Kennedy’s popularity in Massachusetts

I talked with a reporter for the Christian Science Monitor while I was on my Wrigley Field trip on Thursday about Ted Kennedy’s popularity in Massachusetts. I told her that based on my experience, it's more likely for a senator to be wildly popular in a small state than in a large state; what mades Kennedy unusual is that he was an incredibly popular large-state senator. Data from the 2008 Cooperative Congressional Election Study (CCES) bear out this claim.

The CCES asked respondents to rate the approval of their senators on a 1-4 scale (1 = strongly approve, 4 = strongly disapprove). This analysis is based on average the responses within each state for both senators.



The accompanying graph shows the average approval numbers (low numbers are good, high numbers are bad) plotted against (the log of) estimated state population in 2008. I have fit a lowess curve through the data, which picks up the non-linearity of the relationship nicely. (All other methods of curve fitting work here, too, and show roughly the same relationship.)

A couple of things about this graph. First, while there is considerable variation in the graph, on average small-state senators were rated better than large-state senators. Second, note the senators below the line. These are senators who had better approval ratings than you would expect, given the size of the state they represent. There are some notable outliers, including Kennedy, Obama, and Clinton. (Outliers on the other side include Craig, Lieberman, Reid, Coleman, and Dole.)

We can use the vertical distance between the senator and the lowess fit to create a "poplation-corrected approval rating." When we do that, the five most-approved-of senators (correcting for population) were

1. Kenndy
2. Johnson
3. Shelby
4. Sessions
5. Clinton

At the other end, the five least-approved-of senators (correcting for population) were

1. Craig
2. Lieberman
3. Reid
4. Dole
5. Coleman