Ipsos/Reuters Polls (through DNC week): Clinton +6 (tied 4-way)

[Erich Maria Remarque]

Trump gains two points when Johnson and Stein are added lol.

Yeah, that doesn't make a whole lot of sense. Allow people to vote for Johnson and Stein, and somehow MORE people vote for Trump? How does that work, exactly?

This thing remains a mess. The old poll had 4-way with a Clinton lead and 2-way with a Trump lead. The new poll has the opposite.

It's almost like they are polling two separate samples.

Hopefully they get their methodology right.

They do actually use different samples.

Head-to-Head:
(Asked of registered voters, n=1,290)

Including Johnson & Stein:
(Asked of registered voters, n=1,788)

But why? It's weird :/

And they seem to change theirs methodology for Head-to-Head only. What the hell are they doing?!

[Erich Maria Remarque]

According to Nate Silver

https://twitter.com/NateSilver538/status/759104860147122183

You must be logged in to read this quote.

What the hell?

[Erich Maria Remarque]

OK, now Reuters says Clinton is actually up by 6 (2-way, LV)

http://www.reuters.com/article/us-usa-election-poll-idUSKCN1092M5

They really need to get their s--t together.

Well that solves the divergence between 2-way and 4-way.. per the link... Two separate polls.

Why would they have two DIFFERENT samples?

You must be logged in to read this quote.

They probably think, that othervice it might affect the outcome?

If one first chooses X in head-to-head question, one might choose X in 4-way with higher probability than, if there was 4-way question only. Something like this :/