When Stata and Politifact enter a drunken political argument
Earlier today I received a challenge from a Republican to prove that their party has been significantly less honest during the last election cycle. I know this is proving the obvious, but I thought people might like to see the quantitative evidence. In order to do this I applied applied numerical values to Politifact rankings, 0 for true, 1 for mostly true, 2 for half true, 3 for mostly false, 4 for false, and 5 for pants on fire. Sort of a “you lie” index. I then grouped all of the values based on a person’s or group’s affiliation with either the Republicans or Democrats, and threw out any that were not affiliated with a party, leaving a sample size of 195. I then used a χ² test to compare the rates at which the two groups fall into the various categories of lies. The results indicate without a doubt that the Republican sample was heavily skewed toward the pants on fire end of the spectrum. For those that enjoy jargon, the Pearson χ² with 5 degrees of freedom was 130.9639 with p=.000. Also, the highest Pearson value on a category was in pants on fire, indicating that the largest difference was in the likelihood of Republicans over Democrats to make completely outrageous statements. Again, I realize this is completely obvious from the get go, but I found it a good way to win a drunken argument. Hopefully somebody else can make use of these results.