Eric Levine’s blog Content in Reality gave me another variable to measure my database of PolitiFact rulings, which should have been obvious, although I had not really given it a thought before. Maybe it was because of my bleeding heart liberal need to be politically correct. Well heck with that.
He graciously responded to some comments I gave him on his blog concerning the availability of my database, with the pertinent portion for this post:
I will suggest one really important point I made on a previous post about the question over Politifact's ratings meaning republicans lie more or if there is selection bias:
"First off, notice that Politifact has never had an overall Dem vs GOP report card. Ostermeier was the one who did this. So this means that certain conclusions one may be tempted to draw from Ostermeier's study are not ones people may necessarily draw from Politifact itself. This is because all Ostermeier did was total up all statements made by Democrats and Republicans, and compared them. Of course he could have easily done the same with any arbitrary breakdown of statements: Men versus women, majorities versus minorities, southerners versus northerners, etc... If we found similar disparities in the data to the Dem vs GOP data, would we be accusing them of covering "political discourse with a frame that suggests this is the case?”...
I never assigned a variable key for men versus women, etc., because I thought it was not appropriate. But I realized in reading all of Levine’s enlightened writing along with his commentary responding to my conservative counterpart’s “trolling” that it would not be too difficult a matter to assign the male versus female variable to my data: the only time-consuming task would be to make sure gender-neutral names (like Pat, Chris or Alex) are assigned correctly. I didn’t check every every last one, but I know I’m very close. So here it is: of 3,989 rulings through the end of 2011 for individuals, 642 or 16% (1/6) of the rulings were on females. The Truth-o-Meter app breaks down like this:
 |
| Click to enlarge: The Truth-o-Meter male to female looks close to identical except for the Pants on Fire. |
At first glance, it looks like on a percentage basis the Truth Index categories are all quite close. But if you read across from True to Pants on Fire, the female percentages are very slightly lower going all the way to Pants on Fire—where it’s about 3 points higher than the males. On a percentage basis, the female Pants on Fire Percentage is almost 50% (45% to be exact) higher than the males (9.3% versus 6.4%).
But what about the Truth Index—in terms of Republican males and females and Democrat males and females, and those of neither party? Here’s how that looks:
 |
| Click to enlarge: Boys are blue and girls are pink: Pretty close except for Republican girls. |
The indexes across the board fell within a fairly close range with the exception of the Republican females who had the worst Truth Index approaching a “Mostly False” negative 50. And when one thinks of PF Pants on Fire and Republican women, the one outlying outliar who immediately comes to mind is “PolitiFact came out and said everything I said was true” Michele Bachmann.
So, just for sh*ts and giggles, if you back out Bachmann’s rulings, you improve the Republican female average to a negative 39.47. That’s only five points but look at it this way. The percentage of Pants on Fire rulings for Females in total drops from 9.3% to 8.1%, dropping the percentage difference between Pants on Fire for Males to about 27% higher instead of the overall 45% higher. In other words, a lot of bad rulings on one person can have a very large affect on the rulings as a group.
Conclusions? In the same way that my conservative counterpart strains to prove that PolitiFact selects statements with a liberal bias, can it be said from this that they select them with a male bias as well? Indeed, this can almost be seen as a good example of how difficult any bias would be to prove if it indeed existed, as well as the conspiratorial, absurd nature of the “liberal” selection bias claim. To put it in Levine’s words, with this “arbitrary breakdown” I have done, should PolitiFact be “accused of covering political discourse with a frame that suggests” gender bias? I think not.
No comments:
Post a Comment