For REL 502 (that’s the Public/Digital Humanities foundations class), our final project for the semester involves network and content analysis of the #aarsbl17 and #naasr2017 tweets from the AAR/SBL and NAASR conferences, respectively. These conferences took place back to back the weeks before and of Thanksgiving. The first couple of steps of this analysis involved learning what NVivo, TAGS, NCapture, Gephi, and related things even were. (To be honest, I think I’d still be completely lost if I didn’t have a degree in Mathematical Statistics.) The next major step was to get TAGS set up and running to capture all of those tweets while they were going on — which was a minor nightmare, let me tell you. I also ran NCapture every day just so I would have multiple sources in case one spreadsheet failed on me. When I stopped running TAGS, the program had captured 8410 tweets from the #aarsbl17 alone.
So what do you do with 8410 tweets? That’s the question we’re trying to answer now. Over the next couple of days, Sierra, Emma, and I will play around with our data in NVivo, Gephi, and Voyant Tools and see what we find. If I were to guess, I’d say we’re likely to find things we hadn’t imagined we would. This is all very new to all three of us.
But here’s the fun part: while taking a preliminary look at what data we had in class yesterday, Mike noticed something interesting. A lot of people had tweeted “@AARSBL” in their tweets. While the hashtag for the event was #aarsbl17, the AAR (American Academy of Religion) and the SBL (Society of Biblical Literature) are two separate organizations. They have two separate Twitter accounts: @AARWeb and @SBLsite. So what’s this @AARSBL account that everyone was mentioning in their tweets? The account is a parody of the conference itself. It started tweeting on October 1 of this year, although Twitter says the account was created in November 2016. Mostly, it tweeted jokes (that weren’t really that funny, imo) right up until the conference began. It’s most recent tweet:
It begins … bwahaha 👿 https://t.co/4sTBkpHOZY
— AARSBL 2018 Denver (@AARSBL) November 17, 2017
Presumably the people mentioning the account didn’t realize it was a parody. The summary pages of the TAGS sheet for #aarsbl17 say that “AARSBL” was mentioned in every single tweet it captured. This can be read the way we initially did — that every single person had mentioned the account. That is to say, that somewhere in every tweet the characters “@AARSBL” appeared. I quickly ran another TAGS sheet to capture specifically those tweets that had mentioned the account and only got about 100. The more I thought about it, the more I realized this seeming discrepancy was because we had drawn the incorrect conclusion from what the spreadsheet was telling us. When TAGS told us the account had been mentioned so many times, it was actually telling us that the characters “AARSBL” had appeared together in those tweets. So of course all tweets with “#aarsbl17” had “AARSBL” in them. As it turns out, the @ matters.
We could have stuck to our original assumption and been led down a path of inquiry that didn’t make any sense. But by critically reading the spreadsheets and having a basic understanding of how algorithms worked, we realized that people mistook the account for a real one only 100 times, rather than 8400 times. That’s a big difference, statistically speaking. So what’s the lesson from this? Don’t blindly trust data you receive from computers. And certainly don’t blindly trust data you receive from people. It is still kind of funny that so many people (roughly 60, according to my data) mentioned a parody account by accident.