« Twitter roundup | Main | Twitter roundup »

June 10, 2011


Two things:

1) I don't know if Congressional ethics investigations is the appropriate measure. Beaudrot is referring to news coverage not Congressional reaction to scandal. Your earlier choice of Washington Post headlines would seem more appropriate here.

2) "The conservation of headline news" Given a max of one major headline per newspaper/magazine and a fixed number of newspaper and magazines in print. The maximum attention any one event could reach would be the sum of the headlines. Of course blogs violate the "conservation of headline news" How fascinating.

Nice analysis, Brendan.

In my work as a casualty actuary, I dealt with outliers all the time. Sometimes there would be one year with an enormous loss. Including that year in a 5-year average would produce a distorted result. OTOH leaving it out entirely would mean ignoring the very loss that we were providing insurance for. A common solution was to cap each year's results. Thus, a enormously bad year was treated like a moderately bad year.

In Brendan's top graph, the 104th Congress, with 20+ ethics investigations, had less impact on the fitted line than if the 104th Congress had had only 8 or 10 ethics investigations. That seems wrong. I think he'd get a better result by capping the number of ethics investigations at around 10, rather than totally omitting the outlier.

The comments to this entry are closed.