|
|
For the past
three years, I have been studying how the methods used to present police
photographic lineups affect eyewitness identification accuracy,
especially in regards to the similarity of the faces presented within the
lineup. My interest in this topic originated from the results of a meta-analysis Dr.
Ebbesen and I conducted on lineup identification experiments.
Participants in these studies typically view a videotaped crime
reenactment and then later attempt to identify the perpetrator from a
number of distractor faces. About 95% of the time, the distractor faces
presented in the lineup task were selected as stimuli because, in the
opinion of the investigator, they appeared “similar looking”
to the face of the “perpetrator.” I realized that some of the
inconsistent findings reported in the literature on the relationship
between lineup member similarity and identification accuracy might have
occurred because researchers are using differing standards to create
lineups. In addition, I wondered whether identification performance in a
sequential lineup (a lineup that is presented to a witness one photograph
at a time) might be affected by lineup member similarity. Thus, I decided
that my contribution to this research area would be to discover whether
lineup member similarity influences the decision strategy participants
adopt in simultaneous and sequential lineups.
Since beginning
work on this topic, I have shown that lineup member similarity ratings do
not predict lineup choices made by mock witnesses (i.e., people who do
not view the perpetrator and who are asked to pick the person out of the
lineup who seems to stand out the most) who view actual police lineups.
This work led me to appreciate the complexity involved in measuring face
similarity, and to use composite drawing software as a baseline method
for systematically varying the number, type, and spatial configuration of
features across faces in a lineup to determine their effects on accuracy.
Analysis of these data showed that the decision standard participants
adopt is affected by lineup member similarity (Flowe & Ebbesen, 2006;
Flowe, Ebbesen, & Besemer, 2006). Additionally, lineup member
similarity affects decision criterion placement not only in simultaneous
lineups, but in sequential lineups as well, suggesting that when viewing
a sequential lineup, participants compare faces in working memory. I plan
to extend these findings this Fall by determining whether sequential
lineup procedures might be improved by including specific types of
criterion setting instructions, and whether relative comparisons can be
prevented by including a visual mask between faces in the series, and if
so, to describe how identification performance is affected.
I am also
currently working on developing a theory that describes the relationship
between lineup member similarity and recognition accuracy in simultaneous
and sequential lineups. Experts in the field have suggested that
investigators should select lineup foils on the basis of an
eyewitness’ description of the perpetrator rather than selecting
foils that match the perpetrator’s photo. The idea is that lineups
created on the basis of matching the perpetrator’s description will
protect innocent suspects while allowing for the identification of the
guilty. My dissertation research, however, suggests that selecting foils
on the basis of an eyewitness’ description may actually work to
increase false alarms to innocent suspects. If the similarity of lineup
members created using the match description strategy is low, then
innocent suspects might be erroneously identified more often using the
match description compared to the match photo strategy. To further
investigate this question, I have just completed data collection on a
project that examines eye-movement patterns during showups and sequential
and simultaneous lineup tests. These data will be analyzed to determine
the effects of facial feature similarity on identification decision
processes. A planned followup study is to examine the similarity of the
lineups produced using the match suspect and match photo strategies, and
to determine what features are typically matched in constructing these
lineups, and how these features affect accuracy.
Also on my
research agenda is to provide an answer to the question of where in a
sequential lineup should the suspect be placed. Empirical work on this
issue is scant, and current police guidelines simply indicate that the
suspect should not be placed first or last in the series. Currently, I am
testing a mathematical model of sequential lineup identification that
includes suspect position as a parameter to determine whether the
position of the suspect in the lineup affects accuracy. Toward this end,
we have amassed a dataset that includes accuracy rates and similarity
measures for more than 80 lineups. The results thus far demonstrate that
participants lower their decision standard as they progress through the
series of faces in the lineup. Consequently, false alarms increase as the
“innocent suspect” is placed in later positions.
Additionally, the hit rate decreases when the “perpetrator”
is placed earlier in the series. This paper will be submitted to Psychological Review in December
2006.
Another line of
work I am pursuing involves examining factors that affect the
identification performance of real world eyewitnesses. Dr. Ebbesen and I
have compiled extensive information from over 1100 actual cases involving
eyewitnesses. As such, we will be able to verify whether the effects
reported in the literature of laboratory eyewitness identification
studies present in real world cases. Additionally, we are currently
finalizing a manuscript in which we compare the parameters (which include
variables such as exposure to violence, characteristics of witnesses and
perpetrators, and lineup identification conditions) of real world cases
to those used in laboratory investigations. The goal of the project is to
describe areas in which further laboratory research is needed so as to
increase the generalizability of work we do in the lab to real world
cases. Our findings will be submitted for publication November 2006.
My research
interests also extend beyond gaining an understanding of the decision
processes that operate during eyewitness identification. I am examining
on a phenomenon that I term the “Santa Clause Effect”, which
refers to the belief held by prosecutors that juries are less likely to
return guilty verdicts during the December holiday season. I am comparing
the rate at which juries returned a guilty verdict, the rate at which
prosecutors issue charges, and the severity of the probation
officers’ sentencing suggestion over the course of the year. The
dataset I have accumulated is over a 5 year period, and this work is
scheduled to be completed in March 2007. In addition, I am also
interested in testing evolutionary models of rape that purport that women
avoid public places while ovulating as a counteradaptation to rape. The
data I have collected on this issue involved women keeping diaries of
their whereabouts as a function of menstrual cycle phase. I anticipate
that I will be able to resume work on this manuscript next spring.
Finally, Dr. Ebbesen have a paper in press at Law and Human behavior that
examined the relationship between women’s past sexual history and
false rape allegations (Flowe & Ebbesen, 2006), which was based on
research that I conducted under the support of a National Science
Foundation Graduate Fellowship Award.
|
|
