I’ve noted a curious trend in Gallup’s daily tracking poll: the margin between Barack Obama and John McCain seems to be correlated to the day of the week the poll is released.
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 understates 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.
Below is a chart showing how Barack Obama’s lead has fluctuated over the last three weeks of polling. The blue line is the margin between the two candidates. The red line indicates how many days of weekend polling are included in the sample. For example, a three-day tracking poll like Gallup uses would include three days of weekend polling in a sample where the last day of polling was a Sunday and which was released on a Monday. A poll where the last day of polling was on a Thursday would have had its three-day sample collected on Tuesday, Wednesday, and Thursday, and thus would have zero days of weekend polling.
When we overlay the two lines we see a high degree of correlation. Tracking poll results that didn’t include any weekend polling tend to show John McCain closing the gap, while those conducted over the weekends show a widening Obama lead. Mathematically, it works out to 42% of the variance in the daily poll results is related to how many days of weekend polling were included within the sample.
Granted, this is based on a three week sample, so it’s not a big enough data set to make any firm conclusions, but what I’m seeing in the results tells me that from October 8th to October 22nd, Obama’s lead was unchanged and that the only variance in the data was related to the day of the week that the polling was conducted. In other words, Barack Obama probably maintained a steady two-point lead using Gallup’s traditional turnout model during the period.
Beginning 22 October Barack Obama appeared to widen his lead since his midweek trough wasn’t quite as deep as it had been the previous two weeks. The timing indicates that the Colin Powell endorsement gave Obama a modest bump.
However, that trend seems to have reversed itself beginning this past weekend. The second chart shows what the average margin looked like by day of the week. In it you can see that Barack Obama’s lead climbs on the weekends and falls during the week.
Poll releases that included a heavy weekend sample had shown Obama’s lead jumping to an average lead of about 6 points. This Monday’s poll release of a sample taken Friday, Saturday, and Sunday peaked at only 5 points. And Tuesday’s poll release which included two days of weekend polling showed a margin of only 2 points, a steep decline from the two previous weeks’ Tuesday margin of 7 and 6 points respectively.
Whether it’s Joe the Plumber, or Joe the Gaffer, in each of the last three days John McCain appears to have cut into Barack Obama’s lead. It will be interesting to see if the weekly trend continues. If it does, John McCain can expect even better poll numbers over the next three days since each of those days’ poll releases will include a smaller weekend sample.
The Wednesday poll results are out. They includes polling conducted from Sunday October 26 through Tuesday October 28. Gallup’s traditional model now shows Obama +3. The past two Wednesday poll releases had him up 5 points and up 3 points respectively, so today’s results don’t necessarily show an improving McCain trend, but they do continue the weekly trend of a smaller Obama lead in poll results that include fewer weekend samples.
This report feels about right. Summary: Obama has peaked and is coming back down; the remaining undecideds are demographically likely to swing to McCain. Turnout is everything.,
Gallup’s traditional model now has Obama +5. Thursdays have the lowest average Obama lead. Either he is gaining support in the final week before the election or the weekday/weekend disparity I thought I had identified is gone.