Armchair Analysis of the #educhat Twitter Conversations

Date

Categories
Emerging technology

Last night was the forth installment of #educhat, an impromptu Twitter-based real-time conversation amongst educators world-wide. I decided to take a closer look at measuring how widespread the participation in the conversation really is. The results were interesting.

Results

For the scheduled one-hour conversation, over 600 tweets were published. Including the natural "conversational overflow" (e.g., the 90 minutes following), that number rises to nearly 800. The number of participants for the forth edition of #educhat was 81. While the average number of tweets per author was 7.5 overall, nearly two thirds of the volume of tweets came from just 15 author (18.5% of the total). On the other end of the spectrum, there were 47 contributors that had only one tweet (58% of the total). The figure below is from the TwitPoll used in previous #educhat conversations for participants to self identify their role in education.

Method

To ascertain these results, I used the Twitter API to extract all tweets sent since the start of the #educhat conversation (5:30 p.m. PDT, Monday, April 20, 2009) through 8:30 p.m. (which included the official conversation hour and two hours of follow up and side discussions). Twitter returned the full set of tweets, authors, and date/time stamps in XML format.

For this analysis, I extracted the authors for each tweet, which by counting, gave the total number of tweets. From this, I then collapsed by author to get a tweet subtotal for each author, from which I could easily tally the total number of unique contributors. The frequency of each author was given from the subtotal and used to calculate the average number of tweets per author (7.5) as well as the group's overall standard deviation (11.4).

Discussion

It's interesting to see the group dynamics of participation. Like listservs, online communities, and other forums of public group participation, the bulk of the discussion is carried by 20% of the group. That is, all it takes is about one in five participants of a group to foster the discussion for the benefit of everyone else. Likewise, over half the group only made a single tweet on the #educhat discussion. That percentage of lurkers grows dramatically if you include those that author a total of two tweets in the #educhat discussion: 72.8% are lurking, contributing just 8.9% of the total tweets.

One phenomenon to explain the large group of lurkers could be the difficulty in understanding how to participate. In several of the past #educhat sessions (and this one too) there's a handful of participants that tweet something to the effect of, "I'd like to participate in #educhat, but I don't know how." Essentially, the method of participating in #educhat is misunderstood, and that there's some expectation that they need to "signup" or "login" instead of simply tweeting with the #educhat hashtag.

While this might explain some percentage of the large crowd of lurkers, it isn't sufficient. Observation suggests that the lurker population is indeed high, as the only time they might contribute is during the first request of the moderator (@educhat) to introduce themselves. After that, several may just sit back and read the mesmerizing display of tweets cascading down their monitors.

There are several aspects worthy of further study about real-time conversations using Twitter. For example, taking a more qualitative look at the themes, trends, or side conversations. In addition, it would be interesting to see if the amount of participation changes for repeat users. For example, would a first timer only post one or two tweets, but on their next opportunity, be a more prolific contributor? Likewise, how does past participation experience influence promotion among their followers? And of course, do the overall patterns support existing theories of social interaction? These are just some of the questions to consider for further study.