In addition to the usual caveats about any survey results (this was self-reported; the site didn't say how participants were selected; it's not clear if Match controlled for confounding factors like age), there's the biggie: correlation is not causation.
Dr. Ford, whose earlier work had found no correlation between unusual names and lower academic achievement once you controlled for confounding variables, told me he was impressed with Dr. Figlio's research: I am inclined to conclude that there may in fact be some (very) specific circumstances in which the "social stimulus value" of a name is more than trivial.
A significant association of abruptio placentae (odds ratio (OR) 61.0, 95% confidence interval (CI 3.4 1084), delivery at <31 weeks of gestation (OR 19.0, CI 2.8 128.8), and low Apgar score (<7) at 5 min (OR 70.8, CI 16.5 304.9) with increased risk of poor outcome was found in the logistic regression model that controlled for confounding effects.
We controlled for confounding effects of passive or evocative gene-environment correlations by adjusting models for parental histories of psychopathology, and by demonstrating that genotypes did not influence the exposure to abuse.
Observational studies varied in their attempts to control for confounding; some reported unadjusted estimates whereas others controlled for confounding.
However, many of these studies have suboptimal methodological qualities and have not adequately controlled for confounding clinical factors [ 8].
The covariates controlled for confounding effect included age, sex, dyslipidemia, diabetes, hypertension, NIHSS on admission, WBC count, leukocyte ROCK activity, hsCRP, IL-6, and TNF α.
Further, our sensitivity analysis, where we adjusted for comorbidities in a time-varying manner, did not change the results substantially, suggesting the design itself adequately controlled for confounding.
We have controlled for confounding from alcohol consumption (g/month), level of education, and smoking habits (non-smokers, former smokers, current smokers).
Each of these nine indicators was considered one-at-a-time in the base model that controlled for confounding (described earlier).
Based on the strength of association of the crude analysis (adjusted for level 1 variables), we chose EFI4 for further analysis and controlled for confounding variables from within the same category.
