It took me much longer than I had hoped but today I finally finished a first draft of a visualization of some of the #sachat data I’ve been working with:
The method of analysis used in this video is dynamic topic analysis (DTA). DTA was developed by Dr. Susan Herring and more information about the method can be found in one of her papers published in 2003. This method of visualizing DTA data was created by Andrew Kurtz and Dr. Susan Herring. Andrew’s original Java tool isn’t working for me any more so I developed the graphs in this video using Excel macros to generate R scripts (which has been an adventure because this is the first time I’ve used R).
The music in this video is Tutto L’Amor Perduto by Giorgio Costantini. It is available under a Creative Common license at BeatPick.com.
I created this and publicly posted it for several reasons. First, this really is a first draft and I would love feedback so I can improve it. I’ve already noticed a few small mistakes that will be corrected in the next version. I’ve also received some feedback and suggestions for possible improvements. If you have some, please let me know!
Second, I hope that some in the #sachat community find even this very rough first draft interesting, informative, and possibly even useful. Even though I’m comfortable studying a group that is so very public with their actions and membership, I still believe that I should give back to that community in ways that are appropriate and helpful. It just seems like a nice thing to do and it’s a small way of showing my appreciation to them.
Finally, I’m interested in seeing if there is interest in helping me continue this kind of work. One of the reasons why this is only a rough draft is that I’m the only one who has analyzed these data. DTA is a specialized form of content analysis and, like any content analysis, it should be performed by multiple persons to ensure the codes are being applied consistently (which is why good content analysis studies report interrater reliability figures to help bolster the credibility of the findings). This analysis – and it should hold up well even when other coders are added – shows that this particular use of Twitter is moderator-led discussion with coherent threads of discussion. I need to analyze a few other #sachat sessions to ensure this is consistent for other sessions. I also need to analyze some other Twitter data so I have some useful points of comparison.
I think this use of Twitter is fairly unusual and it would be great to be able to publicly discuss that with confidence. This is a wonderful example of a group of people using a very limited tool to do very good things that transcend (my) expectations and it should be represented in the research literature.