III. Illusion of Confidence
A. Aspects of the Illusion of Confidence
1.
Inaccuracy in Confidence Estimates
2.
Judging Others by their Confidence
B. Preliminary Examples of Inaccuracy
1.
Zaragoza et al.'s Conceptual Elaboration Study (from Unit II)
2.
Franks & Bransford's Prototype Card Study (from Unit II)
3.
Talarico & Rubin (mentioned in Chapter 2)
4.
Kruger & Dunning: The "Dunning-Kruger Effect" (from Chapter 3)
C. And in Judging Others
1.
Johnson, Levenkron, Suchman, & Manchester (from the text)
2.
Jansen et al. (2011)
3.
Group Dynamics & Polarization (e.g., Anderson & Kilduff, 2009)
a. Social Comparison
b. Diffusion of
Responsibility (Latane)
c. Self-Serving Bias: Attributing
successes and failures to internal vs. external causes
D. Confidence as an Attribution?
Some Proposed Mechanisms
1.
Prospective Confidence & Availability (e.g., Nelson & Dunlovsky
2.
Processes of Recall
a. Speed of Retrieval (e.g., Nelson & Narens)
b. Fluency of
Retrieval (e.g., Kelley & Lindsay)
c. Justification of Retrieved Info (e.g.,
Koriat, Lichtenstein, & Fischoff)
d. Confidence as
Familiarity: Signal Detection Theory
i. Noise Distribution
ii. Signal + Noise Distribution
iii. Sensitivity (d') and Bias (β)
iv. Putting it Together: Hits,
False Alarms, Misses, & Correct Rejects
3. Contents
of Recall
a. Vividness (e.g., Talarico & Rubin)
b. Amount/Quality of
Detail (e.g., Chandler)
E. Comparing Retrospective &
Prospective Memory: Busey et al.
F. If You Believe You've Recalled
Fully... (Brewer et al.)
G. Some Final Comments on Confidence and
Eyewitness Memory
1.
Garrett (2011): High-Confidence Witnesses in DNA-Overturned Cases
2.
Sporer, Penrod, Read, & Cutler (1995) (from Chapter 3)
a. r = .41 between confidence and accuracy in
lineup experimentd
b. High-confidence
witnesses have 7-% accuracy (note room for error!): Low-confidence
wittesses have 30% accuracy
3. Wixted,
Mixes, Clark, Gronland, & Roediger (September, 2015):
Distinguishing Initial
From Later
Confidence (note the literature on memory change over time!)
a. Distinguishing initial confidence from
courtroom confidence
b. Initial confidence
much more accurate (though note apparent exceptions like the Jennider
Thompson case)
c. But note initial confidence can be quickly
altered by inferential feedback (e.g., "yeah, it's him;" "a known
troublemaker;" etc.)
d. Some of their
recommendationas (based also on other reports like the National Academy
of Sciences)
i. Have an uninformed administrator
during the lineup, not the PI
ii. Video the initial confidence
iii. Exclude later expressions of
confidence, including at the time of the trial (i.e., in the courtroom)
4.
Between-S and Within-S Correlations