pageok
pageok
pageok
You Can't Tell a Conservative Without a Score Card.--

I have written a draft op-ed analyzing Jack and Jeanne Block's study of the correlation between the political orientation of young adults and evaluations of them made by their nursery school teachers about 20 years earlier. I have two copies of the study itself, one downloaded from the journal's website and an earlier draft kindly sent me by Jack Block. But I didn't ask Block's permission to put up a copy here, so if you want a copy, you may get one from Michelle Malkin's website. Until I figure out where to place my op-ed (or whether I even have the time to do more than a cursory job of shopping it around), I will post here only some of my observations.

Do whiny, deviant, insecure nursery school kids grow up to be conservatives? And do fluent, resourceful, self-reliant pre-schoolers grow up to be liberals? The answer to both questions would appear to be "Yes" according to a new study to appear in the Journal of Research in Personality. Its authors are the eminent psychologist Jack Block, an emeritus professor at the University of California, Berkeley, and his deceased wife and colleague, Jeanne Block.

The study claims that those who were relatively conservative at age 23 were originally evaluated by their nursery school teachers as uncertain, guilty, and rigid. The supposedly conservative boys "were especially viewed as deviant from their peers and sensitive to being different." As youngsters, the conservative girls "tended toward indecisiveness, were easy butts of peers, and were quiet, neat, compliant, fearful and tearful." On the other hand, the kids who later became relatively liberal were viewed by their teachers mostly in positive terms, such as "resourceful, autonomous, expressive, and self-reliant."

So far, public responses to the Block study have been mixed.

The Toronto Star, which broke the story in the press, gave the study respectful coverage, but also included the derogative comments of Arizona State psychologist Jeff Greenberg: "I found it to be biased, shoddy work, poor science at best." But the only criticism attributed to Greenberg and mentioned in the story is less than compelling: "He thinks insecure, defensive, rigid people can as easily gravitate to left-wing ideologies as right-wing ones. He suspects that in Communist China, those kinds of people would likely become fervid party members."

Yet assessing conservatism and liberalism in Communist countries is not terribly meaningful because Communist dictatorships are usually considered left-wing by those on the right, and right-wing by those on the left. Even if the Blocks' results were not generalizable to countries like China, that would mostly affect the domain over which their results would apply, not their validity in the United States. So Greenberg's criticism does not go to the heart of the Blocks' thesis.

Responses from bloggers to the Block study have ranged from generally accepting on liberal sites to skeptical or dismissive on most libertarian or conservative sites. Some insightfully mentioned that the new Block nursery school study reminded them of the May, 2003 article in the Psychological Bulletin by a group headed by Professor John T. Jost that reviewed and analyzed data on conservatism. The Jost study also found conservatives to be rigid, fearful, and lacking in intellectual complexity.

Just from what is in the Blocks' article, there is scant reason to think that those whom Block calls "conservatives" are really conservatives, rather than Democrats or political moderates. Those whom the Blocks think of as relatively "conservative" on many of their measures of conservatism are expressing views that would mark them as Democrats or political moderates (not conservatives) in the general public. The study itself acknowledges that the most conservative third or half of their 95 subjects from Berkeley and Oakland were not really conservatives, admitting that there were "relatively few participants tilting toward conservatism."

That the new Berkeley study has severely mismeasured conservatives and liberals is shown by several odd findings, including the strong relationship they claim between their liberal-conservative scale and their subjects' IQs. The Blocks report a correlation between IQ and their conservatism-liberalism scores of .36 for males and .24 for girls. Just for comparison, in the 1996-2004 General Social Surveys the correlation between political party identification (as a Republican, Independent, or Democrat on a 7-point scale) and political orientation (as a conservative, moderate, or liberal on a 7- point scale) is only about .30. And the correlation between a man or woman being a conservative and opposing abortion is only about .17-.25. So if the Blocks' measure of conservatism were to be generalizable to the general public, then a low IQ would have to be a generally better predictor of whether a man is a conservative than whether he was a Republican or whether he opposes abortion—which is highly implausible.

What the Block study really shows is that kids who grow up to have low IQs struggled even in nursery school, while kids who grow up to have high IQs did well even as youngsters. Since in the general public (as shown by General Social Surveys and American National Election Studies), adult conservatives are consistently better educated than the average citizen and (as shown in the GSS) do better on IQ-style vocabulary and reasoning tests, the results of the Block study are generally inapplicable to typical political conservatives, liberals, and moderates in the United States.

With the thousands of studies that have been done on conservatives using representative samples of the general public over the last five decades, we actually know quite a bit about them. It is time for researchers and the editors of scholarly journals to use a little common sense and ordinary skepticism when looking at statistical relationships that are so farfetched that they are almost certainly wrong. And when they encounter such a relationship, they should take the next step and try to determine where the study went awry.

Sadly, of course, many of the questions that the Blocks used to mismeasure political orientation have been widely used by other researchers in social and political psychology—despite the tendency of self-identified conservatives in the general public to be more likely than nonconservatives to give supposedly "liberal," tolerant answers to many of these sorts of questions. Researchers often admit that some of the most widely used indices and scales were developed by first writing down sets of (mostly negative) stereotypes of conservatives. The last prominent study of conservatives that had similar measurement problems was a 2003 Psychological Bulletin article by a group of academics led by Professor John T. Jost (now of NYU). Andrea Irvin, a student at Berkeley, offered an amusing, if somewhat unfair assessment of the Jost study: it was "about as scientific as phrenology."

UPDATE: I updated this post in a new one.

Randy R. (mail):
It sure is an interesting study, and I am willing to bet there is a grain of truth in there. Whether it's the whole truth, I just don't know.

I have found, however, that many conservatives (and I count conservative religious people among them) are indeed different from liberals. Conservatives often have a lot of respect for rules, authority, trust in institutions (such as marriage, church and so on), dislike people who question things too much. They also like top-down decision making without dissent, where everyone is 'on board' with the leader. Conservatives respect power and like to set rules for everyone to follow, because that how people should act.

Liberals, on the other hand, like to question authority, question the rules, have skepticism of social institutions. Liberals distrust being told what to do or believe, and think dissent is a virtue because it help address the real problems. Liberals like to be creative and respect artists. Liberals like to set rules for everyone to follow, because it's what is good for them

In government, this plays out in two ways. Republicans tend to set policy from the highest levels, and really don't care about your opinion on the matter. Everyone else is supposed to follow along without complaint, dissent or even questioning. Democrats believe they should listen to everyone, get all input, consider and weigh options and then try to formulate a policy that most everyone can agree on.

Which is better? it depends on the situation! Where you need swift or decisive action, the repub way works best, when you are trying to solve a complex problem, the liberal way works best. Sometimes neither works. But I think it is in the nature of people that they tend to gravitate towards one or the other views.
3.23.2006 2:07pm
Nathan Hall (mail):
Randy,

I think you overstate your case as regards dissent. Have you seen the way Democrats tread dissenters in their ranks? Ask Joe Lieberman or Cristopher Hitchens about it sometime.
3.23.2006 2:13pm
JohnAnnArbor:
Randy's comment is a study in projection and stereotyping.

Democrat LOVE business regulations and assail anyone who so much as asks questions about them.
3.23.2006 2:19pm
Nunzio (mail):
I think Randy R. must be a conservative because he doesn't question the Block study but acquieces in the face of authority from an institution.
3.23.2006 2:19pm
FXKLM:
If you define conservatives as authoritarians, you're going to reach very different conclusions than if you define conservatives as supporters of limited government. I think the latter definition is probably more consistent with most people's view of the political spectrum in modern America.

Have you looked at the Kerlinger article that is used to define liberalism and conservatism? If not, you probably should before you send this out for publication. It sounds like you're just looking at the 10 issue tests.
3.23.2006 2:23pm
Guest2 (mail):
As one of William S. Gilbert's characters observed:

I often think it's comical (fa la la, fa la la)
That nature always does contrive (fa la la la)
That every boy and every gal
That's born into this world alive
Is either a little Liberal
Or else a little Conservative!
3.23.2006 2:28pm
M. Brown (mail):
And just look at how tolerant of dissent and what lovers of independent thinking those liberal professors are in their little campus fiefdoms. "Liberal" and "open-minded?" An oxymoronic combo if there ever was one.
3.23.2006 2:32pm
James Lindgren (mail):
FXKLM:

Thanks for the suggestion.

I have looked at over 2 dozen studies of representative samples of the general public since 1952 that ask questions from the Adorno F-Scale (Authoritarian Personality Scale). Despite the Jost group's use of this scale as a measure of conservatism itself, I have yet to find a study of representative samples of the general US public that shows conservatives (or Republicans) as scoring significantly higher on average than both moderates and liberals. Usually, the highest F-scale groups are moderates and Democrats (particularly nonliberal Democrats), since these are the poorly educated groups.

I have asked other scholars (including John Jost) for any such studies, and so far no one has been able to identify any.
3.23.2006 2:32pm
Thorley Winston (mail) (www):
I think Randy R. must be a conservative because he doesn't question the Block study but acquieces in the face of authority from an institution.


ROTFLMAO!!
3.23.2006 2:36pm
Kovarsky (mail):
I don't want to defend the test's methodology generally, because I have serious questions about sampling and the control group here. I also have problems with types of "tests" that were administered to create the X-axis.

That being said, yesterday I saw a considerable number of posts asking me to explain what I was calling the "Berkeley Bias" ("BB"), and I never got one. Of course I got some people calling me a stupid liberal twit and castigating me for being unable to see the "obvious bias." I'm going to go through the exercise one more time though. I'm interested in knowing what effect the Berkeley situs had on the RELATIVE levels of conservativism WITHIN THE SAMPLE. These criticisms seemed to be premised on the idea that three year olds were capable of self-identifying conservative or liberal and capable of feeling alienated or angry about that identification at age 3.

I understand that there are serious flaws with the study. I just think the "oh it was at Berkeley" critique to be a particularly poor thought out one. Is there a STATISTICAL phenomenon here that I'm missing.
3.23.2006 2:41pm
Abdul (mail):

Conservatives often have a lot of respect for rules, authority, trust in institutions (such as marriage, church and so on), dislike people who question things too much. They also like top-down decision making without dissent, where everyone is 'on board' with the leader. Conservatives respect power and like to set rules for everyone to follow, because that how people should act.


A friend of mine at law school who is an avowed liberal and I (avowed conservative) have discussed this prospect and arrived at the opposite conclusion from you. My friend said that conservatives tend to have an Anglo common law approach to institutions, while liberals have a French code law approach.

The common law respects institutions, just like the conservatives you mentioned, and defers to tradition almost to a fault. However, this is generally empowering, allowing people to order their affairs in private without government intrusion. For example, the conservative owns a gun to be his own home defense instead of relying on the police; the conservative gets married to provide a stable environment for children rather than relying on the state to provide child care. You can see the "wisdom of crowds" approach in everything from federalism to school vouchers. Admittedly the analogy fails when applied to pro-life and sexual privacy issues which are items of religious faith among many.

In contrast, liberals would prefer if everything were run by a comittee of experts. The liberal wants his child in public schools and day care because too many people can't be trusted to raise their own kids right. In those schools, we have to restrict or remove access to soda because people can't be trusted to brush their teeth or run around the block to work off the extra calories. Besides, it's not fair that the child of the rich family gets a better private education than the person who's only fault in life was to be born in a poor household, so a committee of experts will balance things out so costs and rewards are born equally by all.

These philosophical approaches both have advantages and disadvantages. Metric was invented by a French committee of experts and is more sensible than Imperial measurement, developed through an Anglo common-law respect for tradition. However, the French language is managed by a committee of experts and is sinking in relevance in international business and scientific affairs, while English--accepting the wisdom of crowds--grows by common consensus and becomes more relevant in international affairs.
3.23.2006 2:43pm
te (mail):
Hmm, a whiney thousand word essay about why it is silly to think that conservatives are whiney?

Seems like a dandy idea.
3.23.2006 2:58pm
Mike BUSL07 (mail):
You gotta love the irony. This guy is saying that people who disagree with him are psychologically inferior, all to stroke his own ego. How insecure of him.
3.23.2006 3:09pm
Nathan Jones (mail):
The use of a .1 significance level by the authors leaves much to be desired.

I can't speak for psychology but the conventional standard of significance in major social science journals is to conclude that there is statistically significant evidence of a relationship between observed outcomes and inputs (reject the null hypothesis) only with significance scores below .05. Using .1 significance levels allows a researcher to find "evidence" of something despite the fact that there is up to a 10% probability that these results are only being observed due to pure chance alone.

Even work using conventional significance levels should be closely scrutinized since the claims are rather far-fetched. Using .1 rather than .5 significance is a step in the wrong direction. They should be using .01 significance for something this controversial. The authors also don't mention whether these are one-tailed or two-tailed significance tests, so the probability of a false positive here could be quite high.

A two-tailed significance test at p<.1 can be rationally ignored by skeptical readers. In technical terms, a one-tailed p<.1 significance test means they "got nuthin'"

My two cents.
-nj
3.23.2006 3:09pm
James Lindgren (mail):
Kovarsky wrote:

I'm interested in knowing what effect the Berkeley situs had on the RELATIVE levels of conservativism WITHIN THE SAMPLE. These criticisms seemed to be premised on the idea that three year olds were capable of self-identifying conservative or liberal and capable of feeling alienated or angry about that identification at age 3.


First, of course, the political orientation was assessed at age 23, not at 3.

Second, as I noted, the study itself admits that there were "relatively few participants tilting toward conservatism." So it's not just speculation. If you want to say something about what conservatives (who are now about 35-45% of the population) believe, then you need a substantial enough sample of them to generalize about them.

This is particularly true when the groups that tend express the attitudes that researchers mistakenly call "conservative" are actually in the center or center-left of the spectrum (eg, moderate Democrats). Without a representative sample on the right, then the study would be merely comparing liberals to centrists (moderate Democrats), not liberals to conservatives. Since the outliers on many supposedly conservative views are in the middle of the spectrum, the relationships are not linear and a skewed sample will yield false conclusions about differences in the general public when applied to a wider spectrum. I think that is the statistical explanation you requested.

For example, if I had the data (which Jack Block is unable to share at the moment), I might find that the sample is so skewed that the observed differences are actually between (well educated) liberal Democrats and (poorly educated) nonliberal Democrats. That would certainly be more consistent with national opinion studies and with block's published methods and results than the interpretation of the results given in the paper.
3.23.2006 3:10pm
guest1:
What the Block study really shows is that kids who grow up to have low IQs struggled even in nursery school, while kids who grow up to have high IQs did well even as youngsters. Since in the general public (as shown by General Social Surveys and American National Election Studies), adult conservatives are consistently better educated than the average citizen and (as shown in the GSS) do better on IQ-style vocabulary and reasoning tests, the results of Block study are generally inapplicable to typical political conservatives, liberals, and moderates in the United States.

Jim, the study in question, Randy's, and several other comments demonstrate one thing: everyone wants to think the group he belongs to is somehow better than the other group.
3.23.2006 3:14pm
Bruce:
Jim, could you give more complete cites to the "General Social Surveys and American National Election Studies" that show that "adult conservatives are consistently better educated than the average citizen"? I'd like to see those. Also, if conservatives are above average, who's below average? Liberals? Moderates? Both?
3.23.2006 3:16pm
Bimp (mail):
Does anyone else see the irony here: namely, that the "resourceful, autonomous, expressive, and self-reliant" people need study after study after study to proove that they're better than conservatives?
3.23.2006 3:24pm
Kovarsky (mail):
James,

First, of course, the political orientation was assessed at age 23, not at 3.

I'm sorry if I was unclear, but I was responding to a criticism of the study that said, effectively, that "of course these kids were the more alienated, defensive, and less confident ones - they were conservatives in berkeley!" The implicit is that Berkeley's liberal bias is capable of polarizing 3 year olds - and that it was not in fact an independent variable. I certainly wasn't confusing the age at which the reported political data was observed.

Second, as I noted, the study itself admits that there were "relatively few participants tilting toward conservatism." So it's not just speculation. If you want to say something about what conservatives (who are now about 35-45% of the population) believe, then you need a substantial enough sample of them to generalize about them.

It's unclear to me whether you are talking about the independent or dependent data here. Of course it matters if there are "relatively few participants tilting towards conservatism" if that is a characterization of the time-series data observed at 3. But if you're talking about the observations at 23 - in other words if you're saying that the independent variable has a disproportionately high fraction of N that is liberal - well I don't understand why that matters as long as you have enough conservative data points to be statistically significant. The question, then, is "how many conservative data points" (observed at age 23) not "what fraction of the data points are conservative" (again, this is an important question if you are talking about the data observed at 3).

This is particularly true when the groups that tend express the attitudes that researchers mistakenly call "conservative" are actually in the center or center-left of the spectrum (eg, moderate Democrats). Without a representative sample on the right, then the study would be merely comparing liberals to centrists (moderate Democrats), not liberals to conservatives. Since the outliers on many supposedly conservative views are in the middle of the spectrum, the relationships are not linear and a skewed sample will yield false conclusions about differences in the general public when applied to a wider spectrum. I think that is the statistical explanation you requested.

But the data subjects self-identified? And that is IN ADDITION to being subject to political spectrum sampling that was pretty widely accepted? Maybe people who grew up in Berkeley are more likely to self-identify in a certain direction than are people who grew up elsewhere (or maybe they're more likely to self-identify near the wings of the distribution, whatever side they're on). But that seems to me to be a different objection than the one you're making.

To sum up, I was responding to an earlier criticism of the study, not defending the study generally. That criticism relies on an assumption - that Berkeley's leftist politics renders the data observed at 3 to be a non-independent variable - that I do not understand. I understand your point to be that there may not be enough conservative data points N observed at age 23 to say anything meaningful about the study, or that perhaps there's something about Berkeley that made the timid students grow up conservative and the confident ones grow up liberal (I think that is the most persuasive criticism, incidentally, but those are not the criticisms I meant to address in my post.
3.23.2006 3:34pm
Jim Hu:
It seems intuitive to me that the ends of the political spectrum would fit the "liberal" personality features, and that those with low self-esteem would gravitate to centrist positions as being a place to minimize the interactions that lead them to whiny, insecure, unhappy reactions.

Another question I wondered about from the news coverage (haven't read the paper) is whether there is an observer effect from the teachers evaluating the kids. The kids might not have had defined political philosophies at that age, but the teachers could have been reacting to cultural markers related to the home environments of the kids (I am assuming that home environment is positively correlated with later political affiliation, David Horowitz notwithstanding)...including parent-teacher interactions, but also including phenotypes associated with class, race, ethnicity, religion, political affiliation etc. Growing up Asian in suburbia, I can easily believe that the teachers treated students differently wrt expectations, some of which become self-fulfilling.
3.23.2006 3:42pm
Taimyoboi:
"That criticism relies on an assumption - that Berkeley's leftist politics renders the data observed at 3 to be a non-independent variable - that I do not understand."

Kovarsky,

I think the unstated link here is that the overwhelming number of families in the Berkeley area (something like 90%) are pretty liberal.

Since parents' political orientations are the most decisive factor in determining the child's, not controlling for that as an independent determinant would distort any conclusions you could draw about the child's behavioral dispostion being correlated with future political orientation.
3.23.2006 3:58pm
James Lindgren (mail):
Bruce:

The GSS and ANES can be downloaded from the ICPSR at Michigan (if you have a University account). In any event, you can run simple (unweighted) analyses of the GSS yourself at this site (the ANALYSIS tab):

http://webapp.icpsr.umich.edu/GSS/

The raw (un-recoded) variables of interest are EDUC, DEGREE, YEAR, POLVIEWS, and PARTYID.

They do not publish their results as a full "study" report, except for simple frequencies in the codebook, which is online at the site above through 2000. They simply release the data.

I will try to run numbers for you later, but in short, in the combined 1996-2004 GSS, Democrats and Independents have meaningfully less education than Republicans, and moderates have meaningfully less education than liberals or conservatives. In some GSS analyses I've done, the best educated subgroups are conservative Republicans and liberal Democrats, which is why they score similarly on a substantial minority of questions that are supposed to identify conservatives.

The ANES generally shows a bigger (though less stable) educational advantage for conservatives (over moderates) than the GSS.

This will all be in my Ph.D. thesis in a year or two.
3.23.2006 4:08pm
Truesilver (mail) (www):
I don't know about the rest of it, and I'm not a statistician, so my understanding might well be off. But the thing that occurs to me in reading over the study is the sample-size. It is bothersome to me to note that the study-as-paper cites, on page 3, that the original sample size was only 128 children, only 104 of whom were examined in the follow-up to the study. This seems extraordinarily small for a study that purports to apply to the whole of the United States, almost 300 million people.

I do understand that a survey of just under 6,000 is believed to have a sufficient sample-size, but even that bothers me a little. Professor Lindgren, would you mind awfully explaining to the poor layfolk among us what is required of sample-size for a study to be considered "generally applicable?"

This small sample concerns because it seems clear that even a sample-size of several thousands doesn't necessarily get past the opinion/development biases in a particular locale, in this case, the Berkeley-Oakland area.

I don't intend to reiterate Kovarsky's comments, though I admit mine are in a similar vein. A sample as small as this one seems largely un-indicative in this case, especially without some corroborating studies from other regions in the nation, including more Conservative regions of the country.

I may be wrong in my concerns about the sample size, but I'm not one who's big on trusting statistics generally, especially when they talk about intelligence or politics.
3.23.2006 4:14pm
James Lindgren (mail):
Kovarsky wrote:

But the data subjects self-identified? And that is IN ADDITION to being subject to political spectrum sampling that was pretty widely accepted?


----
Thanks for your detailed response.

But self-identification was only one of the more than 50 questions they used to classify their subjects on political orientation. It may have contributed as much as one-sixth to the ultimate lib/con scores, but the other 5/6ths of the scale would tend to skew the results toward what those other scales measure. The reason that I requested Block's data was precisely to examine the self-id data.

On your second point, I can only restate what I wrote. The use of commonly used scales that don't measure what they are supposed to measure is precisely why the Blocks' study went awry. That was my main point. The problem is with the measurement of who is a conservative, not with researcher mendacity.
3.23.2006 4:25pm
tefta (mail):
I remember reading an analysis of voters that proved liberals were smarter than conservatives because those who voted for Democrats had far more advanced degrees than those who voted for Republicans. What they neglected to clarify was that those degrees were in teacher ed, and other pseudo disciplines like women studies, black studies, etc. What a joke.

Sure people who buy what the leftwing moonbats are selling, something that's been proved wrong time and again over the past one hundred years and has caused poverty and suffering round the globe are smart and we who have somehow contrived to keep the U.S. prosperous and thriving, are dumb.
3.23.2006 4:38pm
Kovarsky (mail):
James,

So even on the residual 5/6, my question is this: I understand that the "standard measures" of LIB/CON might skew towards one side or the other, but really does that mean for the results within the sample? In other words, what does it mean for the study's claim that "look, we're just talking about relative levels of right v. left here;" then all it's saying that the more confident, the more likely to be liberal within the sample, without respect to whether you actually score as a liberal on a dichotomous variable?
3.23.2006 4:43pm
FXKLM:
Kovarsky: The conservative population of Berkeley is extremely limited. That means that people who are more conservative than the average Berkeley resident are probably not actively conservative. They're just uninterested in politics and intellectually unconservative. With a population sample like this, I don't think it's possible to distinguish between the effects of liberalism as compared to conservatism and the effect of political activism as compared to political passivity.

I think the more serious problem with basing the study in Berkeley is that the people doing the personality analyses are Berkeley graduate students and presumably tilt very heavily to the left. That's obviously a problem in the part of the study that deals with personality traits at age 23. Leftist interviewers are probably going to take a harsher view of the personality traits of subjects who disagree with them.

I think it may also reflect a bias in the evaluation of personality traits of toddlers. I don't think it's implausible that there are personality traits of toddlers that correlate in some way with their political views as adults. If that's true, I think it's reasonable to suspect that Berkeley graduate students in education would have an unflattering interpretation of the behavior of proto-conservative toddlers. That doesn't mean that the toddlers had well formed political views that offended the reviewers. It may simply mean that liberals and conservatives have radically different world views that are formed in part before kindergarten. And of course if it isn't true that there are behavioral differences between future liberals and conservatives as toddlers, the whole study is wrong anyway.

I think the Berkeley bias of the subjects, the interviewers, and author all raise very legitimate questions about the value of the study.
3.23.2006 5:12pm
Jim Hu:
Kovarsky,

Maybe I'm missing your point, but it seems to me that James is saying:

If you plot self-confidence etc on the Y axis, and political orientation on the X axis, I think James is saying that the true distribution is shaped like a U, and the sample cuts off one half, which gives it a monotonic positive correlation. So it's true within the sample, but so what?

In other words "true within the sample" is trivial, meaningless, and in the context of the loaded content, disengenuous at best.

I actually think the true distribution is more like "_M_", where the "_"s are the moonbats and wingnuts, and the M is the people James is talking about. But that's just an intuition...no data or statistical analysis.
3.23.2006 5:20pm
KMAJ (mail):
This 'study', a term to be used loosely, seems more geared towards a preconceived conclusion intended to stroke the egos of liberals and prop up an argument of liberal elitism. How many of the 104 follow ups self-identified themselves as liberal/democrat ? What were the parameters given for that designation ? The statement that the vast majority self-identified as liberal, not surprising in the Berkeley community, gives a very stilted and unscientific result, locally and especially nationally.

The fact that it relies on non-expert evaluations of personalities, pre-school teachers are not psychology majors or doctors, thus raising even more questions of veracity. This study violates so many the the scientific principles necessary for a reputable study, that to even take it seriously would raise questions of ones ability to question authority, the one of the premises of the study.
3.23.2006 5:20pm
Kovarsky (mail):
FXKLM,

I think the Berkeley bias of the subjects, the interviewers, and author all raise very legitimate questions about the value of the study.

I don't think the author's bias is that much of a handicapp if his study is transparent. We can just critique the methodology and look for biases in it. But I do think the subjects and interviewers are fair game.

I think you make good points. Some relevant questions I have involve: where the 104 23-year old data points now reside and how long (the longer in Berekeley, the less likely the data is to be unbiased); similar to Nathan's criticism above, what are the confidence intervals and how narrow are they (Nathan instead seemingly just identifies some threshold beyond which the data is "unuseable"); etc.

But this I cannot accept:

I think it may also reflect a bias in the evaluation of personality traits of toddlers. I don't think it's implausible that there are personality traits of toddlers that correlate in some way with their political views as adults. If that's true, I think it's reasonable to suspect that Berkeley graduate students in education would have an unflattering interpretation of the behavior of proto-conservative toddlers.

That strikes me as a stretch. It requires some pretty fantastic assumptions: (1) that behavioral psychologists already knew what the predictors of conservatism were (a condition that seems unlikely given the mission statement of the study!); (2) more specifically, that they knew how to identify these traits in 3 year olds; (3) that if these traits appeared in three year olds, the REASON that trait was there had something to do with the political alienation the 3 yearld experienced in the playpen; etc. etc. You get my point. I think the study has problems, but the argument about the insidious identification and manipulation of "proto-conservative" indicators in 3 year old toddlers strikes me as very unlikely to be the source of those problems.
3.23.2006 5:26pm
Clayton E. Cramer (mail) (www):
Michelle Malkin is pointing out that it appears the children studied were at a rather special school--one open only to the children of UC faculty and staff. Gee, can you see any reasons why this might render the applicability of the results questionable?

I mean Bezerkeley is weird enough. I have a relative who grew up there who stopped using LSD when he was 12--and the peer pressure problems that he suffered for doing so were tremendous.

Bezerkeleyites used to talk about "conservatives" on the city council--and by that, they meant councilmembers who were openly Stalinists. It isn't called the "unlocked ward" by even other Bay Area liberals for nothing.
3.23.2006 5:36pm
Erik Voeten (mail):
I agree that the sample size and the Berkeley location make generalizations problematic but the measure that they use for locating individuals on a liberal-conservative continuum is actually rather thorough and, I must say, much better than what most people use. Moreover, the authors are rather upfront about the limitations of their study.

Your conclusion on IQ may be so, but it does not follow from the presented results. Correlations of .25 or .30 are really not that strong. You would really need to look at the individual data to make such statements.

In all, this is a study with clear limitations but I don't agree that it is executed in a shoddy manner or that it is so irrelevant that it should not have been published.
3.23.2006 5:38pm
Mikeyes (mail):
This is a very technical paper and the authors use a number of tests and measurements that are totally unknown to me. I suspect that a fair critique of the method and results (which are the crucial aspects of any study)will have to be done by those familiar with the tools used.

Long term studies are often heroic in an administrative sense since they strive to hold together a group of inidviduals over a long period of time. Lost subjects may skew the results, the initial choice of tools that may now be invalid is a problem and low numbers introduce bias. The advantage of long term studies is that they can help determine if there is any predicatibility in the variables studied. But if the measuring tools are invalid, then the study is, too.

I suspect that like a lot of others commenting, that the sample size is small and the sample may be biased. At least too biased to draw the very general conclusions that seem to upset a lot of readers. Besides, the authors were measuring "relative conservative" and "relative liberal" trends on a continuum (VC trivia, can you name two other non technical words ending in -uum?) and making the leap from their terms to your terms is fraught with danger. Just as an example, which brand of conservative does one mean? Goldwater conservative? Libertarian? Catholic? The same could be said for the label of liberal.

So this is most likely a tempest in a teapot, the study will be critiqued on it's merits, if it has any, and while it will be cited for political purposes, most likely will not change the political landscape.
3.23.2006 5:42pm
Kovarsky (mail):
Jim,

If you plot self-confidence etc on the Y axis, and political orientation on the X axis, I think James is saying that the true distribution is shaped like a U, and the sample cuts off one half, which gives it a monotonic positive correlation. So it's true within the sample, but so what?

Maybe I'm misunderstanding Jim, but I believe Jim and I are in agreement that the self-confidence of the three year old is the independent variable on the X-axis, and that political orientation is on the Y-axis.

In other words "true within the sample" is trivial, meaningless, and in the context of the loaded content, disengenuous at best.

This is where I think Jim and I disagree. If you assumed that the rest of the test were conducted rigorously, with proper controls, etc., I do think that the observation within the sample means something, as long as the sample is statistically significant.

The point isn't the "absolute value" of the orientation - it's whether you are more likely to be liberal on the basis of certain observed traits. It doesn't matter whether the Y-axis captures that increased likelihood by moving upwards from "moderate conservative" to "moderate liberal" or upwards from "extreme conservative to moderate conservative" or "sean hannity to bill o'reilly." The point is the relative movement along the Y-axis.

I concede Jim's point that whatever 5/6 of the LIB/CON variable consists of "orthodox testing methods" that are biased in some way, that doesn't really indict any relative observations in the sample.
3.23.2006 5:42pm
Clayton E. Cramer (mail) (www):
Liberal doesn't mean quite the same thing in Bezerkeley that it does in other places. When I worked in Marin County in the early 1980s, one of the other engineers lived in Bezerkeley. He would bring the Bezerkeley student newspaper with him every day. It was always amazing to see what passed for appropriate newspaper content. My favorite example was a classified ad from the Bezerkeley student paper that he kept up on his wall, offering rental of sheep, "No VD, guaranteed."

Perhaps the ad wasn't serious. But it says something about the Bezerkeley community that the student newspaper in that time period (late 1970s, early 1980s) would consider that amusing or appropriate.
3.23.2006 5:42pm
JosephSlater (mail):
I think it would be better for American political discourse in general if folks didn't try to attribute differing opinions (at least within a fairly broad range) to some combination psychological problems, irremediable ignorance, bad faith, and/or bad motivations. It's all just ad hominem.
3.23.2006 5:43pm
JosephSlater (mail):
" . . . some combination OF psychological problems . . . ."
3.23.2006 5:44pm
Jay Arthur (mail):
Before anyone dismisses the previous reference to Michelle Malkin's post simply because it comes from her, I suggest following the links and deciding on the veracity of the claims yourself.

I'll be the first to agree that much of what Malkin posts on her site is selective rabble-rousing, but the reason I regularly check sites like hers (and also left-leaning sites just as shrill) is that sometimes they make valid points with corroborating information that nobody else has. This looks to me like one of those times.
3.23.2006 5:48pm
Randy R. (mail):
This is so much fun, to be the first one to post. It frames the debate for quite a while!

What's funny is that people read into my post that I must be a liberal, or something, even though I took pains to say that neither is necessary correct, unless the situation demands it. In other words, each has its advantages and disadvantages based on what the situation is, and in fact a single situation might demand a combo of the two different styles.

Yet -- people here still took pains to ridicule one side as being the dumbasses. Anyone who thinks their side is better than the other is all situations, or even most, is a fool, and he fools no one but himself. In other words, get over yourselves and try to have the open mind that you claim you have.
3.23.2006 5:49pm
Mike BUSL07 (mail):

Anyone who thinks their side is better than the other is all situations, or even most, is a fool

holy overstatement, batman! Why bother aligning yourself politically, if your group is wrong most of the time?
3.23.2006 5:52pm
FXKLM:
Kovarsky: I don't think the interviewers were necessarily consciously discriminating against proto-conservative toddlers. It may be that some kids have an innate respect for property rights and an aversion to sharing. As an adult, those behavioral traits might be expressed through opposition to welfare. As a toddler, they might be expressed through a stronger desire to hold onto your own toys and weaker desire to grab someone else's. A conservative interviewer might be impressed with the kid's willingness to leave other kids and their toys alone, whereas a liberal interviewer might think that the kid isn't socializing properly. I'm suggesting that there might be some connection between toddler behavior and adult political values that might not be consciously perceived by the interviewers.

My deeper point is that the study is damned either way. If there is no connection between toddler behavior and adult political preferences, the whole conclusion of the study is wrong. If there is a connection, the political bias of the interviewers raises some questions about the validity of the methodology.
3.23.2006 5:52pm
Cabbage:
Let me get this straight. We potentially have a situation (especially if Malkin's claim is accurate) where in this group of 104 subjects there are 95 extreme-leftists and 9 slightly less extreme leftists and this supposedly gives us an insight into the differences between liberals and conservatives?
3.23.2006 5:58pm
Cabbage:
It's a little like comparing 5 year olds to 6 year olds to gain an insight into the differences between children and adults...
3.23.2006 6:01pm
Kovarsky (mail):
Sorry that last email was to Jim Hu, not to Jim Lindgren.
3.23.2006 6:07pm
Michael Livingston (mail):
An interesting idea, but I'm skeptical. When I was in college in the 1970s there was a lot of work on how people supported Nixon because they had "authoritarian personalities" . . . their parents hit them when they were little, they were afraid to express emotion, etc. Then the people who did these studies got a little older, acquired a little more power, and became . . . well, more authoritarian themselves. It is an interesting and provocative subject, but one very difficult to separate from political bias. If the authors can look at today's kids and predict their future politics, I would give it more credence.
3.23.2006 6:08pm
James Lindgren (mail):
Jim Hu,

Thanks. That is in part what I am saying.

The relationship between political orientation and some other measures is U-shaped, with moderates as the outlier. If almost all of one sample is from one half of the political curve and almost all of another sample is from the other half of the political curve, then you would reach opposite conclusions by analyzing either skewed distribution.
3.23.2006 6:11pm
Kovarsky (mail):
Cabbage,

A better way to think about it is that you have people who are distributed across a "political orientation index" from 1 to 100. The 95 versus 9 number is a red herring - the point is, more liberal, more likely to have been confident. You get the 95 versus 9 number because somebody set the switch at an index of 95 - that's for expository simplicity, but it doesn't diminish the directional inference of the results.

Also, I'm happy to stand an fall on what I say, and I'm wrong a lot of the time, but it really irritates me when people don't read the things they're critiquing first (I'm not talking to you at this point Cabbage). A prime example is someone upthread who said that the kids were all from a privileged school (obviously trying to create the inference that these kids were all wealthy kids of berkeley profs). Please read the study:

Subjects were participants in the Block and Block Longitudinal Study of Cognitive and
Ego Development at the University of California at Berkeley, begun in 1969 (see Block,
1993; Block &Block, 1980b) for comprehensive descriptions of the study.
Subjects initially (about 1969--1971) were attending two diVerent nursery schools and
resided primarily in the urban areas of Berkeley and Oakland, California; they were heterogeneous
with respect to social class and parents' educational level. At age 23 (about 1989),
104 of the original 128 subjects were intensively assessed. Of these, usable data for the present
analyses were available for 95 subjects, 49 females, and 46 males.
3.23.2006 6:18pm
Kovarsky (mail):
Ahhh... I see the problem. The authors are inverting the axes. Jim Hu was right about that; I was wrong. The authors are treating the political identification (CRQ and CAQ) as the independent variable. That's unfortunate. The study should be to what degree 3 year old behavior predicts political orientation, not vice versa. Good point, Jim Hu - and Jim Lindgren - I'm sorry if I mis-stated a consensus.
3.23.2006 6:21pm
James Lindgren (mail):
Kovarsky,

Let me try again with an analogy.

Let's say that people who watch TV 3 hours a day are better educated and know more about the world than those who don't watch TV at all. But those who watch TV 8 or more hours a day have less education and know less than those who watch only 3 hours a day.

Then one Researcher studies the relationship between TV watching and education, but includes only a small percentage of the sample who watch only 1 or fewer hours of TV a day. He would conclude that TV watching went DOWN with increasing education.

Now assume that a second Researcher uses a sample that includes only a few people who watch more than 3 hours a day. That researcher would conclude that TV watching went UP with increasing education. Both opposite conclusions can't be right.
3.23.2006 6:30pm
Bruce:
"This will all be in my Ph.D. thesis in a year or two." Ah, cool. I look forward to it, thanks. As you may have guessed, I don't really want to run the numbers myself.
3.23.2006 6:41pm
Kovarsky (mail):
Jim,

I was operating under the erroneous assumption that the X axis was confidence (the 3 year old stuff, whatever you want to call it) and the Y-axis was political orientation. Apparently the study inverts them, which is unfortunate. I don't know why they stated their data that way.

I hope you can see now where the confusion was.

I understand that you are saying the sample is biased such that you only get the 1/2 the U-shaped part of the curve. That, of course, would not have been a problem if the study had presented the data on the axes as I believed them to have been presented (with political orientation on the Y, not the X).

I'd be interested in seeing whether other studies that explored this relationship flipped the axes, or what Blocks' data would look like if you did.
3.23.2006 6:56pm
Le Messurier (mail):
It's really quite pointless to try and analyze the validity of this study. The only real test of its validity, and the only worth while approach is to attempt to replicate it in a different population. Everything else is statistical ononism.
3.23.2006 7:37pm
BU2L (mail):
Le Mess, your point is well taken, but it's "onanism," after Onan, the biblical character.
3.23.2006 7:54pm
lpdbw:
onionism?

I do not think that word means what you think it does.

Try Onanism. Mark Twain had something to say about that.
3.23.2006 8:03pm
Le Messurier (mail):
BU2L

Yup, I looked it up to be sure when I was posting and STILL misspelled it! But cut me some slack. I'm on Social Security and have forgotten how to spell most words.
3.23.2006 8:03pm
BU2L (mail):
Le Mess, sorry, it's just one of those words with a colorful etymology. couldnt help myself. :)
3.23.2006 8:09pm
Le Messurier (mail):
Eta who?
3.23.2006 8:24pm
Ray (mail):
A couple of things in highlight fashion:

The first thing that occurred to me even before I finished reading the post, was that the typical nursery school teacher is who? I'll leave that open to speculation. I teach algebra and physics in a small, private K thru 12 school, and I don't necessarily value the political or psychological observations of our pre-K and K teachers. In short, those responsible for the observations are from a very narrow ideological slice of the population. Think of asking the same type of studies be done by Pentecostal preachers, or military officers, or whatever group where a random sample would produce an ideological average that was markedly different than the population as a whole.

Also, those putting ideologies on a Cartesian coordinate system reminded me of the ideological differences I noticed whenever I had to sit through a week's worth of personality testing at Morgan Stanley. It was basically the Myers-Briggs type testing with a few little twists so the consultants doing the seminar could claim some sort of originality.

Essentially (and the lingo will vary depending on what flavor of MBTI one is familiar with) those that are more "feeling" or "expressive" tended to be more liberal in their politics. Even those who were avowed conservatives, as they were more likely to be the ones who would hold certain conservative views, but were still easily swayed by the liberal in sheep's clothing types like President Bush and John McCain.

While those that were "thinking" or "driving" which is to say more logical, were more apt to be not just conservative, but were more principled in their ideology. Thus, these were the anti-McCain types who held their nose and voted for Bush.

Thus I would tend to think that cognitive ability plays a smaller part than how a person gathers information.
3.23.2006 9:54pm
Ray (mail):
As for the skewed identifications of what it means to be conservative/libertarian versus liberal, I think a fairly accurate picture can be taken from economics.

It is the libertarian who espouses a free-market; letting people be free to make their choices, both good and bad. They love their individualism and recognize that taking away self-responsibility is taking away liberty. This ideological view point demands of the viewer self-confidence.

While it is the liberal who naturally gravitates to a more collective system that, while less free, is seemingly more safe for the whole. Individualism is bad; seen as selfishness, callousness. The naturally self-confident are repulsed by such an ideology.
3.23.2006 10:10pm
Robert Schwartz (mail):
So what? Who cares? What else is new? It's junk social science. recent
George Will nailed it a while ago
:

"Professors have reasons for their beliefs. Other people, particularly conservatives, have social and psychological explanations for their beliefs."

As long as Democrats see the people who do not support them as mental defectives, they will be a minority party.
3.23.2006 11:17pm
Ray (mail):
It may seem an obscure, or odd thing to pop in my mind, but when I think on these fundamental differences of ideologies, I think of Rosseau. In short, his radical view held that man was, at least in infancy, tabula rasa, a blank slate. And if one could get to a child soon enough, then avarice, ambition, and strife could be avoided in man.

Which is to say what the poster above just stated; the Left essentially thinks people are defective, and that if they could just get the right strings of power, they could re-write that slate of human kind.
3.24.2006 12:03am
Ubertrout (mail) (www):
I'm not so quick to say the study is meaningless...I just don't think it proves quite what the author intended to prove. One, of course, it demonstrates that much of the social sciences are radical liberal policy cloaked in the garb of science. This is not necessarily the state of the social sciences, but seems to be what it is in practice.

More importantly, I'm rather willing to accept the conclusions of the authors, as applied to an ultra-liberal place like Berkeley. The individuals who were somewhat awkward and didn't fit in well to the group were less vulnerable to concerns about feeling normal regarding their political outlook, since becoming an outcast or an outsider was already their fate. I would also be willing to beleive that for an outsider to become a conservative in an ultraliberal place is something that could be done, at least in part, as an angry affirmation of one's social standing.

The fallacy of the authors, to my mind, is that they apparently thought this mechanism was a function of the ideologies themselves, rather than the setting. However, I'd be willing to bet that the same mechanism works in reverse to create liberals in ultra-conservative places.
3.24.2006 8:34am