I am not a statistician, so my math could be wrong. If so someone please correct it. I actually feel kind of stupid making this post because I know someone who actually understands statistics is going to be bothered by all the errors and assumptions I’ve made.
I added a TL;DR synopsis at the bottom if you want to get a rough idea of what this is about. I used a mix of washington post, New York Times and wikipedia to get stats for this post.
What I did was I compared the net difference between the republican vs the democrat in both the 2012 and 2016 election. The reason I picked 2012 was because I felt that was more accurate than 2008, which was more heavily a democratic election cycle.
So in 2012, the democrat had a 3.9 point net margin over the republican (51.1% vs 47.2%). In 2016 the democrat had a 0.3% margin over the republican (47.6% vs 47.3%). This means national polls moved 3.6% to the right. Numbers are still coming in, so that value could change be a few tenths of a percent (it was a 0.2% margin when I started researching this post). Unless otherwise mentioned, all stats below are not factoring in the 3.6% move to the right. So if I say Rhode Island moved 12 points to the right, I mean it moved 12 points to the right (but after subtracting the national movement to the right, it arguably only moved 8.4% to the right, 12-3.6=8.4). But I rounded numbers for simplicity so when factoring in national polls I moved things 4 points.
So is it safe to say that if a state moved 3.6% points to the right from 2012 to 2016, that that means the state is pretty much the same from 2012 to 2016? Meaning, the net difference change on the state level matched the net difference on the federal level? Example, SC moved roughly 4 points to the right from 2012 to 2016, but so did national polls. So I don’t think there is a major difference in the electorate there from 2012 to 2016.
Here is what I found by researching exit polls:
The west coast, southwest swing states and southern swing states became more democratic either before or after factoring in national polls (at least in presidential voting, no idea about other races). CA moved 5% to the left (9% after factoring in national polls). NC moved 2% to the right from 2012 to 2016, but after factoring in national polls and their ~4% move to the right, that is a 2% move to the left for NC correct?
The northern Midwest and northeast swung heavily to the right. IA moved the most, 16% to the right.
Examples:
NH moved 6 points to the right from 2012 to 2016 (which works out to about 2.4 points after factoring in the fact that national polls moved 3.6% to the right). That is why NH was a fairly reliable state in 2012 but in 2016 was too close to call. It went from a 6 point margin for Obama in 2012 to a razor thin tie in 2016.
But VT also moved 6 point to the right. But it is so deeply blue nobody noticed.
NY moved 7 points to the right. NJ & ME moved 12 points to the right. Pretty much all of new England moved 6-12% to the right (2-8% after factoring out national polls).
Democrats generally rely on the west coast, northern Midwest and northeast as their firewall to give them a base of 246 EVs (242 if you exclude NH), they then try to get southern swing states and the southwest to get to 270. As I said earlier, the west coast, southwestern swing states and southern swing states moved further to the left (their state level party difference was smaller than the national party difference).
However the northeast and northern Midwest moved further to the right.
There was some ‘good’ news, but who knows how reliable it is.
From 2012 to 2016, the margin in TX, VA & GA moved 7%, 1% & 2% towards Hillary. Factor in national polls and that is an 11%, 5% and 6% move to the left. NC and FL only moved 2% to the right, which factoring in the national gap (rounded up to 4%) means NC & FL moved 2 points to the left from 2012 to 2016.
CO & NM moved 2% to the left after factoring in national polls (4 minus 2). AZ moved 9% to the left. Nevada moved 1% to the right, not sure what that is about.
FL went from being 4% more GOP than national polls in 2008 (Obama won national polls in 2008 by 7%, he won Florida by 3% in 2008) to 3% more GOP than national polls in 2012 (Obama won national polls by 4%, he won Florida by 1% in 2012) to 1% more GOP than national polls in 2016 (national polls were roughly equal, Trump won FL by 1%). Of course things can and do change, in 2004 FL was 2% more GOP than national polls, but was equal to national polls in 2000.
Another thing I want to discuss is exit polls and education:
I tried breaking exit polls down in 3 ways. By race, education and income. For whatever reason the race and income numbers don’t match the national shift, but that could be because I do not have the correct info from 2012. Or maybe people lie about how much money they make.
What I did was I took the net difference between R and D voting, and multiplied it by the % of the electorate in each group when comparing 2012 to 2016. So if whites w/o a college degree voted a 10% higher net vote for the republican in 2016 (maybe 60-40 in 2016 vs 55-45 in 2012), and make up 20% of the electorate, I multiplied 10*0.2=2% shift in national polls towards Trump. However I did not have the % of the electorate by race or income from 2012, but I assume the numbers barely changed (in 2012 non-whites were 28%, whites 72%. In 2016 non-whites 30%, whites 70%).
Income:
3.96 T under 50k
.62 H 50-99k
2.97 H 100k+
0.37 T shift total.
So those numbers do not match national polls. Maybe I did something wrong. But people under 50k moved to the right by 4 points, but it was mostly negated by people who claim to earn 100k moving 3% to the left.
Race:
White 0.7 T
Black 0.84 T
Latino 0.88 T
Asian 0.44 T
Other 0.03 T
2.19 T for non-whites. 2.99 T total for all races.
This kind of matches the 3.6% shift to the right, but still not exactly.
Education and race comparing 2016 vs 2012.
White non college grad 4.76% T
Non white college grad 0.91% T
Non white non college grad 1.76% T
White college grads 3.7% H
This one matches fairly well. Non-whites moved national polls 2.67% to the right, whites moved them 1% to the right. This roughly matches the 3.6% move to the right in national polls from 2012 to 2016.
So whites w/o a college education broke heavily for Trump. But whites with a college education broke heavily for Hillary. Combined they mostly cancelled each other’s influence on national polls out.
So here are my questions:
[ul]
[li]Why did the democratic party seem to gain ground on the west coast, southwest and south, but lose so much ground in the northern Midwest and northeast? It isn’t just the swing states in these regions where the democrats did worse. They did worse in virtually all the deep blue states in the northeast and northern Midwest too. PA moved 6% to the right (2% after factoring in national polls) which turned it blue to red. But NY moved 7% to the right (3% after national polls). DE moved 8% to the right (4% after national polls). RI and ME moved 12 points to the right (12 points is also how much OH moved). We just didn’t notice because that 12 point shift turned Ohio from blue to red, but all it did in RI and ME was turn them from deep blue to slightly less blue.[/li][li]Why did college education make such a huge difference for white people but not for non-whites? For whites, there was an 8.5% difference between college grads and non grads in impact on national polls. For non-whites it was 0.8%, which is 10x smaller. Why did college educated whites abandon Trump but college educated non-whites voted for him?[/li][li]Why did minorities in general break for Trump? When you break the numbers down by race, much of Trump’s gain came from doing better among non-whites. Minorities made up 30% of the electorate, and about 70% of the move to the right in national polls.[/li][li]Hell Asians, who made up 4% of the electorate moved national polls 0.44% to the right. Whites who made up 70% of the nation moved the polls only 0.7% to the right. So Asians had half the impact of whites in moving national polls 3.6% to the right, despite being 18x more rare than whites.[/li][li]Of the 10 states with the highest % who have a bachelor degree, 7 are in the northeast and northern Midwest. But those states all moved further to the right than other regions. What gives? Also minorities are rare in several new England states, but common in others. But VT & NH (which are almost all white) moved as far to the right as PA & NY, which have large minority populations.[/li][li]Why are a handful of states in the democratic firewall in the northeast like MD and MA outliers? Why did RI and ME move 12 points to the right, but MA moved 4 points to the left or MD moved 1 point to the left?[/li][li]What role did voter suppression play? There were voter suppression efforts all over the country, but it was only the northern Midwest and northeast that moved to the right (I didn’t really investigate the plains states or non-swing states in the south). There was probably not much voter suppression in the northeast because they do not have GOP rule of both houses of the legislature and the governorship, but they moved far right. However places like NC actually moved to the left after factoring in national polls despite voter suppression efforts there. WI moved to the right with voter suppression, but so did other states in the region which I don’t think had the same voter suppression efforts.[/li][li]Was it a coalition of whites without a college degree and minorities who moved the northeast and northern Midwest so far right? If so, why didn’t CA move to the right, they have a ton of minorities and white people without college degrees too. This shift to the right seems localized to the northern Midwest and northeast. The plains states moved right too, but I didn’t really read up on them as much. [/li][li]Did people in deep blue states vote Trump as a protest vote, because they figured their vote didn’t matter and their state would vote Clinton no matter what, so they figured why not? If so, why was it only the northern midwest and northeast. Why not the west coast? Why not MD or MA? Why did the swing states in the midwest move so far to the right, but the swing states in the south and southwest moved towards Clinton compared to 2012? [/li][/ul]
Maybe the answers are more regional, but I don’t want to investigate at that level of depth (ie, blacks in Michigan didn’t vote the way blacks in California did).
TL;DR – Comparing polls from 2012 to 2016, the democratic candidate for president actually gained ground in the west coast states (CA, OR, WA), southern swing states or potential swing states (TX, GA, NC, VA, FL) and southwest swing states (CO, NM, AZ, but not NV) either before or after factoring in the fact that national polls moved 3.6% to the right from 2012 to 2016.
The Democratic candidate did far worse in the northern Midwest and northeast. Much of this region moved 6-12 points to the right (2-8% after factoring in the fact that national polls moved roughly 4% to the right). This move 6-12 points to the right meant states that were only ~5-8% or so in favor of Obama in 2012 became red states in 2016 (WI, OH, PA) or became razor thin races (MN, NH). However other states in this region also moved 6-12 points to the right, but they were so deeply blue in 2012 they turned blue in 2016 anyway. Just with smaller margins. DE and NY moved as far to the right as PA, but because they are a deeper shade of blue they didn’t switch sides like PA did.
Non-whites made up roughly 2/3 of the move of national polls by 3.6% to the right. Whites only made up the other 1/3 (I’m having trouble with the numbers though).
White college grads broke for Hillary enough to mostly compensate for white high school grads who broke for Trump.
There was a massive difference between white college grads and high school grads in who they voted for (high school grads moved right, college grads moved left), but minorities didn’t really divide by education. Both college educated and non-college educated minorities moved further to the right.
Pro Trump rust belt and northeast (move towards Trump):
PA 6%
MI 10%
WI 8%
OH 12%
IN 10%
IA 16%
MN 7%
NY 7%
VT 6%
CT 6%
NJ 12%
DE 8%
ME 12%
50 states results (T=Trump; H=Hillary):
AL 6% T
AK 1% H
AZ 5% H
AR 2% T
CA 5% H
CO 2% T
CT 6% T
DE 8% T
FL 2% T
GA 2% H
HI 11% T
ID 1% H
IL 1% T
IN 10% T
IA 16% T
KS 0%
KY 7% T
LA 3% T
ME 12% T
MD 1% T
MA 4% H
MI 10% T
MN 7% T
MS 6% T
MO 10% T
MT 6% T
NE 4% T
NV 5% T
NH 6% T
NJ 12% T
NM 2% T
NY 7% T
NC 2% T
ND 16% T
OH 12% T
OK 3% T
OR 1% T
PA 6% T
RI 12% T
SC 4% T
SD 12% T
TN 6% T
TX 7% H
UT 29% H (outlier)
VT 6% T
VA 1% H
WA 2% H
WV 15% T
WI 8% T
WY 7% T