I think I've mentioned this before, but I'm a professor in journalism and most of my research is on political communication and new media, primarily dealing with political blogs. I've also done some work on other kinds of online communities, however, and I'm in the middle of conducting a study on discussion behavior in SBN's baseball blogs. The first wave of analysis uses game and team factors to predict the number of comments and commenters in each game's game thread. I have everything collected from the NL Central blogs, and thought it might be interesting to see how BCB compared with the rest of the division last year, and what tended to prompt the most discussion.
First, a quick note on methodology. There's a lot of variation in community norms and post organization across the 30 MLB blogs at SBNation. This analysis is based on whatever thread was posted specifically for live discussion of a game, and any overflow threads that were posted while the game was still going. At some blogs that provides a cleanly discrete discussion; at others, there's a bunch of discussion of other games and various non-game topics mixed into those threads after the game ends. There is no simple or reliable way to separate that discussion out, which is basically OK in my analysis, since ultimately it will just be a team/blog characteristic that is controlled for in my analysis. But in the results shown here, it inflates the numbers for Viva El Birdos, in particular.
The outcomes that I'm looking at in my initial analyses are comments, commenters, and comments per commenter, which all vary quite a bit across the six blogs.
Along with a general decline in activity during the course of the season, there are clear team-specific ebbs and flows in discussion density. (Scatterplots are each game's comment totals, curves are smoothed trendlines.)
Comments per commenter:
As for BCB specifically, excluding one-series opponents, we were most active in the threads for games against the Dodgers (median comments = 566, median commenters = 40) and least active in games against the Phillies (comments = 124, commenters = 16). This isn't too surprising considering how those games went, and my big analytical model predicting total comments bears this out. If you're not familiar with reading statistical tests, variables with p values of .05 or less are considered significant; we can be 95% sure that those results are not occurring by chance, but represent real phenomena.
|Previous game's result||.950||.332|
|Result * Previous result interaction||.122||.727|
|Weekend * Night interaction||1.209||.274|
|Brewers overall record coming in||16.213||.000|
|Brewers record in last 7 games||.335||.564|
|Brewers place in division||5.038||.027|
|Brewers games back in division||.814||.369|
|Margin of victory by winning team||10.026||.002|
|Total runs scored||8.196||.005|
One thing worth noting is that the distribution of these relationships seems to vary by team, and the overall results will differ from these. For example, the interaction between weekday/weekend and day/night is significant in my preliminary overall tests -- that is, there's a difference between day and night games during the week that doesn't manifest for weekend games. I should be done collecting data on this project in a couple weeks, with an eye toward getting the paper submitted somewhere this spring.