Thursday, September 10, 2015

How to Mislead with Spiffy-Looking Graphs, or The Importance of Good Education

I ran across this graph today on my FB feed:

Now, I cannot easily or quickly verify the source of this data or figure out who the "Andrew Barr" named here is. But let me take these numbers at face value for the purposes of argument.

The data appears to make a remarkable comparison. At first glance, it appears to show that measles as a public health problem had largely died out long before the vaccine was introduced in the United States. This observation, if true, would call into question the relationship between vaccination and the suppression of the measles disease - precisely what the poster of this graph wanted to do.

But notice what is actually being compared here. This is not a graph of measles cases. It's a graph of mortality, normed per 100,000 population. It's not a graph of how much measles there is in the US population, but rather a graph of how deadly the disease is in a proportional sense over time.

In that light, it's not at all surprising to see the decline. Advances in modern medicine have permitted us to render many things which used to be fatal now much less so. Today, people who contract measles are much less likely to die than in those who caught the disease 100 years ago. That's a great thing, and the decline in mortality does in fact have little to do with vaccination.

The rate of measles, on the other hand, is another story entirely. Here's a graph of the actual number of measles cases in the US, in thousands, from 1954 to 2008:

This graph is based on CDC data; I retrieved it here. If you want to argue that CDC data are falsified by some devious government conspiracy, then the conversation is over and you might as well go elsewhere. Otherwise, we can continue.

Note what this graph shows. In the latter half of the 1950's and the early 1960s, the number of measles cases did decline, but was still up over 400,000 cases per year. From 1964 to 1968, the years following the introduction of the vaccine, that number collapses, and with two brief blips over 50,000 in the 1970s never recovers again. Today the number of cases is negligible, essentially reduced to a statistical zero.

The argument that "the measles vaccine didn't do anything" falls apart the moment you show the second graph. Of course, folks who want desperately to believe that vaccination is bad won't show you this graph - they won't even find it themselves.

This is why education is so important. We need to teach our children how to read graphs, how to understand the difference between a graph of total numbers and one normed per 100,000 population, and how to question the data presented to them and the conclusions drawn therefrom. There are lots of charlatans and snake-oil salesmen out there with pretty graphs and thundering voices trying to get us to buy their wares. The defense against this is strong education, from elementary school all the way up to college. We don't have to fall for this kind nonsense, however pretty it is.

1 comment:

  1. Wishing I were teaching a methods class now to show my students. Actually, I think I'll show my students anyway.