statistical breakdown of voter preferences

[bkfizz02]

I have been trying to understand group voting preferences in the United States and was looking for some statistics on this subject. I have failed to find useful information on google, probably because I can't put together a good word combo.

Anyways, I'm trying to answer the following type of question:
If an African-American is running against a Caucasian (American), what percentage of the African-American vote will he get?
What if the African-American is a Democrat and the Caucasian is a Republican? What if it is reversed?
If two African-Americans run against each other, how does the African-American vote break down?
What if two Caucasians run against each other?

I am also interested in this more generally, considering race, gender and perhaps social factors. If anyone can point me in the right direction here, I would appreciate it.

Thanks.

[DVaut1]

[ QUOTE ]
I have been trying to understand group voting preferences in the United States and was looking for some statistics on this subject. I have failed to find useful information on google, probably because I can't put together a good word combo.

Anyways, I'm trying to answer the following type of question:
If an African-American is running against a Caucasian (American), what percentage of the African-American vote will he get?
What if the African-American is a Democrat and the Caucasian is a Republican? What if it is reversed?
If two African-Americans run against each other, how does the African-American vote break down?
What if two Caucasians run against each other?

I am also interested in this more generally, considering race, gender and perhaps social factors. If anyone can point me in the right direction here, I would appreciate it.

Thanks.

[/ QUOTE ]

There's no 'general' answer to this question. When Michael Steele (a black Republican) runs against Ben Cardin (a white Democrat), the results are not at all similar to when Barack Obama (a black Democrat) runs against Jack Ryan (a white Republican) or Alan Keyes (a black Republican) -- if we attempted to extrapolate "group voting" preferences from these results, there's a whole host of variables to consider, not the least of which is that Illinois and Maryland are two different states comprised of entirely different constituencies.

I think you're having trouble finding the information you're looking for because you're asking the wrong kinds of questions.

To get you going in the right direction, type "Pew" into Google. Good luck.

[bkfizz02]

I appreciate that the individual results of an election have a large variance ; it is specifically this difference that I am curious about. I am looking for averaged statistics grouped by the candidates characteristics.
For example:
National elections: African-American Male (D) vs Caucasian Male (R) --> x% of African-American vote for D, y % for R.
Local elections : African-American Male (D) vs African-American Male (R) --> z% D, w% R

My belief/hypothesis is that people will tend to vote for someone whose race/gender/background is the same, rather than for a person who is in the same party. I am trying to quantify this claim with evidence.
I don't think this is too obscure of a question. There surely are voting preference breakdowns along the lines of : African-American voted a% for D and b% for R in national elections in the 90s...I just want a more discrete breakdown.

Thanks.

[sam h]

[ QUOTE ]
I appreciate that the individual results of an election have a large variance ; it is specifically this difference that I am curious about. I am looking for averaged statistics grouped by the candidates characteristics.
For example:
National elections: African-American Male (D) vs Caucasian Male (R) --> x% of African-American vote for D, y % for R.
Local elections : African-American Male (D) vs African-American Male (R) --> z% D, w% R

My belief/hypothesis is that people will tend to vote for someone whose race/gender/background is the same, rather than for a person who is in the same party. I am trying to quantify this claim with evidence.
I don't think this is too obscure of a question. There surely are voting preference breakdowns along the lines of : African-American voted a% for D and b% for R in national elections in the 90s...I just want a more discrete breakdown.

Thanks.

[/ QUOTE ]

Unless you yourself have some statistical skills, then I'm not sure you will find the numbers already crunched in exactly the way you want them.

However, I can tell you based on a fair amount of experience following races and sometimes seeing cross-tabs of specific voting populations that your position does not accord with my experience, at least regarding contests between black republicans and white democrats. In these instances, it usually seems to be the case that a majority of black voters still break democratic. Ties with the democratic party in many black communities are very strong and black voters are very capable of looking beyond the skin color of the candidates.

[DVaut1]

[ QUOTE ]
My belief/hypothesis is that people will tend to vote for someone whose race/gender/background is the same, rather than for a person who is in the same party. I am trying to quantify this claim with evidence. I don't think this is too obscure of a question.

[/ QUOTE ]

It's not an obscure question and your clarification is better suited for an answer. So thank you for that.

FWIW, most of the empirical evidence suggests that partisan identification is the greatest predictor of vote preference -- I point to the work of Wattenberg (link), Miller/Shanks (link), and Rahn (link), which are all quite clear on the subject. This comes with the caveat that "partisan identification" is not synonymous with “party affiliation”, and that younger cohorts in the American electorate do seem to suggest a movement towards non-alignment -- a manifestation of the fact that elections in the US are becoming increasingly candidate-centered. Having said that, that doesn't mean partisan identification is less important -- only that "partisanship" is less strongly correlated with party affiliation than it was in the past.

Understand, however, that these questions are constantly subject to debate, and that all such studies can do is suggest strong correlations.

Having said that, if you're looking for a 'general' response to your hypothesis ("My belief/hypothesis is that people will tend to vote for someone whose race/gender/background is the same, rather than for a person who is in the same party"), the answer is this: your hypothesis is almost certainly wrong based on the best evidence available.

[ QUOTE ]
There surely are voting preference breakdowns along the lines of : African-American voted a% for D and b% for R in national elections in the 90s...I just want a more discrete breakdown.

Thanks.

[/ QUOTE ]

National Election Studies has all of this information from 1948 to 2004 but it won't come pre-packaged to answer your specific questions, so you'll have to do some digging:

http://www.umich.edu/~nes/
http://en.wikipedia.org/wiki/National_Election_Studies

Most of the studies I cited earlier (Wattenberg, Rahn, Miller/Shanks) rely heavily on NES data. There are other election studies, including Annenberg, which are highly respected and utilized by scholars.

Link to Annenberg Public Policy Center:
http://www.annenbergpublicpolicycent...naes/index.htm

As I mentioned before, Pew is another respected research center where you could discover such information:
http://people-press.org/

Specifically, I'd point to "Beyond Red vs Blue ", published in 2005, that lays out some easily accesible information regarding the relationships between American public, political typology, and race/gender/religion.

[Poofler]

Agree you will find a hard time finding studies to tackle something with extreme variables. Maybe you should do a little specific race comparison.

Maryland Senate:
BLACK-Steele(R) v WHITE-Cardin(D):
Steele won 25% of the black vote.
Maryland governor:
WHITE-O'Malley(D) v. WHITE-Ehrlich(R):
Ehrlich won 15% of the black vote.

There are so many variables in play here, but I think most would certainly say some of that extra margin for Steele is pure race-preference. Nothing huge, and usually not enough to get a Black Republican elected, especially with some negative white vote movement.

[bkfizz02]

I think that this post shows an instantiation of the thought I was trying to describe. It seems clear to me now that my hypothesis as stated is completely invalid. It also was a poor expression of the issue I was trying to understand.

My original point was to try do determine what factors would cause voters to swing away from political affiliation, in a statistically significant way.

Suppose that you know that 92% of African-Americans voted Democratic in presidential elections in the last 3 cycles. Then suppose that in the next (hypothetical) election, Democrats only get 75% of the African American vote. This would be a very significant statistical result and I would want to determine what variables were the most strongly correlated (e.g. an African-American Independent candidate, or a Female Republican).

So, reworking my original hypothesis, I am suggesting that factors like race/gender/background can cause statistically significant fluctuations when comparing the voting preferences of certain ethnic/gender/religious groups to historical trends in party affiliation.

Thanks for helping me work out this idea.

[bkfizz02]

Dear DVaut1,
Thanks for the great detailed post with excellent sources. These were the sort of links I was looking for but failed to find on google.

I agree that my hypothesis as stated was not true. I have subsequently tried to rephrase my point in another post. Thanks for your help.
BK

[Poofler]

You also have to watch out for the PC phenomena in a lot of pre-race and exit poll data. Respondents sometimes mislead the pollster to seem PC towards minorities or women.