andrewducker: (Default)
andrewducker ([personal profile] andrewducker) wrote2011-11-25 11:00 am

Interesting Links for 25-11-2011

[identity profile] anton-p-nym.livejournal.com 2011-11-25 03:29 pm (UTC)(link)
Apparently there's a saying going around Microsoft now; bowdlerized, it's, "the average person has one breast and one testicle." It's a caution against relying on statistical models without examining the context.

I strongly suspect that the reason second-child home-births have the same complication rate as hospital births is that those who were most prone to complications (as demonstrated the first time) went to the hospital. This skews the result heavily... indeed, I'm surprised that home-birth for second children doesn't show a greater safety rate because of that bias.

Of course I haven't seen the study design or how they controlled for this so perhaps I'm being unfair.

-- Steve's not a stats genius in any case.

[identity profile] del-c.livejournal.com 2011-11-28 01:16 pm (UTC)(link)
Because complications are rare, pulling complications out of the home birth pool is like pulling tokens out of a bag with a lot of tokens. So removing the mothers who had complications the first time shouldn't affect the home birth statistics positively enough to detect. Which is just what we observe: home births rise to the safety level of hospital births on and after second birth, but do not exceed it, despite the theoretical skew from first-birth complications electing hospital the second time around.