Archive for analysis of covariance

More babies on Valentine’s Day and less babies on Halloween?!

Posted in Kids, Statistics, Travel with tags , , , , , on February 14, 2012 by xi'an

(Source: Levy et al., 2011)

In the weekend edition of Le Monde, more precisely in the Sciences section, I read a report on a 2011 study made by Levy et al. who observed that the birthrate drops at Halloween and surges at Valentine’s Day… The above graph illustrates the fact for Halloween, with a significant [meaning?!] 5.3% decrease for spontaneous births. The increase for Valentine’s Day is 3.6% (still for spontaneous births). Even though those data are the result of a survey of all births in the United States over 11 years, thus unlikely to exhibit sampling biases,  I am fairly bemused both by the phenomenon and by the interpretation made in the study, namely that “pregnant women may be able to control the timing of spontaneous births” (while I find less astounding that “scheduled births are also influenced by the cultural representations of the two holidays“, even though there may be an administrative bias as well). Being unfamiliar with the U.S. procedure for delivery of birth certificates (and how much both Valentine’s Day and Halloween are of a public holiday), I wonder if this may be a reporting rather than biological bias…

Reading now into the paper (thanX, A.!), I see that “these holidays have the advantage of widespread participation, but without ordinarily resulting in the absence of physicians from work, as on certain federal holidays“, so my first idea that the Halloween gap [more pronounced than the Valentine surge] could be due to reduced medical or administrative staff does not seem so likely. The authors mention using an analysis of covariance model to build their significance test, adjusting for weekday and year effect, even though neither the model [what are the covariates?] nor the statistical analysis is provided in the paper. Looking at the equivalent of the above graph for Valentine shows more variability along the window of 15 days used by the authors. It would be fairly interesting to check throughout the years if other variations of that magnitude occur (and if they are always related to culturally significant days), before accepting the conclusion that “pregnant women can expedite or delay spontaneous births, within a limited time frame, in response to cultural representations“…

Follow

Get every new post delivered to your Inbox.

Join 720 other followers