Last week we asked readers how much sleep they lost staying up to watch the political coverage of the Democratic and Republican conventions, and how that compared to the sleep they lost a few weeks earlier watching the coverage of the Beijing Olympics.

Do people stay up later to watch politics, or sports? Or are sports fans less likely to watch politics, and vice versa?

First off, let’s take a look at how much sleep was lost overall.

The Olympics were clearly a much bigger diversion than politics, with an average of an hour lost sleep per night for those who watched during the first week. But among our readers, Obama’s speech also resulted in significantly more lost sleep than any of the other speeches.But what about the larger question: are the people who stay up late for sports also likely to stay up for speeches? Our results don’t support that notion:

There was no significant correlation (positive or negative) between sleep lost watching any of the sporting events and watching any of the candidates’ speeches, or between total sleep lost during sports and sleep lost during politics.

But as many commenters pointed out, the political coverage was shown live, while sports was delayed — only people watching on the U.S. East coast had to stay up especially late to watch the speeches. What happens when we only consider people in the Eastern time zone (still a substantial number of respondents, at 139)?

As you might expect, people on the East coast lost more sleep watching the speeches than people in other places — significantly more in the case of Obama. But there was still no pattern connecting sleep-loss watching sports versus politics.

There were, however, a couple interesting correlations in this group. Women were significantly more likely to lose sleep watching the Olympics than men (r=.22, n=95). Perhaps even more intriguingly, there was a non-significant positive correlation between male gender and sleep lost watching Palin’s speech (r=.17, n=59, p=.19). It’s just a weak trend, but it would be interesting to see if this trend still applied with a larger sample size.