Wednesday, May 07, 2008

Mainstream Media's Election Math Problem: Obama won NC by 14.81%, Clinton won IN by 1.46%

Update
With 100% of North Carolina precincts reporting, here are the results (provided courtesy of CNN):
North Carolina Democratic Primary Results 08 100%

And CNN's updated, but still not complete, results from Indiana:
Indiana Democratic Primary Results 08

And the real math:
Doing the math
Here is my source worksheet in Microsoft Excel format.

So compound rounding for IN now looks ridiculous. If you round each % Won number individually, Clinton's margin of victory is still 2%, but in reality it's just 1.11%. This is just about the best case to prove how silly rounding these results are. The rounding in North Carolina now doesn't matter, rounding each % Won percent before subtracting now doesn't affect the accuracy of the reported 14% difference.

Original Post

I am no math wiz, I could have this totally wrong, but here goes. Here are the results for Barack Obama vs. Hillary Clinton in North Carolina (provided courtesy of CNN):
North Carolina Democratic Primary Results 08

And CNN's results from Indiana:
Indiana Democratic Primary Results 08

After Pennsylvania, when the Huffington Post pointed out that Clinton did not win by 10% like the MSM reported, but something under that, I wondered if they would commit the same crime again for North Carolina and Indiana, and it sure looks like they have:

Doing the math
Here is my source worksheet in Microsoft Excel format.

How are the MSM getting this wrong? They are rounding the % Won column before doing the subtraction, which results in a pretty significant misrepresentation of the results. Remember, according to math rounding rules, you round down any fractional amount of a number that begins with 1,2,3,4, and round up any fractional amount ending in 5,6,7,8,9. In IN, you can make the case through compound rounding that the margin was 2%, but interestingly neither Excel or Numbers will round up to 2% if you tell it to hide all decimal places. If compound rounding is used in NC, the margin is 15%. CNNs Vote % in NC doesn't make any sense, if Obama won 56.68%, you can't round him down to 56%, but round Clinton up to 42% from 41.87%. I don't know what to conclude, but here are the possibilities I see:

  • I am not correctly counting the NC No Preference percentage (I don't think this is the case)
  • A bug in their web site math
  • Intentional distorting of the math
Is it so much to ask that the MSM uses fractional percents? I mean if the american people are smart enough to handle dollars and cents (aka fractional dollars), surely in the 21st century we can manage fractional percents.