How politics makes us stupid.

[y66]

From How politics makes us stupid by Ezra Klein:

 

In the mid-20th century, the two major political parties were ideologically diverse. Democrats in the South were often more conservative than Republicans in the North. The strange jumble in political coalitions made disagreement easier. The other party wasn’t so threatening because it included lots of people you agreed with. Today, however, the parties have sorted by ideology, and now neither the House nor the Senate has any Democrats who are more conservative than any Republicans, or vice versa. This sorting has made the tribal pull of the two parties much more powerful because the other party now exists as a clear enemy.

 

One consequence of this is that Washington has become a machine for making identity-protective cognition easier. Each party has its allied think tanks, its go-to experts, its favored magazines, its friendly blogs, its sympathetic pundits, its determined activists, its ideological moneymen. Both the professionals and the committed volunteers who make up the party machinery are members of social circles, Twitter worlds, Facebook groups, workplaces, and many other ecosystems that would make life very unpleasant for them if they strayed too far from the faith. And so these institutions end up employing a lot of very smart, very sincere people whose formidable intelligence makes certain that they typically stay in line. To do anything else would be upend their day-to-day lives.

 

The problem, of course, is that these people are also affecting, and in some cases controlling, the levers of government. And this, Kahan says, is where identity-protective cognition gets dangerous. What’s sensible for individuals can be deadly for groups. "Although it is effectively costless for any individual to form a perception of climate-change risk that is wrong but culturally congenial, it is very harmful to collective welfare for individuals in aggregate to form beliefs this way," Kahan writes. The ice caps don’t care if it’s rational for us to worry about our friendships. If the world keeps warming, they’re going to melt regardless of how good our individual reasons for doing nothing are.

 

To spend much time with Kahan’s research is to stare into a kind of intellectual abyss. If the work of gathering evidence and reasoning through thorny, polarizing political questions is actually the process by which we trick ourselves into finding the answers we want, then what’s the right way to search for answers? How can we know the answers we come up with, no matter how well-intentioned, aren’t just more motivated cognition? How can we know the experts we’re relying on haven’t subtly biased their answers, too? How can I know that this article isn’t a form of identity protection? Kahan’s research tells us we can’t trust our own reason. How do we reason our way out of that?

 

The place to start, I figured, was talking to Dan Kahan. I expected a conversation with an intellectual nihilist. But Kahan doesn’t sound like a creature of the abyss. He sounds like, well, what he is: a Harvard-educated lawyer who clerked for Thurgood Marshall on the Supreme Court and now teaches at Yale Law School. He sounds like a guy who has lived his adult life excelling in institutions dedicated to the idea that men and women of learning can solve society’s hardest problems and raise its next generation of leaders. And when we spoke, he seemed uncomfortable with his findings. Unlike many academics who want to emphasize the import of their work, he seemed to want to play it down.

 

...

 

At one point in our interview Kahan does stare over the abyss, if only for a moment. He recalls a dissent written by Supreme Court Justice Antonin Scalia in a case about overcrowding in California prisons. Scalia dismissed the evidentiary findings of a lower court as motivated by policy preferences. "I find it really demoralizing, but I think some people just view empirical evidence as a kind of device," Kahan says.

 

But Scalia’s comments were perfectly predictable given everything Kahan had found. Scalia is a highly ideological, tremendously intelligent individual with a very strong attachment to conservative politics. He’s the kind of identity-protector who has publicly said he stopped subscribing to the Washington Post because he "just couldn’t handle it anymore," and so he now cocoons himself in the more congenial pages of the Washington Times and the Wall Street Journal. Isn’t it the case, I asked Kahan, that everything he’s found would predict that Scalia would convince himself of whatever he needed to think to get to the answers he wanted?

 

The question seemed to rattle Kahan a bit. "The conditions that make a person subject to that way of looking at the evidence," he said slowly, "are things that should be viewed as really terrifying, threatening influences in American life. That’s what threatens the possibility of having democratic politics enlightened by evidence."

 

The threat is real. Washington is a bitter war between two well-funded, sharply-defined tribes that have their own machines for generating evidence and their own enforcers of orthodoxy. It’s a perfect storm for making smart people very stupid.

 

The silver lining is that politics doesn’t just take place in Washington. The point of politics is policy. And most people don’t experience policy as a political argument. They experience it as a tax bill, or a health insurance card, or a deployment. And, ultimately, there’s no spin effective enough to persuade Americans to ignore a cratering economy, or skyrocketing health-care costs, or a failing war. A political movement that fools itself into crafting national policy based on bad evidence is a political movement that will, sooner or later, face a reckoning at the polls.

 

At least, that’s the hope. But that’s not true on issues, like climate change, where action is needed quickly to prevent a disaster that will happen slowly. There, the reckoning will be for future generations to face. And it’s not true when American politics becomes so warped by gerrymandering, big money, and congressional dysfunction that voters can’t figure out who to blame for the state of the country. If American politics is going to improve, it will be better structures, not better arguments, that win the day.

[kenberg]

The excerpt shows Scalia, a conservative, supposedly letting ideology trump evidence. True enought, but the report seems to find the problem pervasive. It could be a useful exercise to set out to find examples where views you agree with are supported by arguments that, if you put politics aside, you would find doubtful. it's easy enough to find examples, accurate or not, of careless thinking by people you disagree with.

 

 

Here is another portion of the article [the first part refers to a skin cream "study"]. It matches what I see in everyday life.

 

 

This kind of problem is used in social science experiments to test people's abilities to slow down and consider the evidence arrayed before them. It forces subjects to suppress their impulse to go with what looks right and instead do the difficult mental work of figuring out what is right. In Kahan's sample, most people failed. This was true for both liberals and conservatives. The exceptions, predictably, were the people who had shown themselves unusually good at math: they tended to get the problem right. These results support the Science Comprehension Thesis: the better subjects were at math, the more likely they were to stop, work through the evidence, and find the right answer.

 

But Kahan and his coauthors also drafted a politicized version of the problem. This version used the same numbers as the skin-cream question, but instead of being about skin creams, the narrative set-up focused on a proposal to ban people from carrying concealed handguns in public. The 2x2 box now compared crime data in the cities that banned handguns against crime data in the cities that didn't. In some cases, the numbers, properly calculated, showed that the ban had worked to cut crime. In others, the numbers showed it had failed.

 

Presented with this problem a funny thing happened: how good subjects were at math stopped predicting how well they did on the test. Now it was ideology that drove the answers. Liberals were extremely good at solving the problem when doing so proved that gun-control legislation reduced crime. But when presented with the version of the problem that suggested gun control had failed, their math skills stopped mattering. They tended to get the problem wrong no matter how good they were at math. Conservatives exhibited the same pattern — just in reverse.

 

Added: The site http://test.causeweb.org/wiki/chance/index.php/Chance_News_98

quotes Andrew Gelman along the same lines:

 

"In statistics it's enough for our results to be cool. In psychology they're supposed to be correct. In economics they're supposed to be correct and consistent with your ideology."

[Vampyr]

Here is another portion of the article [the first part refers to a skin cream "study"]. It matches what I see in everyday life.

 

The skin cream thing was really strange. Even if you were not very good at math, how would you approach the question other than by attempting at least a rough idea of the ratios?

[Winstonm]

Thank goodness I'm always right.

[kenberg]

The skin cream thing was really strange. Even if you were not very good at math, how would you approach the question other than by attempting at least a rough idea of the ratios?

 

How indeed? But I have very little trouble believing the results. Even before the advent of calculators, doing simple arithmetic was never really the stumbling block. I think every child in my 8th grade class (by no means in a school for the mathematically talented) could do the arithmetic. What often is not learned is how to see what calculation is relevant.

 

One of my favorite stories, no doubt I told it before, involved a place like McDonalds. In the line next to me a woman is told that her order costs, let's say, $13.57. She gives the cashier a bill, the cashier rings it up, reads the result on the register, and gives the woman $6.43 in change. The woman says "I didn't give you a $20, I gave you a $100 bill". The cashier checks the drawer and apologizes, indeed there is a $100 bill in the drawer. Now what? No one knows how to cancel the orignal entry of $20 in the register. No one knows what to do. The cashier, a young girl, summons an older lady. No idea. A guy is cooking the fries, guys are said to be better at math, they ask him. Nope, not this guy. I usually mind my own business but they need help. I explain that since the customer has given the cashier $80 more than was entered into the register, she should get $80 more in change than what the register reads. Whew. Everyone agrees and the day moves forward. Everyone there, I am sure, could subtract 20 from 100 and get 80. It did not occur to anyone to do so.

[barmar]

I don't find the skin cream versus gun control results very surprising. One of them is emotionally laden, and we're less likely to behave rationally in that mode. There's no skin in the game (pardon the pun) in the first test.

[billw55]

On the skin cream thing, they should have presented a third option, that the cream had no effect. Personally I didn't think the data were very convincing either way. But the question was: did the cream make the rash better or worse? - committing the fallacy of false dichotomy. I wonder if they presented the gun control part of the experiment in a similar fashion?

[barmar]

On the skin cream thing, they should have presented a third option, that the cream had no effect. Personally I didn't think the data were very convincing either way. But the question was: did the cream make the rash better or worse? - committing the fallacy of false dichotomy. I wonder if they presented the gun control part of the experiment in a similar fashion?

The question wasn't about causality, it was about correlation. They didn't ask "did the cream make the rash better or worse?", they asked "Were people who used the create more likely to get better or worse than those who didn't?" Since there were no patients whose rash stayed exactly the same, it wouldn't make any sense to include this as an option when the question is worded this way. You're the one who made the fallacy of assuming correlation equals causation.

 

Of course, so did the participants in the study. When they were given the version that involves gun control, they automatically translated the correlation to causation. Then they allowed their political bias to cloud their response to the question about correlation.

[kenberg]

I see it as simple: Looking at the table of results, would you be more inclined or less inclined to use the cream. There could be all sorts of things wrong with the cream study or the gun study, but in so far as you plan to make use of the results shown in the table, in which direction would it influence you?

 

Apparently, and I agree with bar that it is not so surprising, if there is ideology involved then people tend to interpret the table in whichever way it is that favors their ideological position, and this includes people who are capable of making a sensible interpretation of the table when it is about hand cream. And, I gather, the article is saying that it doesn't really matter which ideology you favor. Your ideology distorts your ability to make sensible use of the table.

 

To put it another way, a person presented with a table might say "I am not even going to look at it, much less tink about it, that's not how I decide". At least on some subjects, I might agree with this approach. It's an entirely different thing to look at the table and no matter what is there, decide that it favors your position. I certainly know people who think this way. And I think we all can fall from grace on occasion.

[billw55]

The question wasn't about causality, it was about correlation. They didn't ask "did the cream make the rash better or worse?", they asked "Were people who used the create more likely to get better or worse than those who didn't?" Since there were no patients whose rash stayed exactly the same, it wouldn't make any sense to include this as an option when the question is worded this way. You're the one who made the fallacy of assuming correlation equals causation.

 

Of course, so did the participants in the study. When they were given the version that involves gun control, they automatically translated the correlation to causation. Then they allowed their political bias to cloud their response to the question about correlation.

I quote:

 

Please indicate whether the experiment shows that using the new cream is likely to make the skin condition better or worse.

(emphasis mine)

[kenberg]

I suppose I agree that they could have posed the questions better, but I doubt that that is the explanation for what happened.

 

Let's start with the skin cream where we have

 

****************Better ******** Worse

 

Use ************ 223 ***********75

 

Don't use ******* 107 ********** 21

 

One could reasonably complain that "Don't use" says only what they did not do, but doesn't say what they did. Probably most people would take it as meaning "Did nothing, just let it run its course". Maybe not, but probably this is how they took it.

 

 

So I gather that quite a few saw that 223 is larger than 75 and said the cream is great. You would hope that they would say that those who did nothing got better in about 5 out of 6 cases and worse in 1 out of 6, while those who applied the cream got better in about 3 out of 4 cases and worse in about 1 out of 4. A reasonable conclusion is that, at least as far the cream figures are accurate, it seems that the cream actually interferes with recovery. Most people don't think much here about correlation versus causation, nor do I think they should here unless there are further studies of causal issues with the cream. A sensible person would look at the table and say "Apparently just leaving it run its course is a better idea than using the cream".

 

But the real point of the study/article, as I get it, is not how people go at this decision making with creams but rather that they change their approach when they have a strong interest that the conclusion come out one way or the other. A person with a rash wants it to go away, his sole interest is in choosing the approach most likely to make this happen. People examining data on the effects of gun laws want a study that will show that they are right. If they favor gun control they want a study to show that it reduces gun violence (fewer guns). If they favor gun rights, they want the study to show that allowingpeople to carry guns reduces violence (through concern of a criminal that an intended victim might be armed).

 

Perhaps there will be further studies on this, it perhaps could be useful, but I am with barmar in not finding this surprising.

 

A long time ago I heard a debate on NPR where exactly this sort of thing arose. I forget the details but the subject matter was how to treat felons. Two people were arguing from the same data drawing totally different conclusions. One was saying that such and such approach helped in 50% of the cases, the other person had very different percentages from the same data. I faulted the host for not speaking up and saying something such as "It cannot be the case that both of your percentages describe the same situation. Please, each of you, tell me what you are dividing by what to get these figures". My guess is that neither one of them would have been able to answer. They had their percentages, they repeated their numbers loudly and endlessly, they never gave any indication that they had any idea where the percentages actually came from.

 

If you want a certain conclusion, you can always find something to divide by something to support that conclusion.

[billw55]

lots of other stuff and:

 

So I gather that quite a few saw that 223 is larger than 75 and said the cream is great. You would hope that they would say that those who did nothing got better in about 5 out of 6 cases and worse in 1 out of 6, while those who applied the cream got better in about 3 out of 4 cases and worse in about 1 out of 4. A reasonable conclusion is that, at least as far the cream figures are accurate, it seems that the cream actually interferes with recovery.

Some people will reach this conclusion, which could be reasonable. I was thinking that at this sample size, the difference between 5 out of 6 and 3 out of 4 does not sound statistically significant. Hence, I would conclude that no effect of any kind has been demonstrated, for better or worse (although a negative effect has not been ruled out).

 

Although my sense of statistics might be off, maybe someone can run the actual math and come up with a figure for significance here? I admit I often have a skeptical bias and tend to think "that doesn't prove anything at all" perhaps more often than is really the case. I would have very much liked to take the other test - with fake gun control data - to see how I fare on it.

 

But your overall point is valid. Rational persuasion (i.e. based on data) is futile with a substantial majority of people. Emotional persuasion, on the other hand, is much more effective. I find this kind of sad.

[kenberg]

I am not a statistician, but generally I think that a posterior statistical analysis is not really kosher. It falls into the "I was just thinking about you when you called, we must be connected by esp" trap. Unusual things happen all the time. So going around and finding tables of data, and then reading significance ito those tables that seem off of average, is not the way to go.

 

You can say something such as "If the statistician designed this experiment to test the null hypothesis that the cream has no effect, would he then have rejected it at, say, the 5% level?". That sort of calculation could be done but I think I won't.

 

Helene, or some other statistician, may want to explain that I don't know what I am talking about here, but I think I am at least sort of right.

[barmar]

I quote:

Oops. The test phrased the question two different ways, and I only saw one of them. You quoted the question above the table. Below the table, it says:

What result does the study support?

• People who used the skin cream were more likely to get better than those who didn't.
  
• People who used the skin cream were more likely to get worse than those who didn't.
  

So I guess the testers were guilty of causing the participants to conflate correlation and causation. But I'll bet they would have gotten the same results without this priming, because it's already well known that people do this. Statisticians are a little better at avoiding it, but lay people rarely do.

[blackshoe]

How indeed? But I have very little trouble believing the results. Even before the advent of calculators, doing simple arithmetic was never really the stumbling block. I think every child in my 8th grade class (by no means in a school for the mathematically talented) could do the arithmetic. What often is not learned is how to see what calculation is relevant.

 

One of my favorite stories, no doubt I told it before, involved a place like McDonalds. In the line next to me a woman is told that her order costs, let's say, $13.57. She gives the cashier a bill, the cashier rings it up, reads the result on the register, and gives the woman $6.43 in change. The woman says "I didn't give you a $20, I gave you a $100 bill". The cashier checks the drawer and apologizes, indeed there is a $100 bill in the drawer. Now what? No one knows how to cancel the orignal entry of $20 in the register. No one knows what to do. The cashier, a young girl, summons an older lady. No idea. A guy is cooking the fries, guys are said to be better at math, they ask him. Nope, not this guy. I usually mind my own business but they need help. I explain that since the customer has given the cashier $80 more than was entered into the register, she should get $80 more in change than what the register reads. Whew. Everyone agrees and the day moves forward. Everyone there, I am sure, could subtract 20 from 100 and get 80. It did not occur to anyone to do so.

Heh. I too have a MacDonald's story. I was in Albuquerque, so it had to be early to mid 70s. The local MacD's had just got some new-fangled cash register that did all your thinking for you. Problem was, it wasn't working. The probably high school aged girl at the register had to hope, I guess, that she could figure it out. My bill was something like $6.24. I gave her a ten, a one, and a quarter. She looked at me. She looked at the cash register. She looked at the money in her hand. Then she looked back at me and started to cry. I told her, gently, to just give me a five and a penny. She looked dubious, but she did it. I still don't understand how anyone can get through - or almost through - high school and not be able to figure that out.

[barmar]

Just so you know that not all hope is lost, last night I stopped at my favorite tacqueria on my way home from the bridge club, for my weekly burrito dinner. The bill was $9.10, and I gave the young woman, who (like most employees there) speaks with an accent, $20.10. She entered $20 into the Cash Tendered amount in the register, so it told her to give $10.90 in change. But she was astute enough to add 10 cents this and give me the proper $11 change.

[kenberg]

Just so you know that not all hope is lost, last night I stopped at my favorite tacqueria on my way home from the bridge club, for my weekly burrito dinner. The bill was $9.10, and I gave the young woman, who (like most employees there) speaks with an accent, $20.10. She entered $20 into the Cash Tendered amount in the register, so it told her to give $10.90 in change. But she was astute enough to add 10 cents this and give me the proper $11 change.

 

It sounds as if she already knew she would be giving you back the $10 and just entered the 20 to get the stupid register to open up. It's always good to hear such things. I suppose the accent was appropriate for the burrito?

 

I have to think that emotions, attitude, confidence, call it what you will, plays a role here. Or perhaps sort of "It's all in the way you were brought up". I have a friend who often has had me help him with really extremely simple things on the computer.Ok, computers can be tricky little devils. But he absolutely refuses to allow me to show him how to do it, he insists that he cannot possibly understand, and the only solution is for me, or someone, to do it for him. Of course he could learn to do it, he insists that he cannot. it's not exactly simply a way of getting others to do things for him, although sometimes iot feels like that. Rather it is a self-image that he insists on keeping intact.

 

Anyway, good for her. My bet is that her future will soon move beyond selling burritos.

[barmar]

It sounds as if she already knew she would be giving you back the $10 and just entered the 20 to get the stupid register to open up. It's always good to hear such things. I suppose the accent was appropriate for the burrito?

It's on the border between Cambridge and Somerville, MA. I think there's a big Portugese community in this area, not Latin American (hrothgar could presumably back me up or correct me). I'm not sure I can distinguish the accents.

I have to think that emotions, attitude, confidence, call it what you will, plays a role here. Or perhaps sort of "It's all in the way you were brought up". I have a friend who often has had me help him with really extremely simple things on the computer.Ok, computers can be tricky little devils. But he absolutely refuses to allow me to show him how to do it, he insists that he cannot possibly understand, and the only solution is for me, or someone, to do it for him. Of course he could learn to do it, he insists that he cannot. it's not exactly simply a way of getting others to do things for him, although sometimes iot feels like that. Rather it is a self-image that he insists on keeping intact.

Very possible. My sister is like that, too. She's as smart as anyone else, but totally hates any technology more complicated than a TV or basic cellphone. She had a laptop computer, but whenever something would go wrong with it she would just put it aside and not bother fixing it for months, because she didn't like using it in the first place. She refuses to even try to learn anything more about using the PC at work than she absolutely needs to get her job done, because she insists that she can't understand it.