November 4, 2016

State of the race - Nov 4

Yesterday I updated my view of the state of the race for the presidency and it looked like a tie as far as the electoral college.  Today I went back to the RCP average of polls and I recalibrated my weighted average of polls to see if I'm in the same place as my electoral college view of how tight this race is.   It seems to be the case that it is very tight. As you can see below:

Click to enlarge.
Previously I had a two week view which I've changed to a weekly view this time around.  This view takes a weighted average of the polls, as they are not all equal.  The approach is as follows - if a poll has 2000 respondents and another poll has 1000 respondents I've pro-rated the voter percentages Trump and Clinton for each poll as a percentage of that poll but the overall voter total would be 3000 so a poll that has Clinton ahead by 4 points but over 1000 people polled would net her less voters than the poll that had her behind 3 points over 2000 voters. 

Additionally I've excluded polls that do not indicate their margin of error as those polls are not transparent and therefore subject to manipulation. I've excluded any polls with less than 800 respondents and polls with margins of error greater than 4.5%.  I've also excluded all polls with a margin of 6 or more (the last measure has been more fluid than the others as the gap has been larger than that at times).

For November-ending polling only 1 poll has been excluded as a result of those criteria - a Reuters/Ipsos poll that has Hillary Clinton +6. It seems like an outlier, though I could be wrong. That poll has consistently showed Clinton at the high end of the spectrum and seems out of alignment with most polls that range from Trump +4 to Clinton +5 (most at Clinton +1 to 3).

One other note - as I've capped the outlier to reflect the current situation, previous weeks have had polls excluded that previously did not have those same polls excluded - my Excel spreadsheet is not that dynamic (yet).  Therefore weeks prior to November were further apart than they appear above.

With that exclusion my weighted average of polls indicates an absolute tie in total voters allocated across 5 November polls. It's a true toss up right now in my estimation, excluding any possible momentum.

Bear in mind that this does not take into account polling methodologies which I do not have the bandwidth to review, and some polls that I have included demand further review.  Nor does this take into account momentum.  The polls today might not reflect next Tuesday.

However, even the polling as it exists, shows a tight, tight race.




Related Posts Plugin for WordPress, Blogger...

Share This