New review in @TrendsCognSci “The Psychology of Fake News” w @DG_Rand

We synthesize research on belief in, sharing of, & interventions against misinformation, with a focus on false/misleading news…

And now a thread that synthesizes our synthesis!
We make three major points in the paper, which I will summarize here. However, there are several other elements to the paper that may be of interest. E.g. a short review on the prevalence of fake news, a discussion of the heuristics that people use, such as familiarity, and more!
The first major point is that, contrary to narratives that focus on political partisanship/motivated reasoning, we find that a lot can be explained by mere lazy thinking (overreliance on intuition).
For eg, individual differences in reflective thinking (via the Cognitive Reflection Test, + others) are consistently associated with an increased ability to distinguish b/w true/false news *regardless* of whether the news is consistent *or* inconsistent with political ideology
In a combined analysis of 14 studies (N>15k). The effect of cognitive reflection is 2x larger than the effect of political consistency on discernment (which, anyways, shows that people are *better* at distinguishing between true/false news if it’s ideologically *consistent*)
There's also experimental evidence that supports the conclusion that analytic thinking is associated with increased truth discernment and not increased polarization (as assumed by motivated reasoning accounts). See:
Of course, if one looks at *overall* belief, people find politically consistent news more plausible regardless of if it’s true or false. So, people believe things that are consistent with their ideology and (separately) being reflective is associated with more *accurate* beliefs.
Importantly, the general bias toward believing things that are consistent with one’s ideology is not evidence of a causal role of political identities per se. Can’t go into detail with so few characters, but the short story is that there are many confounds
The 2nd major point is that social media sharing does not necessarily imply belief. People are often quite good at distinguishing between true/false news when asked to do it directly. However, when it comes to sharing, they barely do so.

I.e., RT!=endorsement is actually true
This indicates that, again, the spread of fake news may be driven (to some extent) by mere inattention to accuracy. People may be getting distracted from thinking about accuracy when deciding what to share (Note: People *say* that accuracy is important to them)
This brings me to the third major point: Points 1&2 indicate that maybe people can make better choices if they slow down and consider accuracy before sharing. And, in fact, there is good evidence that this is the case See:
The broader conclusion is that interventions against misinformation should be informed by an understanding of the underlying psychological mechanisms. Things that intuitively seem that they may work may not be effective (and vice versa).…
For eg, ppl are surprisingly good at distinguishing between high/low quality news sources (when asked to do so – doesn’t mean they do it in practice, see…) & crowdsourced judgments of news headlines could be used to inform algorithms
Another example is that fact-checks that directly follow a news headline actually work better than ones that come directly before the headline (a lot of people would assume that preparing people mentally for falsehood works better, but it doesn’t)
There is still so much to learn about this topic, though. For the psychologists in the crowd, misinformation reveals a lot about how our minds work. And, at the same time, it’s a nascent area where we need research to inform policy. Hopefully, this review is outdated in 5 years!
This paper is a capstone of the amazingly fun collaboration between me and @DG_Rand (+ many others) that started back in 2016. We're still going and hopefully will be for a long time! A frequently updated list of our misinfo (and related) projects is here:…

• • •

Missing some Tweet in this thread? You can try to force a refresh

Keep Current with Gordon Pennycook

Gordon Pennycook Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!


Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @GordPennycook

18 Mar
Warning: Sentimentality ahead

I'm not someone who publishes papers in Nature. I'm just not.

And it's not just this paper, of course. This is just the thing that caused me to reflect on my life and how absolutely bonkers this all still is for me.

So, I thought I would share.
First, I need to give props to @DG_Rand. He is an absolute hero. You know how some PI's just slap their name on work that has been done by junior collaborators? Not Dave. If anything, he takes LESS credit than he deserves. He's also just the best person.

But, anyway, back to me
I grew up on a farm in northern Saskatchewan. It failed & forced my parents to work several jobs. As a kid, I (and my 4 siblings) helped my parents do janitorial work from Grade 1 to Grade 9. This was not at all abnormal to me: On a farm you do chores, so that became our chores. (I'm on the far right)
Read 15 tweets
18 Nov 20
There has been a surge of behavioral research on misinformation & "fake news". To synthesize things, @DG_Rand & I wrote a systematic review:

We take a cognitive/social psych perspective, but we tried to cast a wide net for the review. Feedback welcome!
Sorry to those who retweeted an earlier version of this tweet that I deleted because the image preview was too zoomed in
There's too much in the review to cover in a tweet thread, but here are some of the take-aways that we thought to be particularly important...
Read 9 tweets
8 Nov 20
We're likely to face an unprecedented situation where the incumbent refuses to concede. Although it may not be necessary, things would certainly be easier if Republicans viewed the election as legitimate.

How uphill of a battle will this be? Well, I ran a study with @DG_Rand...
Study was run on Prolific & Lucid on Friday. In total, we have 509 Biden voters & 218 Trump voters. The samples are *not* nationally representative and a bit small. But, some fairly clear results came out.

More info on the sample:
A key initial Q is about people's priors. Do Trump voters believe it is *unlikely* that Biden won?

The answer is yes.

Reminder: This study was run on Friday when Biden was already well ahead & very likely to win. That he would win the popular vote was *never* in question.
Read 18 tweets
7 Nov 20
The following may be of interest to those who use Prolific and/or Lucid for surveys.

Ran a study yesterday about election-related opinions (plus some other stuff - data is a bit depressing, coming tomorrow) using Lucid & Prolific's "nationally representative" sample function...
Both sources use quota-matching to filter people into studies who match U.S. demo's on age, gender, ethnicity, and (for Lucid) region. However, there were some notable differences and similarities between the samples.

(Note: Target N for each was 500, study was ~10 min long.)
I included a very simple initial attention/bot check: "Puppy is to dog as kitten is to _____?" with an open-ended text box to respond. This came at the very start. Two other fairly simple attention checks came later in the survey. We also asked directly if ppl responded randomly.
Read 8 tweets
6 Nov 20
I do hope that someone is keeping a list of elected Republicans who a) supported Trump's baseless attacks on US democracy, b) said nothing, or c) repudiated him.
And for (c), if they did so *before* the election was called for Biden or after
Read 10 tweets

Did Thread Reader help you today?

Support us! We are indie developers!

This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!

Follow Us on Twitter!