Aschwin de Wolf approvingly quoted Jonathan Bowden writing “Those who don’t lie down and die soon discover that happiness and intellect are at opposite sides of the pole.” I thought I heard someone debunk that with the GSS before, and set about googling. I replied by linking a comment from Secular Right that I hadn’t actually read before, but I figured I should present the actual results of a GSS analysis.
Frequency Distribution | |||||
---|---|---|---|---|---|
Cells contain: -Column percent -Weighted N |
HAPPY | ||||
1 VERY HAPPY |
2 PRETTY HAPPY |
3 NOT TOO HAPPY |
ROW TOTAL |
||
WORDSUM | 0 | .8 60.7 |
.7 89.5 |
1.8 41.9 |
.8 192.2 |
1 | 2.0 152.1 |
1.4 176.5 |
3.6 83.8 |
1.8 412.3 |
|
2 | 3.4 265.8 |
3.0 384.5 |
4.3 100.0 |
3.3 750.3 |
|
3 | 5.8 445.7 |
5.6 714.2 |
9.7 224.5 |
6.1 1,384.4 |
|
4 | 9.2 712.8 |
10.4 1,321.1 |
13.0 299.9 |
10.2 2,333.8 |
|
5 | 15.9 1,230.2 |
16.5 2,101.5 |
18.4 424.1 |
16.5 3,755.8 |
|
6 | 21.1 1,627.4 |
22.6 2,881.3 |
19.8 455.8 |
21.8 4,964.4 |
|
7 | 16.5 1,274.9 |
16.3 2,075.8 |
11.3 260.0 |
15.9 3,610.7 |
|
8 | 11.4 876.2 |
10.2 1,297.7 |
7.9 183.4 |
10.3 2,357.2 |
|
9 | 8.1 623.3 |
7.8 999.0 |
5.9 135.6 |
7.7 1,757.9 |
|
10 | 5.8 444.0 |
5.6 717.6 |
4.3 98.1 |
5.5 1,259.6 |
|
COL TOTAL | 100.0 7,713.1 |
100.0 12,758.6 |
100.0 2,307.0 |
100.0 22,778.7 |
|
Means | 6.06 | 6.05 | 5.42 | 5.99 | |
Std Devs | 2.15 | 2.08 | 2.27 | 2.13 | |
Unweighted N | 7,234 | 12,774 | 2,593 | 22,601 |
UPDATE: Not everyone is familiar with the meaning of GSS color codes, so here goes:
Color coding: | <-2.0 | <-1.0 | <0.0 | >0.0 | >1.0 | >2.0 | Z |
N in each cell: | Smaller than expected | Larger than expected |
UPDATE 2: Tyrosine suggested that intelligence may interact with neuroticism to produce unhappiness. The GSS does not include a personality test, but the ANXIOUS variable lists how many of the last seven days the respondent felt anxious/tense. I have lumped responses 0-3 together as the less anxious group and 4-7 as the more anxious group. For those who would like to do something similar themselves, enter the following in the Control field: ANXIOUS(r: 0-3 “Few days anxious”; 4-7 “More days anxious”) The tables for the two groups are below.
Statistics for ANXIOUS = 1(Few days anxious) | |||||
---|---|---|---|---|---|
Cells contain: -Column percent |
HAPPY | ||||
1 VERY HAPPY |
2 PRETTY HAPPY |
3 NOT TOO HAPPY |
ROW TOTAL |
||
WORDSUM | 0 | .5 | .3 | 3.8 | .6 |
1 | 2.3 | 1.3 | 6.7 | 2.1 | |
2 | 3.2 | 3.3 | 5.8 | 3.5 | |
3 | 5.4 | 3.7 | 3.8 | 4.3 | |
4 | 8.8 | 9.8 | 15.4 | 9.9 | |
5 | 14.7 | 20.0 | 19.2 | 18.1 | |
6 | 23.5 | 24.6 | 16.3 | 23.6 | |
7 | 14.7 | 16.6 | 10.6 | 15.4 | |
8 | 14.4 | 10.2 | 8.7 | 11.5 | |
9 | 7.7 | 4.6 | 4.8 | 5.7 | |
10 | 5.0 | 5.6 | 4.8 | 5.3 | |
COL TOTAL | 100.0 | 100.0 | 100.0 | 100.0 | |
Means | 6.11 | 6.01 | 5.20 | 5.98 | |
Std Devs | 2.11 | 1.94 | 2.50 | 2.06 | |
Unweighted N | 225 | 401 | 65 | 691 |
Now for the more anxious:
Statistics for ANXIOUS = 2(More days anxious) | |||||
---|---|---|---|---|---|
Cells contain: -Column percent |
HAPPY | ||||
1 VERY HAPPY |
2 PRETTY HAPPY |
3 NOT TOO HAPPY |
ROW TOTAL |
||
WORDSUM | 0 | 1.0 | .0 | 1.6 | .5 |
1 | .0 | 1.1 | .0 | .7 | |
2 | 1.0 | 3.6 | 7.9 | 3.7 | |
3 | 8.3 | 6.9 | 1.6 | 6.5 | |
4 | 10.4 | 13.5 | 12.7 | 12.7 | |
5 | 18.8 | 18.9 | 11.1 | 17.7 | |
6 | 11.5 | 24.4 | 17.5 | 20.5 | |
7 | 17.7 | 11.3 | 4.8 | 11.8 | |
8 | 10.4 | 7.3 | 25.4 | 10.6 | |
9 | 16.7 | 9.8 | 11.1 | 11.5 | |
10 | 4.2 | 3.3 | 6.3 | 3.9 | |
COL TOTAL | 100.0 | 100.0 | 100.0 | 100.0 | |
Means | 6.30 | 5.82 | 6.32 | 6.00 | |
Std Devs | 2.16 | 2.00 | 2.40 | 2.10 | |
Unweighted N | 53 | 153 | 43 | 249 |
I also ran a multiple regression for HAPPY on WORDSUM and ANXIOUS. To do one of those instead of cross-tabs, go to “Analysis” at the top and select the kind of analysis you want. Results are below:
Regression Coefficients | Test That Each Coefficient = 0 | |||||
B | SE(B) | Beta | SE(Beta) | T-statistic | Probability | |
---|---|---|---|---|---|---|
WORDSUM | -.022 | .009 | -.076 | .032 | -2.386 | .017 |
ANXIOUS | .052 | .009 | .188 | .032 | 5.870 | .000 |
Constant | 1.802 | .061 | 29.338 | .000 |
Color coding: | <-2.0 | <-1.0 | <0.0 | >0.0 | >1.0 | >2.0 | T |
Effect of each variable: | Negative | Positive |
You may be initially confused by those results, due to the way HAPPY is coded. Remember that 1 is “Very Happy” while 3 is “Not too happy”. So a “positive” correlation means less happy, as we might expect from the anxious.
On an unrelated note, Matt Steinglass recently used an analogy of Hollywood screenwriter : liberal :: investment banker : conservative. I’m skeptical i-bankers are that conservative, but don’t feel like investigating it right at this moment. Looking up occupational codes for the GSS is a bit of a hassle, I left some links I used to look them up before here. If anyone wishes to do some unpaid labor by investigating them, that would merit kudos.
April 16, 2010 at 11:39 pm
Pretty unreadable and WTF does the colors mean?
April 17, 2010 at 7:17 am
I think very high intelligence has a real dark side if and when one is neurotically anxious. If you are a bit depressive, without anxiety — or just happy and well — it is a plus because you can have fun being an intellectual.
April 17, 2010 at 12:39 pm
Before the brawl starts, I’ll simply note without the benefit of GSS that you and Aschwin de Wolfe are both right.
April 17, 2010 at 1:49 pm
If intelligent people have better situational awareness they can dumb themselves down in order to better enjoy themselves. That way they can occupy two different worlds. Dumb people don’t have that option.
April 17, 2010 at 2:04 pm
they can dumb themselves down
Well, no, ignorance cannot be learnt.
But I suspect the whole point is moot, it is just that some happy wannabees are such bliss ninnies that it brings disrepute to happiness:
http://www.happiness-project.com/
April 17, 2010 at 2:24 pm
Well, no, ignorance cannot be learnt.
What I mean is that the non-ignorant know how to feign the appearance of ignorance, at least better than the dumb can fake smarts.
April 17, 2010 at 3:06 pm
This data appears far too vague to support any conclusion. My own leading theory would be that when answering a survey people with a larger vocabulary also have meaningfully different knowledge and thought process and the internal scale they use for happiness differs. Perhaps they think about “happiness” more globally and are comparing to a larger number of options. Perhaps they view the word with a different implicit time horizon. Perhaps they view the modified “pretty” to encompass a larger range. My leading theory has to remain very uncertain though, as other theories like “they actually do differ in their happiness” also can’t be ruled out.
There are other possible theories as well, like that there is a sampling bias. Maybe people who score higher on WORDSUM, and are happy, are more likely to complete the survey. Maybe people with a larger vocabulary who are at home and available to take a survey are more likely to be happy. A higher unemployment level among those with a smaller vocabulary might cause such a thing. I haven’t really looked into this, but a quick glance at http://www.norc.org/NR/rdonlyres/21C53AAC-1267-43B6-A915-A38857DC9D63/1281/AppendixA.pdf does nothing to convince me that there’s no bias here.
More theories: People who are tired are more likely to both be unhappy and to do worse on a vocabulary test. People who are unhappy care less about getting the answers right and don’t try as hard, leading to worse scores. Unhappy people may have different word associations in general, which are more likely to be counted as wrong (afraid instead of animal for beast, for example, see WORDA as I can’t figure out how to link to it). Maybe a combination of various different effects, not related to actual happiness.
This is, of course, a problem with surveys in general. They just don’t tell you as much as you might think they do.
April 17, 2010 at 3:53 pm
“What I mean is that the non-ignorant know how to feign the appearance of ignorance, at least better than the dumb can fake smarts.”
It’s not how you look, it’s how you feeeeeel, as sung by Eddie Vedder (but it might be a Neil Young song).
April 17, 2010 at 4:03 pm
Apparently those with wordsum scores of ten lie from the 94.5th to the “100”th IQ/g percentiles (assuming it gives a perfect ranking for psychometric g, which of course it doesn’t).
I wonder what you would see if you only looked at those above the 99.9th percentile. They are a tiny subset of the wordsum=10 people and probably haven’t significantly affected the mean or SD of the wordsum=10 people.
April 17, 2010 at 10:24 pm
It’s not how you look, it’s how you feeeeeel,
Exactly but the key to happiness is probably the belief that you are “doing the right thing” and you can’t control you beliefs, belief isn’t really voluntary.
This may explain the appearances, the intelligent and knowledgeable have more difficulty thinking they are “doing right” while the stupid have it easy, at least until they crater badly when meeting an unexpected reality.
April 18, 2010 at 3:57 am
You know it’s funny, you never hear anybody say “As an idiot, I’m very happy. I find it to be a worthwhile trade-off”.
April 18, 2010 at 12:40 pm
[…] TGGP – “Ignorance is Not Bliss” […]
April 18, 2010 at 1:48 pm
One thing that hasn’t been made explicit: intellect and intelligence are not the same thing. While some above average level of intelligence is surely required for a person to engage intellectual pursuits, raw brain power is at best a necessary but not sufficient condition for the cultivation of the cognitive/temperamental disposition that I think is much closer to the spirit of Bowden’s declaration. At least in theory, an intellectual thinks deeply about issues and ideas. Not everyone with a high IQ does this.
I assume that the the 1-3 Wordsum scorers are relatively dull. As Linda Gottfredson has more than proven, life is difficult for stupid people (what with their myopic time preferences and consequent bad choices), and difficult lives can be fraught with frustration and disappointment. But if you look at the steep part of the curve, i.e., the 4-7 Wordsummers, who constitute most people, you see a lot of self-reported positive well-being, that tracks off conspicuously as people become fewer and smarter, and proportionately more likely to be intellectuals. This suggests to me that the question, nuanced as it is, remains open.
April 18, 2010 at 5:28 pm
Kevembuangga & Tyrosine: I have updated the post in response.
Savrola: Brown-nosing can really only be effectively done one person at a time.
Dain: An interesting point. Girls in particular are known for “playing dumb”. That’s discussed in Erving Goffman’s “The Presentation of the Self in Everyday Life”.
Kevembenggua: I have sniped at the joy-mongers here before.
User: There are many possible reasons one can think up to reject a survey. As surveys go, the GSS is generally considered the “gold standard”. One might give little credence to surveys generally (and I’m certainly more inclined to take that positivist/behaviorist route than Bryan Caplan), but then what evidence do we refer to? Bowden, for example, presents none at all.
Google was no help in determining whether those lyrics came from Pearl Jam or Neil Young. I certainly can’t recall the song.
Tyrosine: That’s a fairly fall segment of the population, so I think we’d need some different sampling to get that kind of data.
Kevembuangga: It’s also an empirical question whether the more intelligent believe they are doing the right thing. I’ll try to see if there’s a GSS question which corresponds to that.
Christopher: Could be related to the “unskilled and unaware of it effect”, but on the other hand, that study may be wrong. I am reminded though of the person from “Brave New World” who says how happy they are to be their particular caste rather than the ones above or below.
Chip Smith: I think Paul Johnson beat you to it by defining intellectual as bad person. Of course Bertrand Russell noted earlier that the common person’s conception of an intellectual is a man who cheats on his wife.
I think this quote from Alan Crowe is relevant. Of course, since Bowden started it, I put the burden of proof on him!
April 18, 2010 at 5:45 pm
The song is “Fuckin Up.” I just googled three or four sources saying Pearl Jam covered it from Neil Young.
Here’s Young playing it, with the same melody as Pearl Jam:
Of course Bertrand Russell noted earlier that the common person’s conception of an intellectual is a man who cheats on his wife.
What’s he mean exactly? That the intellectual cheats and can use his great mind to think up a “great, airtight” justification for why it was inevitable?
April 18, 2010 at 5:50 pm
Call me crazy but I don’t think that’s the best of Neil Young’s songs. I’ll take this any day:
One time I tried sampling some of his more acclaimed later songs on the web. Pretty underwhelming.
April 18, 2010 at 5:59 pm
Some of you probably know the name Lewis Terman from Steve Hsu’s discussion of his study of IQ 140+ subjects. Here is a discussion of his finding about “maladjustment” in these subjects. I believe this webpage was cited before by Caldeonian.
Terman’s own data shows that there is a definite connection between measured intelligence and mental and social maladjustment.
There’s a lot of data table in that article. Note that some of the intelligence scores which they present look like IQ scores, but are not — I think they are scores, normed to 100, for some special intelligence test given only to high-IQ samples.
http://www.prometheussociety.org/articles/Outsiders.html
I haven’t scrutinized the article and can’t vouch for it. I don’t know how Terman defined “maladjustment.”
April 18, 2010 at 6:05 pm
To put it more briefly and concretely, I think the “maladjusted” people with a mean “CMT-A” score of 108 in those tables probably have an IQ of like 155 or something, not 108. The CMT-A appears to be a “super IQ test” that Terman gave to subjects with IQ 140+.
April 18, 2010 at 6:53 pm
His name was William James Sidis, and his IQ was estimated at between 250 and 300 [8, p. 283]. At eighteen months he could read The New York Times, at two he taught himself Latin, at three he learned Greek. By the time he was an adult he could speak more than forty languages and dialects. He gained entrance to Harvard at eleven, and gave a lecture on four-dimensional bodies to the Harvard Mathematical Club his first year. He graduated cum laude at sixteen, and became the youngest professor in history. He deduced the possibility of black holes more than twenty years before Subrahmanyan Chandrasekhar published An Introduction to the Study of Stellar Structure. His life held possibilities for achievement that few people can imagine. Of all the prodigies for which there are records, his was probably the most powerful intellect of all. And yet it all came to nothing.
If you don’t believe that it came to nothing, just sample one of of Sidis’ books. The one I looked at was “scientific,” yet profoundly unempirical and verging on nonsense. The guy clearly developed some kind of mental pathology, much too subtle and rare to be diagnosed.
April 19, 2010 at 10:10 am
TGGP,
Have you altered your beleifs about happiness research (and therefore happiness surveys) since your thread debate at Wilkinson’s over the FLDS?
Thanks for posting a link to that at Hanson’s, BTW. It was a good conversation, and I pretty much agreed with you throughout for what it is worth.
April 19, 2010 at 10:36 am
TGGP,
There are many reasons to reject many surveys. If they’re good reasons then you should reject the surveys in question. Bad data is bad data, no matter how much you want it to tell you something. If your keys aren’t under the streetlight you know a little, but but you can’t possibly claim you know where they are.
This GSS doesn’t appear to be useless, but what is presented here, and in other cases I’ve come across, is at best suggestive. Sometimes it may have enough to answer a particular question, sometimes not. It might be fun to speculate on, but it doesn’t really tell you very much.
If we don’t have any reliable evidence to refer to, but we feel the need to come to some conclusion anyway, then we should rely on common wisdom, logic, and just impressions. Evidence is better if there’s enough of it to be convincing, but sometimes there just isn’t. Sometimes we just have to admit our own ignorance.
April 19, 2010 at 11:17 pm
I read the line about adulterous intellectuals not many days ago, but I can’t seem to find it now. It might not have been Bertrand either.
stephen,
I can’t recall the exact state of my mind back then, so I could have unconsciously shifted my views, but I’m certainly not aware of any major change.
User,
All evidence may be “suggestive” depending on how high you’d like your standards to be. I’ll admit that it would be possible to have much more convincing data, but the GSS is easily at hand and so I exploit it for all its worth.
I don’t put much stock in “common wisdom, logic, and just impressions”. Formal logic works for artificially constructed mental systems, but often when people use the term “logic” in other contexts it means evidence free musing without constraints toward truth. Because of how unmoored human thought can be from reality and our tendency to deceive ourselves, I prefer empiricism to rationalism. Which is not to say I carry water for “irrationalism”, which is too often what “common wisdom” amounts to.
Remaining agnostic is indeed the recommended course of action much of the time, which I suppose we don’t take often enough. I guess I was irked that you started imagining possible reasons to doubt the GSS without knowing enough specifically about the survey to say whether it has such defects. Generalized suspicion of surveys isn’t so bad in my book though.
See also “the Cavuto“, FUD and privileging the hypothesis.
April 20, 2010 at 12:26 pm
I agree with you that “suggestive” is somewhat ill-defined and can be used to mean a wide variety of somewhat different things, but to stick with my theme, we work with the tools we have. If I could think of a more precise word I would.
I also agree that logic often means evidence free musing, but I would argue that more often people think they’re being logical when they reason a lot from a little evidence, rather than none. There is in either case little limit to forming elaborate incorrect theories if predictions aren’t often verified. I would also argue that empiricism has to be subservient to rationalism. Evidence means nothing until you reason about it. We don’t just want to know that sometimes in a supersaturated vapor we get weird spirals, we want to know how subatomic particles interact.
I don’t carry water for “irrationalism” either, but I will for “common wisdom”. Common wisdom is a collection of memes that have gone through quite a selection process. They aren’t only selected for correctness, but it is one of the main components of their fitness. I think lacking a strong reason to challenge it, common wisdom should be deferred to. I suppose if I want to convince you I should gather up a long list of different bits of “common wisdom” and calculate a fraction correct. That would be subject to bias in the picking though.
I’m mildly amused by your privileging the hypothesis link, because I think (I’m not positive) that you’re implying that’s what I’m doing, and I shouldn’t. My position is that that’s what you’re doing. That there is some direct causal link between two thing that are correlated on a survey seems to me a hypothesis on pretty much the same footing as a slightly more complicated connection (much more complicated and I’d be willing to get out Occam’s Razor) or some bias in the sample. I’m offering caveats, sure, but I don’t think they’re on the level of “the Cavuto”. Sample bias is one of the largest problems in statistics, and you can’t read about statistics without seeing “correlation doesn’t imply causation” enough for it to get tiresome. The things I’m bringing up aren’t far fetched “maybe, if the conditions were just right” kinds of things, they’re the basic reasons why people often get fooled by statistics. I offered other theories not to privilege them, but to attempt to point out how you were privileging the theory of a direct causal link.
Sorry, I didn’t mean to irk you. I think I may have been hammering too long now, leading it to become tiresome, so if you’re interested in continuing the discussion I’ll probably continue to respond, but since I don’t demand the last word I’ll probably only do so if you indicate you’d be interested.
April 21, 2010 at 8:23 pm
I’m not that irked now, it made for an interesting discussion.
To defend myself against charges of “privileging the hypothesis”: I didn’t pluck the relation out of thin air. Bowden made a statement about the relation of happiness and intellect which I decided to investigate. I didn’t move on to claims about causality (which is why your remark about it surprises me), I just said the evidence doesn’t support his hypothesis, including with the suggested control variable. Nowhere did I go beyond that evidence on the basis that I can imagine some possibility.
Is it “a little evidence”? A sample size of 22601 seems like a decent amount.
It is going too far to compare you to Cavuto, but I always found that bit funny.
April 20, 2010 at 11:32 pm
We don’t just want to know that sometimes in a supersaturated vapor we get weird spirals, we want to know how subatomic particles interact.
Uh! Oh! Really?
Why do you think it matters more to know about the subatomic particles rather than about the “weird spirals”?
Because, for idiosyncratic, historical, biological, evolutionary, whatever reasons, the
“weird spirals” are the primary evidence we are confronted with and interested in.
The subatomic particles are only a mean to an end, that is, making better predictions of the behavior of the “weird spirals” and they may not (always) be the best epistemological approach to such an end.
I am thinking about things like ancient Chinese medicine which are an horrendous crapload of nonsense on a theoretical perspective and yet get results sometimes far better than modern “rational” approaches (not to speak of disasters like Vioxx).
Full blown reductionism may be helpful in well constrained domains like physics, not so much in more scruffy and complex ones like medicine economy and… politics!
(also genetics: http://www.nature.com/news/2010/100331/full/464664a.html , via GNXP, a nice illustration of the problems with reductionism: complexity explosion)
P.S. My own blog post “Objects as epistemological artifacts” may (?) help in understanding the rationale behind my position, I am not a “realist”.
April 21, 2010 at 10:18 am
I don’t disagree with the points you’re making. I was attempting to point out that evidence is meaningless until we reason about it. I choose to mention cloud chambers in particular because I think that’s a good example where the bare evidence (“weird spirals”) requires a great deal of reason to be applied before any meaning can be extracted. No one, seeing a cloud chamber, without prior knowledge of physics (at least in the theoretical sense), would have any clue what caused the tracks. To extract any information from them you have to reason about pre-established concepts like centripetal acceleration, and electric charge (or at least 1/r^2 forces) and so on.
Subatomic particles are only a means to an end, but we reason about the evidence in cloud chambers to form a model which allows us to predict things, theories not restricted to just what happens in cloud chambers. We need the evidence, but it’s useless until we use reason to form theories. An empiricism which denies that rationalism is “above” it is an empiricism in which nothing useful can ever be learned, only memories of sensations retained.
I’m not sure why you mention “full blown reductionism”. You seem to be following up on what I said, but I’m not sure how. I’d like to try to communicate accurately, so I’d appreciate it if you could tell me what I said to make you think I was making about about reductionism instead of about rationalism and empiricism. Are the concepts simply too heavily entwined? I confess that although I’ve read these terms (empiricism, rationalism, reductionism, epistemology) and think I have a good idea of what they mean I don’t know their whole history, nor did I look up the exact meanings before posting.