A short Twitter history of the 2021 Evidence and Implementation Summit

by Jon Scaccia April 2, 2021

The thunder from down under rumbled this week at 2021 Evidence and Implementation Summit hosted by CEI and Monash University. In a different world and time, we would have been there in person.  Alas.

We presented on using NLP methods to synthesize scientific evidence (so basically the methodological rationale for PubTrawlr), but that’s another post. Today, we’re going to focus on the EIS2021 Twitter chatter.

Twitter has a pretty straightforward API, and there are straightforward r packages to interface with it. I pulled all the tweets over the last week with the preferred conference hashtag “#eis2021.”  I then did some basic bag of words and sentiment analysis on them.

N-grams

There are a few phrases that really stick out here.  First, “evidence base.” As a field, implementation science needs to be constantly reminded that we are trying to implement what has been actually been shown to work.  Second, the strong connection between implementation science and implementation practice.  The folks from TCI are making great progress really emphasizing the practical nature of implementation science that moves away from research and into what actually makes a difference on the ground.  Third, it’s clear that Sue Dopson and Rinad Beidas made an impression.

The network graph is also illuminating; again for the connection between implementation science and implementation practice.  Not sure what to make of the Penn Medicine merry-go-round, though…

Sentiment Analysis

Let’s take a look at the emotional content of the tweets.  Sentiment Analysis examines specific words and sorts them based on the corresponded emotions might be. For this first figure, I used the bing lexicon, which assigns words as positive or negative.

This isn’t that bad!  We see a lot of commentary on the quality of the sessions, which appeared to be good.  My hunch is a lot of the negative words are associated with the content of the session.  This makes sense.  Inequity is an emerging priority for social services research, and more people are beginning to incorporate it into their work.

I also used the NRC lexicon, which assigns words into more complex emotions.  This didn’t work nearly as well here.  First off, “sue” is a person in the context; not an act. And who’s tweeting about elves?

Let’s Sum Up

The Twitter record of the first implementation conference of the year seems positive. We see a good emphasis on implementation practice (which, you know, is the whole point of PubTrawlr), along with an acknowledgment of the role that equity plays in the current landscape.

One down, 3ish to go?