Matching Items (6)
Filtering by
- All Subjects: psychology
Description
Evaluators of eyewitness evidence (e.g., judges, jurors) often must determine whether an eyewitness’s identification of a police suspect is accurate or mistaken. It has recently been argued that a particular class of variables—suspect-bias variables—pose a unique threat to the reliability of eyewitness identification evidence. Unlike “general impairment” variables that generally impair eyewitness identification accuracy (e.g., poor viewing conditions, biased lineup instructions), suspect-bias variables produce a suspect-specific bias that increases the risk of confident misidentifications of innocent suspects. The goal of this research was to examine evaluators’ sensitivity to suspect-bias variables compared to general impairment variables, and to test whether sensitivity to suspect-bias differs as a function of whether the suspect-bias variable is under the control of the legal system (system suspect-bias) or outside of the legal system’s control (estimator suspect-bias). Participant-evaluators (N = 214) read eight crime vignettes paired with one of four different eyewitness variables (system suspect-bias, estimator suspect-bias, general impairment, or no-variable control) and rated the accuracy of each eyewitness. Evaluators also explained the reasoning for their accuracy rating, and their explanations were coded for mentions of procedural suggestion, eyewitness memory strength, memory contamination, and general eyewitness (un)reliability. Evaluators appear to be more sensitive to general impairment variables than to suspect-bias variables. This finding is alarming, as suspect-bias variables pose a greater threat to eyewitness reliability than general-impairment variables. Implications for the collection and evaluation of eyewitness evidence are discussed.
ContributorsKulak, Kylie (Author) / Smalarz, Laura (Thesis advisor) / Salerno, Jessica (Committee member) / Schweitzer, Nick (Committee member) / Arizona State University (Publisher)
Created2022
Description
It has recently been argued that high-confidence eyewitness identifications are highly likely to be accurate regardless of the quality of viewing conditions experienced by the witness. However, new evidence suggests that evaluators of eyewitness identification evidence (e.g., jurors) do not trust highly confident eyewitnesses who experienced poor witnessing conditions. In fact, contextual information about poor witnessing conditions decreases evaluators’ belief of eyewitnesses to a greater extent for highly confident witnesses than for moderately confident witnesses. Why is the effect of witnessing-condition information greater for evaluations of high-confidence witnesses than for less confident witnesses? The current research tested the possibility that information about witnessing conditions influences evaluators’ perceptions of how well-calibrated a witness’s identification confidence is with the eyewitness’s accuracy. Using a paradigm adapted from the confidence calibration literature, I conducted an experiment to test this calibration account of the finding that witnessing condition information has a stronger effect on perceptions of highly confident witnesses than moderately confident witnesses. Although the results replicated the differential effects of witnessing condition context on perceptions of highly and moderately confident eyewitnesses, they failed to yield support for the confidence calibration hypothesis, potentially because the confidence calibration manipulation was ineffective. Directions for future research are discussed.
ContributorsLebensfeld, Taylor Cameron (Author) / Smalarz, Laura (Thesis advisor) / Salerno, Jessica (Committee member) / Arndorfer, Andrea (Committee member) / Arizona State University (Publisher)
Created2022
Description
The purpose of this research was to investigate the effect of the type of crime (namely, its perceived immorality) a juvenile is suspected of on how juvenile suspects are perceived (in terms of moral character, immaturity, and suggestibility) and, in turn, interrogated. I expected act-person dissociation to influence that effect. To that end, perceptions of crime (i.e., immorality, seriousness) were also investigated. The study was first conducted with law enforcement officers (n = 55), then replicated with laypeople (n = 171). Participants were randomly assigned to one of three crime conditions: robbery, sexual assault, and murder. In each condition, participants read a probable cause statement involving a 15-year-old suspect. There were several key findings: (1) Murder was the most serious crime, whereas robbery and sexual assault were more immoral. (2) Act-person dissociation did not occur. (3) Participants were more likely to endorse the use of psychologically coercive tactics on the juvenile suspected of sexual assault than the juvenile suspected of murder. (4) The more favorably participants perceived a juvenile’s moral character, the less likely they were to endorse the use of psychologically coercive interrogation tactics. (4) Participants who more strongly agreed that juveniles are more immature and suggestible than adults were less likely to endorse the use of psychologically coercive tactics, more likely to endorse the use of tactics that encourage compliance with interrogators, and more likely to adhere to the PEACE model of juvenile interrogations. The implications and limitations of these findings are discussed, along with potential directions for future research.
ContributorsFaison, Lakia (Author) / Mickelson, Kristin (Thesis advisor) / Smalarz, Laura (Committee member) / Salerno, Jessica (Committee member) / Arizona State University (Publisher)
Created2021
Description
In the legal system, the prediction of a person’s risk of committing a crime has mostly been based on expert judgment. However, newer techniques that employ machine learning (ML)—a type of artificial intelligence—are being implemented throughout the justice system. Yet, there is a lack of research on how the public perceives and uses machine learning risk assessments in legal settings. In two mock-trial vignette studies, the perception of ML-based risk assessments versus more traditional methods was assessed. Study 1 was a 2 (severity of crime: low, high) x 2 (risk assessment type: expert, machine learning) x 2 (risk outcome: low, high) between-subjects design. Participants expressed ethical concerns and discouraged the use of machine learning risk assessments in sentencing decisions, but punishment recommendations were not affected. Study 2 was a within-subjects design where participants were randomly assigned read through one of three crime scenarios (violent, white-collar, sex offense) and one of three risk assessment techniques (expert, checklist, machine learning). Consistent with Study 1, participants had ethical concerns and disagreed with the use of machine learning risk assessments in bail decisions, yet their own decisions and recommendations did not reflect these concerns. Overall, laypeople express skepticism toward these new methods, but do not appear to differentially rely on ML-based versus traditional risk assessments in their own judgments.
ContributorsFine, Anna (Author) / Schweitzer, Nicholas (Thesis advisor) / Salerno, Jessica (Committee member) / Smalarz, Laura (Committee member) / Arizona State University (Publisher)
Created2021
Description
Since the advent of DNA analysis, organizations such as the Innocence Project have been able to exonerate people who were wrongfully convicted of crimes, often due to erroneous forensic evidence analysis. In many cases, analytical techniques, such as fingerprint analysis, toolmark analysis, or hair comparisons have been cited as nearly infallible sources of evidentiary fact. However, these methods rely on subjective interpretation by a forensic examiner and lack stringent, quantitative methods for ensuring reliability and accuracy. For most of these methods, the examiner is supplied only with the unknown sample from the crime scene, and a known sample from a suspect. This, combined with the influence of psychological factors such as confirmation bias, has resulted in the need for a reliable mechanism of ensuring the efficacy of a particular type of analysis as well as the objectivity, and competence of the analyst. One proposed method to resolve these issues is the use of a filler-control method, in which analysts are given an “evidence line-up” containing at least three samples: the unknown sample from the crime scene, a sample from the suspect, and at least one filler sample from an individual who is not involved in the investigation. This method provides a reliable method for estimating error rates for an analyst and can provide the analyst with direct feedback about their performance to accurately gauge their competence. This method also helps to prevent the introduction of confirmation bias, as the source of the samples is unknown to the analyst. The goal of the current research is to test the capacity of a filler-control method to lead to better confidence-calibration of examiners’ match judgements when compared to the conventional method. The hypothesis of this experiment is that participants using the filler control method will have improved performance and increased confidence calibration due to receiving feedback over the course of the trials when compared to participants using the traditional method.
ContributorsRocha, Bethany (Author) / Smalarz, Laura (Thesis director) / Kukucka, Jeff (Committee member) / Barrett, The Honors College (Contributor) / School of Mathematical and Natural Sciences (Contributor)
Created2022-05
Description
Risk assessments are key legal tools that can inform a number of legal decisions regarding parole sentencing and predict recidivism rates. Due to assessments being
historically performed by humans, they can be prone to bias and have come under various
amounts of scrutiny. The increased capability and application of machine learning
technology has lead the justice system to incorporate algorithms and codes to increase
accuracy and reliability. This study researched laypersons’ attitudes towards these algorithms
and how they would change when exposed to an algorithm that made errors in the risk
assessment process. Participants were tasked with reading two vignettes and answering a series
of questions to assess the differences in their perceptions towards machine learning and
clinician-based risk assessments. The research findings showed that individuals lent more trust
to clinicians and had more confidence in their assessments when compared to machines, but
were not significantly more punitive when it came to attributing blame and judgement for the
consequences of an incorrect risk assessment. Participants had a significantly more positive
attitude towards clinician-based risk assessments, noting their assessments as being more
reliable, informed, and trustworthy. Participants were also asked to come to a parole decision
using the assessment of either a clinician or machine learning algorithm at the end of the study
and rate their own confidence in their decision. Results found that participants were only
significantly less confident in their decision when exposed to previous instances of risk
assessments with error, but that there was no significant difference in their confidence based
solely on who conducted the assessment.
ContributorsMa, Angeline (Author) / Schweitzer, Nicholas (Thesis advisor) / Powell, Derek (Committee member) / Smalarz, Laura (Committee member) / Arizona State University (Publisher)
Created2023