Thread by @kdd0211

[1] Read entire thread. @ThirdWayTweet just came out with another blatantly partisan paper supposedly claiming a relationship between red states and homicide rates over last 20 years -
thirdway.org/report/the-two…

@Heritage @tzsmith @cullystimson @parkersheppard

[2] Several months ago, alongside my colleagues @tzsmith and @cullystimson, we ripped apart another blatantly partisan paper @ThrirdWayTweet The Red State Murder Problem thirdway.org/report/the-red… with The Blue City Murder Problem heritage.org/crime-and-just… @Heritage

[3] Both "papers" are laughable and a case study about how to blatantly mislead with statistics.

[4] If the authors were interested in any meaningful analysis, they could have begun by computing correlation coefficients using either vote percentages per election or Cook PVI
@CookPolitical and assessing associated statistical significance.

[5] Exclusionary analysis should be based on counties with high homicide rates, not high populations. The authors should learn what influential observations are -en.wikipedia.org/wiki/Influenti…

[6] This is great data ripe for un-supervised machine learning techniques such as k-means analysis. Sadly, the authors put partisanship ahead of a great opportunity to engage in good and informative analysis - en.wikipedia.org/wiki/K-means_c…

[7] The authors talk about factors impacting homicide rates. Great question but where is their regression analysis? #whatajoke.

[8] Don't worry though - My colleagues and I are already working on doing a proper analysis of this data and associated questions. Stay tuned.
@Heritage @tzsmith @cullystimson @parkersheppard