Monday, January 14, 2013

Hiding the facts behind graphs

The Daily Kos is one of my everyday reads.  Not because I agree with their views, but because they best represent the current thoughts of the liberal hoi polloi.  They do have the occasional well thought out articles  which force me to modify and burnish my world view.  But for the most part, they have the same role and one-sided world view that the right-wing sites do, it is just a mirror image.

So, when this graph came up on their site as a means of convincing me that we should immediately start hiring more government employees, I got a big giggle when the following chart came to the surface as a piece of evidence for more cops and government employees.

OK, so following the 1981 recession, the public sector employment went up to 103% of the starting point.  The next point that the cherry picked is the 1991 recession when it went up to 103% of the starting point of that recession.  The 2001 recession was brutal with the poor guvmint workers in that they only gained up to 101% of their recession starting point.

So lets do some basic math.  

100 x 1.03 x 1.03 x 1.01 = 1.07.  

That means that recessions usually cause the number of government employees to go up.  The last three recessions have netted a nice little 7% increase in total government employment in the period 1981 to 2007.

The folks over at the KOS are whining because since 2007, the guvmint employess (or as I call the The Daily Kos' "Base") has actually gone down 2%.  So, back to math.  

100 x 1.03 x 1.03 x 1.01 x 0.98 =1.05

So even the the rapacious and draconian cuts in government employment since 2007 (2%!!!!!) leaves us with a 5% increase in total government employment in the ten or so years of the thirty-two tear period that this graph digests.  

So what does the whole picture look like?

So, what I see is in 1981, the number of government employees was around 15 million, in 2008 it was around 22.5 million.  The implication from the Daily Kos article is that they only grew around 7%, when in fact, they grew around 50%.  

Big difference there.  The take home lesson of this little morality play is that when you see someone cherry pick from a time frame and then start presenting little games like percentage lost/gain, you can be utterly assured that they are trying to hide the real data.

1 comment:

Craig Cavanaugh said...

Lies, damn lies, and statistics. I never trust statistics, because it is far too easy to skew the numbers in the way the statistician wants, and it is virtually impossible to independently verify those numbers...