Political Polling Sept 19, 2016

[StatsCougar]

In the space of a week, Donald Trump as gone from no chance of winning to little chance of winning. The results of this week's simulations has Mr. Trump winning 4.5% of the time, with .5% chance of a tie. (Still no states won by Gary Johnson.)

This shift puts HRC average electoral college count to 301, a drop of 34 votes. The shift happens as Texas moved more into the firm red category (40% to 35%) and several other states moving to be more neutral (Ohio [42% vs 40%] and Florida [42% vs 42.1%]). Even with these shifts, the polls are not strong enough to overcome the built in advantage that the Democrats have with the coastal states.

The question can be asked whether HRC health is part of the shift in the polls. (I have read results that say no, but it could be sub-conscious part of how voters think.) The recent events on American soil should also shake up the polls, likely against HRC.

With the first Presidential debate a week a way, polls will hit hard to look at a before and after picture of voters views. For Trump to tighten up this contest, he will need to sway voter in the battleground and light blue states to abandon HRC (whether to vote from him or vote third party.)

Notes: Graphics will follow, once the issue with uploading pictures has been resolved. There was a small change in the methodology as well, in that polls that look at registered voters instead of likely voters are now weighed less heavily.

[scott715]

Things are starting to get interesting.

[SpiffCoug]

In your methodology, how many polls still have pre-9/11 & the HRC physical collapse data. It seems like that even has really changed things.

[hawkwing]

Attachments should be working now. If my work around is working consistently.

[StatsCougar]

In your methodology, how many polls still have pre-9/11 & the HRC physical collapse data. It seems like that even has really changed things.

Most of the polls are pre-collapse, as I'm trying to keep a month's worth of polling. Part of the reason for this is some states aren't polled all that often (Alaska for example), so . But to help compensate for that, I am using a weighting scheme. The older a poll is, the less weight it will carry.