Statistical analysis shows Romney's new Twitter followers are fake | Hacker News Reader

Statistical analysis shows Romney's new Twitter followers are fake

theatlantic.com

"We did not find a similar pattern in Barack Obama's recent followers. "

Why would you expect the patterns to be remotely similar? Obama has been the President for over 3 years. Romney has been a presidential candidate for a few months.

"Our test is based on the underlying assumption that the followers of Twitter accounts tend to display a some kind of general indegree distribution."

Unwarranted assumption. It could be that many new people are joining Twitter JUST to follow the Romney campaign.

"Selected the twenty accounts closest in size to both Romney and Obama,"

I'm not sure why anyone would consider this a valid sampling technique.

"The median number of followers for Romney's new followers was 5, whereas the median for the comparison group was 27. This represents a stark, and statistically significant difference. If you are a statistics nerd, like us, you might want to know that the p-value on this was 0.0000. "

Not if the "comparison group" is completely bogus.

A better comparison group would be people who have gone from (relative) obscurity to extreme prominence over a short period of time. Sheer number of followers doesn't seem to be important (particularly since I would expect most of the accounts with that many followers to belong to people who have a) been prominent for quite some time and b) have been on Twitter for quite some time).

The graphs comparing the "indegree" distribution of each candidate's new followers to that of comparable accounts would be much more informative if they showed the distribution of distributions.

In other words, show the range of indegree distributions from the 20 comparison accounts (not just their average).

Good points.

Why not just name the comparison accounts?

This is awesome. The Atlantic just made up for its reddit spamming (sort of).

Reply on news.ycombinator.com