19:00:24 .@NHLaVa Some of the Google results are really out there. Many w/crazy small samples. Some massive outliers. Makes me wish I weighted. :-) in reply to NHLaVa
19:18:57 .@NHLaVa Undecided & Other presumably, but yup. As I said, a lot of these Google ones are odd. Note their Maine sample was only 88 people. in reply to NHLaVa
19:47:00 .@NHLaVa Yup! Election Graphs tries to include everything and stay simple (so no weights), but Google makes me want to question that! :-) in reply to NHLaVa
00:46:44 If anybody has unrounded 8/19 & 8/26 Ipsos numbers, I’d love to have them. Ipsos API only has curr week, & I didn’t know about it in time. in reply to ElecCollPolls
00:48:35 Anyway, starting to add the 9/8 Ipsos numbers momentarily. Unrounded. :-) 9/2 numbers will revise to unrounded versions as I do this. in reply to ElecCollPolls
00:50:34 To be clear, I actually mean the 9/9 Ipsos numbers, including data through 9/8. I was inconsistent in my usage the last few tweets. :-) in reply to ElecCollPolls
00:55:54 Note that raw Ipsos data reports to 3 places, & that will be tweets here & used to get margin, but I also round to nearest tenth in display.
00:57:57 Given margins of error, even tenths are not really significant, let alone 1000ths. But useful 2 keep decimals when subtracting 2 get margin. in reply to ElecCollPolls
17:57:11 Full 3P results logged: Google w/3P in VT from 2016-08-10 to 2016-08-16 – Clinton 60% to Trump 9% to Johnson 11%
17:58:44 I believe that is the 1st state poll anywhere I have seen showing Johnson in 2nd. Not enough to make me activate “3P has chance at EV” plan. in reply to ElecCollPolls