• Blog

    The Return of Dan Ariely: The Survey Results Are In

    by Adam Taggart

    Monday, February 7, 2011, 7:32 PM

Nearly a month ago, friend-of-the-site and renowned behavioral economics researcher Dan Ariely invited our readers to take part in one of his experiments. Readers were asked to fill out a survey that asked questions about a hypothetical country and the manner in which they would prefer to see wealth distributed within it.

Nearly one thousand of you completed the survey and the ensuing discussion on the site was animated, to say the least. Many railed against the injustice of confiscatory tax codes and the lack of social mobility proposed within the study.

Well, the results are in, and Dan explains below what he was testing for, how the survey was structured (Surprise! Many of you saw different versions than your peers), and the conclusions you helped his team reach regarding people’s perception of wealth inequality. Suffice it to say, you’re an admirably fair-minded bunch.  Here are the results:


Thanks to all who took part in our survey a few weeks ago about wealth inequality. We received 937 complete responses, and from the discussion board it looks like many of you had strong opinions on this important topic. Now, as promised, here are the results.

The Premise

As you recall, this survey was based on philosopher John Rawls’ concept of justice. Rawls famously defined “justice as fairness,” and proposed the “veil of ignorance” thought experiment to help determine what constitutes true social justice. The thought experiment goes like this:

Instead of trying to consider what society should look like in an abstract sense (i.e., for everyone else), you will decide what kind of society you personally would want to live in. 

Here’s the catch: before you decide, you must don a veil of ignorance that hides from you all information about your personal characteristics and situation. That is, you might be wealthy beyond your wildest dreams or in complete poverty, predisposed to good health or have a severe disability, but you won’t find out until after you design your society.

Think of it like this — the stork is about to randomly drop you somewhere in the world, and you have to choose, mid-flight, what kind of a world you want to be dropped into.

What this means to you is this:  You must decide on a world that you will be happy to live in no matter where you end up. 

The Experimental Design (What You Knew)

Rawls’ theory can be used to consider almost any type of moral issue, but we were primarily interested in your opinions of wealth inequality. We gave nine different pie charts, each split into a maximum of five slices to represent quintile distributions of wealth. The pie charts ranged from very equal to very unequal. We can’t precisely say perfectly unequal or perfectly equal since we’re dealing with quintiles, but that’s the idea. 

The first pie chart had five equal slices, meaning that each 20% of the population has exactly 20% of the total wealth. The next seven charts had increasingly unequal slices. In the second chart, for example, the top 20% of people had 30% of the wealth, and the bottom quintile had 15%. The ninth pie chart was as unequal as possible — the top 20% had 100% of the wealth, and the other 80% had nothing. 

Most of these charts were made up so that we could gather opinions about the full range of possible wealth distributions, but a few were based on real data from different countries. The most equal wealth distribution data from any country belongs to Spain, which resembles distribution #5 in this study. Wealth distribution data is incredibly hard to measure and to obtain, so Spain may or may not really be the most equal country in the world. Distribution #6 below looks like the wealth distribution in Norway, and Distribution #7 resembles that in the United States. 

You rated each distribution on a sliding scale from 0 – 100, indicating how much you would or wouldn’t want your country’s wealth distribution to look similar to each example.  A rating of 0 meant “I don’t want it to look like this,” and a rating of 100 meant “I want it to look like this.”

The Experimental Design Continued (What You Didn’t Know)

All of you rated the same nine distributions, based on the prompt that you were about to join a new country and would be randomly assigned to some place in the country’s wealth distribution. 

For some of you, this is all you were told about the new country. This was the control group. Now the other 5/6ths of you received a little bit more information about the new country.

We manipulated two variables between subjects: 1) the degree of social mobility in the new country, and 2) the effects of redistribution on taxes. 

Variable #1: Social Mobility

This variable had three different levels, so you received one of these three variations:

Variation 1: There is Social Mobility

In the social mobility condition, we said this about the new country:

There is social mobility, so wherever you start after being randomly assigned is not necessarily where you will stay for the rest of your life.  Based on your skills, abilities, and opportunities, your situation can change.

Variation 2:  There is No Social Mobility

In the no-mobility condition, we said this instead:

There is no social mobility, so wherever you start after being randomly assigned is where you will stay for the rest of your life.  

Variation 3: Control

The control condition did not mention social mobility at all. We assume that this led people to assume some degree of mobility, since responses resemble those in the social mobility condition (Variation 1).


Variable #2: Taxes

In addition to these three variations on social mobility, there were two conditions regarding the presence of tax information:

Variation 1: Tax Information is Provided

In the taxes condition, we said:

You may have to pay taxes. The wealthier you are, the more money you will pay. 

In this country, taxes are the only way to achieve a more equal wealth distribution.  That is, if you see a distribution that of wealth that is very equal, then the government of that country collects more money from the wealthy to redistribute to the poor.  What this means to you is that if you are wealthy, a more equal distribution means that you are paying more in taxes, and if you are poor, a more equal distribution means that you are getting more.  

Variation 2: No Tax Information Provided/Control

In this condition, we did not include any information about taxes at all.


The tax information was intended to remind people that equality isn’t necessarily fair, and that the cost of redistributing money to the poor is often higher taxes on the rich. We predicted that the conditions with tax information provided would favor more inequality than those without the tax prime.

We also predicted that inequality is perceived to be more permissible when social mobility is high. This led us to expect the no social mobility conditions to rate the more equal distributions higher than people in the conditions with social mobility.

Based on past studies, we expected that social mobility would affect the results more than the tax information, but we did expect that the tax information would lead to somewhat lower ratings for equal distributions,



We had 849 people rate pie charts, excluding those that gave a rating of zero for everything (this would mean that either they disliked everything equally, or that they didn’t answer the question since zero was the default answer – and either reason wouldn’t provide meaningful results).

Here’s a brief summary of the interesting results, followed by a chart of average ratings for each distribution:

  • Surprisingly, in all 6 conditions, the equal distribution had the highest average rating.
  • The middle-range of distributions received similar ratings from all conditions; the manipulations seemed to primarily affect ratings at the extremes (very equal and very unequal).

Regarding Social Mobility

  • With no social mobility, the more equal distributions are rated higher, which makes sense.
  • With no social mobility, the perfectly unequal distribution is rated higher, which is the opposite of what we predicted. We’re interested to hear readers’ ideas about this – what might explain why, without social mobility, ratings for the most unequal distribution increased?
  • On the whole, the control group, which did not mention social mobility at all, and the condition that explicitly mentioned social mobility had similar average ratings for all nine charts. This makes sense, since the control group most likely assumed some degree of social mobility existed.

 Regarding Taxes

  • In the condition with social mobility, the information about taxes lowered ratings of the three most equal distributions. In the control condition, we see the same result, but the differences are smaller. This is consistent with our hypothesis that taxes would make equality less desirable.
  • Taxes had the greatest effect on ratings for those in the control group with an annual household income of $100,000-250,000. For the most equal distribution, the average rating without taxes was 54; with taxes this decreased to 24. On the other end of the spectrum, ratings for the two most unequal distributions more than doubled, from an average of 6 without taxes to 12 (2nd most unequal) and 16 (most unequal) with taxes.



Next, we looked at the ranking of distributions, not the actual rating values to get an indication of which distribution people liked best without the potentially confounding effects of differences in people’s individual ranges of rating values. The table shows the percentage of people in each group who ranked each distribution first. To be conservative since the results were skewed to the equal end, any ties went to the less equal distribution (so if charts #3 and #4 were both ranked 100, we coded this as distribution #4). In all six conditions, more than 40% of people liked the perfectly equal distribution best, and we found this very surprising. We’ll be interested to see what readers think about this – would you expect such stable preferences across the conditions?


We also asked several demographic questions, and we appreciate that most people responded to all of these questions. The most significant demographic variables were household income and fiscal conservatism. The fiscal policy question was answered on a sliding scale from 0 (very liberal) to 10 (very conservative). 780 people answered this question, with a mean response of 4.9 and a median of 4.5. We split the responses at the median into two groups so that we could compare the more liberal half of our sample with the more conservative half of the sample. Here, we see a large difference in rankings – 58% of liberals rated the perfectly equal distribution the highest, compared with 30% of conservatives. However, liberals and conservatives alike favored the more equal distributions.  


Taken as a whole, the results suggest to us that there is much more agreement than disagreement about wealth inequality. Across differences in wealth, income, education, political affiliation and fiscal conservatism, the vast majority of people (89%) preferred distributions of wealth significantly more equal than the current wealth spread in the United States. In fact, only 12 people out of 849 favored the US distribution. The media portrays huge policy divisions about redistribution and inequality – no doubt differences in ideology exist, but we think there may be more of a consensus on what’s fair than people realize.

– Dan Ariely


If you haven’t yet heard Chris’ interview with Dan, you’re missing out on an excellent and fascinating exploration of the emerging and increasingly influential field of behavioral economics. If you find yourself with a free 40 minutes at some point this week, we highly recommend listening to the podcast.

Related content
» More

No Comments