2014 :: 2013
2012 :: 2011 :: 2010 :: 2009
2008 :: 2007 :: 2006 :: 2005
2004 :: 2003 :: 2002 :: 2001



Know what I find quite annoying? ("No, tell us. Or don't. Just quit asking rhetorical questions. They're pointless and stupid. Get on with it, then.") I keep seeing people cite studies whose results point to the idea that people who get more than eight hours of sleep a night are more likely to die than people who don't, using such study results to try and tell me I shouldn't sleep so much or that my sleeping so much means that I obviously have some health problem, that I'll consequently die from. The problems with using such studies to try and prove anything are manifold:

1. Such a study would obviously be correlational. That means that there's no direct causal link between sleeping more and dying sooner.

2. Besides the study just being correlational, there are many things that could serve as confounds here. Here's a somewhat involved example: Say someone's doing a lot of manual labor, perhaps as a construction worker. That just might make it necessary for him to sleep more at night in order to recuperate fully from the day's work. Perhaps he drinks a bit before bed or takes a sleeping pill/antihistamine, too, to ease the pain of a day's hard work and help him get to sleep. These things might make it so that guy sleeps more than eight hours—and then the guy might die the next day when he's got a hangover and a crane drops a ton of sand on his head. Voila—excess sleep and death are now both characteristic of this subject's existence.

Should this guy happen to be in the study, say to pick up a bit of extra cash, he's just become another statistic in the correlation between death and excess sleep—yet his sleeping patterns weren't what caused his death. His sleeping patterns were the result of several factors, namely A. being tired from manual labor, B. taking a sleeping pill/antihistamine, which could, the next day, have impaired his judgment, making him more likely to be involved in a fatal accident, and C. drinking alcohol, which, depending on how much he drinks, might have decreased his life expectancy anyway, and which certainly may have decreased his judgment capabilities. You see all the potential problems here?

Then you have other possible confounds, like depression, alcoholism, substance abuse, sleeping disorders, etc., any and all of which might constellate in a given subject to create a condition where s/he'd simultaneously be at higher risk for death and have a greater need for sleep, or at least have drastic changes in his/her sleeping cycle, yet the risk for death and sleep cycle changes wouldn't necessarily have any causal relation.

3. Some people simply need to sleep more, or sleep at different hours, than the mythical "average" subject. (Tangent: To take a point from Peter Ho Davies' story about "the great detective," the most apt murder victim to use for the "perfect crime," an untraceable crime, would be a guy who is average in every way—and that would surely drive up the chances of dying in general, should a criminal exist who would want to commit an untraceable, pointless crime.) The very measurement of an "average" amount of sleep implies that there are people who fall above and below this value.

Now, of course, a given study may have taken these confounds into consideration when recruiting volunteers, or may have at least written them into the results if their volunteer selection wasn't that discerning—but even if that's the case, quoting such studies out of hand to make a thin point about my sleeping habits (or anyone else's habits, for that matter) is rather pointless.

1:48 am, March 20, 2004 :: the jablog years

You should follow me on Twitter.