Polling Update # 20 All Quiet, except for the outliers.

This week saw a slight dip for Clinton in the RCP average, but with time running out her odds of winning increased nonetheless.  At an earlier stage in the race, the time remaining mattered less, just as being up 7 points in a football game with a half to play is not that much different from being up 7 with three quarters to go.  However now that we have reached the last two weeks Hillary gains more from eating up one of the three remaining weeks than she loses from a slight narrowing of the margin. So let’s look at the numbers. Clinton now leads in the two-way race: 47.8 % to 42.3%, down from last week when she led 48.8% to 41.8%. In the four-way, the race is now 44.9% to 39.9%, down from 46% to 38.9% last week.  The primary driver behind this decline is the new Investor’s Business Daily tracking poll, which shows a tie in both contests and is new this week.  One poll showing a tie can move the average by about a point. For example, if you had 5 polls, each with a 5-point lead, the poll average margin would be 5%.  If you then added one that was tied, the average would drop to 4.2%. There is no reason to discount the IBD poll, but there is no reason to think it is special either.  Before this is over, we might go back and check to see if lopping off the best and worst poll for each side would yield a better result or if pure averaging is better. Either way the advent of this outlying poll is responsible for most of the average change, and that’s why we called the week quiet with the exception of outliers. The one other thing to watch is that there are still somewhere between 7% and 10% undecided, depending on whether we use the 2 or 4 way.  10% left is basically nowhere near enough to overcome a 6% deficit (Trump would need 80% of them).  But if the undecided simply don’t show or tip one way or another that could go a long way toward determining the size of the spread. For now the point is clear: Clinton leads and there is very little time left on the clock.