View Single Post
Old 09-17-2006, 12:39 PM   #35
meretricious dilettante
trisherina's Avatar
Join Date: Jan 2003
Posts: 11,068
I'm not sure what point you're trying to make.

Mine is that (a pro-breastfeeding site) is attempting to establish breastfeeding as the causal explanation for reduced SIDS incidence in breastfed children. That's poor science of the sort that pisses me off, mostly because it is i) so exceedingly common and ii) so blithely accepted without analysis by the average reader. Of the three elements of causation (association, temporal order, and elimination of alternatives or "no spuriousness"), only association (the two phenomena occur together or appear to act together -- in this case breastfeeding and reduced SIDS incidence) applies to that "careful study of statistics."

Temporal order is absent (to establish this, you'd have to at the very least contrast a period of time in which breastfeeding was at an all-time low and SIDS at an all-time high with a period where there was more breastfeeding and less SIDS, OR more ideally you could take non-breastfed children in a family who were collapsing of SIDS and make their mother breastfeed subsequent sibs and show that their little lives were saved).

Also absent is the elimination of plausible alternatives, which I prefer termed "no spuriousness" -- plausible alternatives in this case include how the breastfed vs. the nonbreastfed infants were placed to rest, education and income and smoking factors in the infants' mothers (because these factors show positive and negative association with infant care/vigilance variables), bundling and swaddling, and family medical/psychiatric histories, just to name a few plausible alternatives I can think of off the top of my head.

As for a statistically significant association all on its own as a strong argument for causality, well, let me tell you about the way my dog makes the sun come up every morning by stretching and snuffling on her dog bed.

Source: my brain and Neumann's Social Research Methods, Qualitative and Quantitative Approaches, 2006 (6th ed.).
trisherina is offline