Statistics - as one hopes more and more of the public is coming to understand - can be deceptively simple-looking, but, perhaps for that very reason, so easily misunderstood. It would be all too easy for someone reading the article in Tuesday's Citizen by Professors Panagopoulos and Weinschenk, to take away just what is proclaimed mistakenly in the headline - "Clinton's Comey claim doesn't add up."
This is not at all what they were able to conclude. The accurate banner would have been "Clinton's Comey claim not proven."
What the article shows is that using the approach they chose (most of which one cannot really discern from what's in the Citizen) they were unable to show statistically significant changes in the immediate days after Comey's announcement, to affect the outcome. Given, however, that the margins in the swing states were themselves so small, finding cause and effect relations would indeed have been unlikely without much closer polling supported by interviews. Add to that the professors somewhat gob-smacking revelation that they worked with a national poll and not one focused on the swing states, and the unnewsworthiness of their findings stand out. Their conclusions and a couple of bucks will get you a so-so cup of coffee.
- Norman Dale
Prince George
