follow CCP

Recent blog entries
popular papers

Science Curiosity and Political Information Processing

What Is the "Science of Science Communication"?

Climate-Science Communication and the Measurement Problem

Ideology, Motivated Cognition, and Cognitive Reflection: An Experimental Study

'Ideology' or 'Situation Sense'? An Experimental Investigation of Motivated Reasoning and Professional Judgment

A Risky Science Communication Environment for Vaccines

Motivated Numeracy and Enlightened Self-Government

Ideology, Motivated Cognition, and Cognitive Reflection: An Experimental Study

Making Climate Science Communication Evidence-based—All the Way Down 

Neutral Principles, Motivated Cognition, and Some Problems for Constitutional Law 

Cultural Cognition of Scientific Consensus
 

The Tragedy of the Risk-Perception Commons: Science Literacy and Climate Change

"They Saw a Protest": Cognitive Illiberalism and the Speech-Conduct Distinction 

Geoengineering and the Science Communication Environment: a Cross-Cultural Experiment

Fixing the Communications Failure

Why We Are Poles Apart on Climate Change

The Cognitively Illiberal State 

Who Fears the HPV Vaccine, Who Doesn't, and Why? An Experimental Study

Cultural Cognition of the Risks and Benefits of Nanotechnology

Whose Eyes Are You Going to Believe? An Empirical Examination of Scott v. Harris

Cultural Cognition and Public Policy

Culture, Cognition, and Consent: Who Perceives What, and Why, in "Acquaintance Rape" Cases

Culture and Identity-Protective Cognition: Explaining the White Male Effect

Fear of Democracy: A Cultural Evaluation of Sunstein on Risk

Cultural Cognition as a Conception of the Cultural Theory of Risk

« Roadtrip | Main | Those of you waiting for Science of Science Communication Session 4 reading list & questions -- your wait is over! »
Thursday
Feb162017

Politically biased information processing & the conjunction fallacy

So everyone probably is familiar with the “conjunction fallacy.”  It figures in Tversky & Kahneman’s famous “Linda  problem”:

 Linda is 31 years old, single, outspoken, and very bright. She majored in philosophy. As a student, she was deeply concerned with issues of discrimination and social justice, and also participated in anti-nuclear demonstrations.

Which is more probable?

1. Linda is a bank teller.

2. Linda is a bank teller and is active in the feminist movement.

According to T&K (1983), about 85% of people select 2. This is a mistake, in their view because “Linda is a bank teller” subsumes all the cases in which she is a bank teller and thus logically includes both the cases in which she is a “bank teller active in the feminist movement” and all the cases in which she is a “bank teller not active in the feminist movement.” On this reading, belonging to class 2 cannot logically be more probable than belong to class 1.

Nevertheless, people make the mistake because 2 is more concrete and conveys a picture that is more vivid than 1.  Those who over-rely on heuristic, “System 1” information processing are thus likely to seize on it as the “right answer.”  Individuals who score higher in conscious, effortful, “System 2” processing tend to be more likely to supply the correct answer (Toplak, West & Stanovich 2011).

What happens, though, when the individual actor featured in the problem behaves in a manner that evinces bad character, and the more vivid “choice 2” includes information that he possesses certain political outlooks?  People tend to attribute bad character to those who disagree with them politically. So will the likelihood of their picking choice 2 be higher if the actor’s political outlooks differ from their own?

We wanted to figure this out. So in our variant of the “Linda problem,” we informed our subjects, approximately 1200  ordinary people, that

Richard is 31 years old. On his way to work one day, he accidentally backed his car into a parked van. Because pedestrians were watching, he got out of his car. He pretended to write down his insurance information. He then tucked the blank note into the van’s window before getting back into his car and driving away.

Later the same day, Richard found a wallet on the sidewalk. Nobody was looking, so he took all of the money out of the wallet. He then threw the wallet in a trash can.

We then assigned them to one of three conditions:

“Which of these two possibilities do you think is more likely?

1. ex-felon condition

Which of these two possibilities do you think is more likely?

(a) Richard is self-employed ____

(b) Richard is self-employed and a convicted felon ___

2. procontrol.  

Which of these two possibilities do you think is more likely?

(a) Richard is self-employed ____

(b) Richard is self-employed and a very strong supporter of strict gun control laws? ___

3. anticontrol

Which of these two possibilities do you think is more likely?

(a) Richard is self-employed ____

(b) Richard is self-employed and a very strong opponent of strict gun control laws? ___

The motivation to test this proposition originated in a cool article by Will Gervais (et al. 2011), who found that when “Richard” is described as an atheist, people are more likely to display the “conjunction fallacy” than when he is described as an “atheist” or as a “rapist”; we adapted the “Richard” vignette from their study.

What did we find?

Well, first of all, the probability of the conjunction fallacy was highest, regardless of political outlooks, when Richard was described as a convicted felon.  Moreover, this bias grew in magnitude as subjects became more right-leaning in their politics.

But when Richard was described as either a "strong opponent"or a "strong supporter" of gun control laws, left-leaning subjects were slightly more likely to display a bias congenial to their political outlooks. Right-leaning ones displayed no meaningful bias in their appraisals. 

So there you go. Make of this what you will!

References

Gervais, W.M., Shariff, A.F. & Norenzayan, A. Do you believe in atheists? Distrust is central to anti-atheist prejudice. Journal of Personality and Social Psychology 101, 1189 (2011).

Toplak, M., West, R. & Stanovich, K. The Cognitive Reflection Test as a predictor of performance on heuristics-and-biases tasks. Memory & Cognition 39, 1275-1289 (2011).

Tversky, A. & Kahneman, D. Extensional versus intuitive reasoning: The conjunction fallacy in probability judgment. Psychological Review 90, 293-315 (1983).

PrintView Printer Friendly Version

EmailEmail Article to Friend

Reader Comments (9)

It's possible that "ex-con" and "self-employed" are already sufficiently politically charged terms, and so interfere with the test.

It's also possible that the non-political linkage between guns and petty crime (Richard's actions) is sufficient to interfere as well.

So I see too many variables...

Did you try a politically neutral test to first see if the conjunction fallacy mindset is more prevalent on one side of the political spectrum? There may also be weird conjunctions vs. non-weird conjunctions (weird -> more likely to trigger system 2), and those linkages could be political even if the independent properties are not.

What would happen if you replace "self-employed" with a politically neutral property, and used some non-crime related political issue as the other property, with no weird linkage between them? Like "Richard is in his 30's and supports/opposes gay marriage".

February 16, 2017 | Unregistered CommenterJonathan

@Jonathan-- thanks.
I don't know the answers to most of your questions. This is the only trial of this.

But it is clear (see the loess plots) that in fact conservs are more likely to commit the fallacy in "ex con" condition. That's "controlled for" in the model that finds that there is an interaction between political outlooks and assignment to the other conditions. In other words, the model measures impact of political outlooks over & above the one observed in the "ex con" condition.

I used the "self-employed" vs. "self-employed and ..." b/c Gervais et al. did & found interesting effects.

February 16, 2017 | Registered CommenterDan Kahan

I suspect the issue is due to the conjunction fallacy fitting stereotypes to behaviours, and conservatives don't hold that particular stereotype about gun control advocates. They might think they're authoritarian controlling bullies, they don't think they're financially dishonest or notably criminal.

I predict that if you changed the story to one about a person demanding 'safe spaces', shutting down free speech from anyone who disagrees with them, and launching twitter mobs against innocent cancer research scientists for alleged sexism, you would more likely get a more significant result. You have to understand what stereotypes are held about gun control advocates to be able to trigger them.

February 16, 2017 | Unregistered CommenterNiV

Suppose "self-employed" is a positive characteristic for conservatives, but not for liberals. Then, conservatives might use "ex-con" more frequently than liberals to adjust for Richard's negative behavior.

In the pro/anti-control cases, conservatives may see neither as adequately providing sufficient negative character, but liberals may see anti-control as possibly indicating someone who wishes to own a gun for nefarious reasons, perhaps because to them, nefarious reasons come to mind more readily than benign ones (such as self-defense or hunting, which may come to mind more readily for conservatives).

These might explain the graphs without supporting the hypothesis that "character correlates with political affiliation" is a system-1 heuristic belief. It's possible that nobody considered Richard's political affiliation at all, even while using their own political beliefs to evaluate Richard.

February 16, 2017 | Unregistered CommenterJonathan

@Jonathan-- Your account is exactly the one that I saw the problem as testing. If what lulls liberals but not conservatives into the fallacy is that liberals more readily see owning a gun as "nefarious," then the result suggests liberals more readily associate being pro-gun with the moral defects that made Richard behave dishonestly. If one wanted to test the hypothesis that liberals associate being conservative w/ bad moral character, then option (2) should be "... self-employed and a conservativve Republican."

I suspect you are having an argument w/ someone who would assimilate this result to the now-popular view that liberals and conservatives hate one another

But what motivated me to formulate the problem was something along the lines NiV is suggesting -- that *gun* attitudes are viewed as cohering with character in opposing ways by left-leaning & right-leaning people. That hypothesis is supported but only to a very modest degree; if the effect had been ideologically symmetric, the result would have been stronger (for me, if not NiV).

But the asymmetry is another interesting thing about the result. It turns on its head the usual "asymmetry thesis," which holds that conservatives are more prone to politically motivated reasoning than liberals.

February 17, 2017 | Registered CommenterDan Kahan

The "want guns -> nefarious reasons" link in liberals might simply be availability bias (more liberals in cities where there is less hunting and more violent crime on the evening news). Consider star anecdote Bernie Sanders - although city born-and-raised, once living in rural areas he associated guns more with hunting. That would still support your hypothesis that "*gun* attitudes are viewed as cohering with character in opposing ways by left-leaning & right-leaning people" - but because of availability, not political reasons.

Perhaps the gun issue is triggering liberty vs. care moral foundations heuristics in conservatives vs. liberals respectively. This would explain the asymmetry in your test - the large effect on liberals (anti-care strongly linked to interpersonal malevolence) and smaller effect on conservatives (neither pro- nor anti-liberty strongly linked to interpersonal malevolence). Perhaps if Richard's actions were more anti-liberty than anti-care, the test would see a larger conjunctive fallacy disparity in conservatives than liberals. Hence, I don't see your test as countering the "usual asymmetry thesis".

On that asymmetry thesis, have you seen: http://www.ssrn.com/abstract=2111700 ?

I suspect (as it certainly sounds like you do) that the science-curious trait might overwhelm all other left-right bias asymmetries, but it might still be the case that science curious is more likely to be found on the left in the US these days (consider science academia as a test case). However, consider a society in which GMOs, nuclear power, vaccines, etc. as socio-political issues outweighed climate change, evolution, fetal stem-cell usage, etc.. That might be a society in which reality would have a conservative bias, and science curious would gravitate rightward.

February 17, 2017 | Unregistered CommenterJonathan

"On that asymmetry thesis, have you seen: http://www.ssrn.com/abstract=2111700 ?"

Interesting paper. It's slightly off-putting at first because it uses terms in a technical sense that could be misinterpreted as culturally judgemental - like "analytic" or "Western". They made a good point about libertarians not being captured by the usual categorisation.

Not particularly impressed by their hypothesis on why conservatives oppose welfare: "Conservatives care deeply about close others, but they may dislike welfare programs because those programs serve strangers or even people from outgroups, and conservatives may see individuals on welfare as weighing down the system." (Conservatives care about the poor, too. They just don't think welfare is the right solution.) But presumably the authors are liberal - it's not very surprising.

However, the symmetry thesis is not about saying there are *no* differences between liberal and conservative thinking - it's about saying neither is more irrational than the other. The paper doesn't contradict that.

"I suspect (as it certainly sounds like you do) that the science-curious trait might overwhelm all other left-right bias asymmetries, but it might still be the case that science curious is more likely to be found on the left in the US these days (consider science academia as a test case)."

I offered the alternative hypothesis that science curiosity simply measured the extent to which people liked and trusted science documentaries as a source for their opinions. The polarisation is less because they simply follow the prevailing views among science documentary makers.

The test case for that I proposed was fracking (one of the topics Dan asked about), where the scientific consensus and best scientific evidence is that it's safe, but the documentary makers and science curious converge on the belief that it's dangerous.

I think Dan's tested whether the left or right was more science curious, and found no significant difference. The lack of intellectual diversity in academia is evidently caused by something else. Anecdotally, I've heard that conservatives feel highly unwelcome in academia (particularly recently, with some extremely hostile behaviour directed against conservatives at universities), and therefore either keep their true political beliefs hidden, or tend to drift out into industrial science.

February 17, 2017 | Unregistered CommenterNiV

@NiV

"I offered the alternative hypothesis that science curiosity simply measured the extent to which people liked and trusted science documentaries as a source for their opinions."

Perhaps there are two different science curious groups - one that watches documentaries and reads pop-sci and stops there, and another that is looking for scientific consensus and/or best scientific evidence.

Also, see http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0166332 - I wonder what the correlation is between science curious and moralizing rationality...

Did Dan find roughly what the percentage of science curious is in the overall US population?

February 17, 2017 | Unregistered CommenterJonathan

"Perhaps there are two different science curious groups - one that watches documentaries and reads pop-sci and stops there, and another that is looking for scientific consensus and/or best scientific evidence."

Or three - there are the ones who read pop-sci and stop, there are the ones who ask what the consensus of experts is and stop, and there are the ones who ask what the best evidence for and against it is.

"Did Dan find roughly what the percentage of science curious is in the overall US population?"

It depends where you set the threshold for what you would describe as "science curious". Dan's metric just puts people on a continuous scale from the minimum score to the maximum; it makes no binary classification into curious/uncurious categories. Dan generally plots against percentiles of the distribution (although they might be multiples of the standard deviation, we're a bit unclear on this point), and the "science curious" are the ones up at the right hand end of the graph. Since 50% of the population are to the right of the 50th percentile, I guess you could say 50% are above average "science curious". :-)

By the way, I've been digging around trying to find if Dan said anything more about the distributions, and while I can't find the distribution plot I thought I remembered, there was a diagram showing the mean of SCS for various social categories, and it turns out SCS was a little higher for liberals, although so far as I can see the difference in mean would be far smaller than the spread. I may have misremembered.

February 18, 2017 | Unregistered CommenterNiV

PostPost a New Comment

Enter your information below to add a new comment.

My response is on my own website »
Author Email (optional):
Author URL (optional):
Post:
 
Some HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <code> <em> <i> <strike> <strong>