Effects of Lineup Member Similarity on Criterion Placement in Simultaneous and Sequential Lineups He

1
Effects of Lineup Member Similarity on Criterion
Placement in Simultaneous and Sequential
LineupsHeather D. Flowe Ebbe B.
EbbesenUniversity of California, San Diego
2
Acknowledgments
Professor Ebbe B. Ebbesen, Chair Professor
Nicholas Christenfeld Professor Garrison W.
Cottrell Professor Hugh Mehan Professor John T.
Wixted

• Anneka Besemer
• Crystal Fecht
• Marimer Santiago
• Jinny Burns
• And a small army of research assistants

3
Mistaken Identification

• Mistaken eyewitness ID often cited as 1 reason
  for erroneous convictions.
• The advent of DNA technology has allowed for
  conclusively determining whether a suspect was
  falsely identified.

4
Relative versus Absolute Responding

• Relative
• Face that is relatively the most similar to
• the culprit chosen
• Absolute
• Test face is strictly compared to culprit in
  memory

5
Do previously seen faces affect sequential IDs?

• Implication of absolute model is that lineup
  foils will not affect accuracy.
• Thus, similarity structure of the lineup should
  have little consequence in ID decisions.
• However, though direct visual comparisons impeded
  by sequential procedures, is it possible that
  meta-comparisons in working are memory made?

6
Effects of Similarity on Positive ID Criterion
Placement

• In the verbal learning literature, similarity of
  distractors on final recognition task influence
  criterion placement (Benjamin, 2005 Benjamin
  Bawa, 2004).

7
Effects of Similarity on Positive ID Criterion
Placement

• Similarity manipulations and lineup
  identifications (Clark Davey, 2005)
• Target IDs reduced when foil similar to perp is
  presented before the target (experiment 1)
• In target removed lineups, both simultaneous and
  sequential participants shifted their choice to a
  foil (experiment 1 2)
• Next best alternative chosen at a higher rate in
  removed lineups if presented later in the
  sequence (experiment 2)

8
Experiment 1

• Does foil similarity affect criterion placement
  in simultaneous and sequential lineups?
• Does order of the suspect in the lineup affect
  identifications depending on foil similarity?

9
Experiment 1

• Does foil similarity affect criterion placement
  in simultaneous and sequential lineups?
• Low compared to high similarity foils decrease
  the judgment standard, thereby increasing overall
  choice rates and positive identifications of the
  suspect.

10
Experiment 1

• Does order of the suspect in the lineup affect
  identifications depending on foil similarity?
• If suspect placed late rather than early in the
  lineup, high similarity lineups will lead to
  fewer positive IDs of suspect.
• In low similarity lineups, positive IDs will be
  greater if suspect is placed late rather than
  earlier in the sequence.

11
Experiment 1Design and Procedure
Study Session 12 faces presented to participants
(n294) for 10 sec each
Test Session
Target Similarity (identical or feature
substituted) Foil Similarity (feature matched or
random) ID Procedure (simultaneous, sequential,
or showup)
12
Experiment 1Face Stimuli
13
ResultsIs accuracy affected by presenting foils?

• Both hits and false alarms were reduced when
  foils were introduced, regardless of whether the
  lineup was presented simultaneously or
  sequentially.

Showup n134, Simultaneous n94, Sequential n93
14
ResultsSuspect Choices

• In simultaneous lineups, rate of picking target
  and look-a-like is higher if the foils are low
  rather than high in similarity.
• In sequential lineups, rate of picking
  look-a-like is higher if the foils are low rather
  than high in similarity.

15
ResultsFoil Choices

• Rate of picking foils did not differ depending on
  whether or not they matched the study face.
• Lineup presentation procedure did not affect rate
  at which foils were chosen.

16
ResultsLineup Rejections

• Sequential participants rejected the lineups at a
  higher rate overall than simultaneous
  participants.
• Rejections significantly higher overall for
  matched compared to random lineups.

17
ResultsOrder Effects

• Matched lineups
• Sequential participants missed the target more
  often when presented late in the sequence (Early
  M.48 versus Late M.36)
• Random lineups
• Suspect choices higher if presented later rather
  than earlier in the sequence (Early M.24 versus
  Late M.30)

18
Experiment 1Conclusions

• Criterion placement is affected by similarity
  characteristics of the lineup
• Innocent look-a-like chosen more often in low
  compared to high similarity lineups
• True for both simultaneous and sequential lineups
  

19
Future Directions

• Are sequential witnesses affected by the
  similarity of the first face to the culprit, or
  only after several faces in the sequence have
  been predicted?
• How similar to the culprit does the innocent
  suspect have to be in order for the similarity
  effects to hold?
• Are the effects we reported stronger if
  foil-culprit similarity is increased further?

20
Conclusion

• Simultaneous and sequential witnesses affected by
  similarity characteristics of the lineup in much
  the same way.
• Future work examining lineup member similarity
  should take into account the position of the
  suspect in the lineup, as order effects may arise
  as a result of similarity manipulations.

21
(No Transcript)
22
Extras
23
Experiment 2

• When the target is removed, how is the
  distribution of choices affected?
• Is the most popular foil the same lineup member
  in both procedures?

24
Experiment 2Predictions

• When the target is removed, how is the
  distribution of choices affected?
• Under the relative judgment model, removing the
  target should shift choices to the next best
  lineup member.
• An absolute model predicts that lineup rejections
  will increase when the target is removed.

25
Experiment 2Predictions

• Is the most popular foil the same lineup member
  in both procedures?
• If so, suggests that the test faces are
  considered in a similar manner in simultaneous
  and sequential lineups.

26
Experiment 2Design and Procedure
Study Session 12 faces presented to participants
(n48) for 10 sec each
Test Session
All lineups were blank Foil Similarity (feature
matched or random) ID Procedure (simultaneous or
sequential)
27
Experiment 2Design and Procedure
Similarity Assessment Another group of
participants (n272) rank ordered the foils with
respect to the study face from most to least
similar. Comparison Method (perceptual or
memory) controlled between participants
28
Experiment 2Full versus Removed LU Results
29
Experiment 2Ranking Results

• Rankings predicted the most popular foil (in both
  full and removed lineups) equally well in the
  simultaneous and sequential conditions.
• Memory and perceptual rankings predicted equally
  well.
• Agreement across the two lineup presentation
  conditions in identifications of the most similar
  foil was strong (r.56).

30
Experiment 2Conclusion

• Simultaneous participants appear to shift their
  choice to the most similar foil and other foils.
• Sequential participants appear to shift their
  choice to other foils.
• The ranking data suggest that information from
  the test faces is considered in a similar manner
  in simultaneous and sequential presentations.

31
Implications for Lineup Construction

• How should distractors be selected?
• Luus and Wells (1991)
• Matching foils to suspect photo creates lineup in
  which foils are too similar to the suspect
• If suspect is guilty, hits will be compromised
• If suspect is innocent, false alarms elevated
• However, results inconsistent across studies
  most suggest false alarms increase in description
  matched lineups.

32
Expected Effects of Constructing Lineups Based on
Match Photo Strategy on LU Members Feature
Distributions
From Tunnicliff and Clark (2000).
33
Implications for Lineup Construction

• Results of Experiment 1 might lend some insight
• If description matched produce low similarity
  levels and suspect matched produce high
  similarity levels, criterion placement lower than
  in description compared to suspect matched
  lineups.
• Hits reduced in suspect matched
• False alarms increased in description matched