Breakdown of states from most pro-Hillary to most pro-Bernie
Name |
Estimated Clinton Vote Percentage |
Percentage White |
Percentage Non-White |
Income Average Category |
Percentage White + Liberal Rank |
|
Hawaii |
64.24 |
23 |
77 |
5 |
UNKNOWN |
|
California |
63.32 |
39 |
61 |
5 |
33 |
|
District of Columbia |
62.7 |
35.8 |
64.2 |
5 |
UNKNOWN |
|
Maryland |
61.61 |
53.3 |
46.7 |
5 |
29 |
|
New Jersey |
61.09 |
57.6 |
42.4 |
5 |
26 |
|
Texas |
60.79 |
39 |
61 |
3 |
36 |
|
New York* |
60.71 |
57.2 |
42.8 |
4 |
11 |
(added +5% Clinton since it's her home state) |
Alaska |
60.11 |
57.5 |
42.5 |
5 |
UNKNOWN |
|
Illinois |
58.91 |
62.7 |
37.3 |
4 |
31 |
|
Nevada |
58.35 |
52.2 |
47.8 |
3 |
27 |
|
Arkansas* |
57.89 |
68.7 |
31.3 |
1 |
28 |
(added 5% to Clinton since it's a former home state) |
Georgia |
57.29 |
49.8 |
50.2 |
2 |
34 |
|
Minnesota |
57.18 |
81.9 |
18.1 |
5 |
UNKNOWN |
|
Colorado |
56.94 |
69.4 |
30.6 |
4 |
UNKNOWN |
|
Mississippi |
56.46 |
52.5 |
47.5 |
1 |
37 |
|
North Carolina |
56.42 |
59.4 |
40.6 |
2 |
32 |
|
Virginia |
56.37 |
63.6 |
36.4 |
5 |
23 |
|
Louisiana |
55.99 |
54.6 |
45.4 |
1 |
39 |
|
New Mexico |
55.33 |
39.4 |
60.6 |
1 |
19 |
|
South Carolina |
55.05 |
58.9 |
41.1 |
1 |
35 |
|
Wyoming |
54.8 |
84.1 |
15.9 |
4 |
UNKNOWN |
|
Delaware |
54.7 |
64.1 |
35.9 |
4 |
22 |
|
Alabama |
54.49 |
61.4 |
38.6 |
1 |
38 |
|
North Dakota |
54.34 |
87.3 |
12.7 |
4 |
UNKNOWN |
|
Arizona |
53.92 |
51.7 |
48.3 |
2 |
16 |
|
Connecticut |
53.65 |
69.6 |
30.4 |
5 |
7 |
|
Washington |
53.16 |
71 |
29 |
4 |
UNKNOWN |
|
Florida |
53 |
56.4 |
43.6 |
2 |
15 |
|
Massachusetts |
52.99 |
75.1 |
24.9 |
5 |
4 |
|
Kansas |
52.73 |
77.1 |
22.9 |
3 |
UNKNOWN |
|
Tennessee |
52.63 |
69.9 |
30.1 |
1 |
30 |
|
Nebraska |
52.02 |
81 |
19 |
3 |
UNKNOWN |
|
Rhode Island |
51.08 |
75.3 |
24.7 |
4 |
8 |
|
Idaho |
50.79 |
83.1 |
16.9 |
2 |
UNKNOWN |
|
Utah |
50.44 |
79.7 |
20.3 |
4 |
6 |
|
Oklahoma |
50.26 |
62.5 |
37.5 |
1 |
25 |
|
Iowa* |
49.9 |
87.6 |
12.4 |
3 |
3 |
(Clinton should have gotten 45.79 according to model) |
Pennsylvania |
49.49 |
78.4 |
21.6 |
3 |
12 |
|
West Virginia |
49.4 |
92.7 |
7.3 |
1 |
18 |
|
Michigan |
49.16 |
76.1 |
23.9 |
2 |
14 |
|
Maine |
48.65 |
94 |
6 |
2 |
UNKNOWN |
|
South Dakota |
48.59 |
83.3 |
16.7 |
3 |
13 |
|
Missouri |
48.32 |
80.4 |
19.6 |
2 |
20 |
|
Ohio |
48.31 |
80.5 |
19.5 |
2 |
21 |
|
Indiana |
48.26 |
80.7 |
19.3 |
2 |
24 |
|
Oregon |
47.67 |
77.5 |
22.5 |
3 |
5 |
|
Wisconsin |
46.74 |
82.5 |
17.5 |
3 |
10 |
|
Kentucky |
45.17 |
85.6 |
14.4 |
1 |
17 |
|
Montana |
45.04 |
87 |
13 |
2 |
9 |
|
New Hampshire* |
41.01 |
91.6 |
8.4 |
5 |
2 |
(added +10% Sanders since it's a neighboring state) |
Vermont* |
33 |
93.8 |
6.2 |
4 |
1 |
(added +10% Sanders since it's his home state |
|
After seeing the polls on the GOP side spectacularly fail in Iowa, I got to thinking that there must be a better way to quantify which states are most likely to favor Hillary and which are most likely to end up backing Bernie. Using exit polls from the Iowa Democratic primary (www.nbcnews.com/...) I got around to finding some data points that I thought would be most helpful in determining which states are naturally predisposed to voting for one candidate or the other.
Firstly, it is worth saying that some data is not terribly useful. For example, the exit polls showed a gender gap, with women favoring Hillary and men favoring Bernie, but that is not helpful looking forward: it’s not like there are some states with 90% women and others with 75% men! Iowa also did not do much to clarify how each candidate performs with different racial groups; Iowa is so overwhelmingly White that it is very hard to know for certain which candidate the African American, Latino, and Asian American communities backed ... let alone how strongly. Age is much the same story: Bernie’s strength with younger voters is well documented, but age distributions are simply not that different from state to state (every state has at least some older population, and every state has at least some major colleges and universities).
However, the exit polls did provide one particular data point that I believe could shed light on where the other states might end up in the Hillary v. Bernie race: income. Sanders ran notably strong among those with less annual income, while Hillary won in a landslide among the $100K+ demographic. Unlike age and gender, income can vary widely from state to state. That was the starting point for my analysis.
Another data point that I believe could be useful in determining how each candidate will do is the proportion of White liberals in each state. Nate Silver’s FiveThirtyEight blog has a post that includes a graphic showing 39 states ranked according to the estimated number of Whites who identify as liberals living in them (fivethirtyeight.com/...), which was also used in my analysis.
The table above is organized in the following way: the first column includes the states ranked by what percentage Hillary was calculated to get by my analysis (with an asterisk added if there was a special reason that the percentage for Hillary was adjusted, like in each candidates’ respective home states); the second is the exact percentage Hillary is expected to get in each state; the third shows what percentage of the state overall is White (there were some adjustments made to this number; see below for more); the fourth assigns a number, 1 through 5, to each state based on its average annual income (see below for details; the basic idea is that category 5 states are the wealthiest and category 1 states are the poorest); the fifth includes how each state ranked in terms of its number of White liberals (11 states did not have enough data to make an estimate); and the sixth and final column includes an explanation of adjustments made to the percentage Hillary was estimated to receive.
Now some details to clarify how each value was arrived at. In terms of the third column, percentage White, each states with a White percentage that exceeds 75% was simply assigned its regular value (meaning that no adjustment was made to it, on the presumption that the Democratic primary there will roughly reflect the larger population); for states that have a White population of 60-75% and voted for Romney in ‘12, the White percentage was knocked down by 5% (on the presumption that, like most conservative states, their electorate is more racially polarized and thus the White vote is less likely to take part in Democratic primaries); and for states that have a White population under 60% and voted for Romney, the White percentage was adjusted down 10% (for basically the same reason as the previous adjustment). Based on polling so far, the non-White vote was assumed to be 67% Hillary vs. 33% Bernie. This may actually slightly underestimate Hillary’s potential with the non-White demographic, since she currently polls closer to 75% among Latinos and 80% among African Americans, but I did these calculations on the presumption that Bernie’s support among non-Whites will rise over time (as Bernie has become better known he has slowly been able to lock up the ultra-liberal activist wing of the Democratic Party, who I believe were always going to turn against a moderate Democrat like Hillary — the question was never if Hillary’s large leads in the polls would disappear, but when, as the party’s left wing began its inevitable consolidation behind a less moderate candidate).
The non-White percentage was calculated through some very straightforward math: subtracting the percentage White from 100.
For the income average, each state was categorized on a 1 to 5 scale based on where they placed when ranked by average annual income. Category 5 states are in the top 10 in terms of average annual income; category 4 are states 11-20; category 3 are states 21-30; category 2 are states 31-40; and category 1 includes the 41st state through the one with the lowest average annual income in the nation (Mississippi).
To calculate what percentage Hillary and Bernie got from the White vote in each state, I based it purely on how each state ranked in the above categorization scheme. Among White voters in category 5 states, Hillary was assumed to have won by 10% (she won among the $50K+ demographic by 10% according to Iowa exit polls); in category 4 states, Hillary was assumed to have won by 5%; in category 3 states, Bernie was assumed to have won by 3%; in category 2 states, Bernie was assumed to have won by 5%; and in category 1 states, Bernie was assumed to have won by 10%.
As to how each state ranked in terms of the number of White liberals, every state that placed in the top 10 in that ranking was adjusted to give Bernie +5%; those ranked 11-25 were adjusted to give Bernie +3%; and those that ranked 26-39 were adjusted to give Hillary +1%. Those without any data on this point were not adjusted.
With all of those data points entered, my calculations yielded the table above, ranked in descending order in terms of what percentage Hillary is likely to get in each state. The image at the top of this article conforms to the table: the 25 states most favorable to Hillary are colored blue; the 14 states where Bernie would be expected to win by these calculations are colored red; the gray states are ones where the data indicates Hillary has an edge but that failed to rank in her top 25 most favorable.
The most important point to stress with this analysis is that the percentages estimated for Hillary in each state is almost certain to change. Every election, particularly primaries, is fluid; perhaps Hillary will see a rise in her fortunes among working class Whites, or Bernie will start getting majorities among African Americans — at this point there is just no way to know. Indeed, my analysis has already failed: whereas Iowa was calculated to be a Bernie state (and fairly easy at that — I would have expected Hillary to get only 45.79% based on its demographics), it ended up backing Hillary narrowly. What I think this analysis can reveal, even with its limitations, is where each candidate could expect to run strongest in relative terms.
As an example: Tennessee and Indiana would seem to be almost identical states (in the context of the Democratic primary) from afar. They both voted for Hillary in ‘08 (though Tennessee did so by a much stronger margin), both are GOP-dominated in the general election (though Indiana is considerably more competitive, as evidenced by the fact that it voted for Obama in ‘08 and currently has one Democratic U.S. Senator, neither of which is even remotely true of Tennessee), both have similar population sizes, etc. Yet this analysis reveals that Indiana is in fact substantially more friendly to Bernie than Tennessee, at least from a demographics standpoint.
Indeed, this analysis reveals that there is the potential for Bernie to build a heartland of support in the Midwest, centered on Indiana and Ohio. He may have difficulty in Illinois (with its large non-White community), Minnesota (thanks to its relatively-affluent population), and North Dakota (for the same reason as Minnesota), but the other states in and around that area (other than Iowa, which he has already lost) seem ripe for the picking.
Hillary’s best areas are shown to be concentrated in two places: the Deep South (with its large African American population) and the US-Mexico border region (thanks to its high number of Latinos).
The most competitive “swing” regions according to this analysis are areas that have a mix of conflicting factors that favor Hillary and Bernie. Washington state is a great case in point: it has a large number of affluent voters and a growing concentration of Latinos and Asian Americans, both of which would seem to point toward support for Hillary, but also a significant proportion of White liberals, which would imply that it’s Bernie territory.
I would love to get people’s feedback on this analysis! Again, I want to stress that the percentages estimated for Hillary are very unlikely to actually reflect the exact number she will get. What I think this analysis can show is which states it would make sense for each candidate to target. For example, Texas and Florida might seem like fairly similar states in the context of the Democratic primary: both are in the Sunbelt and have large African American and Latino populations. However, this analysis shows that it would make more sense for Bernie to target Florida over Texas, since the former’s larger population of White liberals provides him with more of a base to build on to get to a majority. For Hillary, Oregon and Washington state might seem like very similar places; indeed, she actually did better in the former during her ‘08 run than the latter. Yet, as shown here, Washington is in fact the Pacific Northwestern state most likely to give her a win despite that region’s large proportion of Bernie-friendly White liberals.