Damien Hirst at Tate Museum

Blog

Random musings, tales of my travel, etc.

Do Educated People Vote For Obama?

I saw this chart showing that the more educated a state's population, the more they voted for Obama. But their methods bothered me. Are these results statistically meaningful? What if most of these states were swing states? So I downloaded the data from the census bureau and correlated it with the election results.

Below are R values (correlation coefficient) for the correlation between the a state’s value for the given variable and the percent of each state that voted for a particular candidate.

Educational Attainment (% of state) Correlation w/ % Obama Vote (R-value)
Graduate or professional degree 0.78
Bachelor's degree 0.50
No high school 0.02
Some high school, no degree -0.19
Associate's degree -0.21
High school graduate or equivalent -0.36
Some college, no degree -0.60
School enrollment (% of state) Correlation w/ % Obama Vote (R-value)
College or higher 0.66
Nursery/Preschool -0.09
High School -0.17
Kindergarten -0.64
Elementary -0.73
Marital Status (% of state) Correlation w/ % Obama Vote (R-value)
Never Married 0.75
Separated 0.02
Widowed -0.09
Divorced -0.42
Married -0.67
Ancestry (% of state) Correlation w/ % Obama Vote (R-value):
Lithuanian, African, Arab, Italian, Ukrainian 0.4 to 0.51
Greek, Polish, Russian, Portuguese, Indian, 0.3 to 0.4
Canadian, Hungarian, French, Slovak, Irish 0 to 0.3
Swedish, Scottish, Norwegian, Czech, Dutch, Swiss, English, German, Welsh 0 to -0.3
Danish, Scotch-Irish -0.3 to -0.4
American -0.54

So we have some expected results:

  • Married couples with young children are more likely to vote for Romney
  • The less American/Western European your heritage the more likely you voted for Obama

But some are pretty unexpected:

  • People who drop out of college vote for Romney more than people who never start college. What?!?
  • The Married/Unmarried divide is stronger than the educational one.

Anyway, it turns out the original article's chart is not that misleading. But now we have some actual numbers that us compare the effect of different variables. Enjoy!

MathAndrew Carman