Ebbesen, E. B., & Flowe, S. (in press). Simultaneous
versus sequential lineups: What do we really know? Law and Human Behavior.
Previous research has shown that participants using simultaneous
lineups rely on relative judgment strategies whereas participants using
sequential lineups rely on absolute judgments and that because of this,
sequential lineups result in fewer false alarms. However, Ebbesen and Flowe suggest an alternate view, that
participants are actually shifting their criteria for making an identification
from a more liberal one in simultaneous lineups to a stricter one in sequential
lineups. Theirs is a criterion-shift
model that is based on signal detection theory.
Ebbesen and Flowe make the claim that all identification
procedures, whether in simultaneous or sequential lineups, are comparison tasks
because participants are comparing photographs to the memory they have for the
face. Thus, a person, when matching faces
to their memory of the target, would have to have a criterion value to which
they would be willing to accept or reject a face. Therefore, they are arguing that participants use an “absolute”
degree-of-match to the information they have in memory.
Dual Process Model?
When presented with a lineup, participants could be making a relative
judgment (this is the guy that is most familiar) and then an absolute judgment
(does this guy surpass my absolute decision criterion?). Alternatively, they could make an absolute
judgment for each picture, and those that pass the absolute decision criterion
(if there are more than one) could be put through a relative judgment process
(which one of these three guys is the MOST like the guy I saw?)
By acknowledging the fact that identification could rely on
two processes, it is important to investigate what role serial position (the
order in the lineup that the target is placed) has on identifications. For instance, in a sequential lineup, if the
target is placed in the last position, but a foil that surpasses the absolute
decision criterion is placed in the second position and the participant accepts
the foil, then the participant would never even see the actual target. To test this, the authors conducted a
Monte-Carlo simulation. The results
suggest that the effect of serial position increases as d’ increases and
decreases as the criterion becomes stricter (see Figure 2, pg. 12). However, the serial position effects are
extremely small, so that could be why empirical research has never shown them
(because they might not have had enough power). Putting both innocent suspects and guilty culprits later in the
sequence decrease the odds that they will be chosen.
Sequential & Simultaneous Lineups & Signal Detection
Theory
By studying Figure 3 (pg. 17) we can see a representation of
sequential and simultaneous lineups in a signal detection model. It is important to note that one effect of
learning on recognition memory (in signal detection terms) is to increase the
variance of the distribution.
A Monte-Carlo simulation was conducted to determine how the
rate of false alarms to a target absent (TA) lineup and hits to a target
present (TP) lineup would change as a function of d’ and criteria placement in
simultaneous lineups. Results indicate
that an increase in the criterion value will result in larger changes in false
alarms than hits. (This means that if
witnesses/participants were instructed to use a stricter decision criterion in
a simultaneous lineup, that there would be a much larger reduction in false
alarm rates in TA lineups than to hit rates in TP lineups.)
A meta-analysis was conducted to determine the effect that
changing from simultaneous to sequential lineups has on hits and false alarms.
Major findings:
The average proportion of false alarms was lower in
sequential TA lineups (than sequential TP & simultaneous TP & TA).
20 of 21 studies found a higher false alarm rate in
simultaneous than sequential lineups and 9 of 13 studies found a higher hit
rate in simultaneous lineups.
Figure 9 shows that hit & false alarms were lower in
sequential lineups, but this effect was bigger for false alarms than it was for
hits (From 12 experiments where all of this information was included.)
The authors conclude from the data in Figures 11 & 12
that the primary difference between simultaneous and sequential lineups is that
participants place their decision criteria higher in sequential lineups (and go
on to say, that while more evidence is needed to verify this conclusion, that
the relative versus absolute model of identifications is unnecessary to explain
differences found in sequential and simultaneous lineups.
The authors point out several Procedural Uncertainties that
might also play a role in the differences between sequential & simultaneous
lineups, other than a stricter criterion)
The authors also point out several External Validity Issues
And finally, there are limitations to the application of
signal detection theory to eyewitness identifications, those being that assumed
distributions of memory strengths are from multiple witnesses (instead of many
data points from one person), criterion placements will be different for
different people, and finally, that this model makes the assumption that the
“degree of match” between someone’s memory of a face and a picture of a face
can be captured on a single dimension.