2 ms·

> Dunning-Kruger's data actually demonstrated that the more confident someone is in their ability, the higher their ability, statistically. It also demonstrate

> Dunning-Kruger's data actually demonstrated that the more confident someone is in their ability, the higher their ability, statistically. It also demonstrated that the worst-performing individuals estimated themselves to perform MUCH better than they did, while the best-performing individuals expected to perform slightly worse than they did. This comports with the average reference to the phenomenon ("stupid people think they're smarter than they are, smart people think they're stupider than they are"). The article you linked for some reason makes this comparison: > So the bias is definitively not that incompetent people think they’re better than competent people. [...] they typically still don’t think they’re quite as good as people who, you know, actually are good. This is just categorically false, and the chart linked not one paragraph above completely torpedoes it. The bottom quartile of people thought that they would perform at around the 57th percentile. They clearly expected to perform better than more competent individuals. Perhaps the author meant to write "not that [they] think they're better than competent people do", which would be correct, but the rest of the conclusions seem to follow off of this flawed reading of the data.

dahart 6y ago
> The bottom quartile of people thought that they would perform at around the 57th percentile. They clearly expected to perform better than more competent individuals. I think you’re misunderstanding what Dunning & Kruger presented, which is precisely why references to this paper need to stop. You cannot read literally into the graph as saying that the bottom quartile of people believed they were at the 57th percentile of ability, that’s not what the experiment measured. Have you read the actual paper? The respondents were 1- all Cornell undergrads volunteering for extra class credit, 2- asked to rank themselves, not to rate their abilities, 3- asked about such “skills” as the ability to get a joke. It seems pretty wild that anyone ever took this paper seriously. The conclusions that it makes in it’s wordy prose are not supported by the data they present. Take a little more time to understand Tal’s primary argument - that this paper most likely is measuring regression to the mean. Because these kids were asked to rank themselves, and they didn’t know how good their peers were, there is good reason to expect a correllation less than 1. Because the authors were barely out of undergrad school themselves at the time, and not studying statistics per se, there is plenty of reason to suspect that the methodology isn’t perfect ... beyond the fact that it wasn’t a random sample of the population, it studied all rich kids who (self-selected) needed extra credit, it included no difficult mental skills, and that followup papers have shown a complete reversal of whatever effect is there, when applied to more difficult skills like law or engineering.

dahart 6y ago

> The bottom quartile of people thought that they would perform at around the 57th percentile. They clearly expected to perform better than more competent individuals. I think you’re misunderstanding what Dunning & Kruger presented, which is precisely why references to this paper need to stop. You cannot read literally into the graph as saying that the bottom quartile of people believed they were at the 57th percentile of ability, that’s not what the experiment measured. Have you read the actual paper? The respondents were 1- all Cornell undergrads volunteering for extra class credit, 2- asked to rank themselves, not to rate their abilities, 3- asked about such “skills” as the ability to get a joke. It seems pretty wild that anyone ever took this paper seriously. The conclusions that it makes in it’s wordy prose are not supported by the data they present. Take a little more time to understand Tal’s primary argument - that this paper most likely is measuring regression to the mean. Because these kids were asked to rank themselves, and they didn’t know how good their peers were, there is good reason to expect a correllation less than 1. Because the authors were barely out of undergrad school themselves at the time, and not studying statistics per se, there is plenty of reason to suspect that the methodology isn’t perfect ... beyond the fact that it wasn’t a random sample of the population, it studied all rich kids who (self-selected) needed extra credit, it included no difficult mental skills, and that followup papers have shown a complete reversal of whatever effect is there, when applied to more difficult skills like law or engineering.