Featured Image: Godvertiser
by Ryan P. Burge, Eastern Illinois University
Back in May, I wrote up a short post that tried to understand how Twitter used the word “evangelical.” In total, I collected about 55,000 unique tweets that covered a decent period of time: February 23rd through April 1st. In reality, I wanted to scrape more tweets but my Google sheet was too full and couldn’t add any more rows. Then, about a week ago I thought that I should start another scrape of Twitter data. This time it turned out much differently. Instead of grabbing nearly six weeks of tweets before my spreadsheet hit capacity, this time it took less than five days. In fact, I grabbed over 67,000 tweets from December 12th to December 17th. Here’s a graph to convey the rapid increase in daily volume, with the most recent few days in red.
The biggest day for volume is not a surprise: December 13th when 25,316 tweets were sent containing variation of the word evangelical*. The day after (12/14), the amount only fell to 15,297. There’s obviously a reason for that: Roy Moore’s loss to Doug Jones in the Alabama special election. The Twitter conversation was likely trying to understand how 80% of evangelicals cast their vote for Moore when he had been accused by several women of sexual misconduct.
Not only was Twitter talking about evangelicals, users were talking about them with a very negative sentiment. I used the “bing” lexicon to score every word from all the tweets time periods. I added together all the positive words and then subtracted the negative words to arrive at an overall daily sentiment score. Obviously the overwhelmingly negative sentiment in the later sample is due to the higher volume of tweets in December, but even when comparing each of the two samples on a normalized scale: 58% of the words in Feb.-Mar. were negative, but that rose to 61% negative words in December. So not only did the volume increase, but so did the vitriol.
The most frequently used positive used in the December tweets was “trump.” Obviously, this is a limitation of natural language processing because the lexicon sees the word as a verb, not as Donald Trump. The second most positive word was “support.” On the negative side, “death” and “threats” led the way. That’s likely from the popularity of this story:
Finally, I wanted to see if there has been a shift in where the term “evangelical” drives traffic online. Most of what Twitter is about is sharing links to content that people might find interesting. In the tweets that I scraped there were hundreds of links to websites/blogs/videos, etc. The graph below contains the most linked domains in February and March on the left panel, along with the most linked domains in December on the right.
Obviously, the frequency of all domains is lower in the first round of tweets than the ones most recently collected. Note, though, that the type of domain has shifted pretty significantly. The links in December are much more focused on “mainstream” websites like the New York Times, Politico, and the Washington Post. There was much less of that happening in the February/March tweets. There are a fair number of explicitly evangelical sites in the earlier sample, but few Christian sites remain popular in both (except for Christianity Today).
As someone who researches evangelicals and voting behavior, I have to admit that I am both delighted and terrified regarding how much attention this topic is getting recently. I am delighted because people are reading and trying to understand evangelicalism with a renewed fervor. But, I am terrified that they will read someone’s “hot take” on the internet that is just not backed up by the evidence. Evangelicals are still likely the most consequential religious force in American politics today, but I worry that people (on both sides) are only reading arguments that reinforce their preconceived notions about conservative Christians. Several scholars of religion and politics (myself included) have noted how evangelical academics have seemed to not necessarily question evangelicals political ideology, but instead cast aspersions on polls that assess evangelical political behavior.
One has to wonder if Trump’s use of the “Fake News” moniker for stories that paint him in a negative light has become part and parcel of other prominent voices in the evangelical Twittersphere. We’ve written about that just recently, and let me reiterate the point now: social scientists have a theoretically and methodologically sound approach to measuring evangelical Christians. Evangelical elites might feel like they have lost control of the narrative. There’s a famous Carl Sandburg quote that says, “If the facts are against you, argue the law. If the law is against you, argue the facts. If the law and the facts are against you, pound the table and yell like hell.” It seems that evangelical Twitter has become a lot less about arguing the facts and a lot more pounding the table.
Full coding syntax for this analysis is available on my Github.