Though promising to “Make America Great,” Republican Presidential candidate Donald J. Trump has spent an inordinate amount of time focusing upon all of the things that ostensibly make America downright terrible. But what impact has Trump’s gloom and doom campaigning had on voting public? Tonight, on the eve of the election, many of us are wringing our hands wondering what to expect from tomorrow’s election. Some data that my colleagues and I have collected may put some liberals’ minds at ease, at least with regard to how well the Trump bluster translates into supporter follow-through.
Since mid-summer, we have been tracking the candidates’ daily social media messaging on Facebook, as well as daily polling and donation data. We’ve watched as Trump emphasized violent crime, economic decline, crushing debt, job loss, racial tensions, and the supposed threats posed by illegal immigrants. We’ve listened as he expressed alarm over the “rigged system” and corrupt media. We’ve tracked his rhetoric, counting each reference to anger, anxiety, conflict, and America’s future. We have done the same monitoring of social media language for his Democratic rival, Hillary Clinton. We then subjected this data to a time series analysis, looking to see what permanent lagged effects this speech seemed to have on poll numbers and donation amounts, using a vector autoregressive (VAR) model.
What our data suggests, is that while much of Trump’s negatively-toned bluster is effective at increasing social media engagement among his Facebook followers, it is much less effective – in fact, somewhat damaging — on his actual polling numbers. Social media content related to anger, anxiety, and negative sentiment drove up his online engagement, but drove down his poll numbers. No such effects were evident in the data for Hillary Clinton. If these polling effects are indicative of actual voter behavior (the burning question), it appears that his threatening speech may have more bark than bite when it comes to making a tangible impact on voter decisions.
But what about his confidence? Does his forceful optimism about winning have a positive impact on followers? We also tracked mentions of achievement related words (win, succeed) for both Trump and Clinton, and mentions of words related to the future. The results again showed a marked divergence – as shown in the above graph. Achievement-oriented language increased social media engagement for both candidates, especially for Trump, but failed to pay-off for him in terms of polling outcomes or donations. In contrast, Clinton’s achievement-focused language resulted in permanent increases in both her polling results and in her campaign donations. Future-oriented language showed the same pattern, with large benefits for Hillary Clinton in terms of engagement, polls, and donations, but non-existent or negative effects for Trump’s outcomes.
So what to make of all this? Sticks and stones may break bones, but words can harm you. For all his bluster, Trump’s language (and his supporters’ resulting enthusiasm), do not appear to land a punch when it comes to concrete outcomes.