12 October 2016

Battleground/National Report, 11 October, 2016 - T-minus 28 days (Clinton 344 / Trump 194)

I did this series in 2008 and 2012 as well, but started earlier (at 42 days out) then.

Tuesday, October 11th, 2016, was exactly 28 days before the General Election on November 8, 2016. One month to go!

Complete polling results for all states for 2016 to-date are here as an EXCEL table for 10/11/2016 in GOOGLE DOCS. All polls are hyperlinked.


2016 is different from 2012 and 2008 in that we are often getting double statistics for the race: 2-way statistics and 3-or-4-way statistics, which means there are often two sets of margins to look at. Realizing that the statistical probability that the truth usually lies somewhat between these two sets of margins, I took the mean of both.

2016 is also different in that at least 3 pollsters have been putting out a 50-state series of polls: UPI, Ipsos/Reuters and Morning Consult (R). This means that states that were hardly polled at all in 2012 have been polled this time around, only the survey group sizes for the various states leaves something to be desired and they tend to be internet-only polls, so sometimes the results seem a little crazy. That being said, I stick with my philosophy that the mean value of ALL polls tends to cancel out both the high and the low (possible) outliers with time.

Let's take a look at how I did the averages quickly, easily and mostly per hand, for example, in the Buckeye State:




So, looking back the last 10-14 days we have (without repeater polls) 8 polls: 

IPSOS/Reuters, UPI/Cvoter (both part of their 50-state series), CBS/YouGov, Target Smart, PPP (D), Monmouth and Qunnipiac. 

Of those 8 polls, 5 poll values are for a 2-way race and 5 poll values are from the 3-way race: 3+3.8+1+2-3/5 = 1.36 (two-way) and 4+3+1+2+2/5=2.4. The mean of 1.36 and 2.4 = 1.88. So, that is the margin value that shows up on the BATTLEGROUND tab.

The LEGEND for the following EXCEL table is HERE. In fact, you need to read it in order to develop an "eye" for the tab:

A B C D E F G H I J K L M
State EV % of EC Polls 11.10.16 10.10.16 Shift 2012 2008 SWING (H-I) End polling 2012 Compare E – K Compare E – GE2012
MS 6 1,11% 3 of 3 11,40
11,40 11,50 13,17 -1,67 10,12 1,28 -0,10
SD 3 0,56% 2 of 2 10,55
10,55 18,02 8,41 9,61 12,00 -1,45 -7,47
MO 10 1,86% 2 of 2 9,00
9,00 9,38 0,13 9,25 10,12 -1,12 -0,38
IN 11 2,04% 3 of 3 8,15
8,15 10,20 -1,03 11,23 11,20 -3,05 -2,05
AK 3 0,56% 2 of 2 7,60
7,60 13,99 21,54 -7,55 22,00 -14,40 -6,39
SC 9 1,67% 2 of 2 5,70
5,70 10,47 8,98 1,49 6,00 -0,30 -4,77
GA 16 2,97% 2 of 2 4,40
4,40 7,81 5,20 2,61 8,25 -3,85 -3,41
AZ 11 2,04% 4 of 4 2,80
2,80 9,04 8,48 0,56 7,67 -4,87 -6,24
IA 6 1,12% 4 of 4 2,10
2,10 5,81 9,53 -3,72 2,14 4,24 7,91
NC 15 2,79% 7 of 7 1,88
1,88 2,40 0,33 2,73 2,40 4,28 4,28
OH 18 3,35% 8 of 8 1,88
1,88 2,97 4,58 -1,61 3,24 -1,36 -1,09
FL 29 5,39% 5 of 5 2,11
2,11 0,88 2,81 -1,93 0,61 2,72 2,99
NV 6 1,12% 4 of 4 2,48
2,48 6,68 12,49 -5,81 4,67 -2,19 -4,20
NH 4 0,74% 3 of 3 4,13
4,13 5,58 9,61 -4,03 2,50 1,63 -1,45
MN 10 1,86% 2 of 2 4,65
4,65 7,69 10,24 -2,55 6,25 -1,60 -3,04
CO 9 1,67% 5 of 5 5,04
5,04 5,37 8,95 -3,58 1,87 3,17 -0,33
WI 10 1,86% 5 of 5 5,69
5,69 6,94 13,90 -6,96 4,63 1,06 -1,25
USA 538 100,00% 12 of 12 5,74
5,74 3,86 7,26 -3,40 1,22 4,52 1,88
VA 13 2,42% 5 of 5 8,58
8,58 3,87 6,30 -2,43 0,77 7,81 4,71
PA 20 3,72% 8 of 8 9,05
9,05 5,38 10,31 -4,93 4,20 4,85 3,67
MI 16 2,97% 3 of 3 9,58
9,58 9,48 16,44 -6,96 4,79 4,79 0,10
OR 7 1,30% 4 of 4 10,78
10,78 12,09 16,35 -4,26 6,00 4,78 -1,31
NM 5 0,93% 3 of 3 11,53
11,53 10,15 15,13 25,28 8,00 3,53 1,38


Since 10/11 is the start date of this series, I left 10/10 blank, but you can see how it works. For the battleground report for 10/12, when it comes out, then the data from 10/11 wil slide over to the right and so on...

Assuming 5% as the boundary for battleground states, at current, 9 states are battlegrounds, meaning that the polling aggregate as I described above is within the 5% margin.  I think it's important first to see two key states are are NOT in the battlegrounds but were in them almost the entire time in 2008 and 2012: Colorado and Virginia. In fact, Hillary Clinton is posting double digit margins in both of those states. However, on the Democratic side, Minnesota is currently within the battlegrounds and on the Republican side, Arizona and Georgia are within the battlegrounds.

Please also notice that Hillary Clinton's national margin is also outside of the battleground boundary. In 2012, it was within the boundary the entire time. This is an important data-point to consider.

If there are two states that look very much like they are going to switch sides this year, pretty much irregardless of national margin, it looks like - at least for now - that North Carolina (15 EV) is moving blue and Iowa (6 EV) is moving red.

It should be noted that no state is at an aggregate of 1.0 or less, so technically, there are no real tossups at this time. There are close races, but the tendency in all cases is very clear to be seen.

You will also see that a number of states are on the battleground chart, states with margins far larger than 5%. Those are states that were once battlegrounds in the past and I deemed it good to start with them in the table, at least for now.

Were the election to be held today and that states were to fall as the polling aggregates indicate, then Hillary Clinton would easily win with 344 electors:



Here the national stats at current:

Nr. NATIONAL POLLING Date Sample MoE Clinton Trump Und/Oth Mar.
Clinton Trump Johnson Mar.

AVERAGE (one week): N/A N/A N/A 48,05 42,24 9,71 5,81
44,78 39,11 6,89 5,67
















Most recent




100 0



0
510 IPSOS / Reuters 11.10.16 1,605 LV +/-3.0 45 37 18 8





508 UPI / Cvoter 11.10.16 1,367 LV +/-3.0 50,23 44,09 5,68 6,14



0
509 NBC / WSJ 11.10.16 806 LV +/-3.45 50 40 10 10
46 37 8 9
502 You Gov / Economist 10.10.16 1,135 RV +/-3.8 48 43 9 5
44 38 5 6
507 Rasmussen 11.10.16 1,500 LV +/-2.5




44 39 7 5
500 NBC Survey Monkey 10.10.16 23,329 LV +/-1.0 51 44 5 7
46 41 8 5
506 USC Dornsife / LA Times 11.10.16 1,400 RV +/-3.0 43,3 45,3 11,4 -2



0
497 Morning Consult (R) 09.10.16 1,549 RV +/-3.0 45 41 14 4
42 38 8 4
492 FOX News 07.10.16 896 LV +/-3.0 48 44 8 4
44 42 6 2
491 Quinnipiac 07.10.16 1,064 LV +/-3.0 50 44 6 6
45 40 6 5
486 PRSI (PRRI) / The Atlantic 06.10.16 609 LV +/-4.0


0
47 41 3 6
482 Farleigh-Dickinson / SSRS 05.10.16 788 LV +/-4.4 50 40 10 10
45 36 11 9



So, instead of doing a blow-by blow of each battleground state, I will let the opening stats stand for themselves and then start to notice the changes with the next report.

However, at the EXCEL document is also a thing called the EVERYTHING TABLE, and it is most enlightening, because it compares the current polling averages to 2012 (actual results, 2012 end-polling averages) and in one way, it kind of proves that the more things change, the more they stay the same. For instance, of the 27 states (including DC) that look like they are in Hillary's column, 19 of the them show polling averages of +8 or more and 16 of them are in double-digits - those are SAFE states. Conversely, of the 24 states that look like they are in Trump's column, 20 of his states are posting +8 or more polling aggregates, 18 of them in double digits - those are also SAFE states. That makes for only about 10 states in the Union that are really under +8. And we see that excepting Virginia and Colorado, the usual suspects for the closest races are in their, especially Ohio, Florida, North Carolina, Iowa and Nevada. New this time around is the true addition of both Georgia and Arizona to the battlegrounds. We also see that in 19 of 27 "blue" states, Clinton's current polling aggregate is better than Obama's end-polling aggregate, calculated on election day, 2012. With Trump it is exactly the opposite: in 19 of his 24 states,his current polling aggregates are less than they were for Romney on election day, 2012. You can see that data in column "I" of the EVERYTHING TABLE

And some end-statistics: I first started collecting 2016 presidential polls at the beginning of October 2013 and since then I have logged 3,946 matchups from 1,859 individual polls. You can find the breakdown to this stuff at the EXCEL table, in the tab that says "poll totals".

For past reference, here is the BATTLEGROUND REPORT from 28 days before the GE 2012, from 10/09/2012And here is the BATTLEGROUND REPORT from 28 days before the GE 2008, from 10/07/2008.

-Statistikhengst

No comments:

Post a Comment

Constructive comments and critique are always welcome. Please keep it polite and respectful.