Much of this material is based on a single, published political science statistical report. The data I found in it proved to be interesting and useful to the purposes I had in mind, which was essentially to look a little more closely into the 1984 Presidential election. The report is entitled, "Personal Economic Well-being and the Individual Vote for Congress: A Pooled Analysis, l980-l990," which appeared in Political Research Quarterly, Volume 49, Number 3, for September, l996, by David W. Romeno and Stephen Stowbaugh, (607-17). The report is essentially a descriptive study, described as to purpose in its
"Summary and Conclusion:"
"Until now, an apparent anomaly existed between aggregate level analysis and individual level analysis of the economy's influence on the House congressional vote. Previous individual level examinations of the economy's influence on the Congressional vote that have turned up null results for the pocketbook thesis have examined single clusters or a series of single elections. By pooling election studies we find that retrospective pocketbook assessments and changes in personal economic conditions have a significant impact on the individual vote for Congress. These results establish a symmetry between Presidential election studies and Congressional election studies regarding the economic considerations voters make in the vote decision; they employ both pocketbook and sociotropic concerns.
"Our findings reveal that under certain conditions (worsened personal financial conditions coupled with unemployment) pocketbook evaluations are about as influential on the individual vote as sociotropic assessments . . . Finally, our findings are compatible with the recent debate over the economy's influence on aggregate Congressional election outcomes . . . [W]hile the economy has a significant influence on the aggregate Congressional vote, it is limited to independent voters residing in open seat districts . . . [W]ith the growth of Independent voters over time, the conditions for the economy's maximal impact on the individual vote has [also, therefore] increased over time."
In conducting their research, Romeno and Stowbaugh employ three major questions: Pocketbook ("We are interested in how people are getting along financially these days. Would you say that you and your family are better off or worse off financially than you were a year ago?"); sociotropic ("How about the economy? Would you say that over the past year the national economy has gotten better, stayed the same, or gotten worse?"); and unemployment ("We'd like to know if you are working now, or are you unemployed, retired, or what?")
On page 609 they describe part of their methodology which is perhaps worth noting: "Here we chose to model the dependent variable as the vote for the Democratic candidate for the House of Representatives. We control for 3 family influences on the Congressional vote: Partisan voting, incumbent voting, and partisan trend. The partisan voting and incumbent voting variables are coded so that we expect positive coefficients. Positive coefficients on the partisan trend variables indicate a pro-Democratic year compared to the base year (l990). The economic variables of interval are all retrospective, two pocketbook ('. . .personal economic well being . . . and . . . whether [you] were unemployed within the past year . . .'), and one sociotropic (an assessment of the national economy over the past year)."
Romeno and Stowbaugh present a graph that shows the confidence intervals for their study. They also note the study's overall confidence level in predicting how a given individual polled would respond as to how he voted on the basis of the variables described above and the questions listed above (in defining the variables), and the answers to those questions in regard to those variables. The largest confidence interval was .055, which was for the l980 election, for all categories of election (Presidential, Senatorial, and House elections, as pertained to the individual's evaluation of the economy on the personal and national level); the lowest confidence interval was in predicting whether the respondent was a member of the incumbent's party in any year (from l980 to l990, for every two years: that is, l980, l982, l984, l986, l988, and l990); in the latter instance the confidence interval was .014.
In further describing their main purpose and conclusions, they say, (on page 611):
"Consistent with recent scholarship at the aggregate level, economic considerations, while statistically significant, do not dominate the individual's House congressional vote.
However, on page 612 they note:
"[As to the influence of the economy on Democratic ideas:] . . . [I]f personal finances have improved under a Republican President, [this] decreased [the] probability of voting for a Democratic candidate from .78 to .76. An improved national economy decreased this probability even further, to .71. If both of these occur, [the] probability of voting for a Democratic candidate decreases to .68 . . . [T]he most meaningful change occurs with Independent voters . . . If [such] a person's economic situation worsens, [the] probability of voting Democratic increases from .47 to .49; if they are unemployed over the past year, their likelihood of voting Democratic increases to .52; if they have experienced [both] a decline in their personal financial situation and unemployment, their likelihood of voting for a Democratic candidate increases to .54. If the national economy got worse [in their view] their probability of voting for a Democratic candidate will increase to about .56; and if they experience all these [negative] aspects [bad personal finances, unemployment, and a bad national economy] their likelihood of voting for a Democrat increases to .64."
On the other hand, an improved personal financial situation decreases the likelihood for a Democratic vote from .47 to .44; an improved national economy decreases the likelihood of a Democratic vote to .37; and if both the individual and national economy improve, the likelihood of a Democratic vote decreases down to .35."
And on page 617, they point out that:
"The synopsis of individual level examinations of sociotropic assessments and pocketbook assessment influences on House voting [that is, previous to Romeno and Stowbaugh's] is that, while National economic evaluations are consistently significant, personal economic evaluations are inconsequential . . ."
They then refer here to a l994 study by, among others, David J. Lanoue ("Retrospective Voting in Presidential Election Years," Political Science Quarterly Volume 47, l93-207):
" ' This should not be construed to suggest that all voters choose retrospectively or even that economic factors are generally decisive in elections. Clearly, neither conclusion is supported by the evidence. It does suggest, however, that many voters may limit their information search to an 'up and down' evaluation of the current administration's economic performance.'"
Interestingly, in contrast to this earlier literature, say Romeno and Stowbaugh:
"[T]here is no case here that sociotropic voting is stronger than pocketbook voting...In l984, our retrospective personal and national measures show approximately equal strength (except in House elections, where pocketbook evaluations dominate)...In l988, sociotropic assessments appear to have a slightly stronger effect than pocketbook evaluations, but the latter is still at least marginally significant in the case of Presidential and Senate election voters."
According to Romeno and Stowbaugh, their overall confidence interval was .16. The confidence level given, as we noted, was approximately 78.6 in predicting how a given respondent would vote based on the analysis of data gained by the polling questions pertaining to the variables of sociotropic, personal, and unemployment-based evaluations by each individual polled.
Lanoue's findings agree with the Romeno and Stowbaugh survey, then, for all elections except the l984 data--and, as a student of that data, I find myself asking: Why all the statistical flukes in l984? Why should people not have voted for President more on the basis of national economic factors than on the basis of personal economic factors in 1984, when they did so in every other presidential election from l980-l990?
Also, why should pocketbook issues dominate House elections in l984 if they didn't in any other House election in Presidential election years?
In other words, why should this election vary from this norm the two ways of, (1) letting pocketbook issues dominate in deciding the House vote, (while letting National economic issues dominate in all other elections) and (2) in letting personal financial considerations dominate over national economic performance in Presidential voting? This would be somewhat interesting, just in itself. However, there are other flukes about l984, as well.
For example, it had the third lowest-turnout in United States history--behind only l924 and l988. Yet it also had the record low post-World War 2 turnout for third party presidential candidates.
This is especially interesting in light of Reagan's low standing in the polls for two solid years among groups of voters (who weren't always re-polled in the third year of his administration), as to his popularity and credibility levels; this is also intriguing in light of data which have consistently shown that voters in general have been disillusioned with the two major parties since l968.
One would think, given all this (and the additional factor of the large Federal deficit during Reagan's first term) that there should have been a sizable increase in, for example, the Libertarian Party vote in l984 from what it had in l980. That vote had been up from l976, and that in turn, up from l972.
Yet this proved not to be the case, but only in l984: by l988 this trend of a steady increase in third party voting was back up. Given the extreme differences in the perceived positions of the two candidates, and their non-positions on a number of others, the door was definitely open for much more third party voting in l984 than was actually seen.
In yet another anomaly about l984, the level of homelessness in the industrial Northeast was recorded as approaching that of the milder days of the Great Depression and was found, during Congressional hearings in 1987, to have increasingly involved non-drug-abusing individuals and families rather than the classic alcoholic or drug-abusing drifter.*
More interesting still is the fact that federal government policy from l983 to l987 was essentially based on the idea of "breaking up" any federal funding for local low-rent housing, and replacing it with local funds.** Major efforts were made to come up with local private funds for the "shadow market" in low-rent housing*** during this time of high layoffs and the replacement of high-paying jobs with extremely low-paying ones in the industrial Northeast. It is also interesting that there was a major banking scandal in the area of funds for housing during this period of time (the "HUD scandal").
In any case, I digress a little bit from the main conclusion that personal thinking about the economy does impact on voter decisions in elections. On that score, Romeno and Stowbaugh's study does seem to make a little more sense of these l984 data in putting them into a more "generalized" conceptual model than the Lanoue study had done. Certainly, l984 seems somewhat less the exception to the rule in their later, broader context than in Lanoue's earlier, narrower one.
*See for example,1987 World Almanac, "Plight of the Homeless in l986," by June Foley 922-23.
**Foley 923: The meeting referred to was held in Philadelphia, (Pennsylvania).
***See "The Shadow Market in Housing," by William C. Baer, Scientific American, Vol. 255, Number 5, November, l987, pages 29-35: the final paragraph article seems chillingly prophetic as to the impact of l980's federal budget cuts on low-rent housing funds.Go back to The George Bush-Undercurrents Website