Last September or so there was a lot of headlines about a survey that claimed people in the US wanted a wealth distribution more like Sweden’s. OK, that was maybe not so surprising.
I recently got a reason to read the survey, and it is interesting. It asked a couple of questions, in the first one they got to choose between three unlabeled wealth distributions, where one was a perfectly even distribution (obvious even when unlabeled) with every quintile having 20% of the wealth, one was a wealth distribution where the bottom quintile had 0.1% of the wealth and the top quintile 84% of the wealth. And the third was a wealth distribution where the bottom quintile had 11% and the top 36%. They asked people which one they preferred.
Well, which one would you choose, if you don’t know anything about economics? Right, the middle one. The one with a completely equal wealth distribution is obviously one extreme. If you don’t know anything about the subject, you will automatically assume the 84% – 0.1% to be extreme as well. But it isn’t. That would be having 100% for the top quintile and 0%for all the others. They didn’t show any such chart, making the question completely biased from the start.
To make matters worse, of these three distributions, only one was real, the 84% – 0.1% distribution. It was the US wealth distribution. The perfectly equal one was obviously unreal, but the third one is also fictional. In the survey they claim it’s Sweden’s wealth distribution, but that is not true. In a footnote (yes a footnote) they admit that it’s not. It is Sweden’s income distribution. Although possibly that isn’t true either. According to data from SCB, the top 20% have 40% of the income and the bottom 20% have 4%of the income. Not 36% and 11% as Norton and Ariely claims.
However, the big problem here is of course that income and wealth distributions are completely different beasts and will look very different. They claim they chose the income distribution to have a bigger contrast, because although Sweden’s wealth distribution is more equal the the US wealth distribution it is still, according to them ”extremely top heavy”. Yes, because that’s how wealth distribution works. Sweden’s wealth distribution is 73% in the top quintile and 0% in the bottom one. And that’s the worlds most equal country.
So Norton and Ariely showed one real and two completely fictional and unrealistic income distributions and asked which one they preferred. The 8% that preferred the US income distribution to the fictional Swedish one are likely the 8% of people that know enough about economy or maths to notice the flaw.
What should Norton and Ariely have done here if they were serious? Well, they probably should have shown them several different income distributions taken from real countries, with two fictional (but not wildly unrealistic) extremes. Why income distributions? Because wealth distributions are by nature extreme. The people in the bottom quintile will not tend to save money, but to spend their income. They don’t really have money over to save. Hence, they will always fall close to zero. That looks much more unequal than what it really is. Income distribution is less unintuitive in that way. Maybe they could have had Sweden, US and Namibias income distributions, for example. There will likely always be a bias towards the middle in such a question, so it’s important to have US as the middle of the road to avoid a bias, and see when people then prefer, more or less income distribution.
Then they got asked what they though the US income distribution was, and then of course put the numbers somewhere in the middle of the option they choose (the unrealistic fabricated option) and the one they thought was extreme (US), proving that they didn’t know what they were being asked. And then they asked what they wanted it to be, and then wanted in average a slightly more equal distribution than they thought they had. In fact, slightly more equal than the completely unrealistic center option they had got, with the top quintile earning 36% of the total and the bottom quintile 11%.
These choices are both influenced by the false data they were fed in the first question, and an effect of the unintuitive maths here. This survey got Americans to claim that the average guy on the top 20% should earn about three times as much as the average guy in the bottom 20%, while in the most equal country in the world, Sweden, he in fact currently earns around ten times as much. This is not an expression of what people really want in a society, it’s an expression of that people won’t understand the issues if they are expressed in unclear maths after being feed misleading information.
I think this survey is interesting and you can learn a lot from it. It would be an excellent case study in how to not make surveys as they pretty much did everything wrong.
So what did this survey show? It shows that the maths involved are unintuitive. Most people in the lower 20% doesn’t save money. They will have very little of the wealth, because even if they have money over at the end of the month they spend it. That’s normal and expected, and every country will work like that, always. It also shows that people in the US believe that the US is more unequal than they would like. That’s it. But it doesn’t show any of the things they claim it shows, much less that Americans prefer Sweden’s income distribution. Maybe they do, maybe they don’t. But this survey doesn’t show it.
Lennart Regebro: Politik utifrån 