UCLA awarded grant to study error rates in fingerprint
Posted: Thu Feb 11, 2010 5:35 pm
http://www.newsroom.ucla.edu/portal/ucl ... 53642.aspx
Last February, the National Academy of Sciences issued a landmark report on the state of forensic science in U.S. crime laboratories, concluding that the accuracy and reliability of nearly all forensic methods, from ballistics to fingerprint analysis, had not been adequately established through rigorous scientific scrutiny.
In response to the troubling findings, the National Institute of Justice, the research and development arm of the U.S. Department of Justice, launched an effort to promote new research on forensic sciences. The agency recently awarded UCLA $866,764 to conduct a comprehensive study of error rates in latent fingerprint evidence.
Led by UCLA School of Law professor Jennifer Mnookin, the project, which began Jan. 1, will establish a scientific method to quantify the accuracy and error rates of latent fingerprint examination, one of the most widely used forensic disciplines.
In addition to Mnookin, a legal scholar specializing in scientific evidence and forensic science, the project's interdisciplinary research team includes UCLA cognitive psychology professor Philip Kellman, latent fingerprint expert David Charlton, and Itiel Dror, a researcher with a substantial track record in forensic science–related research. Both Charlton and Dror are with Cognitive Consultants International.
"This research will make a major step toward providing some of the much-needed information about accuracy and reliability for latent fingerprint identification," Mnookin said. "Even though we have been using fingerprint evidence in court for almost a hundred years, not nearly enough is known about how often fingerprint examiners might make mistakes, or in what circumstances.
"Right now, we do not even have a field-wide error rate, but the truth is that a field-wide error rate for latent fingerprint identification would be too general and too broad to be useful," she said.
The research project, which will be completed in the summer of 2012, is based on the insight that error rates for latent fingerprinting are likely a function of the inherent difficulty of the specific comparison.
"We expect accuracy will be a function of task difficulty, and not all fingerprint identification tasks are created equal," Mnookin said. "We would certainly predict that the error rate for the most difficult fingerprint identifications will be higher than the error rate for easy ones. But right now, there is no validated method for telling the difficult and the easy ones apart. By the end of this research project, we aim to have developed a scientific metric for assessing difficulty that could allow us to take a given pair of fingerprints and associate it with a potential error rate."
Knowledge about error rates is important to the courts in assessing reliability under Daubert v. Merrell Dow, a 1993 U.S. Supreme Court case that determined the standard for admitting expert testimony in federal courts. According to Mnookin, access to this information will help fingerprint examiners and fingerprint experts meet the Daubert criteria, or the state equivalent, and the research will also have broad benefits for the criminal justice system.
"Based on our research, courts could choose to enact different precautions or a different degree of scrutiny for the most difficult fingerprint identifications," she said. "The research will also provide a context for informing juries about the degree of confidence appropriate for any particular identification or exclusion. Depending on what we find, it might have possible effects on how laboratories go about their work as well."
The project, which is based at and managed by UCLA, will have four phases. In phase 1, a fingerprint database will be created; in phase 2, methods to quantify the visual complexity of fingerprints and comparisons will be examined and developed; in phase 3, a cognitive-difficulty analysis will be combined with a visual-complexity analysis to establish and test a difficulty classification scheme for fingerprint comparisons; and in phase 4, the potential for error for each difficulty classification level in the hierarchy will be determined.
The grant from the National Institute of Justice was awarded as part of a competitive solicitation directed specifically at promoting significant new research in the forensic sciences in the wake of the 2009 NAS report.
"There have been a number of challenges to the admissibility of fingerprint identification evidence in courts around the country," Mnookin said. "While nearly all courts have continued to admit fingerprint evidence, there are many scholars, myself included, who believe that it does not yet sit on an adequate research foundation. This research is designed to help develop that necessary research foundation."
Last February, the National Academy of Sciences issued a landmark report on the state of forensic science in U.S. crime laboratories, concluding that the accuracy and reliability of nearly all forensic methods, from ballistics to fingerprint analysis, had not been adequately established through rigorous scientific scrutiny.
In response to the troubling findings, the National Institute of Justice, the research and development arm of the U.S. Department of Justice, launched an effort to promote new research on forensic sciences. The agency recently awarded UCLA $866,764 to conduct a comprehensive study of error rates in latent fingerprint evidence.
Led by UCLA School of Law professor Jennifer Mnookin, the project, which began Jan. 1, will establish a scientific method to quantify the accuracy and error rates of latent fingerprint examination, one of the most widely used forensic disciplines.
In addition to Mnookin, a legal scholar specializing in scientific evidence and forensic science, the project's interdisciplinary research team includes UCLA cognitive psychology professor Philip Kellman, latent fingerprint expert David Charlton, and Itiel Dror, a researcher with a substantial track record in forensic science–related research. Both Charlton and Dror are with Cognitive Consultants International.
"This research will make a major step toward providing some of the much-needed information about accuracy and reliability for latent fingerprint identification," Mnookin said. "Even though we have been using fingerprint evidence in court for almost a hundred years, not nearly enough is known about how often fingerprint examiners might make mistakes, or in what circumstances.
"Right now, we do not even have a field-wide error rate, but the truth is that a field-wide error rate for latent fingerprint identification would be too general and too broad to be useful," she said.
The research project, which will be completed in the summer of 2012, is based on the insight that error rates for latent fingerprinting are likely a function of the inherent difficulty of the specific comparison.
"We expect accuracy will be a function of task difficulty, and not all fingerprint identification tasks are created equal," Mnookin said. "We would certainly predict that the error rate for the most difficult fingerprint identifications will be higher than the error rate for easy ones. But right now, there is no validated method for telling the difficult and the easy ones apart. By the end of this research project, we aim to have developed a scientific metric for assessing difficulty that could allow us to take a given pair of fingerprints and associate it with a potential error rate."
Knowledge about error rates is important to the courts in assessing reliability under Daubert v. Merrell Dow, a 1993 U.S. Supreme Court case that determined the standard for admitting expert testimony in federal courts. According to Mnookin, access to this information will help fingerprint examiners and fingerprint experts meet the Daubert criteria, or the state equivalent, and the research will also have broad benefits for the criminal justice system.
"Based on our research, courts could choose to enact different precautions or a different degree of scrutiny for the most difficult fingerprint identifications," she said. "The research will also provide a context for informing juries about the degree of confidence appropriate for any particular identification or exclusion. Depending on what we find, it might have possible effects on how laboratories go about their work as well."
The project, which is based at and managed by UCLA, will have four phases. In phase 1, a fingerprint database will be created; in phase 2, methods to quantify the visual complexity of fingerprints and comparisons will be examined and developed; in phase 3, a cognitive-difficulty analysis will be combined with a visual-complexity analysis to establish and test a difficulty classification scheme for fingerprint comparisons; and in phase 4, the potential for error for each difficulty classification level in the hierarchy will be determined.
The grant from the National Institute of Justice was awarded as part of a competitive solicitation directed specifically at promoting significant new research in the forensic sciences in the wake of the 2009 NAS report.
"There have been a number of challenges to the admissibility of fingerprint identification evidence in courts around the country," Mnookin said. "While nearly all courts have continued to admit fingerprint evidence, there are many scholars, myself included, who believe that it does not yet sit on an adequate research foundation. This research is designed to help develop that necessary research foundation."