On Friday we added 13 new statewide database, and these new surveys mostly confirm what we already knew. Most represent updates since September and show the same progress for the Democrats we have seen or the last few weeks. Three new surveys in Florida suggest that Obama's lead there may be narrowing. Another new survey in North Dakota indicates that the surprisingly close contest in North Dakota is not a mirage and the national polls suggest a slight uptick in support for McCain nationwide.
We had three new surveys yesterday in all-important Florida from the Times Union and Sun Sentinel/Research 2000 (Obama +4), SurveyUSA (McCain +2) and Democratic polling firm Hamilton Campaigns (Obama +4). These new surveys narrowed Obama's Florida margin on our trend estimate to 4.2 points (49.1% to 44.9%), which represents a 2.6 point drop on the margin from the 6.8 point lead we saw earlier in the week.
Research 2000 released a new North Dakota survey yesterday sponsored by the left of center web-site Daily Kos, showing the presidential contest deadlocked, 45% to 45%. The three new surveys this week in North Dakota are the first since mid-September and one of only six released there since the conventions. The other two give Obama a very narrow lead. When we have only a handful of polls, as in North Dakota, our system defaults to a linear trend estimate (relatively straight lines rather than the curved loess regression trend. Since the new polls indicate a sharp change in the trend lines, each new poll confirms and extends that trend. In most cases, a linear trend line works best in these situations, but in this case it produces an odd quirk: Today's 3.6 point margin for Obama (44.9% to 41.3%) ends up being greater than Obama's 1.6% lead on the average of last three polls (44.6% to 43.0%).
Based on the three new polls, North Dakota certainly looks like a true toss-up, at least for the moment, but in this case our trend estimate likely overstates Obama's current standing there.
As of this writing, Obama's lead on the national trend is just under seven percentage points (although today's tracking surveys, which may be added by the time you read this, will likely change that margin as they usually do). However, we now have enough polls to showing a slight narrowing in the national margin.
Any change in the trend line, as of this writing, does not appear to reflect reactions to Wednesday's debate. Yesterday's new national tracking polls did include interviews conducted on Thursday night, but these represent just a third of the interviews for five of the trackers and a quarter for two more. Unlike my standard table above, the table below shows how yesterday's releases compare to those from the day before. The changes show no consistent pattern. Two showed slight movement to Obama, two showed slight movement to McCain, one showed no change and one (Gallup) showed either no change or a one point improvement for Obama, depending on which population you look at.
If the debates produced a change in vote preference, it was too small to be detected by the first post-debate release of the new tracking polls. We will have a much better sense of any post-debate trends tomorrow, when five of the trackers release their first full samples collected after the debate.