>I'd be curious to see his data as well and why it his conclusions seem so different. OP here - me, too :) Question: are you including every tweet in your ana

>I'd be curious to see his data as well and why it his conclusions seem so different. OP here - me, too :) Question: are you including every tweet in your analysis, or do you ignore some as being irrelevant? I mentioned the example of people complaining about spam from public whois records; if you mark such tweets as negative, I could see that explaining a large part of it. (I ignore them as I don't see it as being directly related to a host's performance, even though I know there's an argument to be made about public and "private" whois records). FWIW I also ignore positive tweets that consist of short generic "great, thanks!" or "Yay, awesome!" - I've found that they most often tend to be talking about some sponsorship/promotion or meetup/party the host has held.