Sorry if I’m being a bit innumerate, but would you mind telling your readers something more about the actual methods you are using to come up with these pretty charts? What, for example, is the “share of conversation” metric for news media based on? Is it word counts? References in headlines? How does your algorithm score a story as being about “politics” vs being about “issues”? How does it count a candidate’s level of news coverage? By word count of the story they are mentioned in? What if more than one candidate is mentioned in a story? How do you weight headlines? Photos? How much of the data set for media coverage is TV vs web (since you include two cable channels)? Same with the share of the Twitter conversation? What is the denominator of that share? Total number of tweets in English? How do you know they are only from US Twitter users?
While I’m as curious as many about the potential of “big data” to help us better understand what public issues ordinary people are interested in, it’s pretty frustrating to be offered such pretty charts with so little explanation not only how to read them but whether the underlying algorithms used to create them make any sense.
Given that you have full access to the Twitter stream, and Twitter wants to be seen as relevant to the news business, could you also share how Twitter chatter about politics compares to non-politics subjects? It may be that the big spikes your charts seem to depict are only big when not compared to the things most people are actually talking about.