Conspiracy theorists love correlating things together. A map of the rise in TV dinners and nasal infections. The travel patterns of Bush administration officials and tornadoes. Corrleation, as any first year knows, is not causation.
Here’s your conspiracist correlation nonsense of the week.
A new study found that for every 25,000 retweets that a known Russian troll account received during the 2016 election, Donald Trump’s poll numbers jumped 1%.
Wait, so all you need to win an election is about 150,000 retweets?
And are these the same polls that predicted a Hillary victory? Those seem like a reliable metric.
Details: The Tennessee-Knoxville study analyzed 770,005 tweets in English from known Russian troll accounts, as well as corresponding poll data from FiveThirtyEight’s archive of multiple polling outlets.
Every 25,000 retweets of Russian accounts correlated to a 1% increase in Trump’s poll numbers one week later.
Given the frequency of tweets from Russian accounts, 25,000 retweets would average around 10 retweets per tweet.
Retweets did not have a similar effect on Hillary Clinton’s poll numbers.
Since the IRA wasn’t really promoting Hillary Clinton, but was promoting Bernie Sanders and Jill Stein, did the IRA tweets correlate with a rise in their poll numbers?
Caveat: Correlation does not always mean causation. If a Trump talking point encouraged a particularly viral Tweet, for example, it may have also encouraged a change in Trump’s polling on its own.
So why report on this nonsense in the first place?
Actual studies long since established that IRA trolls gravitated to trending content in order to gain influence. Discovering that pro-Trump tweets trending correlated with Trump’s poll numbers rising is like learning that more people buy cars during auto shopping season.
Some of said buyers may be Russian spies, but they didn’t cause the trend.
Leave a Reply