Econstudentlog

Are chess players smart?

“Administered the French Wechsler Intelligence Scale for Children (WISC) to 33 tournament-level 8–13 yr old Belgian chess players (4 girls and 29 boys). The mean full scale IQ was 121, verbal IQ was 109 and performance IQ was 129. Results suggest that high levels of general intelligence and spatial ability are necessary to achieve a high standard of play in chess.”

Link.

It’s the only study I’ve found dealing specifically with this subject – there are a few others on spatial ability in particular, but this is seemingly an area where very little work has been done.

I’d have liked some loose IQ-rating relationsship, all that’s freely available is an abstract with the above information and a little more. This study remarks that the study also found that:

“there was some evidence that better players had higher performance IQ scores than the weakest players (top third = 131 vs. bottom third = 124).”

So quite young tournament chess players are significantly smarter than average (an average of 121 is much closer to the average IQ of a classroom filled with university students than it is to a primary school classroom average – and the top third’s score of 131 is genius level if using the standard scale with a st.dev. of 15 – the wisc IV has a mean of 100 and a standard deviation of 15 – link) and smarter chess players play better chess than chess players who’re not as smart. That’s not much, I’d like to know more. To know more about this would make it easier to know exactly when to decide that decreasing returns have kicked in. I don’t expect to ever get a norm. But should I give up the attempts to improve my rating 300 points before I get close to that, or perhaps 6-7-800(?) points before? [update: this question made more sense when I asked it than it does now - see also my update at the bottom of the post; I at this point feel reasonably sure that an ELO rating of 1900-2000 ought at least to be achievable..]

My own estimate, given the chess players I have met and talked to with a chess rating that high, is that I’ve yet to meet a 2000+ guy who could not become a member of Mensa without even trying. There’s no way I’d be able to compete with the really smart ones playing this game. But how close could I get if I worked hard? More data would be nice. And no, I won’t get that answer just by playing and studying – the point is that I’d like to have some idea as to when I’ve reached the MC>MB-point before having spent the many hours working without improving enough to justify the costs. If there were good data available it’d be easier to form realistic expectations conditional on ‘inherent characteristics’ known to be important to performance, and thus not get hurt by a large gap between expectations and results.

Anyway, the closest you can get to an answer to the initial question given the data from the study is this: ‘Young Belgian children who play tournament chess are significantly smarter than average.’ I think, based on my own limited experience, that it’s safe to say that most tournament players with a rating above, say, 1600, have a higher than average IQ – and that players above 2000 are usually smart people, i.e. people with a high IQ. I’d consider a rating of 2000+ to be a stronger indicator of high IQ than a non-specific Master’s Degree.

Update: For what it’s worth, my performance rating in the last tournament I played in ended up at ~1800, after having crossed the 2000-mark for a brief period of time.

About these ads

March 25, 2011 - Posted by | Chess, studies

11 Comments »

  1. Have you followed the “cited by”-feature in Google Scholar? for the article by Frydman?

    Comment by Stefan | March 26, 2011 | Reply

  2. The problem is that almost all chessplayers can get from a rating of 1000 to one of 1200. Most players can get from 1200 to 1300, if they make an effort. The point is that there’s a cut-off somewhere, and you don’t find the cutoff by looking only at players who’re still at the beginning of their careers, in an area where MC are still relatively low; that’s like measuring a lifetime-wage/IQ-relation by looking at people’s wage and IQ at the age of 25. You find it by figuring out where MC=MB in the dynamic setting. In the group of young players with an average IQ above Mensa level, their average rating was 1603 – there’s no way that number will not be much higher ten years later in a follow-up, because that playing strength’s not even close to what they have the potential to achieve. They know this, which again is why they are more active in chess tournaments than the average..

    Maybe the researchers do not completely understand how the chess ratings work, I don’t know but I was thinking along those lines. Chess rating is a measure of your current playing strenght, not a measure of ‘success in chess’. If what you’re interested in is the potential, then a better measure of that is highest performance achieved (highest rating achieved), not current performance (current rating). Ratings change over time, most people get better over time, then gradually reach some equilibrium level and move around that level for years until stagnation kicks in. [Ratings also change over time because of rating inflation, but that's not relevant for studies like this, because the ordering of players is not affected by this phenomenon and it's the ordering we're interested in]

    Comment by US | March 26, 2011 | Reply

  3. I don’t know why I didn’t think of that. It was probably because the initial search gave lots of results but next to nothing remotely relevant, so I discarded that idea out of hand.

    Thank you for reminding me of the existence of that feature. One of the studies that cites the Belgian one found that:

    “Although it is widely acknowledged that chess is the best example of an intellectual activity among games, evidence showing the association between any kind of intellectual ability and chess skill has been remarkably sparse. One of the reasons is that most of the studies investigated only one factor (e.g., intelligence), neglecting other factors relevant for the acquisition of chess skill (e.g., amount of practice, years of experience). The present study investigated the chess skill of 57 young chess players using measures of intelligence (WISC III), practice, and experience. Although practice had the most influence on chess skill, intelligence explained some variance even after the inclusion of practice. When an elite subsample of 23 children was tested, it turned out that intelligence was not a significant factor in chess skill, and that, if anything, it tended to correlate negatively with chess skill. This unexpected result is explained by a negative correlation between intelligence and practice in the elite subsample.”

    It can be found here. I might update the post at a later point to add additional information.

    Comment by US | March 26, 2011 | Reply

    • > “When an elite subsample of 23 children was tested, it turned out that intelligence was not a significant factor in chess skill, and that, if anything, it tended to correlate negatively with chess skill. This unexpected result is explained by a negative correlation between intelligence and practice in the elite subsample.”

      This does not surprise me too much; time is very finite. For example, http://xa.yimg.com/kq/groups/190222/1077302733/name/Kwon is a brain imaging study of Korean Go players; the expert professional Go players averaged IQ of 93, the control group 101. May be related to the Go players neglecting their schooling and other activities…

      Also, somewhere on the _Cambridge Handbook_ http://dl.dropbox.com/u/5317066/cambridge-expertise.pdf , I think, the case is made that IQ influences learning *speed* but has much less of an influence on the ultimate level of expertise.

      Comment by gwern | October 22, 2011 | Reply

      • n = 23 is a very small sample, so I wouldn’t put too much into the result either way but you’re right that it makes sense that the very smart would practise less on average. Anyway, there are lots of mechanisms that could be driving this.

        I don’t buy the case made in that book, unless ‘much less’ means something else to them than it does to me (yes, I know you are paraphrasing). In a model of the world where IQ is almost exclusively about learning speed, human brains can by analogy be likened to computers doing 0/1-computations at different processing speeds due to hardware differences, with software playing only a small role. I like to think that the software part matters a lot too. People with an IQ of 80 can’t become experts in astrophysics and it’s not just the time constraint that’s holding them back. There are too many dots to connect, and you get a much better image of that stuff in Excel than you do in MS Paint.

        Incidentally, I don’t think I’ve said this yet: Thank you for spending time reading my stuff and thank you for your comments!

        Comment by US | October 22, 2011

      • Thank you for the dropbox-link too!

        Comment by Stefan | October 22, 2011

  4. Another bit from the study:

    “Among the children participating in the study there were 23 children (all boys, mean age 11.4, SD = 1.2) who were regularly participating in local, national and some even in international chess competitions. In comparison with the players in the rest of the sample, the players had higher IQ scores (133 vs. 114; t(55) = 4.9, p < .001), were more experienced (5.5 vs. 3.5; t(55) = 5, p < .001) and spent more time playing chess (2.8 vs. 1.9; t(55) = 10.4, p < .001)."

    The authors spend some time doing in-sample analysis of the high rated players (the elite sub-sample part above), I find the 'raw' difference between the 'elite' group and the rest far more interesting. The average IQ of their 'elite' group was 133! That’s insane..

    I would have to agree that IQ far from tells the whole story, because chess is ‘rigged’ – you need to know a lot of stuff to become a good player, and it takes a lot of time and practice no matter who you are. But there’s just no way it’s not a major factor relating to performance one way or the other. One thing they point out in the study as well is that there’s a selection problem, in that bad players tend to stop playing the game over time, so that most adult chess players have similar characteristics making the differences less pronounced than in a sample of children where the never-to-become-very-good players still play; and when the variance of a variable of interest that you’re looking at goes down a lot, it can easily cause you problems. When you look at the sub-sample’s IQ/rating-relationship, you look at the relationship between inherent ability and test scores of Harvard students – they’ve already taken out all the people who didn’t get in to Harvard to begin with when doing that analysis.

    Comment by US | March 26, 2011 | Reply

  5. IQ and chess are both topics I am interested in, even if I do not have much to contribute in terms of insight.

    First off, I think IQ has a ton of predictive ability, and is not widely used because we live in politically-correct times. [Disclaimer: I do, or would have, a high IQ - I took the math/spatial part of a MENSA test when I was 13, and scored the max (164+), while the verbal part was not available in my language. It helps to remember "I used to think the brain is the most fascinating part of my body, and then I thought "Look who's telling me this!"]

    IQ tests are largely illegal as a hiring tool in the US because of “disparate impact” – if you use them, you end up, you end up with a workforce “biased” (that is overweighted compared to the general population) toward white and even more so (shocker!) Asian people, and against blacks and Hispanics. You can only have tests that are “relevant to job performance”. This, to me, is patent BS and example of reverse discrimination. IQ is demonstrably correlated to job performance for a lot of jobs, although I admit to not having seen studies that try to establish a causal link – but then again, such studies get no funding, and carry a significant stigma for the researcher, in terms of social opprobrium and career prospects.

    Your point about survivorship bias in chess is spot on. However, this ties into emotional intelligence, the kissing cousin of IQ. Persisting into an activity, even if the initial feedback you get is discouraging, is a sign of emotional intelligence. The ability and willingness to put your frustration, anger, boredom, etc. on a backburner in the expectation of a longer-term payout counts for a lot, in my book at least.

    But then again, what do I know? My deformation professionelle is to want as many data points and variables as possible, and then sort out what works or not. I believe in statistics, and could not care less about what is fair or not.

    Comment by Plamus | March 26, 2011 | Reply

  6. Count me in as one of the people who’d like to see more studies on intelligence and how such measures influence other variables of interest. That whole area has to a large extent become completely taboo and untouchable, which is a shame. And once again to people in doubt; no, assuming the variable isn’t important doesn’t make it so.

    “The ability and willingness to put your frustration, anger, boredom, etc. on a backburner in the expectation of a longer-term payout counts for a lot, in my book at least.”

    In my book too, but that’s probably because I’m so bad at it – which I probably shouldn’t be telling you as it will decrease my status in your eyes, or so some people would say. I used to think that I used to be good at it, now I’m no longer so sure if I ever was. I guess the higher work ethic I sometimes still tell myself I used to have had a lot to do with the fact that I didn’t really consider the stuff I was doing all that boring/frustrating/ect. – it wasn’t really ‘work’.

    Maybe it’s just that I don’t care all that much anymore. I’ve definitely cared less about everything since my suicide attempt compared to the time before I got sick.

    Comment by US | March 28, 2011 | Reply

  7. “Deliberate practice: Is that all it takes to become an expert?”, Hambrick et al 2013 http://www.msu.edu/~ema/HambrickEtAl13.pdf provides some more discussion and analysis of chess/IQ stuff.

    Comment by gwern | May 22, 2013 | Reply

    • Thanks for the link Gwern – I’ll have a look at it next week.

      Comment by US | May 22, 2013 | Reply


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Follow

Get every new post delivered to your Inbox.

Join 146 other followers

%d bloggers like this: