((1/s)O *fC ) > ((1/t)M * fP), so what's bambi's point?
UPDATE: Someone at Free Republic suggested reducing this to bumper-sticker size. How's this:
You know how Bambi nails Johnny Mac for “voting with President Bush 95% of the time”? Like it’s a bad thing?
Consider the construction: People do not like “P.” If you want to make “M” look badly, associate him tightly with “P.” Do this by showing a correlation of “M” to “P” through variable “v.”
The resulting equation is this: vM = P
It shows the portion of “M” that is equal to “P.” Through this equation, we just need to change Bambi’s words a little: “95% of McCain is the same as Bush.”
Simple enough.
Now assume there are two competing entities, “M” and “O.” The idea is to create a disparaging association. We already know the direction of “O”: Associate “M” with “P.”
“M” decides to use “C” against “O.” “C” is the democrat Congress.
The resulting equation is this: vO = C
I need to distinguish Mac’s and Obama’s voting correlation variable, so I will change them to “t” for Mac and “s” for Obama.
We need one more piece. The assumption is that being associated with “P” or “C” is a bad thing. But how bad? Let’s add variable “f” for favorably rating.
The equations are now:
sO = fC
Now I run headlong into my lack of statistical acumen. I need the equation to yield an answer! So let’s reconfigure:
sO * fC = Y
The result can now tell us if X > Y or X < Y, that is, if Mac’s resulting rating is “better” or “worse” than Bambi’s rating.
I have one more logical issue. The higher the value of “t” (or “s”) leads one logically to presume that it is “more bad.” That is, Bambi states 95% in a way that leads one to believe that if it was 70% it would be better for Johnny Mac. I can achieve this by using the reciprocal of the value of “t.” As Mac’s and Bambi’s correlation of voting with W or Congress decreases, the resulting values of “X” and “Y” will increase.
For example, assume that Mac voted with W 100% of the time, 75%, 50%, and 25%. Assume W’s favorability remained constant at 20%. The equation would yield answers of 20, 27, and 40 (using whole numbers). The less often that Mac votes with W the better; higher the resulting rating, the better.
The final equations are this:
(1/s)O *fC = Y
The next issue is that we want as long a record for “t” and “s” as possible. The more data of voting allegiance, the better, right? So we are going to take the latest information possible. Also, variables M and O remain constant at 1. This leads to the only true variables in the equations being the favorability rating of W and Congress.
Let’s put some data and form to these equations. I am going to use Real Clear Politics for source info, using the “Approve” data, found here for the President and here for the Congress.
The specific values I use for “fP” and “fC” are a straight average for every poll listed by month. For W, the data represents 290 polls; for Congress, 139 polls.
I’ll use Bambi’s claim of 95% allegiance for Mac, and the Open Congress calculation of 96% party allegiance for Bambi. Interestingly, OC claims that Bambi abstains 46% of the time – probably influenced by his running for POTUS since Day 1.
The final equations and fixed values are:
(1/s)O *fC = Y
(1.0526)1 * fP = X
(1.0417)1 *fC = Y
The last step that I want to add is a column that equals Y/X (called “Z”). The thought is that the stronger the argument by Bambi (Mac’s association with a disliked president outweighs his association with a disliked Congress), then the higher the value of Z. The actual calculation is:
If the arguments cancel each other out, then Z would equal 100. If Bambi could cast stones because he doesn’t line in a glass house, then Z would be substantially above 100. To the degree that his house is made of plate glass, the closer Z comes to 100 (and perhaps even following below it in the instance when he has shattered glass around his feet in the same manner Bill Ayers has the flag at his feet).
Date | fP | fC | X (Mac) | Y (Bambi) | Z (Y/X) |
October 2008 | 25.7% | 15.7% | 27.1 | 16.4 | 61 |
September 2008 | 29.3% | 18.9% | 30.8 | 19.7 | 64 |
August 2008 | 30.4% | 17.4% | 32.0 | 18.1 | 57 |
July 2008 | 29.2% | 18.3% | 30.8 | 19.1 | 62 |
June 2008 | 29.5% | 18.5% | 31.1 | 19.3 | 62 |
May 2008 | 29.8% | 18.7% | 31.4 | 19.5 | 62 |
April 2008 | 20.1% | 21.0% | 30.6 | 21.9 | 72 |
March 2008 | 30.7% | 21.4% | 32.3 | 22.3 | 69 |
February 2008 | 32.7% | 24.0% | 34.4 | 25.0 | 73 |
January 2008 | 32.9% | 24.7% | 34.6 | 25.7 | 74 |
December 2007 | 33.9% | 25.0% | 35.7 | 26.0 | 73 |
November 2007 | 32.6% | 22.5% | 34.3 | 23.4 | 68 |
October 2007 | 33.5% | 24.1% | 35.3 | 25.1 | 71 |
September 2007 | 32.9% | 26.0% | 34.6 | 27.1 | 78 |
August 2007 | 32.0% | 22.2% | 33.7 | 23.1 | 69 |
July 2007 | 30.8% | 27.4% | 32.4 | 28.5 | 88 |
June 2007 | 30.9% | 25.4% | 32.5 | 26.5 | 82 |
May 2007 | 33.2% | 33.7% | 34.9 | 35.1 | 101 |
April 2007 | 34.8% | 37.1% | 36.6 | 38.6 | 105 |
March 2007 | 33.9% | 31.7% | 35.7 | 33.0 | 92 |
February 2007 | 33.6% | 34.1% | 35.4 | 35.5 | 100 |
January 2007 | 34.4% | 34.3% | 36.2 | 35.7 | 99 |
Here’s chart of X, Y, and Z. Click to enlarge.
Seems like Bambi’s argument isn’t all that strong, eh?
UPDATE - Post Script - Yes, I clearly could have just said: Bambi votes just as often with Congress, and their rating is substantially lower than W's. But that would not have been any fun. Right?
No comments:
Post a Comment