I think the implication that "blacks aren't real Americans" might be a bit too strong for interpretation. I got a "they like him too much to be objective about his policies", which is less directly offensive but just as patronizing. I never saw articles about how GW Bush's high rating among Texans means they identify with him too much to be objective, but identity politics has been the Republican game for at least the past decade.

Jesus, this is a ridiculous interpretation.

Have you taken a statistics class? The mean of a set of data may be skewed by outliers. When we ask "how do people view Obama's policy on X", the mean is usually a decent approximation to what the polity believes. However, if the populace is polarized on opinion of Obama (as the article asserts), using the mean as the answer may not agree with the majority's opinion of the policy.

That's all. No malicious intent. And please take a stats class.

I'd agree with both Beth and Duran. The implication of the quote is that African Americans, as a subset of the total polled set, skew the results because theirs are so far outside a norm established by the set outside that subset.

In other words, his rating within that group is higher than among the population-at-large, which introduces an error in the overall average. That group could be black, white, short, tall, brown-eyed or blue. They're still members of the larger set ("real" Americans), but distinct for their statistical anomaly in this poll question.

Duran and Gomi: the article says that African-American mess up the poll and make Obama's positions seem "more popular overall than they actually are." Despite what you say that claim is probably false.

If I have 10 kids, 9 of which weigh 100 pounds and 1 of which weighs 500 pounds, their average weight would be 140. Now, you could say that the last kid "skews" the average.

However, that wouldn't change the average. It would make no sense to say that the 10th kid makes their average weight seem bigger than it actually is. The average weight of all ten kids STILL IS 140.

I could say that some segment "skews" the popularity poll but could not at all conclude that he is not as popular as the polls show--at least not without excluding that segment of the population from the poll.

I think Byron York's response to this criticism is pretty good:

"Those numbers raise a question: What if a president were wildly popular with one group, and only middlingly popular with another group and yet was often portrayed as being hugely popular with the whole group? It seems worthwhile to point that out that there are differences within the group -- something that is done all the time with political polls. The president's job approval ratings are what they are -- the Times had him at 69 percent approval -- but the numbers inside the numbers are striking."

I have to agree with Lisa that the article implies blacks are not "average ameicans".

Clearly, not many in power (public or corporate) think black americans are average americans. You can see this by their corporate messages, their "diversity" programs.

If this poll is correct, Obama is more popular in america thanks in part to one groups overwhelming support. But lets not forget that that group has drastically less influence on american media, policy and just about everything else.

So the poll does have a point in that - although average americans are supportive, the people who's voices actually count, are not as supportive.

i just don't understand what picture of america is left over after excluding african american opinion. isn't that equivalent to saying "among non-black americans" why is the opinion of "non-black americans" equated with the Obama's "actual" popularity. what does being "actual" have to do with being "non-black"

everyone's vote is "actual."
so his "actual popularity" includes african american opinions, regardless of whether they like him or not.

I should clarify that, even though the president is black; and even considering the many glass ceilings that have been shattered by non-white-men; It is my belief that land owning white males still have most of the power in America. That is who this story was speaking to.

For those people in that seat of power, black america is not important.

If Hilary had won, women would be described as ruining the average.

"If Hilary had won, women would be described as ruining the average"

but what does "average" mean, when you exclude half the population? it would be more honest to say -- "among men," or "among non-white people."

it is insulting to say that "among men" or "among non-white people" equals average.

Average means "after taking everyone into account..."

another analogy:

it's like trying to calculate the "average" hair length of a class full of kids, and excluding kids with curly hair (because they disrupt the "average")

the resulting "average" is just the mean length of "straight people's hair." Publishing that info in a peer-reviewed hair journal would mislead readers about the actual average length of hair in classrooms.

@MissivesFromMarx

I agree that the average weight would be 140.

However, the implication behind political polls is NOT "average opinion". If 99% of the populace slightly disagrees with proposition X, and 1% very, very, very strongly agrees with X, we would not in any way say that the population supports proposition X.

This is what happens when people use imprecise verbiage to describe mathematical concepts. Someone always ends up pulling the PC card.

@Evan

Blacks as an ethnic group are not average in most interesting demographic statistics. Wow. You even question that? Look at poverty rates, education levels, crime rate, incarceration rate, locality where they live, music and media tastes, political preferences, etc. The differences go on and on.

@MissivesFromMarx,
I'd repeat what Duran said in response, but also this:

Using your example, let's make another kid 500lbs, just to illustrate it more. Say you're a teacher with these ten kids as your class. You tell a fellow teacher: "My students are 180lbs, on average!" That teacher is probably going to ask you what they're eating for lunch, that they're all so heavy. But, in reality, only 20% of your class is heavy, but monstrously so.

The perception of general weight in your class is skewed by a subset of the population who's outside the norm of the superset.

A particularly high opinion of the President among a subset of the respondents can create a higher perception of opinion among the total respondents. But that doesn't mean that subset aren't "real Americans" or even imply that. It just says that their response is outside a general norm of the rest of the poll, skewing the perceived perception presented by the poll.

@ Duran "the differences go on"

you make my point.

That’s all. No malicious intent. And please take a stats class.

This might be the stupidest thing I've ever read on this blog, and a decent amount of stupid stuff comes up here.

If Obama has a 66% approval rating, then Obama has a 66% approval rating. He "actually" has a 66% approval rating. If you feel compelled to interpret this as a "real" approval rating of 60% [from white people], which gets adjusted upwards to 66% because of "outliers" [black people], then you are doing exactly what Lisa is saying in the post: treating black Americans as not-real, not-actual Americans. You are not being statistically sophisticated, you dolt.

There are other ways to slice the data. For example, you can say that Obama has 100% approval from people who approve of him, and 0% approval from people who don't approve of him. I'm not sure which of these numbers is more real, maybe we can do some sort of "average" to come up with an overall approval rating. Also, African-Americans support Democrats by more than 9-to-1 in general, so black support for Obama is not especially out of line compared to black support for Clinton, Gore, and Kerry.

Maybe the author didn't mean to say that black americans aren't real americans, but the choice of words for "actually are" was awful. I can see that the reporter could be trying to show the numbers inside the numbers, as it was said, which is important because just looking at averages can be misleading, but the words used were really, really dubious.

Black people are not outliers. They make up a significant portion of the population.

@Beth T: we didn't hear about how much Texans loved George Bush, but we did hear a whole lot (which polls) about how much evangelicals loved him.

@Ellen: "Black people are not outliers. They make up a significant portion of the population", 12% according to the most recent census. But the article isn't implying that their view is automatically outlier because they are a minority group, but because the percentage of people within that group who have a positive opinion of Obama is an outlier in relation to other groups (which does go beyond "rich white dudes" and also includes hispanics, asians and women).

It is insulting that much of the coverage of 100 days approval coverage ends up suggesting that black people are loyal to Obama even when he's doing poorly out of loyalty, because he is black.

But it makes me uncomfortable that there is also a counter assumption that when people of other races (usually white, since in discussions of race during this election the black-white dichotomy was played up and latinos and asians were often ignored) disapprove of his performance that it is based on his race, rather then his refusal to investigate torture, coverups of information about the cheney torture regime, and his constant caving to the radical right wing extremists in the GOP.

Look at poverty rates, education levels, crime rate, incarceration rate, locality where they live, music and media tastes, political preferences, etc. The differences go on and on.

You have to love how this kind of "different" shit appears when blacks are singled out from the "real" Americans in statistics about voting (you have to know the racial breakdown of the respondents, doncha know?) but once a minority group starts pointing out the statistical differences between their *hiring* rates versus those of whites suddenly everyone's "colorblind" again.

But the article isn’t implying that their view is automatically outlier because they are a minority group, but because the percentage of people within that group who have a positive opinion of Obama is an outlier in relation to other groups (which does go beyond “rich white dudes” and also includes hispanics, asians and women).

You can do whatever you want with data. My entire family and all of my friends have a positive view of Obama too. Maybe we're a group that should be excluded so we don't "skew" the data? Or maybe we should leave out the poor people? I'll bet that the data would look a little more like "real" America without all the riffraff cluttering it up! Maybe we should leave all the racists out? Bet Obama's approval would go up a bit without those people skewing it down.

My point is that you can group the population however you want, and make the data say whatever you want. The kinds of data massaging that we *accept* says more about us than the poll numbers do. And our easy acceptance of data sorted by race makes that whole "colorblind liberal" persona we like so much look like bullshit.

@ Those impugning Lisa's stat skills

"Outlier" is a technical - and relative - term. An observation is only an outlier if it keeps us from accurately estimating the statistic we want to estimate. But keep in mind here we are simply averaging a nationally representative sample of a dichotomous outcome. Every individual is either "approve" or "disapprove." There can be no outliers here - which would be the equivalent of "super-dooper totally approve." This is why the comparison to children's weights is not relevant. If nine out of ten children want to eat lunch, the one child who doesn't want to eat lunch isn't an outlier. That child is a legitimate datum. And the real statistic is 90%.

So do you want to argue that you were calling a group an outlier, not an individual observation? Then perform a logistic regression with some controls, then write back to me. Consider: blacks are overwhelmingly Democratic. And guess what? His black approval rating is only marginally higher than his Democratic approval rating overall.

But does it even make sense to call a group an outlier when that group is part of the population whose opinion we are measuring? Of course not. We can draw any number of artificial boundaries to show that Obama is "less popular than he seems." The black approval rating is also about as high as his approval rating in the Northeast. His black approval rating is also just about as high as his approval rating with 18-25 year olds. And I hear he's pretty popular in urban areas too! So let's drop this whole stupid debate about who is a "real" or "representative" American.

And those of you who think you have something over Lisa should stop acting like you're the second coming of Sir Francis Galton.

I'm confused by some of the comments here about skewing, particularly the ones with the weight examples. Am I wrong in thinking that the poll simply asks, "Do you approve of Obama's policies?" (maybe broken down into specific questions), and then takes the simple yes and no answers and turns them into an approval rating?

How then can a 'strong' approval skew the results? I didn't think the approval rating was weighted by strength of feeling.

"Every individual is either “approve” or “disapprove.” There can be no outliers here - which would be the equivalent of “super-dooper totally approve.” This is why the comparison to children’s weights is not relevant. If nine out of ten children want to eat lunch, the one child who doesn’t want to eat lunch isn’t an outlier. That child is a legitimate datum. And the real statistic is 90%."

YES THANK YOU. THIS.

I am so glad somebody finally said that. I thought I was going crazy reading all the comments and seeing that nobody had pointed this out.

@opminded quoting York's response. When I blogged this, I checked the Times poll. The overall approval rating was 68%. Among blacks it was 96%, among whites 62%. In his original article, York gave much lower numbers (49% among whites), which I could not find in the Times poll he referenced. Sixty-two percent seems to me better than "only middlingly popular" (York's phrase in the excerpt opminded gave). But as I and others pointed out, what is most troubling about York's article is his assumption contained in the word "actually." It's one thing to point out that blacks and whites give different approval ratings. It's quite another to say that the surveys which include blacks are not "actual" ratings.

Jay and Carl, YES! EXACTLY!

Its all the magic of statistics. Now, saying he is not 'actually' as popular is ridiculous. If you are counting all of the country, then you count all of the country. Now, you could say that the numbers are driven by a huge disparity between group X and group Y or some such.
Also, typical and average ARE NOT the same. Say you want an average $ spent on clothes by New Yorkers. You pick 21 people and 20 of the spend $1,000 each month. The 4th person is Elton John (let's say is in the sample) and he spends $190,000 per month. Well, your 21 people spent $210,000 for $10,000 per month. so the AVERAGE is $10,000 per month IF you believe you have a representative sample.

BUT, per your sample, the typical New Yorker spends 1/10th of that ($1,000). 95.2% of the sample spent $1,000.

@ grad student

You use a continuous variable ($ spent) to make your point. Of course outliers exist when we are using continuous variables. But continuous variables are irrelevant here, because average approval rating is calculated from a dichotomous variable.

Dichotomous variables don't have outliers.

About the "disproportionate crime rate vs black population, so yay for racial profiling" argument made by Duran - we know how mixed race people don't get counted in the census and that most black people in this country are actually not 100% black. It could be that when a mixed race "black" person gets arrested for a crime they are counted as "black" due to racial profiling, but when it comes to the census the same people are excluded for being multiracial. So that can explain why the "rates" are disproportional. I never really considered that until now, but I always knew there was something fishy about the "crimes disproportionate to their population" argument. I mean, racial profiling advocates say that it's "disproportionate", indicating that the census is wrong.

Isn't the implication that Black people are voting for Obama because of his skin colour not his policies, and this makes Obama's policies seem more popular than they would if they weren't associated with him?

"The implication here is, of course, that Black Americans aren’t “real” Americans and that including them in opinion poll data is literally skewing the results."

We get that you're trying to be provocative but you're off target completely. The author is simply pointing out that within the overall approval rating there was a dichotomy between blacks and whites. There is no implication, other than in some sociologists' minds, that blacks aren't real Americans.

And most of the links in your post are broken.

Or perhaps the article is saying that Black Americans are far more likely to approve of Obama than White Americans. Having a black man in office has misled many into beliecing that society is more accepting and equal when it is not.Obama's not particularly well suited for the oval office or to promote any kind of necessary change and thisr who have scrutinized his actions know this. Black people seem to just be happy that a charismatic half black man is in office and see that as a triumph, thus overlooking his actual performance. Society has a long way to go before it reaches a point where ethnicity no longer determines how someone is treated and judged. That fundementsl shift had yet to occur. And I am speaking of biases held about everyone.

Your post is like a light in the dark. It's like finding the right path when everything seems confusing.
Chick Fil A

This post is very informative as well people can look at the other restaurant menus which is available.
Daftar menu mixue