1 / 24

The Relationship Between Confidence and Accuracy in Eyewitness Identification

The Relationship Between Confidence and Accuracy in Eyewitness Identification. Dwight J. Peterson Eyewitness Seminar University of Northern Iowa. Importance to Eyewitness ID. Confidence statements from witness are assumed to correlate with accuracy of identification (predictor of accuracy)

lynley
Download Presentation

The Relationship Between Confidence and Accuracy in Eyewitness Identification

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. The Relationship Between Confidence and Accuracy in Eyewitness Identification Dwight J. Peterson Eyewitness Seminar University of Northern Iowa

  2. Importance to Eyewitness ID • Confidence statements from witness are assumed to correlate with accuracy of identification (predictor of accuracy) • Influence on juror’s perceptions of the accuracy of the eyewitness • Wells, Lindsay, & Ferguson (1979) • Many factors which make confidence malleable (subject to change) • Luus & Wells (1994)

  3. Five Criteria for Trustworthiness of the Witness • Based on landmark case: Neil v. Biggers (1972) • Confidence (certainty) of witness • Opportunity to see suspect at time of crime • Degree of attention of witness • Accuracy of witness’ description of suspect • Delay between crime event and confrontation

  4. Confidence • Metacognitive technique used to indicate the degree to which you feel you were accurate in your identification • Involves retrospective (after the fact) assessment of your decision based on your memory for the event • Witness: “I’m pretty sure it was him.” • Cop: “So, we’re feeling pretty good about number 4, eh?”

  5. Accuracy SIGNAL • Did you correctly identify the person you saw commit the crime? • If yes, this is a hit • If no, this is a false alarm • Two other possibilities as well depending on lineup DECISION

  6. Correlation • Relation between two variables of interest • Statistical procedures to assess relation • Involves magnitude (strength) and direction (positive or negative) • Can range from - 1 (perfect negative correlation) to + 1 (perfect positive correlation) • Positive Relation: As X increases, Y increases • Negative Relation: As X increases, Y decreases

  7. Confidence/Accuracy Relation • Variety of results, but generally the finding of a significant correlation between confidence and accuracy in eyewitness identification studies is inconsistent • However, suggestions have been made regarding findings of sig. pos. relations compared to zero relation or neg. relations

  8. Optimality Hypothesis • Ideal conditions in the encoding, storage, and retrieval of information related to a crime event will lead to higher CA correlations • Based on Information Processing Theory • Conditions in which there is much interference will lead to lower CA correlations • Proposed to explain why there are different results from various CA studies • Deffenbacher (1980)

  9. Optimality Hypothesis cont. • Support: • Optimal conditions during experiment found high and significant CA correlations • Low optimal conditions found low CA correlations or significant negative correlation • Lack of support: • High optimal conditions leading to zero CA correlation • Low optimal conditions leading to high significant CA correlation

  10. What should we conclude? • Support or lack thereof for optimality hypothesis indicates a need for caution • Relation between confidence and accuracy should NOT be interpreted as being reliable based on the existing evidence • Deffenbacher (1980)

  11. Other CA Approaches • CA calibration techniques • Arguments made about the limitations of point biserial correlation coefficient • Underestimates CA relation • Point biserial is used when one dichotomous (accuracy yes/no) and one continuous (confidence 0-100) variable are being compared • Argue that calibration using statistics indicating how confidence discriminates accurate from inaccurate ID’s is better • Brewer, (2006); Brewer & Wells, (2006); Brewer, Keast, & Rishworth, (2002)

  12. CA Calibration cont. • Most studies use point biserial • Recently, suggestion made to use calibration procedures • Requires larger sample size • More people = more power • Used in conjunction with other experimental manipulations • Target to Foil similarity • TA and TP lineups • Foil similarity • Lineup instructions • (biased vs. nonbiased)

  13. CA Calibration cont. • Originally used with face recognition paradigm • Weber & Brewer (2003, 2004) • Obvious implications for EW ID research • Different uses and interpretation of results regarding their application • Recent use of calibration in EW ID research • Brewer & Wells (2006)

  14. Other Important Issues • Delay between identification and confidence statement is important • Brewer (2006) • Response Latency • Negative relation between response latency and accuracy • The longer it takes to respond, the lower the accuracy of the response • Especially for positive identifications • Weber & Brewer (2006)

  15. Confidence Malleability • Feedback to witness is important • Hard to disregard something you’ve been told • Semmler, Brewer, & Wells (2004); Wells & Bradfield (1998, 1999) • Getting information about what others say is important • Co-witness ID’s or non-ID’s will influence the confidence of partcipants depending on the information they receive • Luus & Wells (1994)

  16. Confidence Malleability cont. • Certain information will inflate confidence • Told that co-witness made same choice • Certain information will deflate confidence • Told that co-witness rejected the lineup • Different strengths of effects for deflation or inflation in confidence • Inflation effects were lower compared to control than deflation effects compared to control • May be easier to deflate when other information is incorrect and say: “well, now that I know this, I’m not as sure.” • Luus & Wells (1994)

  17. Confidence Malleability cont. • Perceptions of accuracy and confidence for various conditions of Luus and Wells (1994) • Experiment 2 had new sample view tapes of witnesses from Exp. 1 making identifications • Certain conditions of witnesses perceived as being more accurate compared to control group • Certain conditions were perceived as being less accurate than control group

  18. Confidence Malleability cont. • Perceptions of confidence via new sample were higher for the conditions which witnesses received information confirming their ID as being that of a co-witness • Indicates that perceiving accuracy of ID’s may be related to perceiving confidence of person making the ID • Luus & Wells (1994)

  19. Meta-analysis • Sporer et al. (1995) identify many moderator variables of interest to the CA relation • Choice is a focus of their meta-analysis • 30 studies included in meta-analysis • 27 reanalyzed to look at CA correlation for choosers and nonchoosers • Compared confidence for incorrect versus correct choosers and confidence for correct versus incorrect nonchoosers

  20. Meta-analysis cont. • From all studies examined • r = .29 effect size estimate, accounts for 8% of variance • For choosers • r = .41 effect size estimate, accounts for 17% of variance • For nonchoosers • r = .12 effect size estimate, accounts for only 1% of variance • Predictions made more easily for choosers compared to nonchoosers

  21. Meta-analysis cont. • Need to consider both instances in which an ID is made (choosers) or the lineup is rejected (nonchoosers) • Jurors may benefit from hearing about CA relation overall, (taking into account both choosers and nonchoosers) or just one or the other, depending on the situation/case

  22. Meta-analysis cont. • Suggest that jurors should be told by expert witnesses that confidence is only one of the indicators of accuracy • Also, should tell jurors that confidence is far from 100% predictive of accuracy • Should take into account moderator variables • Should point out that confidence should not be the only predictor used to assess accuracy

  23. Conclusion • Need to be careful regarding the CA relation • Although you may think that a confident person is an accurate person, this may be a hasty and dangerous strategy • Experts need to convey different information regarding the CA relation depending on the circumstances surrounding a particular case

  24. Thanks! • Comments, questions, discussion of any points mentioned?

More Related