Analysis of California Propsition 8 Exit Poll Data

In 2008, of all the ballot measures throughout the United States, California's Proposition 8 garnered the most monetary donations, over $74 million ($36+ million for, $38+ million against). More than 13 million Californians voted, with 52.3% supporting prop 8 (page 6 of PDF), thereby declaring, "Only marriage between a man and a woman is valid or recognized in California.". posted exit poll data from over 2000 respondents (which is statistically significant).


At its core, the institution of marriage formalizes certain associations between members of a society. Looking at the historical record, the associations themselves have varied over time as well as the number of members and combinations of their sexes. The data supports common wisdom, contains a few curious details and provides a basis for predicting the results of future votes concerning same-sex marriage. The exit poll question which most decisively indicates that an individual believes marriage is between one male and one female centers around religion:

Religion    % of respondents  Yes on prop 8   No on prop 8
Protestant  (43%)             65%             35%
Catholic    (30%)             64%             36%
Jewish      (5%)              N/A             N/A
Other       (6%)              N/A             N/A
None        (16%)             10%             90%

9 out of 10 people who claim no allegiance to a particular religion believe California should not restrict marriage to specific combinations of gender. In addition, combining the Jewish percentage with Other, the missing yes % for those 11% of respondents can be filled in:

Religion    % of respondents  Yes on prop 8   No on prop 8
Protestant  (43%)             65%             35%
Catholic    (30%)             64%             36%
Other       (11%)             33%             67%
None        (16%)             10%             90%

As Protestants and Catholics constitute Western Christianity, which in turn accounts for about 90% of Christians in the world (most likely more than 90% in California), it is highly unlikely that a significant portion of the "Other" group are Christians. Therefore, further simplification of the data:

Religion      % of respondents  Yes on prop 8   No on prop 8
Christian     (73%)             65%             35%
Non-Christian (27%)             18%             82%

shows people voted for or against Proposition 8 primarily based on whether an individual considered him or herself not simply religious, but specifically Christian. Given another statistic:

Church Attendance % of respondents  Yes on prop 8   No on prop 8
Weekly            (32%)             84%             16%
Occasionally      (44%)             46%             54%
Never             (21%)             17%             83%

One can quite accurately predict whether someone voted yes or no on Proposition 8 with this algorithm.

Does the person consider him or herself Christian?
If no, figure that person voted no.
If yes, ask the person how often they attend church.
  If they answer weekly, figure they voted yes,
  if some other frequency, they likely voted no. 

Based on the exit poll data, this algorithm should lead to a correct prediction 3 out of 4 times.

The California Government is supposed to be secular. If one makes the argument that marriage is a religious institution, then no definition of it should exist in a government document such as a constitution. If a definition of it does exist in a government document, then the content of it should not be influenced by religious beliefs. By that logic, the passing of Prop 8 is a sham.


Looking at the vote by race reveals a curious detail:

Race              % of respondents  Yes on prop 8   No on prop 8
White             (63%)             49%             51%
African-American  (10%)             70%             30%
Latino            (18%)             53%             47%
Asian             (6%)              49%             51%
Other             (3%)              51%             49%

That data can be reduced to:

Race              % of respondents  Yes on prop 8   No on prop 8
African-American  (10%)             70%             30%
Other             (90%)             50%             50%

Why are African-Americans 40% more likely to have cast a vote in favor of Prop 8? Unfortunately one can only speculate as the exit poll data doesn't provide any other data allowing further refinement. However, another curious detail arises when one looks at sex and race:

Sex     % of respondents  Yes on prop 8   No on prop 8
Male    (46%)             53%             47%
Female  (54%)             52%             48%

On the surface, whether one is male or female doesn't seem to play much of a part in how an individual votes. However, adding race into the mix:

Sex and Race    % of respondents  Yes on prop 8   No on prop 8
White Men       (31%)             51%             49%
White Women     (32%)             47%             53%
Black Men       (4%)              N/A             N/A
Black Women     (6%)              75%             25%
Latino Men      (8%)              54%             46%
Latino Women    (11%)             52%             48%
All Other Races (9%)              49%             51%

And calculating the missing black men percentages:

Sex and Race  % of respondents  Yes on prop 8   No on prop 8
Black Men     (4%)              62.5%           37.5%

It is clear that within the other races, the percentage differential between the male and female vote was at most 4%, yet black male and female voters differed by 3 times as much. Why are black females 20% more likely to cast a yes vote on Prop 8 than their male counter-parts?


Whether a person works full time or not also seems to have a stronger than expected effect on their vote:

Do You Work Full-Time?  % of respondents  Yes on prop 8   No on prop 8
Yes                     (59%)             48%             52%
No                      (41%)             57%             43%

What is it about being a full-time employee that causes people to vote against prop 8? The poll data doesn't differentiate further between working part-time and not working at all, but it does provide some data with respect to income:

Vote by Income    % of respondents  Yes on prop 8   No on prop 8
Under $15,000     (5%)              46%             54%
$15-30,000        (10%)             48%             52%
$30-50,000        (15%)             54%             46%
$50-75,000        (19%)             54%             46%
$75-100,000       (17%)             50%             50%
$100-150,000      (17%)             54%             46%
$150-200,000      (7%)              47%             53%
$200,000 or More  (9%)              45%             55%

It's useful to reduce the granularity a bit to more clearly highlight where the votes swing:

Vote by Income    % of respondents  Yes on prop 8   No on prop 8
Under $30,000     (15%)             47%             52%
$30-75,000        (34%)             54%             46%
$75-100,000       (17%)             50%             50%
$100-150,000      (17%)             54%             46%
$150,000 or More  (16%)             46%             54%

As income increases, the vote for or against Prop 8 appears to see-saw! Without any other data we're again left to pure speculation as to what could explain these numbers.


The exit poll data contains one more data point worth exploring.

Vote by Age   % of respondents  Yes on prop 8   No on prop 8
18-24         (11%)             36%             64%
25-29         (9%)              41%             59%
30-39         (17%)             52%             48%
40-49         (22%)             59%             41%
50-64         (26%)             51%             49%
65 or Over    (15%)             61%             39%

Assuming relatively even distribution of population within an age group, using the midpoint of the age range as X and the yes % as Y for the first 4 age ranges, one ends up with a surprisingly linear graph, with a regression equation:

prop 8 graph

To put it more simply, if someone is 49 years old or younger, add 14 to their age and that is the percentage likelihood he or she would have voted yes on Prop 8.

The 50-64 and 65 or Over groups were left off the chart because there appears to be a major difference in the voting behavior depending on whether someone is over or under 50 and there is no longer a linear connection between a person's age and their vote.

Presuming all the other distributions of race, religion, ideology, income, etc. stay relatively the same, and most people who voted in 2008, don't change their vote, one can predict the outcome of similar elections in the future. The Yes percentages for the 18-49 age groups will decrease by roughly 1% a year. It's likely the 50-64 group will see a slight increase in Yes %, while the 65 or Over will see a slight decrease. Presuming those two roughly cancel each other out, in 2 years (the next time Californians will vote on propositions), the numbers will likely look like this:

Vote by Age   % of respondents  Yes on prop 8   No on prop 8
18-24         (11%)             34%             66%
25-29         (9%)              39%             61%
30-39         (17%)             50%             50%
40-49         (22%)             57%             43%
50-64         (26%)             51%             49%
65 or Over    (15%)             61%             39%

Those numbers would lead to a 50.7% Yes vote on a proposition similar to prop 8, a decrease in the yes vote of 0.75% per year. In 2000, California Proposition 22, which initially prevented California from recognizing same-sex marriages, was adopted by 61.4% of voters. Therefore, between 2000 and 2008, the initiative lost support at an average rate of 1.15% per year. It makes sense that the rate predicted above of 0.75% for the future is somewhat smaller since the rate of change in support/opposition of a controversial proposition would be expected to slow down as it approaches equalibrium. Nevertheless, while it's unlikely for a vote in 2010 to come up in favor of legalizing same-sex marriage, unless something drastic happens, it looks like a majority of citizens will likely be in favor of same-sex marriage by 2012.