Bradley Effect

[DukeNukeEm]

A black guy was elected president and it fell right in line with the best polling model there is, I cant imagine why somebody would still find the Bradley Effect important anymore. It just does not exist.

[StrangeSox]

OK. Which proves that in national elections the "bradley effect" is negligible, it doesn't prove there is no bradley effect. To prove a negative, that it doesn't exist, you basically have to prove that no where is there a single person that tells a pollster they are voting for the black candidate and insteads votes for the white. To prove it does exist, you would need to find one person who did. Basically it is usually impossible to prove a negative in an academic sense.

 

The Bradley Effect is a significant variation in the final results from pre-election polls due to racism. If it is such a small number of people that it is negligible, it isn't the Bradley Effect any more. That's just statistical noise at that point.

[Texsox]

The Bradley Effect is a significant variation in the final results from pre-election polls due to racism. If it is such a small number of people that it is negligible, it isn't the Bradley Effect any more. That's just statistical noise at that point.

 

Not exactly. Taking this past election as an example. There are many non-racist conservative voters who voted McCain. The Bradley effect would claim that they would tell pollsters they are voting Obama because they believe they would look better in the pollster's eyes. Or they want to avoid being labeled as racist. The motivation is usually not racism. In fact, a racist person would be much less likely to admit to voting for Obama to anyone. Think in this election someone who is socially liberal but fiscally conservative. That would be the type of person that would give an inaccurate answer to the pollster. Not the racist person who would be very willing to have it known he isn't voting for a black person. Think about how some groups tried to paint all McCain voters as racist. We know that is not true, but why hassle with some pollster about it?

 

The social desirability factor is well researched and accepted. This is just a variant on that. People lie to pollsters all the time, the best example is income.

[lostfan]

Tex, you're really twisting the logic into knots here and getting far, far away from what the original definition is. The Bradley Effect doesn't account for the opinions of the vast majority of non-racist McCain voters because it's assumed they are reporting their votes accurately to the pollsters. The theory suggests that people are secretly racist and lying about claiming they are voting for the black candidate when they know they won't, nothing more. When you add other factors into it then it just dies along with all the other statistical factors. In the states where it's believed race (where it would hurt Obama) was a factor, the polls were accurate.

[Disco72]

Actually if you look at the first post in the thread as well as the standard definition of social desireability, Tex is 100% correct. The "popular" definition of the Bradley Effect is racist voters, but the reality is that that is not true. Tex's explanation of voters saying they'd vote Obama but actually voting McCain is definitely what the Bradley Effect is supposed to be.

 

In either case, it seems that it was not an issue in this election (and perhaps previous elections). However, social desireability in all forms of polling, surveys, etc is a well-established bias that is alive and well. In this election, it was either accounted for or small enough to fall within the standard margin of error.

[lostfan]

Maybe we're really saying the same thing. In any case what I'm trying to say is that the most accurate polling model from this election cycle (538) was as close to dead-on as polls are ever going to get, and there was no Bradley Effect, or anything else. If anything, that "social desirability" translated into actual votes.

[Texsox]

Tex, you're really twisting the logic into knots here and getting far, far away from what the original definition is. The Bradley Effect doesn't account for the opinions of the vast majority of non-racist McCain voters because it's assumed they are reporting their votes accurately to the pollsters. The theory suggests that people are secretly racist and lying about claiming they are voting for the black candidate when they know they won't, nothing more. When you add other factors into it then it just dies along with all the other statistical factors. In the states where it's believed race (where it would hurt Obama) was a factor, the polls were accurate.

 

The theory is people will say they are voting for the black candidate because they believe it will make them seem more socially desirable to the pollster. They want the pollster to believe they are a "good person" and not swayed by race. A racist would not have that concern. The people that would have that concern are the people who are not racist but do not want anyone to think that they are. With all the talk about a vote for McCain is a vote for racism, some people would want to avoid that.

 

A racist person would not believe that he would appear more socially desirable by voting for a black candidate. In fact, they would believe it would make them less socially desirable.

 

The Bradley effect theorizes that the inaccurate polls were skewed by the phenomenon of social desirability bias.^[7][8] Specifically, some white voters give inaccurate polling responses for fear that, by stating their true preference, they will open themselves to criticism of racial motivation.

 

Who would fear that criticism? A racist? Or a non racist person? Did you ever try to reason with a racist person? Do you think they really care if they are criticized? They do not believe they are wrong, and in fact, it is the other people who are wrong and just do not understand.

[Disco72]

Maybe we're really saying the same thing. In any case what I'm trying to say is that the most accurate polling model from this election cycle (538) was as close to dead-on as polls are ever going to get, and there was no Bradley Effect, or anything else. If anything, that "social desirability" translated into actual votes.

 

Yes, we agree that the polls were accurate and thus, there was no noticeable Bradley Effect in this election. Like Tex, I was trying to clarify what the Bradley Effect actually is since it seems to be misreported or misunderstood by many people (not necessarily you lostfan).

[lostfan]

Yes, we agree that the polls were accurate and thus, there was no noticeable Bradley Effect in this election. Like Tex, I was trying to clarify what the Bradley Effect actually is since it seems to be misreported or misunderstood by many people (not necessarily you lostfan).

I probably did misunderstand because it kept getting repeated over and over and eventually I forgot what it actually meant.

[Texsox]

I probably did misunderstand because it kept getting repeated over and over and eventually I forgot what it actually meant.

 

A better way to think of this, and perhaps it is a good example, you bump into Athomeboy in a dark alley and he asks you who you are voting for. Of course you would have said Obama, even if you were voting McCain. Not because you were a racist, but because it is what you believe the pollster is wanting to hear.

 

I'm certain almost everyone has done this a time or two. A candidate knocks on your door during a Sox game. Damn right I'm voting for you, now get on down the road and reach those undedecided. I remember when Nielsen of the TV ratings was switching from pen and notebook reporting to computer based monitoring in homes. They mentioned that they adjusted their numbers when it was self reporting because X% of families will self report Disney and Masterpiece Theater when they were really watching WWF and Dukes of Hazard. They were excited about testing their hypothesis about how many were giving false information.

 

I find polling fascinating. The science keeps getting better. The results more accurate.

[lostfan]

I got you Tex, you're right, I just thought I was arguing against something else at first.

[Texsox]

I got you Tex, you're right, I just thought I was arguing against something else at first.

 

I guessed that. But racism is such a charged topic that I wanted to clarify. The fact that it is so charged, is exactly why their could be this kind of effect.

 

And we really do not know if

 

A. People were accurate in their responses

 

or

 

B. Pollsters accurately factored in for a Bradley Effect.

 

I'm am hoping that it is A and believe it probably is.