4 ms·

Until another replicated study comes along to dispute this, I'll take the word of the author and the editors at JAMA over your 'feel' that this isn't statistica

Until another replicated study comes along to dispute this, I'll take the word of the author and the editors at JAMA over your 'feel' that this isn't statistically significant. "There was a significant interaction (P = .02) between child sex and MUFSG (6.89; 95% CI, 0.96-12.82) indicating a differential association between boys and girls. A 1-mg/L increase in MUFSG was associated with a 4.49-point lower IQ score (95% CI, −8.38 to −0.60) in boys, but there was no statistically significant association with IQ scores in girls (B = 2.40; 95% CI, −2.53 to 7.33)." 1. https://jamanetwork.com/journals/jamapediatrics/fullarticle/2748634 https://jamanetwork.com/journals/jamapediatrics/fullarticle/...

dtech 7y ago
I browsed the article and have a basic understanding of statistics, no need to get patronizing. The article does not state the variance of this IQ test and does not state how many subgroups they analyzed and whether they corrected for that or if they a-priory were planning on analyzing gender. Picking subgroups until you've found one that is statistically significant is a well known technique in P-hacking. Other problems: - The IQ score for girls is 5 points higher on average than for boys in the study group - Fluoridation has the opposite effect for girls, the average IQ was 1.5 points higher. It is really strange for a chemical to have a completely different effect on young girls and boys. Statistical variance is much more likely - The average IQ score in the sample was 107, well above the average 100 - Parent IQ was not a variable, only the very roughly correlated high school diploma or not and household income. "Extraordinary claim require extraordinary evidence", and this study seems to at best be an indication that further research is needed rather than iron-clad proof of this effect.
- word-reader 7y ago
  It's really not that strange for a chemical to have different effects on girls and boys, anything that affects sex hormones at all or sex-specific development is going to have a different effect.
  - shawnz 7y ago
    Even in 4 year olds, though? And such that it produces the exact opposite effect compared to the control?

dtech 7y ago

I browsed the article and have a basic understanding of statistics, no need to get patronizing. The article does not state the variance of this IQ test and does not state how many subgroups they analyzed and whether they corrected for that or if they a-priory were planning on analyzing gender. Picking subgroups until you've found one that is statistically significant is a well known technique in P-hacking. Other problems: - The IQ score for girls is 5 points higher on average than for boys in the study group - Fluoridation has the opposite effect for girls, the average IQ was 1.5 points higher. It is really strange for a chemical to have a completely different effect on young girls and boys. Statistical variance is much more likely - The average IQ score in the sample was 107, well above the average 100 - Parent IQ was not a variable, only the very roughly correlated high school diploma or not and household income. "Extraordinary claim require extraordinary evidence", and this study seems to at best be an indication that further research is needed rather than iron-clad proof of this effect.

word-reader 7y ago
It's really not that strange for a chemical to have different effects on girls and boys, anything that affects sex hormones at all or sex-specific development is going to have a different effect.
- shawnz 7y ago
  Even in 4 year olds, though? And such that it produces the exact opposite effect compared to the control?