider the circumstances surrounding the killing in decid ing if the killing was Murder in the second degree or whether the killing was excusable or resulted from justifiable use of deadly force. To prove the crime of Second Degree Murder, the State must prove the following three elements beyond a reasonable dou bt: 1. Ms. Bennet is dead. 2. The death was caused by the criminal act of Michael Smith . 3. There was an unlawful killing of Ms. Bennet by an act imminently dangerous to another and demonstrating a depraved mind without regard for human life . An "act" in cludes a series of related actions arising from and performed pursuant ordinary judgment wou ld know is reasonably certain to kill or do serious bodily injury to another, and 2) is done from ill will, hatred, spite, or an evil intent, and 3) is of such a nature that the act itself indicates an indifference to human life. In order to convict of Se cond Degree Murder, it is not necessary for the State to prove the defendant had an intent to cause death. The defendant has entered a plea of not guilty. This means you must presume or believe the defendant is innocent. The presumption stays with the def endant as to each material allegation in the indictment through each stage of the trial unless it has been overcome by the evidence to the exclusion of and beyond a reasonable doubt. A reasonable doubt is not a mere possible doubt, a speculative, imaginary or forced doubt. Such a doubt must not influence you to return a verdict of not guilty if you have an abiding conviction of guilt. On the other hand, if, after carefully considering, comparing and weighing all the evidence, there is not an abiding convict ion of guilt, or, if, having a conviction, it is one which is not stable but one which wavers and vacillates, then the charge is not proved beyond every reasonable doubt and you must find the defendant not guilty because the doubt is reasonable. It is to t he evidence introduced in

PAGE 124

124 this trial, and to it alone, that you are to look for that proof. A reasonable doubt as to the guilt of the defendant may arise from the evidence, conflict in the evidence, or the lack of evidence. If you have a reasonable doubt, you should find the defendant not guilty. If you have no reasonable doubt, you should find the defendant guilty. To overcome the defendant's presumption of innocence, the State has the burden of proving the crime with which the defendant is charged was c ommitted and the defendant is the person who committed the crime. The defendant is not required to present evidence or prove anything. The evidence which has been admitted to show other crimes, wrongs, or acts allegedly committed by the defendant will be c onsidered by you only as that evidence relates to proof of motive or intent of the part of the defendant. It is up to you to decide what evidence is reliable. You should use your common sense in deciding which is the best evidence, and which evidence shou ld not be relied upon in considering your verdict. You may find some of the evidence not reliable, or less reliable than other evidence. You should consider how the witnesses acted, as well as what they said. Some things you should consider are: Did the w itness seem to have an opportunity to see and know the things about which the witness testified? Did the witness seem to have an accurate memory? Was the witness honest and straightforward in answering the attorneys' questions? Did the witness have some in terest in how the case should be decided? Does the witness's testimony agree with the other testimony and other evidence in the case? Has it proved that the witness had been convicted of a crime? Was it proved that the general reputation of the witness for telling the truth and being honest was bad? You may rely upon your own conclusion about the witness. A juror may believe or disbelieve all or any part of the evidence or the testimony of any witness. The defendant in this case has become a witness. You s hould apply the same rules to consideration of his testimony that you apply to the testimony of the other witnesses. Also remember that whether anyone else should be prosecuted and convicted for this crime is not a proper matter for you to consider. The po ssible guilt of others is no defense to a criminal charge. Your job is to decide if the government has proved this defendant guilty. Do not let the possible guilt of others influence your decision in any way. These are some general rules that apply to your discussion. You must follow these rules in order to return a lawful verdict. You must follow the law as it is set out in these instructions. If you fail to follow the law, your verdict will be a miscarriage of justice. There is no reason for failing to f ollow the law in this case. All of us are depending upon you to make a wise and legal decision in this matter. This case must be decided only upon the evidence that you have heard from the testimony of the witnesses and these instructions. This case must n ot be decided for or against anyone because you feel sorry for anyone, or are angry at anyone. Remember, the lawyers are not on trial. Your feelings about them should not influence your decision in this case. Your duty is to determine if the defendant has been proven guilty or not, in accord with the law. It is the judge's job to determine a proper sentence if the defendant is found guilty. Your verdict should not be influenced by feelings of prejudice, bias, or sympathy. Your verdict must be based on the e vidence, and on the law contained in these instructions.

PAGE 125

125 Deciding a verdict is exclusively your job. I cannot participate in that decision in any way. Please disregard anything I may have said or done that made you think I preferred one verdict over anoth er. You may find the defendant guilty as charged, or guilty of such lesser included crime as the evidence may justify or not guilty. In closing, let me remind you that it is important that you follow the law spelled out in these instructions in deciding yo ur verdict. There are no other laws that apply to this case. Even if you do not like the laws that must be applied, you must use them. For two centuries we have lived by the constitution and the law. No juror has the right to violate rules we all share.

PAGE 126

126 APPENDIX D POST TRIAL QUESTIONNAIRE FOR PILOT STUDY Instructions: Please circle the best answer to the following questions. Act as if you were a real juror who just watched the case. 1. Do you find the defendant guilty of the crime of murder in the first deg ree? Guilty Not Guilty 2. I believe the defendant should be found guilty of the charges brought against him. Strongly Disagree Disagree Slightly Disagree Slightly Agree Agree Strongly Agree 3. How confident are you in your above verdict decision? 0% -5 -10 -15 -20 -25 -30 -35 -40 -45 -50% -55 -60 -65 -70 -75 -80 -85 -90 -95 100% 4. Please explain, in as much detail as possible, your reasoning for selecting the above verdict decision. 5. Please explain, in as much detail as possible, how you weighed the evi dence to come to the above marked decision? Instructions: Please circle the best answer to the following questions. Act as if you were a real juror who just watched the case. 6. The confession information the investigator collected was handwritten by th e suspect. Yes No 7. The confession information was collected without coercion or force from the investigator. Yes No 8. The confession information was elicited after a short interrogation (less than 3 hours). Yes No

PAGE 127

127 9. The investigators who collected the confes sion from the suspect were blind to the guilt of the suspect (they were not the same officers who had completed the earlier stages of investigation ) . Yes No 10. The eyewitness was told the actual perpetrator may or may not be present in the lineup before maki ng her choice. Yes No 11. The eyewitness testified that he completed the lineup task 2 days after seeing the crime . Yes No 12. after she made her lineup selection. Yes No 13. The ey ewitness made her lineup selection using a sequential lineup (showing 1 picture at a time ) . Yes No 14. The eyewitness overheard people talking about the person who confessed before making her lineup selection. Yes No 15. The eyewitness was sheltered from hearing any information about the case prior to completing her lineup selection. Yes No 16. Yes No 17. Yes No 18. Do you have any suspicions as to the true purpose of the study? Yes No If so, please describe what you belie ve the study is examining:

PAGE 128

128 APPENDIX E POST TRIAL QUESTIONNAIRE FOR THE CURRENT STUDY Instructions: Please circle the best answer to the follo wing questions. Act as if you were a real juror who just watched the case. Verdict Questions 1. Do you find the defendant guilty of the crime of murder in the first degree? Guilty Not Guilty 2. How confident are you in your above verdict decision? 0% -5 -10 -15 -20 -25 -30 -35 -40 -45 -50% -55 -60 -65 -70 -75 -80 -85 -90 -95 100% 3. I believe the defendant should be found guilty of the charges brought against him. 1 2 3 4 5 6 Strongly Disagree Disagree Slightly Disagree Slightly Agree Agree Strongly Agree 4. W hat is the probability that you believe the defendant is actual guilty of the charges brought against him? 0% -5 -10 -15 -20 -25 -30 -35 -40 -45 -50% -55 -60 -65 -70 -75 -80 -85 -90 -95 100% 5. Please explain, in as much detail as possible, your reasoning f or selecting the above verdict decision. 6. Please explain, in as much detail as possible, how you weighed the evidence to come to the above marked decision? Confession Evidence Questions 7. I believe the confession is the strongest piece of evid ence in this case. 1 2 3 4 5 6 Strongly Disagree Slightly Slightly Agree Strongly

PAGE 129

129 Disagree Disagree Agree Agree 8. If I were the police investigator, I would have collected the confession information differently. 1 2 3 4 5 6 Strongly Disagree Disagree Slig htly Disagree Slightly Agree Agree Strongly Agree 9. I believe the confession was elicited under optimal conditions. 1 2 3 4 5 6 Strongly Disagree Disagree Slightly Disagree Slightly Agree Agree Strongly Agree 10. I thought the police officer explaining the de untrustworthy. 1 2 3 4 5 6 Strongly Disagree Disagree Slightly Disagree Slightly Agree Agree Strongly Agree 11. I am certain that the suspect confessed because he is truly guilty. 1 2 3 4 5 6 Strongly Disagree Disagree Slightly Disa gree Slightly Agree Agree Strongly Agree 12. I believe that jurors should consider the confession when determining the 1 2 3 4 5 6 Strongly Disagree Disagree Slightly Disagree Slightly Agree Agree Strongly Agree 13. I believe the confession in this case is a weak piece of evidence. 1 2 3 4 5 6 Strongly Disagree Disagree Slightly Disagree Slightly Agree Agree Strongly Agree 14. The information used to describe the confession evidence was helpful in deciding whether to believe the evid ence or not. 1 2 3 4 5 6 Strongly Disagree Disagree Slightly Disagree Slightly Agree Agree Strongly Agree 15. I used my knowledge about the confession when deciding on a verdict. 1 2 3 4 5 6 Strongly Disagree Disagree Slightly Disagree Slightly Agree Agree Strongly Agree Eyewitness Identification Evidence Questions 16. I believe the eyewitness identification is the strongest piece of evidence in this case. 1 2 3 4 5 6

PAGE 130

130 Strongly Disagree Disagree Slightly Disagree Slightly Agree Agree Strongly Agree 17. The eyewit ness identification evidence was not collected correctly. 1 2 3 4 5 6 Strongly Disagree Disagree Slightly Disagree Slightly Agree Agree Strongly Agree 18. I believe the identification was made under the best possible conditions. 1 2 3 4 5 6 Strongly Disagre e Disagree Slightly Disagree Slightly Agree Agree Strongly Agree 19. I thought the witness seemed untrustworthy. 1 2 3 4 5 6 Strongly Disagree Disagree Slightly Disagree Slightly Agree Agree Strongly Agree 20. I am certain that the person selected from the line up is truly guilty. 1 2 3 4 5 6 Strongly Disagree Disagree Slightly Disagree Slightly Agree Agree Strongly Agree 21. I believe the eyewitness identification should be considered by the jury when 1 2 3 4 5 6 Stro ngly Disagree Disagree Slightly Disagree Slightly Agree Agree Strongly Agree 22. I believe the eyewitness evidence was a high quality piece of evidence. 1 2 3 4 5 6 Strongly Disagree Disagree Slightly Disagree Slightly Agree Agree Strongly Agree 23. The details about the lineup were important in deciding whether to believe the evidence or not. 1 2 3 4 5 6 Strongly Disagree Disagree Slightly Disagree Slightly Agree Agree Strongly Agree 24. When choosing a verdict, I used my knowledge about the lineup. 1 2 3 4 5 6 Strongly Disagree Disagree Slightly Disagree Slightly Agree Agree Strongly Agree General Evidence Questions 25. I believe confessions are usually unreliable. 1 2 3 4 5 6 Strongly Disagree Disagree Slightly Disagree Slightly Agree Agree Strongly Agree 26. I wou ld not rely on confessions if I were deciding on verdict in a real criminal trial. 1 2 3 4 5 6

PAGE 131

131 Strongly Disagree Disagree Slightly Disagree Slightly Agree Agree Strongly Agree 27. I believe confessions can be coerced out of a suspect. 1 2 3 4 5 6 Strongly D isagree Disagree Slightly Disagree Slightly Agree Agree Strongly Agree 28. I think confessions are rarely problematic. 1 2 3 4 5 6 Strongly Disagree Disagree Slightly Disagree Slightly Agree Agree Strongly Agree 29. Suspects only confess to a crime when they ha ve guilty knowledge of the crime (meaning they actually committed the crime in question). 1 2 3 4 5 6 Strongly Disagree Disagree Slightly Disagree Slightly Agree Agree Strongly Agree 30. If I were hearing this case in court, I would want to hear the informat ion provided with the eyewitness testimony. 1 2 3 4 5 6 Strongly Disagree Disagree Slightly Disagree Slightly Agree Agree Strongly Agree 31. Regardless of my reaction to the eyewitness in this trial, I generally think that eyewitnesses are unreliable. 1 2 3 4 5 6 Strongly Disagree Disagree Slightly Disagree Slightly Agree Agree Strongly Agree 32. Eyewitnesses base their testimony only on their memory for the crime. 1 2 3 4 5 6 Strongly Disagree Disagree Slightly Disagree Slightly Agree Agree Strongly Agree 33. Ey ewitness testimony can be influenced after a lineup identification has been made. 1 2 3 4 5 6 Strongly Disagree Disagree Slightly Disagree Slightly Agree Agree Strongly Agree 34. testimony of what they saw. 1 2 3 4 5 6 Strongly Disagree Disagree Slightly Disagree Slightly Agree Agree Strongly Agree Reliability Questions 35. How reliable do you think the eyewitness is in this case? 1 2 3 4 5 6 Very Unreliable Very reliable

PAGE 132

132 36. How reliable do you think the detective is in this case? 1 2 3 4 5 6 Very Unreliable Very reliable 37. In general, how reliable do you think eyewitness testimony is? 1 2 3 4 5 6 Very Unreliable Very reliable 38. In general, how reliable do you think detectives are? 1 2 3 4 5 6 Very Unreliable Very reliable Demographic Information Questions 39. The confession was handwritten by the suspect. Yes No 40. The confession was collected without coercion or force from the investigator. Yes No 41. The confession was given after a short interrogation (less than 3 hours). Yes No 42. The investigators who heard the confession did not know about the guilt of the suspect (they were not the same officers who had completed the earlier stages of investigation ) . Yes No 43. The eyewitness was to ld that the actual perpetrator may or may not be present in the lineup. Yes No 44. The eyewitness completed the lineup task 2 days after seeing the crime . Yes No 45. The police officers said the eyewitness made the lineup selection . Yes No 46. During the lineup, t he eyewitness saw 1 picture at a time. Yes No 47. Before making the lineup selection, t he eyewitness overheard people talking about the person who confessed. Yes No 48. The eyewitness did not hear any information about the case prio r to completing the lineup selection. Yes No 49. the confession. Yes No 50. Please circle your gender.

PAGE 133

133 Male Female 51. Please circle the racial category you identify with most. White Black Hispanic Asian Other 52. Please write your age here: 53. Please circle your school classification. Freshman Sophomore Junior Senior Graduate 54. Yes No 55. Were you given any false information in this case description? Yes N o If yes, what was false? 56. Do you believe what you were told about this case? Yes No If no, what was not believable?

PAGE 134

134 APPENDIX F PILOT STUDY DEBRIEFING Thank you for completing our study. This current study is pilot testing materials to ensure th ey are strong enough to use in a larger study later this semester. While participating, you read 1 of 6 versions of a trial summary with varying levels of evidence quality. One of the factors we are looking for in this pilot study is to ensure our verdic t distribution for the summaries with high quality evidence have a great proportion of guilty verdicts than the summaries with only low quality evidence. Thus, we are hoping to see more people say the defendant is guilty in the conditions with confession and eyewitness evidence collected under ideal conditions (i.e. no coercion), and more people say the defendant is NOT guilty in the conditions where the confession or eyewitness evidence was collected poorly (i.e. with coercion). We are also interested in how mock jurors consider potentially contaminated evidence (evidence that was collected with prior knowledge of other evidence, thus possibly biasing their interpretation of any subsequently learned evidence). For this pilot study, we are testing to ensur e the manipulations are noticeable, meaning that most of you are able to identify whether a piece of evidence was collected after already having knowledge of another piece of evidence. More specific to this study, we are making sure that students can reme prior to making their lineup selection. You should know that any and all results from this study will only be reported in group form, meaning your individual responses will never be identified. If you would like to have your data not included in the study, please inform the researcher now. Last, please do not tell any other students about the study procedures or hypotheses so they can begin their participation with a clean, unbiased slate. Thank you again for your participation.

PAGE 135

135 APPENDIX G CURRENT STUDY DEBRIEFING Thank you for completing our study. While participating, you read 1 of 6 versions of a trial summary with varying levels of evidence quality. Everyone heard confession evidence, how ever some heard about a high quality confession and others heard about a low quality confession. Again, everyone heard about eyewitness evidence, however some heard high quality and some heard low quality eyewitness evidence. We also included the potenti al for contamination in some conditions; this is when the potential exists that the confession information was heard by the eyewitness before making their identification, and thus may have influenced the identification. Finally, we varied the presentation order across all conditions (meaning that some of you heard about the eyewitness then the confession and others heard about the confession then the eyewitness) to ensure the timing of evidence was controlled for. We expect that jurors are able to identify and properly weigh high versus low quality evidence (meaning they will correctly give a guilty verdict when high quality evidence is presented with both the confession and eyewitness and correctly give a not guilty verdict when low quality evidence is pre sented with both the confession and eyewitness), however we do not yet know how community members (thus possible future jurors) account for the potential contamination that can occur this study is examining just that. We also do not know how the quality of evidence might affect that ability to appropriately weigh potentially contaminated evidence. So, thank you for helping us figure it out! You should know that any and all results from this study will only be reported in group form, meaning your individ ual responses will never be identified. If you would like to have your data not included in the study, p lease inform the researcher now. Last, please do not tell any other students about the study procedures or hypotheses so they can begin their participation with a clean, unbiased slate. Thank you again for your participation.

PAGE 136

136 LIST OF REFERENCES Ask, K. , Rebelius, A., & Granhag, P. (2008). The 'elasticity' of criminal evidence: A moderator of investigator bias. Applied Cognitive Psychology , 22 (9), 1245 1259. The Benjamin N. Cardozo School of Law at Yeshiva University. (2008). The Innocence Project. www. innocenceproject.org. Boring, E. G. (1930). A new ambiguous figure. The American Journal Of Psychology , 42 444 445. Bornstein, B. (2004). The impact of different types of expert scientific testimony on mock Psychology, Crime and Law , 10(4), 429 446. Cooper, J., Bennett, E. A., & Sukel, H. L. (1996). Complex scientific testimony: How do jurors make decisions?. Law And Human Behavior , 20(4), 379 394. Charman, S. D. (2013). The forensic confirmation bias: A problem of evidence integ ration, not just evidence evaluation. Journal Of Applied Research In Memory And Cognition , 2(1), 56 58. Davis, J.H. (1973). Group decision and social interaction: A theory of social decision schemes. Psychological Review , 80, 97 125. Dror, I. & Charlton, D . (2006). Why Experts Make Errors. Journal of Forensic Identification. 56(4), 600 617. Dror, I., Charlton, D., & Peron, A. (2006). Contextual Information Renders Experts Vulnerable to Making Erroneous Identifications. Forensic Science International, 156, 7 4 78. Dror, I. E., Péron, A. E., Hind, S., & Charlton, D. (2005). When Emotions Get the Better of Us: The Effect of Contextual Top down Processing on Matching Fingerprints. Applied Cognitive Psychology , 19 (6), 799 809. Elaad, E., Ginton, A., & Shakhar, G. (1994). The effects of prior expectations and outcome knowledge on polygraph examiners' decisions. Journal Of Behavioral Decision Making , 7(4), 279 292. Fulero, S. M. (2009). System and estimator variables in eyewitness identification: A review. In D. A. K rauss, J. D. Lieberman (Eds.) , Psychological expertise in court: Psychology in the courtroom, Vol. 2 (pp. 57 78). Burlington, VT US: Ashgate Publishing Co. Gabbert, F., Memon, A., & Allan, K. (2003). Memory conformity: Can eyewitnesses influence each oth er's memories for an event?. Applied Cognitive Psychology , 17(5), 533 543.

PAGE 137

137 Hasel, L.E. (2012). Evidentiary independence?: How evidence collected early in an investigation influences the collection and interpretation of additional evidence. In Lynn Nadel & Walter Sinnott Armstrong (Ed.) Memory and Law. New York. Oxford University Press. Hasel, L. E., & Kassin, S. M. (2009). On the presumption of evidentiary independence: Can confessions corrupt eyewitness identifications?. Psychological Science , 20 (1), 122 1 26. Hasel, L. E., Vogler, V. & Lusizzo, M. (2013) manuscript under review. Helgeson, J. G., & Ursic, M.L. (1993). Information Load, cost/benefit assessment and decision strategy variability, Journal of the Academy of Marketing Science , 21, 13 20. Heuer, L. , & Penrod, S. D. (1994). Trial complexity: a field investigation of its meaning and its effect. Law and Human Behavior , 18, 29 52. witne ss relationship in susceptibility to misinformation. Acta Psychologica , 127 (2). pp. 476 484. Horowitz, I. A., FosterLee, L. & Brolly, I. (1996). Effects of trial complexity on decision making. Journal of Applied Psychology, 81, 757 768. Inbau, F. E., Rei d, J. E., Buckley, J. P., & Jayne, B. C. (2013). Criminal interrogation and confessions (5th ed.). Burlington, MA: Jones & Bartlett Learning. Kassin, S. M. (1997). The psychology of confession evidence. American Psychologist , 52 (3), 221 233. Kassin, S. M. (1998). More on the psychology of false confessions. American Psychologist , 53(3), 320 321. Kassin, S. M. (2012). Why confessions trump innocence. American Psychologist , 67(6), 431 445. Kassin, S. M., Bogart, D., & Kerner, J. (2012). Confessions that corr upt: Evidence from the DNA exoneration case files. Psychological Science , 23 (1), 41 45. Kassin, S. M., Drizin, S. A., Grisso, T., Gudjonsson, G. H., Leo, R. A., & Redlich, A. D. (2010). Police induced confessions: Risk factors and recommendations. Law And Human Behavior , 34 (1), 3 38. Kassin, S. M., Dror, I. E., & Kukucka, J. (2013). The forensic confirmation bias: Problems, perspectives, and proposed solutions. Journal Of Applied Research In Memory And Cognition , 2 (1), 42 52.

PAGE 138

138 Kassin, S. M., & Kiechel, K. L. (1996). The social psychology of false confessions: Compliance, internalization, and confabulation. Psychological Science , 7 (3), 125 128. Kassin, S. M., Meissner, C. A., & Norwick, R. J. (2005). 'I'd Know a False Confession if I Saw One': A Comparative Study of College Students and Police Investigators. Law And Human Behavior , 29 (2), 211 227. Kassin, S. M., & Sukel, H. (1997). Coerced confessions and the jury: An experimental test of the "harmless error" rule. Law and Human Behavior , 21(1), 27 46. Kien zle & Levett (2013). Manuscript under review Kovera, M., & McAuliff, B. D. (2000). The effects of peer review and evidence quality on judge evaluations of psychological science: Are judges effective gatekeepers?. Journal Of Applied Psychology , 85(4), 574 5 86. Kovera, M., McAuliff, B. D., & Hebert, K. S. (1999). Reasoning about scientific evidence: Effects of juror gender and evidence quality on juror decisions in a hostile work environment case. Journal Of Applied Psychology , 84(3), 362 375. Leeper, R. R. (1935). A study of a neglected portion of the field of learning the development of sensory organization. The Pedagogical Seminary And Journal Of Genetic Psychology , 46 41 75. Leo, R. A., & Drizin, S. A. (2010). The three errors: Pathways to false confessi on and wrongful conviction. In G. Lassiter, C. A. Meissner (Eds.) , Police interrogations and false confessions: Current research, practice, and policy recommendations (pp. 9 30). Washington, DC US: American Psychological Association. Leo, R. A., & Liu, B. (2009). What do potential jurors know about police interrogation techniques and false confessions? Behavioral Sciences & the Law , 27, 381 399. Levett, L. M. (2012). Co witness information influences whether a witness is likely to choose from a lineup. Leg al And Criminological Psychology , 18(1), 168 180. Levett, L. M., Danielsen, E. M., Kovera, M., & Cutler, B. L. (2005). The Psychology of Jury and Juror Decision Making. In N. Brewer, K. D. Williams (Eds.), Psychology and law: An empirical perspective (pp. 365 406). New York, NY US: Guilford Press. Levett, L. M., & Kovera, M. (2008). The effectiveness of opposing expert witnesses for educating jurors about unreliable expert evidence. Law And Human Behavior , 32(4), 363 374. Levett, L. M., & Kovera, M. (2009) . Psychological mediators of the effects of opposing expert testimony on juror decisions. Psychology, Public Policy, And Law , 15(2), 124 148.

PAGE 139

139 Nickerson, R. S. (1998). Confirmation bias: A ubiquitous phenomenon in many guises. Review of General Psychology, 2, 175 220. Paterson, H. M., Kemp, R. I., & Ng, J. R. (2011). Combating co witness contamination: Attempting to decrease the negative effects of discussion on eyewitness memory. Applied Cognitive Psychology , 25 (1), 43 52. Pennington, N., & Hastie, R. (1993 ). The story model for juror decision making. In R. Hastie (Ed.), Inside the juror: The psychology of juror decision making (pp. 192 221). New York, NY US: Cambridge University Press. Perillo, J. T., & Kassin, S. M. (2011). Inside interrogation: The lie, the bluff, and false confessions. Law And Human Behavior , 35 (4), 327 337. Rodriguez, D. N., & Berry, M. A. (2010). System and estimator variables, eyewitness confidence, and the postidentification feedback effect. American Journal Of Forensic Psychology , 2 8 (2), 17 37. Sommers, S., & Douglass, A. (2007). Context matters: alibi strength varies according to evaluator perspective. Legal and Criminological Psychology , 12(1), 41 54. Sweeny, D. J. (1995). An Analysis of Harmless Error in Washington: A principled process. 31 Gonz. L. Rev. 277. Thibodeaux, Damon v. Louisiana, (1997). http://www.innocenceproject.org/Content/Damon_Thibodeaux.php. Thompson, W. C. (1997). A Sociological Perspective on the Science of Forensic DNA Testing. U.C. Davis Law Review , 30(4) 11 13 1136. Wallace, D., & Kassin, S. M. (2012). Harmless error analysis: How do judges respond to confession errors?. Law And Human Behavior , 36(2), 151 157. Wells, G. L., Small, M., Penrod, S., Malpass, R. S., Fulero, S. M., & Brimacombe, C. E. (1998). Eyew itness identification procedures: Recommendations for lineups and photospreads. Law And Human Behavior , 22(6), 603 647. Wilford, M. M., & Wells, G. L. (2013). Eyewitness system variables. In B. L. Cutler (Ed.) , Reform of eyewitness identification procedur es (pp. 23 43). Washington, DC US: American Psychological Association. Yarmey, A., Yarmey, M. J., & Yarmey, A. (1996). Accuracy of eyewitness identifications in showups and lineups. Law And Human Behavior , 20 (4), 459 477.

PAGE 140

140 BIOGRAPHICAL SKETCH Megan Kienzl e is currently an Instructor at The College at Brockport (SUNY), but will transition into a tenure track Assistant Professor after the completion of this doctoral degree. Originally from Iowa, Megan graduated from Cedar Falls High School in 2004 and moved to Ames to attend Iowa State University f or her undergraduate educ a tion. Majoring in p sychology, she took a juvenile delinquency class her freshman year, added a criminal justice s tudies minor, and from then on has worked to combine her love of psycholog y and interest in criminology. During undergraduate, she worked in many research labs, two of which helped to refine her interest in psychology and law; the first was working with Dr. Douglas Epperson on validating the Juvenile Sex Offender Recidivism Risk Assessment Tool (JSORRAT) to predict future offending in juveniles with sexual offense histories. During her senior year, Megan worked in the lab of Dr. Gary Wells to examine eyewitness identification issues (specifically, the post id entification feedbac k effect). The time spent in both labs proved invaluable in After graduation in the spring of 2008, Megan came to the University of Florida to attend graduate school in c riminology. Working with Dr. Lora Levett, she has continued her research in psychology and law. During her time at UF, Megan has had the opportunity to work on two NSF grants with Dr. Levett (the first on co witness influence, the second applying the story model of juror de cision making to the group (jury) context). I n 2011, Megan received her MA after completing her thesis on alibi generation and evidence interaction (funded by t he APLS Grants In Aid program). She taught three courses while at UF ( Death Penalty , Psycholog y and Law , and Law and Society ). She w as also actively involved in service activities during graduate school,

PAGE 141

141 particularly with the department graduate group the Syndicate; she was on the Executive Board for the Syndicate for four years. While at Brockport , she has had the opportunity to teach four new courses ( Investigations, Police Process, Introduction to Criminal Justice, and Research Methods ). She i s also actively pursuing research with her new colleagues and finishing up several projects that began while in graduate school.