Misleading graphs from the WSJ to hide reality

Fun With Charts: Making the Rich Look Collectively Poor
[Via Daring Fireball]

Same data, two different ways of presenting it, two different stories.


(I hope this is a little clear. Since the whole purpose of misleading with numbers is to take advantage of the innumeracy of people, I am afraid some of this will not make sence. But I am attempting to help people learn what is happening and why people lie with graphs)

Another in the long line of lying with numbers. The fact that the WSJ knowingly created a misleading graphic demonstrates that they are not serious about the facts but about representing a definite view. Making the data fit that view is the purpose of the chart.

What is shown at the link is that it is perfectly possible to use the same data to misleadingly present the exact opposite view as the WSJ. That is what can often be done when data is misused – it can be made to represent almost anything.

The second graph is exactly the same, he just moved the ‘everything higher’ filter to the left at 200k. Very different chart.

The point is to demonstrate just how misleading the graph is by its choice of bins along the x-axis.

The WSJ purposefully manipulated the data to present a distorted view. This would have been tossed out by any peer review process because it is misleading.

That is why it is so easy to identify denialists and liars – they knowingly misuse data, creating representations that are simply false.

The first hint is that we expect each bar to represent different levels of the same thing. But each bar represents quite different intervals. First we go up by 5K. Then 10k. Then 25k. Then 100K, 300K, 500K, 3 million, and then 5 million. Then everything higher.

Why would anyone decide to completely change the interval instead of using the same interval? Most graphs that want to provide enlightening examination of the data use things like 5 bins called  quintiles, to represent the data. Thus each bucket represents 20% of the population.

What is the reason for changing the intervals for each step along the x-axis? That should make one wonder.

The easiest way to determine if a graph is misleading is to look at what is really being displayed. Why is total income on one axis while household income is on the other? What does this really tell us, without knowing how many households are at each point?

Because, total income is the product of the income and the number of people  with that income. Y= A * X where X is the income and A is the number of people. But A changes for every bar along the X. It is not constant but a variable. The y value is directly dependent on two values – X and A –  but only one of those values is graphed – X. And A changes for every different group of x values.

An effective 2D graph usually shows the connection of one value with one variable. Here, they are using data that includes 2 vaiables but they only chose to graph one.

Why? To mislead. Because the height of the peak is not only dependent on the x value but on some other value (A – the number of people with that income) which is not shown.

The y values contain another variable that they have simply not chosen to include in the graph. Why would they take data based on two variables but only graph one?

Because that would ruin their whole point.

Let’s try this approach that includes all the values but reduces them to one variable not two. Let’s recalculate the numbers but take the variable A into consideration by using percentages of the whole population for each x. Instead of having this weird x axis, lets simply split up the whole range of incomes in the US into 5 buckets – each bucket has 20% of the total population. Let’s call those lower, lower middle, middle upper middle and  upper classes. This provides us with even spacing along the x-axis.

In 2006, the ranges for each bucket were:

Lower – $0-$18500
Lower middle – $18500-$35000
Middle – $35000-$55000 
Upper middle –  $55000-$92000
Upper – $82000 and up

Now, for each bucket we can total up all the incomes and divide by the population to get the y value. Since now BOTH values are dependent on the total population – the x is made up of fractions of the total population  as is the y –  the A variable  is canceled out and we can directly examine what percentage of the total income in the USA is found in each group of incomes.

Essentially, in this approach we have removed the dependence of x and y on the variable A. Here is what some of the data looks like:

my graph 2

So, you can see that the upper 20% of our population – the ones who make the most – are responsible for 48% of the total income. The lower 20% of our population only represent 4.8 % of the total income. Thus,  the top 20% represent 10 times more total income than the bottom 20%.

Here is the data above shown as a pie chart :

my graph 3
Screen shot 2011 05 13 at 2 29 23 PM

48% of the pie is given to to 20% of the people. One out of every five people in the USA get only 4% of the pie.

No wonder the WSJ wanted to break out the A value when they graphed their data. By drawing a more legitimate graph, we can see that the top 20% of income earners in the US has almost the same total income as the entire other 80%.

This is termed income inequality, something the WSJ journal probably does not really want people to see..

And it only gets worse if total wealth, not total income, is graphed.

And here is the wealth chart:

my graph 4

The lower and lower middle classes hold so little wealth in the USA that it is simply not visible on the pie. 40% of the people in the US simply get no pie at all.

There are countries with more equitable quintiles. Here is the income chart for Sweden:

my graph 4

Not only is Sweden’s income more evenly distributed but a survey of Americans prefers that income distribution over than our own –  92% to 8%. Yep, 92% of the people in the US would rather live inSweden than the US – al least as far financial equality is concerned.

And what was really fascinating was that this preference cut across both parties, sexes and even wealth. 89% of the people making over $100,000 a year – which would put them in the upper class – preferred Sweden’s distribution. And when given a chance to create more equitable ratios, every group took from the top and gave to the bottom, holding the middle the same.

In fact, the wealthiest in the survey were willing to reduce their piece of the pie by one third in order to create a more ideal society.

The huge wealth inequality is what the WSJ does not want us to see. When people do see a better – less misleading – representation, they do not like it.  Even most wealthy people.

2 thoughts on “Misleading graphs from the WSJ to hide reality

Comments are closed.