Legislative Responsiveness: A Class Bias?

Richard Florida makes the claim that the American states are becoming increasingly conservative (http://www.theatlantic.com/politics/archive/2011/03/the-conservative-states-of-america/71827/). This trend, he goes on to argue, has been exacerbated by the economic troubles over the last few years. Evidently: “Conservatism, at least at the state level, appears to be growing stronger. Ironically, this trend is most pronounced in America's least well-off, least educated, most blue collar, most economically hard-hit states. Conservatism, more and more, is the ideology of the economically left behind.  The current economic crisis only appears to have deepened conservatism's hold on America's states. This trend stands in sharp contrast to the Great Depression, when America embraced FDR and the New Deal.”

He goes on to make the claim: “But the much bigger, long-term danger is economic rather than political. This ideological state of affairs advantages the policy preferences of poorer, less innovative states over wealthier, more innovative, and productive ones.” Supposing that poverty spurs conservatism (at least in recent years) and, given the recent economic crisis, the states are becoming more economically disadvantaged, it follows some logic that policy outcomes are likely to become increasingly conservative. However, policy outcomes are ultimately decided at the individual level, by congressional vote and consent of the executive. A more precise examination of this supposed issue would examine the extent to which members of Congress respond to constituent opinion by economic group. That is, are members of Congress more responsive to particular economic groups of the citizenry? In his highly discussed book, Larry Bartels (Unequal Democracy, 2008, Ch. 9) showed that congressional votes tend to co-vary more significantly and robustly with citizen opinion of the wealthy. The validity of this claim ultimately brings into question some of Richard Florida’s claims. Policy may be increasingly conservative, but such outcomes may have little to do with economic hardship and the movement of those “left behind” into the conservative camp.

The Cooperative Congressional Election Study (2006) provides an opportunity to replicate, albeit imperfectly, some of Bartels’ findings. The survey asked respondents their opinions on a series of Senate roll call votes, including, but not limited to: (1) support the ban on late term abortions, (2) oppose a plan to offer illegal immigrants a pathway to citizenship, and (3) support cuts to the capital gains tax. Answers to these questions were aggregated by economic group (low income earners, middle income earners, and high income earners) and state (using sample and population proportion weights). Senator votes on these issues were then regressed on citizen opinion by economic group, controlling for party identification of the office-holder. As an additional check, Senators DW-NOMINATE scores from 2006 were regressed on mean state ideology by economic group (See Table below). In two of the models, the probability of a Senator voting in a conservative direction increased significantly only as citizen opinion of the wealthy became more conservative (Abortion and Capital Gains). These models support the findings of Bartels. The final model shows a significant relationship between citizen opinion of the middles class and Senator ideology on roll call votes. All of the models cast doubt on the extent to which the poor, who although may be an increasing majority in many states, have an influence on legislator decision-making.

Senator Roll Call Votes and Citizen Opinion by Economic Group, 2006
Abortion Cap. Gains Immigration NOMINATE
Mean Opinion L 2.341 -3.532 2.772 0.275
(3.481) (4.879) (2.232) (0.179)
Mean Opinion M 3.107 5.465+ 1.509 0.302*
(3.284) (2.827) (2.212) (0.131)
Mean Opinion H 7.521+ 4.440+ 1.053 0.179
(4.143) (2.465) (1.890) (0.111)
Republican 4.318** 5.367** 2.374** 0.745**
(0.909) (1.041) (0.598) (0.041)
Constant -3.510 -3.999 -3.328 -0.437
(1.177) (1.991) (0.655) (0.024)
N 90 100 100 100
Chi-Squared 27.523** 48.265** 28.779**
BIC 72.424 68.346 118.756
R-Squared 0.881
Note: Data are derived from the CCES 2006 and Poole and Rosenthal’s DW-NOMINATE senator estimates (voteview.com). Respondents are asked in the CCES to report their opinions on these specific roll call votes (coded here for each respondent before deriving means -1 oppose, 0 indifferent, 1 support). Mean opinion is derived for economic groups using sample and population weights (See Bartels 2008). Low income earners (L) are those who report a household income below $30,000 a year, middle income earners are those who report income between $30,000 and $70,000, and high income earners (H) are those who report incomes above $70,000. Mean opinion in the final equation is based on ideological self-placement (-1 liberal, 0 moderate, and 1 conservative). Roll call votes are coded in conservative direction (1 support 0 otherwise). Columns 2-4 are based on logit equations; column 5 is based on OLS (dependent variable is DW-NOMINATE scores for each senator ranging from -1 to 1). Legend: +p<0.10,  *p<0.05, **p<0.01; BIC = -2*ln(likelihood) + ln(N)*k. Standard errors, clustered by state, are in parentheses.