The data came from the Genetic Disease Screening Program at the California Department of Public Health. These are TSH (thyroid stimulating hormone) levels of California newborns for the period 2009-2012. Mangano, Sherman, and Busby used this same dataset for their paper on congenital hypothyroidism which was incomprehensibly botched. Dr. Yarnold and I made comments on this paper here. (http://optimalprediction.com/wp/airborne-radiation-and-congenital-hypothyroidism-of-california-newborns/)
While Mangano et al. created arbitrary classifications of hypothyroidism, which are not used in practice, these results reflect TSH levels before and after Fuku. TSH is implicated in many diseases, not just hypothyroidism. It affects the entire hypothalamus-pituitary-thyroid axis. For one thing, increased TSH levels cause the newborn to absorb more iodine-131 than average TSH newborns would.
This is also important because it involves the effect of all species of radioisotopes, I-131, Cs-137, Pu-239, etc. etc. We don’t know if only I-131 affects newborns, in fact Dr. Bandazhevsky has shown that Cs-137 affects thyroids too.
The data have been split up into three seasons:
1. January 1 – March 16
2. March 17 – June 30
3. July 1 – December 31
It involves virtually the entire population of California newborns for the years 2009-2012.
AFTERFUK is an indicator variable, which equals 1 if the birth was after Fukushima, and 0 otherwise. “IF 4.5 < TSH THEN AFTERFUK = 1" means than if TSH is greater than 4.5 (actually 4.99 in the original data), then the newborn is classified as being born after Fukushima. ODA uses maximum-accuracy classification - no model of this form achieves higher classification accuracy. No distributional assumptions are involved.
All three seasons, and the combined data also, show increased TSH levels after Fukushima, and all are statistically significant by two-tailed UniODA at the p<.01 level.
The ESS levels vary from 3.90% to 8.02%. The effect is weak, but I am not a thyroid doctor. Since over two million newborns are involved, even a small effect has consequences for many newborns. I would think that something that would affect even 100 newborns is a big huge deal.