Tuesday, May 10, 2005

Mr Drum, we can all play with meaningless numbers

The Washington Monthly's Kevin Drum claims Democratic Presidents have a record of stronger economic growth than do Republican Presidents. Besides a sample size much too low to come anywhere near approaching statistical significance, it fails to take into account other factors like military campaigns, technological innovation, and that 800-pound gorilla--the US Senate.

The Senate has more clout in controlling legislation than does the President. Even Drum would not deny that. We know where this is going--since 1968, (I will look at data going all the way back to the 40s in the near future) using data from the US Census, we find that when the US Senate is controlled by Republicans, economic growth per capita is 34% greater than under a Democratically-controlled Senate.

(Red shows years that the Senate was controlled by Republicans, Blue shows years of Democratic control)

Year AvgInc. Change over last
2003 23276 -0.17156%
2002 23316 -1.81083%
2001 23746 -0.51948%

2000 23870 1.81710%
1999 23444 3.38684%
1998 22676 3.11491%
1997 21991 3.87322%
1996 21171 2.52797%
1995 20649 1.58910%

1994 20326 2.77076%
1993 19778 3.67458%
1992 19077 -0.90900%
1991 19252 -1.95060%
1990 19635 -2.49292%
1989 20137 2.69788%
1988 19608 2.18887%
1987 19188 2.68101%
1986 18687 4.06527%
1985 17957 3.18929%
1984 17402 4.61076%
1983 16635 1.50720%
1982 16388 0.03052%
1981 16383 -0.49803%

1980 16465 -2.16875%
1979 16830 1.50784%
1978 16580 6.98155%
1977 15498 3.16869%
1976 15022 3.46443%
1975 14519 0.10342%
1974 14504 -2.33654%
1973 14851 3.39762%
1972 14363 7.01088%
1971 13422 2.97683%
1970 13034 0.79654%
1969 12931 5.51612%

(All income stated in 2003 dollars)

There's the data, and here's the verdict (stated in average annual per capita income growth between 1968-2003):

Reps avg Dems avg
2.23405% 1.67040%

This is just a first briefing. I will expound on it in the near future. The abstract is thus: stats claiming ceteris paribus when dealing with something as enormous as the Presidency of the United States, especially when lacking tremendously in sample size, are dubious at best and should be taken with a grain of salt. Don't buy Drum, and don't buy me.


jeffreyok said...

Those numbers are meaningless. So are the Washington Monthly numbers. On such a broad-ranging scale it is easy to find numbers to fit some supposed trend, and then claim that one is a dependent and the other an independent. Or that they are even correlated at all.

Billy Bob said...

Meaningless? Broadrange scales tend to be much more reflective of reality than merely viewing on a micro level.

crush41 said...


That was my intention. It is dishonest for the Washington Monthly to "surmise" that voting for a Democratic or Republican President is meaningfully related to economic growth without taking any other factor into account.

Just as easily we can show that voting for a Republican Senator will "strengthen" the economy. There are far too many factors here to make that sort of judgment. A much more in depth analysis is required, along with an explanation as to why Republican Senates lead to a stronger economy.

It is fathomable that a split of power between the Executive and Legislative tends to lead to a stronger economy, but from the data presented by Drum we can not tell anything.

billy bob,

Absolutely, provided that they take into account other meaningful factors. In this case, we would obviously need to look at the Legislative branch during the Presidency, and then there are a host of other factors that would skew the data (especially given such a small sample size) like techonological innovation, age composition of the society, etc etc.