I love when people talk about this book. I have gifted this book at least two of my friends.
Regression to the mean is the ultimate tool for a balanced approach as mentor, parent or a manager. Even in life, when things get crazy, I used to convince myself that is is one of the extreme datapoint, and eventually it will get better. But on a meta level this makes me worried about the world. People like Steven Pinker convinced us that our world now is much peaceful that at any point in history. But what if the momentary peace, which seems waning nowadays, is another datapoint? Then what comes next is probably the extreme point on the other side of the regression line . It always gives me a scary chill.
I think trends still exist, right - if you take the weight example, you can certainly have a causal effect on your weight over time, it's not all random fluctuations. I don't think the 2000 year long trend towards peace is random!
Good to read this piece Amrita, though I have concluded that one cannot overcome biases, or some biases. The way I think of it is: biases are like hardware on a computer, way more difficult to alter, than fallacies which are like software, which can be reprogrammed.
False dilemma fallacy is one of them which I think we can fight against. Not so sure about confirmation bias!
Agreed, but do you think being aware of the biases can help us make better decisions even though our perceptions are subject to biases? Something like "I'm pretty sure the F train has had a delay every time I've ridden it, but it could be confirmation bias, because I wouldn't remember the times when I rode it without incident."
Oh yes, absolutely! When I tried correcting my classmates on their beliefs, I started catching myself wanting to cherrypick data to support my belief that they belief is wrong. I've started catching myself earlier now 😄
Okay, so then I think we are in agreement except for some semantics. I'd call 'catching yourself' fighting the bias, but I can also see how one might not want to call it that because it's pretty much impossible to not be biased in the first place
Regression to the mean is such an interesting concept. Had no idea about it until I read this piece. Thanks for explaining it so clearly for a no-math-please brained person like me.
Nice read - at a good time for me as I did a short 2-minute piece earlier on when using your gut is mostly fine, basically: (1) time pressure; (2) expert environments; (3) similar situations; and (4) ‘good enough’ decisions.
Agreed, but I'll raise a caveat for the expert context - I think the examples you shared (firefighter, chess player etc.) are valid, but a lot of policy 'experts' go with their gut/lived experience instead of data, falling prey to a lot of biases that Kahneman outlines. But in general, yes, one has to choose what to think about carefully
Nice article, Amrita. I was a fan of the book but turns out many of the studies discussed there don't replicate. Here's a good summary - https://replicationindex.com/category/thinking-fast-and-slow/. They highlight the chapters that are particularly problematic.
Oooh thank you for sharing this! Good to know. It definitely lowers my opinion of the book a bit. And in general, I'm aware of replication problems in psychology but perhaps haven't been vigilant enough when reading individual studies. Is there a good way to quickly look up how a particular study fares on this metric?
Specifically re: anchoring, I see that this article gives chapter 11 an R-index of 44 for the 37 studies cited (terrible) - but the anchoring bias itself is quite robust, no? Is the replication about the specific value of the anchoring index found in a particular setting or the fact of anchoring itself?
Yeah, I don't think there's a quick way to tell how replicable a finding is. Not to my knowledge at least. You'd have to see if multiple labs have replicated the novel finding but even then you run into the issue of only significant results getting published so the effect might not be as robust as it may appear based on a lit review. But the field is getting better in this regard with the adoption of open science practices.
Re: the anchoring effect, I think it's robust. I got it for Amitabha's course project lol when we took it. Maybe it has to do with more nuanced aspects of the effect like the anchoring index you mentioned but I'm not a 100 percent sure.
I love when people talk about this book. I have gifted this book at least two of my friends.
Regression to the mean is the ultimate tool for a balanced approach as mentor, parent or a manager. Even in life, when things get crazy, I used to convince myself that is is one of the extreme datapoint, and eventually it will get better. But on a meta level this makes me worried about the world. People like Steven Pinker convinced us that our world now is much peaceful that at any point in history. But what if the momentary peace, which seems waning nowadays, is another datapoint? Then what comes next is probably the extreme point on the other side of the regression line . It always gives me a scary chill.
I think trends still exist, right - if you take the weight example, you can certainly have a causal effect on your weight over time, it's not all random fluctuations. I don't think the 2000 year long trend towards peace is random!
Excellent writing, Amrita. Sharp, clear, fun!
Thanks Rohan! :)
Such a fun piece, Amrita! Lovely, fluid writing. I hadn't read this book, but I'm now tempted.
Yay! Mission accomplished. I basically wanted to stand on a soapbox and tell people to read this book.
Ahaha. Will pick it up!
Good to read this piece Amrita, though I have concluded that one cannot overcome biases, or some biases. The way I think of it is: biases are like hardware on a computer, way more difficult to alter, than fallacies which are like software, which can be reprogrammed.
False dilemma fallacy is one of them which I think we can fight against. Not so sure about confirmation bias!
Agreed, but do you think being aware of the biases can help us make better decisions even though our perceptions are subject to biases? Something like "I'm pretty sure the F train has had a delay every time I've ridden it, but it could be confirmation bias, because I wouldn't remember the times when I rode it without incident."
Oh yes, absolutely! When I tried correcting my classmates on their beliefs, I started catching myself wanting to cherrypick data to support my belief that they belief is wrong. I've started catching myself earlier now 😄
Okay, so then I think we are in agreement except for some semantics. I'd call 'catching yourself' fighting the bias, but I can also see how one might not want to call it that because it's pretty much impossible to not be biased in the first place
Absolutely we’re in agreement. Some people say “Ravan” others say “Ravana”!
😂
Regression to the mean is such an interesting concept. Had no idea about it until I read this piece. Thanks for explaining it so clearly for a no-math-please brained person like me.
That is so good to know :)
Very intelligent women tend to be married to men who are less intelligent than them.
I can relate, my wife is one of the most rational people I have ever met.
Such a great piece! It’s one of my favorite books, and it’s what first introduced me to the world of psychology and mental models!
Thank you, Tapan!
Nice read - at a good time for me as I did a short 2-minute piece earlier on when using your gut is mostly fine, basically: (1) time pressure; (2) expert environments; (3) similar situations; and (4) ‘good enough’ decisions.
https://ondeciding.substack.com/p/when-to-trust-your-gut
Agreed, but I'll raise a caveat for the expert context - I think the examples you shared (firefighter, chess player etc.) are valid, but a lot of policy 'experts' go with their gut/lived experience instead of data, falling prey to a lot of biases that Kahneman outlines. But in general, yes, one has to choose what to think about carefully
Interesting read. I did not finish this book earlier, now I should pick it up and finish it!
Nice article, Amrita. I was a fan of the book but turns out many of the studies discussed there don't replicate. Here's a good summary - https://replicationindex.com/category/thinking-fast-and-slow/. They highlight the chapters that are particularly problematic.
Oooh thank you for sharing this! Good to know. It definitely lowers my opinion of the book a bit. And in general, I'm aware of replication problems in psychology but perhaps haven't been vigilant enough when reading individual studies. Is there a good way to quickly look up how a particular study fares on this metric?
Specifically re: anchoring, I see that this article gives chapter 11 an R-index of 44 for the 37 studies cited (terrible) - but the anchoring bias itself is quite robust, no? Is the replication about the specific value of the anchoring index found in a particular setting or the fact of anchoring itself?
Yeah, I don't think there's a quick way to tell how replicable a finding is. Not to my knowledge at least. You'd have to see if multiple labs have replicated the novel finding but even then you run into the issue of only significant results getting published so the effect might not be as robust as it may appear based on a lit review. But the field is getting better in this regard with the adoption of open science practices.
Re: the anchoring effect, I think it's robust. I got it for Amitabha's course project lol when we took it. Maybe it has to do with more nuanced aspects of the effect like the anchoring index you mentioned but I'm not a 100 percent sure.
Haha! Fun. Yeah, I want to understand exactly what the R-index means at some point. Useful reference.