Casually Explained · Transcript · · Updated
Casually Explained: The Dunning-Kruger Effect – Transcript
Video by Casually Explained · Watch the original on YouTube
Summary
AI-generated from the transcript. Verify details against the source video.
- The Dunning-Kruger effect is commonly described in pop culture as stupid people massively overestimating their intelligence and smart people underestimating themselves.
- The original study by David Dunning and Justin Kruger was titled Unskilled and Unaware of It: How Difficulties in Recognizing One's Own Incompetence Lead to Inflated Self-Assessment.
- The actual Dunning-Kruger effect shows that a skilled person knows they are skilled, an average person knows they have average skill, and a person with no skill thinks they have average skill.
Use this video with Pro
Reading, copying and transcript downloads are free. Create saved AI results for this video with a signed-in Pro account ($9/month).
Transcript
Search ignores accents. All transcript text stays visible.
This video is sponsored by Incogny. The Dunning-Kruger effect is one of the most well-known and relatable psychological biases and one of my favorites because despite being referenced in pop culture nearly all the time, if you even skim the actual paper, there's no way anyone has actually read the thing because you'll immediately realize that our homies Dunning and Kruger were really just pranking students and calling it research. If you're unfamiliar, the Dunning-Kruger effect is if I got my exam back and the professor said to me, "Well, you thought you did better than 95% of the other students, but you
actually only got two out of 10 questions correct." I then think to myself, "Wow, everyone else must have done really badly." In pop culture, the Dunning-Kruger effect is that stupid people usually massively overestimate their own intelligence and smart people tend to underestimate themselves. This is most commonly expressed with this chart, which says that when you know very little about something, you massively overestimate your own ability and think you're a genius, and it takes significantly more knowledge on the subject to realize you don't actually
know very much at all. Then you start a podcast about here. A somewhat recent example of this, according to Neil deGrasse Tyson, was Terrence Howard, who published a 26-page paper and multiple 4-hour-long Joe Rogan dissertations on why 1 * 1 = 2. Nearly everyone, of course, was like, "Bro, that doesn't even make any sense. 1 * 1 is obviously 1. It's just one group of one." Which itself isn't the Dunning-Kruger effect, but it is the midwit meme because anyone who's taken university math knows that
if an exam said, "Prove 1 * 1 = 1" and gave you an entire blank page to answer, you know you're completely The original paper by homeboys David and Justin was called Unskilled and Unaware of It: How Difficulties in Recognizing One's Own Incompetence Lead to Inflated Self-Assessment, which is already funny because it means one of the guys in the lab had the idea, then went to the other and was like, "Bro, want to help me with this? You'd be perfect." In the original paper, they administer a battery of tests to participants, then get them to
self-evaluate how they thought they did. Most importantly, they were like, "First category, humor." Wherein there were actually some people who thought, "Yeah, I think I probably got 60% on the funniness test." And then the researchers were like, "Sorry, all wrong." Which in itself is hilarious.
The details are even better because the methodology was to get students to rate jokes from 1 to 11 and then compare their rankings to professional comedians and comedy writers. I presume they chose 1 to 11 instead of 1 to 10 to make the math easier, but they literally don't even mention why and don't do it again.
So, it turns out they were just making a Spinal Tap reference. 13,000 citations, by the way. They then were like, "Yes, we even included some classics from Jeff Rovin's Really Silly Pet Jokes." And when they sent them out to be rated by the pros, there was a total of eight comedians that responded. But the researchers were like, "Yeah, all the comedians ratings correlated positively with the others except for one. So, we got rid of his responses." Which immediately caused me to think this person was Shane Gillis on his 32nd Bud Light calling all the jokes gay. It's also the classic lesson no
matter what level of research you're doing, whenever you have a statistical outlier that doesn't fit with your assumptions, just get rid of it. Because they literally compared what professional comedians thought versus some random them test subjects. That means that when some innocent 19-year-old student read the joke, and this is an actual joke, by the way, "What is as big as a man but weighs nothing?"
"His shadow." They thought, "Well, that that was pretty clever. I'll I'll rate this joke a five." The administrators were then like, "No, you idiot. That one was a 1.3." To make matters worse, here's an actual joke that was literally the highest rated in the whole study and given a 9.6 by alleged comedians.
"If a kid asks where rain comes from, I tell him God is crying. And if he asks why God is crying, I tell him probably because of something you did." While it's now obvious where millennials get their anxiety from, in this situation, the student is probably thinking, "Um I'm rating this one a three. What Why are you instilling the fear of God into a child?" But at this point, David Dunning is high-fiving Justin Kruger in hysterics being like, "Yo, that's the
un-skilled and unaware of it effect for you. Whoa, these kids, man. In classic Neil deGrasse Tyson fashion, the actual results of the study are not the Mount Stupid chart he was referencing. And what the chart actually looks like is this. This is the line that represents where your actual test scores rank you compared to other students, and this is how the students rank their own performance. What is often called the Dunning-Kruger effect are these two gaps. The bottom performers overestimated their performance, and the top performers underestimated. This whole chart is misleading though,
because it makes you think the bottom performers got zero out of 10, and the people up here got 10 out of 10. But it's actually all relative performance. So it's possible the people at the bottom got six out of 10, and the people at the top got eight out of 10, and the chart would still look the same. But good news, they thought of that. Here's the results from the logic puzzle study, which had objectively right or wrong answers. The underlying shape of the chart is basically the same, so I'm just going to erase it like all good academics and chart the more interesting information, which is the actual test scores everyone got out of 10. So in
green, I'm going to chart what scores everyone thought they got, and now I'm going to chart in red what their actual scores were. For reference here, nearly the entire bottom quartile got zero, and more than the entire top quartile got 100%. This chart to me is the real juice of the entire study, because it's what us laymen psychologically relate to when we think of the Dunning-Kruger effect.
I've never met a student who thought they didn't at least have a chance of passing exam, and then they got zero. When I've taken a class for the second time and knew all the answers, I thought I'd get about 90% because I figured I'd probably screw something up, but then one time I didn't. Frankly, if you stare long and hard into this abyss, it explains just about everything. Matches on dating apps, what you get paid, modern politics, and of course whether or not you just self-identified as insulted or flattered. While this is what we all relate to, it's actually the case that none of the things we've
looked at so far are actually the Dunning-Kruger effect. Back in 2020, I remember seeing a pretty viral article that said the Dunning-Kruger effect is probably not real, which is not only a cautious and self-aware headline, but one where the original researcher went on to explain that you can get the same chart pattern where both over and underperformers look miscalibrated as a result of statistical noise. And while this is a legitimate explanation of the entire chart shape and applies directly to the overperformers in the study, the
underperformers are addressed regarding this issue in the original paper. In the study I just mentioned, people in the bottom 25% who got on average 0.3 out of 10 on the logic test, which is worrying given that they're Cornell University students, thought they'd gotten about five out of 10 correct. The administrators then gave them a 10-minute training session on how to do the logic puzzles they just did, and suddenly they looked at their own answers again and estimated they'd gotten about one right. Meaning Dunning-Kruger had humbled our future world leaders into thinking they're now
only three times as good as they are. Statistically though, this basically collapses the perception gap and largely shows that when someone overestimates their own low level of ability, it isn't statistical noise or regression to the mean, it is actually because they do in fact have an idea of what they're doing, but that idea is completely incorrect.
Nevertheless, this is what the Dunning-Kruger effect actually is. A skilled person knows they're skilled, an average person knows they have average skill, and a person with no skill thinks they have average skill. This too is extremely relatable because it means that the actual Dunning-Kruger effect isn't that idiots are walking around thinking they're geniuses, it means that in an area where you're hopelessly incompetent, people genuinely think it'll make a difference that they're 260. And it takes acquiring a bit of skill to realize you didn't have any to begin with. The issue here is that if
you ask 99.9% of us if we've ever actually truly and honestly skill checked ourselves with much of anything, the answer is usually no. Take the guy who posted the now removed I would have survived the Titan submersible implosion Reddit thread. He went on to say, "Yeah, they say it would be scientifically impossible, but I would have found an air bubble or something."
Initially, everyone was vehemently arguing, "You would experience over 300 atmospheres of pressure. You'd be crushed like a tank rolling over an aluminum can." And he was like, "I don't I don't see how a tank would get crushed by rolling over an aluminum can." And then they were like, "No, if you were the can, the water would feel like a tank rolling over you." And he was like, "Mhm, I mean, I've done a lot of swimming and I just I just don't really feel the water could do that to me." And then another person would be like, "You You would be over 10,000 ft underwater.
Even if you weren't crushed, you'd have to hold your breath for at least 2 hours to swim to the surface." And he'd be like, "Nah, I'd find a shorter way." Early commenters became enraged and fought tooth and nail to convince this guy he was wrong, but it had no effect because he was just way too high. Moral of the story, never attribute to stupidity what could be adequately explained by stupidity. Now, you might be thinking, "Shame that post was removed. I was going to go read it to fact-check all the details you mentioned." And while you can definitely
trust me, this is exactly what your enemies are thinking about you and your personal information. Which brings us to today's sponsor, Incogni. Data brokers can collect things like your name, address, phone numbers, and other personal information, and Incogni helps remove your personal data from over 420 harmful broker sites. And other than learning that Elon Musk is now writing Incogni sales copy, Incogni helps remove your personal data by automatically handling the removal requests and follow-ups for you. And like half of YouTube, Incogni has been supporting my
channel for so long, we've all started to wonder, "I'm over Incogni, when are we getting Incogni 2?" And I'm happy to announce the new Incogni Unlimited also includes their new exposure scanner, which can help uncover additional places where your personal information may be lurking online, where it could fall into the hands of scammers, hackers, or EDP 445. This is important because it reminds me of the time when I was a year into YouTube and some guy made a casually explained face reveal video, sleuthed my personal life, got everything right except my actual name,
and then just doxxed some other dude. There's also now an iOS and Android app, and every month Incogni gives you a risk assessment report so you can actually see what's going on, which is useful because it allows me to see the incredible danger I'm in at nearly all times, which makes me feel alive. So, if you'd like to have Incogni help you remove your personal data from the internet, you can go to incogni.com/casually or use code casually to get 60% off an annual plan.