Presidential Poll Analysis: Weekly Trends

I was surprised today to see two different sources independently reach a conclusion I thought was common knowledge. I thought it was a well known fact that Democratic candidates poll better during the weekend and Republican candidates poll better during the work week. Apparently this fact wasn't as well known as I thought.
Bob Krum performs an analysis of Gallup results looking for this trend. (Hat tip: Instapundit.)

Pollsters have long known that the demographic makeup of a random telephone sample changes from night to night. Those with families are less likely be available at home during weekend polling. Since that demographic tends to vote more Republican, weekend polling often understate support for the GOP. Polling companies use weighting to attempt to balance unbalanced weekend samples. But the Gallup traditional daily tracking poll doesn’t seem to have balanced the disparity enough.

So is Gallup not accounting for this known effect on purpose? Hard to tell of course but it is suspicious. Steve Schippert reaches similar conclusions at Wizbang.

Notice the trend that appears in the Gallup image: McCain gains consistently once the vast majority of those polled actually have to go to work - and it tightens most midweek. The dates marked in the image are Sundays, consistently right in the middle of Obama's biggest trended edges.

...

Take from it what you will, but I would suggest that when people are actually at work, doing what they do, earning what they earn, these working people trend toward McCain most. Would also suggest the saavy marketing/media practice of Friday releases to own a three day cycle has been successfully captured by the Obama campaign. Friday's news stews during his biggest margins.

The commenters over at Hedgehog report have discussed this effect for years. They know that a 3-day tracking poll released on Monday is going to favor the Democratic candidate--it always has. Still I'm glad to see this phenomenon is being understood and explain to more people.