Whiny conservative kids, whiny liberal adults
A lot has been written on the blogosphere today how the Berkeley study, written about in the Toronto Star, claims that “whiny kids tended to grow up conservative, and turned into rigid young adults who hewed closely to traditional gender roles and were uncomfortable with ambiguity.” On the other hand, the confident kids turned out liberal and were still hanging loose, turning into bright, non-conforming adults with wide interests. The girls were still outgoing, but the young men tended to turn a little introspective.”
This is the biggest bit of scientific poo-poo I’ve seen in a long time.
- What was their methodology for defining whiny children?
- What was their definition of a rigid adult? Someone who votes Repubican, someone who is pro-life, somene who doesn’t cheat on their wife?
- Are boundaries necessarily a bad thing? I have boundaries. I don’t drink and drive. I don’t park in handicapped parking spots. I don’t steal things from other people. Rigid? Perhaps. Maybe what the study defines as rigidity is actually self-control and maturity.
Further down in the story, the paper mentions this:
Part of the answer is that personality is not the only factor that determines political leanings. For instance, there was a .27 correlation between being self-reliant in nursery school and being a liberal as an adult. Another way of saying it is that self-reliance predicts statistically about 7 per cent of the variance between kids who became liberal and those who became conservative. (If every self-reliant kid became a liberal and none became conservatives, it would predict 100 per cent of the variance).
A 7% correlation is garbage and just shows how bad social science is. What other correlations were there that might have been more predictive? Growing up in a single-parent home? Aggression? High school graduation rate?
Here’s how my wife can tell the difference between a conservative man and a liberal man.
TrackBack
TrackBack URL for this entry:
http://www.mcculloughsite.net/cgi-bin/mt-tb.cgi/496
Comments
Since nobody posts here, I gather you’re lonely.
I just want to correct an item in your post, as well as make a point:
1.The article mentions a 27% correlation, not a 7% correlation. I have no idea if you have any training in statistics or not, although your comments leave the clear impression that you want people to think you do, but the first thing in statistics is just to get the numbers right.
2.I haven’t seen the study, and, obviously, neither have you. Of course, I haven’t commented on it. It seems to me though, if you want to make a judgement on the study, you should actually see it. Judging a study by reading secondary sources about it is unsound methadology.
Posted by: Adam T | March 22, 2006 03:08 AM
Since nobody posts here, I gather you’re lonely.
I just want to correct an item in your post, as well as make a point:
1.The article mentions a 27% correlation, not a 7% correlation. I have no idea if you have any training in statistics or not, although your comments leave the clear impression that you want people to think you do, but the first thing in statistics is just to get the numbers right.
2.I haven’t seen the study, and, obviously, neither have you. Of course, I haven’t commented on it. It seems to me though, if you want to make a judgement on the study, you should actually see it. Judging a study by reading secondary sources about it is unsound methadology.
Posted by: Adam T | March 22, 2006 03:08 AM
No, I have not read the original article, but I would love to. I can, however, go by the comments in the article by Jeff Greenburg, a social psychologist at the University of Arizona. The article states that there is a 0.27 correlation coefficient between being self-reliant in a nursery school and being a liberal as an adult.
Yes. my knowledge of statistics isn’t as strong as it was 20 years ago, but, as you may know, the strength of the relationship between X and Y is often expressed by squaring the correlation coefficient and multiplying by 100. The resulting statistic is known as variance explained (or R2). Example: a correlation of 0.5 means 0.52x100 = 25% of the variance in Y is “explained” or predicted by the X variable.
Since the correlation coefficient is 0.27 the variance explained is 0.27×0.27×100, which happens to be 7.3%. Thus, the strength of the relationship of being a whiny kid and growing up to be a conservative is 7.3%, which I rounded off to 7%. Only 7% of the adult liberalness is predicted by a child’s whininess.
To me, that’s junk science.
¿Comprende?
Posted by: Michael McCullough | March 22, 2006 03:54 PM