Research Projects:
On this page you'll find a summary of the research projects I'm currently working on. For more information, feel free to contact me directly. Please do not cite anything on this page.
Ross, S. J, & Malpass, R. S. “An evaluation of the field studies on eyewitness identification procedures.”
The recent publications of lineup evaluations from Hennepin County, Minnesota and Illinois have sparked much controversy over the relative effectiveness of sequential and simultaneous lineup procedures. At the base of this controversy is the contention that field evaluations are highly sensitive to aspects of the study environment and, therefore, researchers must be diligent to ensure that proper field evaluation techniques are being utilized when conducting this research. However, commentators have overlooked that we now have nine published reports evaluating eyewitness identification practices in actual police investigations. These reports provide us with an opportunity to evaluate the influence of various system and estimator variables across time and jurisdiction. The present research was conducted to compare, and combine where possible, the results from these studies to determine the influence of specific system and estimator variables on eyewitness identification rates. Results from this project were presented at the “Off the witness stand” conference at John Jay College in March, 2007. Additional results will be presented at the annual meeting of the American Psychology-Law Society in Jacksonville, FL in March 2008. In addition, we submitted a manuscript of this research to Law and Human Behavior. Please feel free to e-mail me with any questions about this research or for an advanced copy of the manuscript.
Ross, S. J. & Pickel, K. L. "Evaluating the ‘suggestiveness’ of showups: Going beyond the 'lack of protection' aspect."
Currently, there is a fairly large body of research that has examined the efficacy of showups (presentation of a single suspect to eyewitnesses for the purpose of identification). Contrary to the intuitive beliefs of researchers and other legal scholars, this research has suggested that showups are not more suggestive than lineups and, in fact, lead to more accurate eyewitness identifications than lineups (for a review see Steblay, N., Dysart, J., Fulero, S., & Lindsay, R. C. L. (2003). Eyewitness accuracy rates in police showup and lineup presentations: A meta-analytic comparison. Law and Human Behavior, 27(5), 523-540.). This research, however, only looked at one aspect of showups that was believed to be suggestive by the courts; lack of protection. In comparison to lineups, showups offer no protection to an innocent suspect when he/she is presented to an eyewitness that feels compelled to choose anyone. In a properly constructed lineup, the likelihood of misidentifying an innocent suspect will be equal to 1/k (where k equals the number of individuals presented in the lineup). In a showup, the likelihood of misidentification by an eyewitness that feels compelled to identify anyone is also 1/k, however k always equals 1 and therefore the likelihood of misidentification is equal to 100% in a situation where the eyewitness feels compelled to pick anyone. According to the previous research, this "lack of protection" does not appear to increase the likelihood of misidentification. However, drawing ultimate conclusions that showups are more accurate than lineups is troublesome since the research only focused on one specific aspect of showups that was believed to be suggestive by the Courts. The Courts have also proposed that "custodial influence" (the presentation of an individual in the custody of police officers) may be unduly suggestive and therefore bias the eyewitness' identification. The present research was designed to assess whether custodial influence increases the likelihood of identification of a suspect in a showup, regardless of whether the suspect is the perpetrator of the crime. Results of this research suggest that custodial influence increases both the likelihood of identification of a suspect, regardless of guilt or innocence, as well as the confidence the eyewitness has in his/her identification. I am currently in the process of preparing a manuscript of this research to send out for review.
Ross, S. J. “Toward creating a ‘better’ measure of legal authoritarianism: A factor analysis of 4 current measures.”
This research is intended to assimilate the current legal authoritarianism measures with the goal of understanding the underlying factor structure across all measures. An understanding of the underlying factor structure may help explain recent findings of poor convergent validity across the measures (Ross, 2005; Ross & Malpass, 2006a) as well as assist with the development of a single, parsimonious measure. Data collected from previous research I have conducted is currently being used to assess the factor structure inherent in these measures and to compare that structure across Anglo and Hispanic cultures. To date, the data includes responses on these measures from 696 Anglo participants and 210 Hispanic participants. With the assistance of Dr. Roy Malpass and Dr. Ozzie Morera, we have just begun the process of conducting the initial factor analysis and will then determine model fit by using confirmatory factor analyses.
Ross, S. J., Morera, O. F., & Hosch, H. M. "Determining the factor structure of the Revised Legal Attitudes Questionnaire."
Since the development of the Legal Attitudes Questionnaire (LAQ) by Boehm in 1968, psycholegal researchers have used the measure to assess the construct of legal authoritarianism in research participants and jurors. While the LAQ has been used in numerous research projects, it is interesting to note that the factor structure of the measure has never been confirmed. According to Boehm (1968), the measure consists of three factors relating to authoritarianism, anti-authoritarianism, and equalitarianism. However, no research has ever confirmed this factor structure. Kravitz, Cutler, & Moran (1993) revised the LAQ into a more internally consistent measure and conducted the only known factor analysis of the questionnaire. They, however, did not believe that the measure assessed three independent factors. Instead, they attempted to fit a single factor model to the data believing that the singular factor was "authoritarianism". Goodness of fit indices did not support a single factor model and they report that numerous exploratory factor analyses did not reveal an interpretable factor structure. The purpose of this current research is to attempt to determine the factor structure of the revised LAQ developed by Kravitz et al. (1993). To date, we have conducted confirmatory factor analyses of three hypothesized factor structures. The analyses suggest that a structure proposed by Couch & Sundre (2001) is the most appropriate factor structure for the measure. We are currently writing up a manuscript for submission.
Ross, S. J. & Morera, O. F. “A cross-cultural comparison of the RLAQ: Is the factor structure the same for Anglo and Hispanic respondents.”
This project is an extension of the work that we have done on the RLAQ measure of legal authoritarianism. As stated earlier, we have been able to confirm the factor structure of the RLAQ in a primarily Anglo sample. This project is intended to compare the factor structures of the RLAQ across Anglo and Hispanic samples. We used a MIMIC approach to confirmatory factor analysis and discovered that the factor structure fits equally well in Hispanic and Anglo samples, however there are sample differences on the latent variables. We have just finished conducting these analyses and are in the process of preparing writing a manuscript for submission.
Ross. S. J. “A re-examination of the authoritarianism-verdict association: Is the reliability of legal authoritarianism measures influencing the strength of the association?”
This research was developed to assess whether the diminishing reliability of legal authoritarianism measures may explain the findings of a previous meta-analysis that I conducted which showed a decrease in the strength of the association between legal authoritarian attitudes and juror’s verdicts. A meta-analysis published in 1993 (Narby, Cutler, & Moran) reported that legal authoritarian ideology was positively correlated with jurors’ verdicts such that those jurors who held a more authoritarian attitude were more likely to convict. Last year, I conducted a meta-analysis of the studies conducted since the published 1993 meta-analysis and discovered that the strength of the association has appeared to weaken (Ross & Malpass, 2006b). Therefore this project was developed to serve two purposes. First, it was intended to update the original 1993 meta-analysis with all relevant research that has since been conducted. In addition, the observed effect sizes would be corrected for the reliability of the measures used. These "corrected" effect sizes will then allow for assessing whether the diminishing reliability has influenced the observed effect sizes. To date, I have finished coding the previous studies and have calculated an updated meta-analysis without correcting for the reliability of the measures. I am currently in the process of adjusting the observed effect sizes for the reliability of the measures. Once the effect sizes have been adjusted, the results will be compared and a manuscript will be completed.
Ross, S. J. & Tafoya, M. R. “Are jurors influenced by the subliminal presentation of emotionally-valenced words?”
This research was developed based upon a “call to arms” from legal scholars. The attorney’s for Michael Skakel are currently appealing the conviction of their client on the basis that the prosecution unfairly manipulated the jury during their (prosecution’s) closing argument. As you may recall, Skakel is a relative of the Kennedys and was convicted in for the murder of a teenage neighbor in 1980. Skakel’s attorneys feel that the prosecution’s use of graphic photos and a multimedia presentation during their closing argument manipulated the jury and primed them to vote guilty. Legal scholars have since argued that the appellate court has no grounds to rule on this appeal since social science has not directly tested the influence that multimedia presentations may play in juror’s verdicts. Therefore, this research was developed to determine whether the subliminal presentation of emotionally-valenced words during a closing argument could influence jurors’ verdicts. Mock-jurors read a transcript of a trial during which emotionally-valenced words are subliminally presented during a closing argument and then render verdicts, assess punitive damages, and complete attitude questionnaires. Harmon Hosch is my faculty advisor on this project. M. Ruben Tafoya is an undergraduate student who has assisted with data collection. To date, we have collected responses from 283 participants and we are currently collecting manipulation check data to assess the effectiveness of our manipulations. While no results have been finalized, preliminary results of the project were presented as a poster at the most recent AP-LS conference in St. Petersburg, FL (Tafoya, Leyva, Ross, & Hosch, 2006) and other results of this project have been accepted as a paper presentation for the 3rd International Congress of Psychology and Law being held from July 3-8, 2007 in Adelaide, Australia..
Malpass, R. S., Ross, S. J., Topp, L. D., & Susa, K. J. “Developing a face domain with known suspect-filler similarities.”
While much research has been conducted in order to better understand what factors influence eyewitness identifications, we have come to realize that we don’t know very much at all about the stimuli (i.e. lineups) that we are using in relation to one important variable – suspect-filler similarity. The primary goal of this project is to assemble a set of about 100 faces which represent a large range of degrees of similarities between the target and each face. Once we know the quantitative similarities (similarity scale values) of the faces in the domain we can construct lineups that conform to specific suspect–filler similarity requirements. The ability to develop lineups with specific suspect-filler similarities will enable us to examine specific questions pertaining to lineups in general. Also, this ability may allow us to clarify and explain the reasons behind contradictory findings across some studies. The face domain was developed using a Q-sorting task. In this task, individuals sort faces in relation to how similar they are to a target individual. Upon completion of this sorting task, we will have a Likert-type range of faces from 1 (very dissimilar) to 9 (very similar). The faces that make up the domain will consist of those that have high reliability/agreement within each level of the Likert-type range.
In the past year we have developed this domain to use as stimuli in other projects within the laboratory. Results of this project have indicated that inter-rater reliability is quite high for dissimilar faces, but lowers as the degree of similarity between faces increases. In other words, we tend to agree on what is “dissimilar”, but we do not agree on what makes faces “similar”. We are currently using this domain to assist us in understanding what types of faces construct “good” lineups and what information witnesses tend to make in mock-witness evaluations.