Wednesday, September 16, 2020

Anger increases susceptibility to misinformation: Anger did not affect either recognition or source accuracy for true details about the initial event, but suggestibility for false details increased with anger

Greenstein, M., & Franklin, N. (2020). Anger increases susceptibility to misinformation. Experimental Psychology, 67(3), 202–209. https://doi.org/10.1027/1618-3169/a000489

Abstract: The effect of anger on acceptance of false details was examined using a three-phase misinformation paradigm. Participants viewed an event, were presented with schema-consistent and schema-irrelevant misinformation about it, and were given a surprise source monitoring test to examine the acceptance of the suggested material. Between each phase of the experiment, they performed a task that either induced anger or maintained a neutral mood. Participants showed greater susceptibility to schema-consistent than schema-irrelevant misinformation. Anger did not affect either recognition or source accuracy for true details about the initial event, but suggestibility for false details increased with anger. In spite of this increase in source errors (i.e., misinformation acceptance), both confidence in the accuracy of source attributions and decision speed for incorrect judgments also increased with anger. Implications are discussed with respect to both the general effects of anger and real-world applications such as eyewitness memory.

Discussion

Anger is an approach-oriented emotion adapted to guide
behavior under circumstances that involve time pressure
and consequences for safety. Because rapid cognition and
disinhibition would support effective action under these
circumstances, we made three predictions for how anger
would impact outcomes in a misinformation paradigm as
follows:
1. Reduced skepticism would increase susceptibility to postevent misinformation.
2. Disinhibition would be manifested as increased confidence.
3. More streamlined cognition would lead to faster responding.

We found evidence for all three of these outcomes.
When time is of the essence, there is value in suppressing
self-doubt to act quickly and decisively. While the experiment
did not call for urgent action, the cognitive adaptations
associated with anger should, as observed,
impact cognition pervasively, suggesting that these findings
reflect a broad cognitive style associated with anger.
Furthermore, anger increased suggestibility for schemairrelevant
and schema-consistent details, demonstrating
its broad impact on cognition.
Interestingly, anger did not seem to impair memory for
events that actually did occur as it affected neither recognition
memory nor source accuracy for details actually
present in the original event. Instead, anger impaired the
ability to dismiss errors that were subsequently suggested.
This is consistent with the characterization of anger as
streamlining cognition in support of action rather than
additional reflection. Coupled with the observed rapid and
confident memory decisions, this points to a constellation
of risks associated with anger’s impact on memory.
Because anger affected confidence and accuracy in
opposite directions, it affected the confidence–accuracy
relationship. The two are traditionally moderately correlated,
as observed in the neutral condition. Anger, however,
led to the opposite pattern where increased
confidence was associated with decreased accuracy. To
the extent that people use expressed confidence to judge
the reliability of other people’s memory (Wells et al., 1979),
this may be problematic. Because anger did not impair
participants’ source accuracy for events that actually had
occurred in the Film, outside observers with corroborating
evidence of those film details would have a basis for accepting
the participants’ confident additional claims. This
is precisely the sort of situation that jurors are exposed to
in a courtroom.
The current work has clear implications for witness
memory. Crimes can induce anger (Matsumoto & Hwang,
2015) and are associated with a risk of highly consequential
memory impairment (Deffenbacher et al., 2004).
Following an incident, witness memory is subject to
postevent input, such as cowitness accounts or leading
questions, which can distort memory (e.g., Roediger et al.,
2001; Wade et al., 2002). With each discussion or interview
comes renewed opportunity for misinformation effects,
which our results suggest will disproportionately
impact angry witnesses. To the extent that an angry witness
becomes more prone to errors of action (incorporating
postevent information into memory) rather than
inaction (rejecting new information), their memory reports
may become particularly unreliable.
Indeed, for at least three reasons, the observed effects
may underpredict those that would occur following a
crime. First, criminal cases may involve more repetition of
and elaboration upon postevent misinformation than occurred
in this experiment. For instance, criminal cases that
lead to prosecution generally involve multiple interviews
of the same eyewitness (e.g., by police officers, detectives,
and prosecutors). Separate from these, witnesses often
also relate details of the incident to cowitnesses, family,
and friends and potentially receive incorrectly suggested
details from these sources as well. Second, the anger
experienced in a criminal case is likely directed at the
source of the memory (the perpetrator) rather than at an
incidental target (the experimenter). Although this work
demonstrates that anger affects processing of content
unrelated to the source of the anger, it is possible that the
effects increase for related content. Third, the anger experienced
by many victims and witnesses would likely
exceed that of our laboratory participants. Inasmuch as the
degree of anger that affects the tendency to fall prey to
these biases, the potentially angrier victims and other
witnesses may be impacted more than our participants
were. It is also possible for the above factors to interact
with one another, further increasing risk in real-world
situations.
With regard to schematicity, participants in both
emotion induction conditions accepted more schemaconsistent
than schema-irrelevant misinformation. The
streamlined cognitive processing style associated with
anger did not increase preexisting tendencies to incorporate
this type of information into memory (Kleider
et al., 2008), although this may also reflect the high
plausibility of both schematic and schema-irrelevant
items. While this work intentionally sought to use only
highly plausible misinformation, future research should
continue to explore this question using less plausible
misinformation.
The current findings, coupled with the frequency with
which anger is experienced (Matsumoto & Hwang, 2015),
call for a greater understanding of its effects on memory.
Much is already known about memory’s vulnerability to
misinformation, and the current work finds that anger can
increase the frequency of these errors as well as one’s
confidence in them. Applied to criminal contexts, a richly
detailed witness report, combined with high confidence in
the associated memory, can contribute to heightened
perception of credibility in the eyes of cowitnesses, investigators,
judges, and jurors (Wells et al., 1979). Thus,
the dangers of anger, particularly in the justice system,
where the errors have real consequences, appear to be
more serious than previously understood.

Replication and Extension of Alicke (1985) Better-Than-Average Effect for Desirable and Controllable Traits

Replication and Extension of Alicke (1985) Better-Than-Average Effect for Desirable and Controllable Traits. Ignazio Ziano, Pui Yan (Cora) Mok, Gilad Feldman. Social Psychological and Personality Science, September 16, 2020. https://doi.org/10.1177/1948550620948973

Rolf Degen's take: https://twitter.com/DegenRolf/status/1306294580925607937

Abstract: People tend to regard themselves as better than average. We conducted a replication and extension of Alicke’s classic study on trait dimensions in evaluations of self versus others with U.S. American Mechanical Turk workers in two waves (total N = 1,573; 149 total traits). We successfully replicated the trait desirability effect, such that participants rated more desirable traits as being more descriptive of themselves than of others (original: η2p = .78, 95% confidence interval [CI] [.73, .81]; replication: sr 2 = .54, 95% CI [.43, .65]). The effect of desirability was stronger for more controllable traits (effect of Desirability × Controllability interaction on self–other-ratings difference; original: η2p = .21, 95% CI [.12, .28]; replication: sr 2 = .07, 95% CI [.02, .12]). In an extension, we found that desirable traits were rated as more common for others, but not for the self. Thirty-five years later, the better-than-average effect appears to remain robust. All materials, data, and code are available at https://osf.io/2y6wj/.

Keywords: better-than-average effect, self-evaluation, comparative judgment, replication


Mortality risk halves during the period of incarceration, with large declines in murders, overdoses, and better diet

Norris, Samuel and Pecenco, Matthew and Weaver, Jeffrey, The Effect of Incarceration on Mortality (July 2, 2020). SSRN: http://dx.doi.org/10.2139/ssrn.3644719

Abstract: This paper analyzes the effect of incarceration on mortality using administrative data from Ohio between 1992 and 2017. Using event study and difference-in-differences approaches, we compare mortality risk across incarcerated and non-incarcerated individuals before and after pre-scheduled releases from prison. Mortality risk halves during the period of incarceration, with large declines in murders, overdoses, and medical causes of death. However, there is no detectable effect on post-release mortality risk, meaning that incarceration increases overall longevity. We estimate that incarceration averts nearly two thousand deaths annually in the US, comparable to the 2014 Medicaid expansion.

Keywords: Incarceration, health, mortality, crime


2.2 Direct effects of incarceration on mortality

The first column of Panel A of Table 1 estimates the DiD specification from Equation 3 on
mortality risk, measured in deaths per hundred thousand individuals annually. The coefficient
on “Incarcerated in quarter,” corresponding to β in the previous section, measures the direct
effect of incarceration. We estimate that incarceration reduces mortality risk by 365 deaths per
hundred thousand (p < 0.001) relative to the post-release mean of 622.4. This is a nearly sixty
percent reduction in mortality risk and is approximately equal to the difference in mortality
between smokers and non-smokers aged 45-54 (Banks et al., 2015).

We use detailed cause of death information to understand the factors underlying this effect.
The most common non-medical cause of death in our sample is overdose (29% of post-release
deaths), which approximately halves during the period of incarceration, declining by 99.8
deaths per hundred thousand (column 2).10 This reduction may reflect addiction treatment
or more difficulty obtaining narcotics while incarcerated.11

Contrary to popular portrayals of correctional facilities, murder and suicide are greatly
reduced during the period of incarceration, though not completely eliminated (columns 3 and
6).12 Murder is particularly important since it is the third most common risk-factor in this
sample (19.4% of deaths post-release), highlighting the dangerous environment faced by the
criminally-involved outside of correctional facilities. The presence of correctional officers and
lack of access to firearms while incarcerated both likely play a significant role; firearms are
involved in 85% of homicides and 38% of suicides in our sample.

Inmates are constitutionally guaranteed medical care, and there may be changes to diet
or lifestyle that affect mortality risk. We find a large reduction in deaths (55 per hundred
thousand) due to medical causes during the period of incarceration (column (4) of Table 1).
Panel B of Table 1 finds the gains mostly come from reductions in heart disease, infection, and
non-classified causes. Even if the quality of prison medical care is suboptimal, many inmates
receive better care than they would otherwise. For example, 79.9% of inmates with persistent
medical problems reported being examined by a medical professional upon intake and twice
as many inmates with serious mental health conditions receive psychiatric medication during
incarceration as compared to prior to arrest (Wilper et al., 2009).

In summary, we find that incarceration dramatically reduces mortality.