Well speaking of diet, in the vast amount of stuff Mrs. Rainy and I read before the birth of our first, there was at least some speculation about availability of fresh fruits and vegetables in the mother’s diet (i.e. summertime) at an important time in the child’s brain development. So someone in the medical community is looking at it from this angle.
Unfortunately, I can’t supply a cite, but if memory serves, summer coinciding with the last trimester was thought to be best.
If you read through back issues of The Skeptical Inquirer, you will learn that the so-called “correlation” in the so-called “Mars Effect” has been compellingly reduced to the level of noise.
And just FYI, I’ve already agreed with Marley, and he/she with me.
The difference in cutoff age has already been discussed. It is not the month/day that influences that, it’s school policy. I don’t know why you’re going on about it.
Please cite the evidence that the birth month and day are an inicator of success and test scores, as the OP asked.
The idea that the month/day of a person’s birth effects their life in and of itself is pure astrology.
I don’t know what you’re going on about either. The OP asked whether, empirically, there is a correlation between birthdate and “measures of success”. You seem to acknowledge above that the answer could be “yes” without astrology being true, so how can you accuse the OP of asking whether astrology is true or suggesting that it is? Furthermore, the mechanism suggested by the OP is clearly non-astrological.
10 percent difference in risk between high and low is very low and most likely just a random fluke.
I think that a phenomenon called the birth month fallacy is important for this discussion. This site has an example of what it is:
If you follow the link the site has more examples of how the fallacy works.
I do think that there are certain things that are affected by when you are born. Sports may be one such thing. Since a lot of competetive sports are based on people of the same age competing against eachother, a person born at the beginning of a year will more likely be bigger and more developed than someone born in December. This will of course become less important the older you get, but it could probably have an effect in sports that get competetive at an early age.
By definition, astrology is the notion that the position of the stars and planets and heavenly bodies have an effect on earthly events in a person’s life. One’s birthday is only used to calculate the approximate positions of those heavenly bodies at the person’s precise moment of birth; the date of birth itself is not significant except as a reference point.
If one says that people born in September and October are better athletes because of school policies, or because of availability of summer produce then it is not astrology; if one says those same people are better athletes because of the moon and planets then it is.
Well, first of all linking schizophrenia, which probably have lots of different factors that contribute to the risk, with just one the time of year born is hard. You have to keep all other factors constant, something we do not know if they have done correctly. If you do not know what the other factors are that influence the development of a disease. How do you know what factors to hold constant?
Secondly in a retrospective epidemiolgical study a relative risk of 1.1 is generally not considered significant. A 10 percent higher chance of a rather rare event is still a rather rare event.
I also refer you again to the Birth Month Fallacy:
So if Vitamin D deficiency and the lack of vitamins from fruit and vegetables in winter pregnancies have been considered as possible culprits, is it reasonable to suggest that Seasonal Affective Disorder in wintertime expectant mothers could produce stress hormones which might affect the unborn child? That’s always been my favourite explanation - could someone blow it out of the water please?
The null hypothesis would be that the many factors that contribute to risk are independent of birthday. If independence of rates from birthday is rejected by the data, this suggests that some risk factors are birthday-dependent. This is not some kind of artifact, it is what the study set out to find. Nothing in the above doesn’t justifies the “random fluke” claim.
You can say that 10% is uninteresting (I don’t see what the rarity of schizophrenia has to do with it), but that has nothing to do with statistical significance and cannot justify the “random fluke” assertion. Effect size and statistical significance are different things. You may think that a 0.1% excess of heads over tails is not interesting, but it would be a mistake to declare it a “random fluke” without knowing how many coin flips I had performed.
I understand that it is bogus to look at the incidences, pick the highest month, and declare the difference to be real. This just points to a more general fact that every competent researcher (and, one hopes, most referees) knows: that it is necessary, when analyzing data, to perform appropriate statistical tests. You seem to be asserting that with an effect size of “only” 10% it would be impossible for such tests to show a statistically significant effect. This is clearly incorrect: with sufficient sample size, it can be significant. The assertion that a 10% difference is probably a “random fluke”, which I take to mean the result of sampling variance due to finite sample size, is unfounded.