The Record of American Democracy, 1984-1990
(ROAD)
[Named "Best Political Science Research Website" by the American Political Science Association (CMS), 1999]
Gary King and
Bradley Palmquist
With Contributions From Project Leaders: Greg
Adams,
Micah
Altman,
Kenneth
Benoit,
Claudine Gay,
Jeffrey B. Lewis,
Russ Mayer,
and Eric
Reinhardt
The Record of American Democracy [A Brief Description]
by Gary King - Bradley Palmquist
Published in PS: Political Science and Politics
and Sociological Methods and Research
A massive new aggregate data set on American politics is now
available. Our Record Of American
Democracy
(ROAD)
data include election returns, socioeconomic summaries,
and demographic measures of the American public at unusually low
levels of geographic aggregation. The NSF-supported ROAD project
covers every state in the country from 1984 through 1990 (including
some off-year elections). One collection of data sets includes every
election at and above State House, along with party registration and
other variables, in each state for the roughly 170,000 precincts
nationwide (about 60 times the number of counties). Another
collection has added to these (roughly 30-40) political variables an
additional 3,725 variables merged from the 1990 U.S. Census for
47,327 aggregate units (about 15 times the number of counties) about
the size one or more cities or towns. These units completely tile the
U.S. landmass. This collection also includes geographic boundary
files so users can easily draw maps with these data.
We find it remarkable that the electoral record of the world's leading
democracy is routinely lost or discarded. Election returns in the
U.S. are collected by precinct and passed on to county offices in
every state. In these county offices, the official electoral record
then gets stuffed under desks, recycled, occasionally put into
archives, or most often discarded. For the first time, a substantial
piece of the entire electoral record of American democracy has been
preserved. We hope someone (or our elected officials) takes on the
task of institutionalizing the formal preservation of this record.
For now, we hope the scientific community will take advantage of this
unprecedented opportunity.
The ROAD data represent an opportunity for political scientists,
geographers, quantitative historians, sociologists, and others to
learn about electoral behavior, the political characteristics of local
community context, electoral geography, the role of minority groups in
elections and legislative redistricting, split ticket voting and
divided government, elections under federalism, and numerous other
topics of central importance to many disciplines.
Some examples:
- With few exceptions, scholars until now have had access to
district-level (i.e. state, county, or constituency) electoral
information at best, usually for only one office at a time.
Presidential election results broken down by congressional districts
are impossible to obtain except for a few recent years, and are of
dubious quality; more detailed disaggregation is usually
unobtainable. In contrast, our data can provide presidential (and
other) election results broken down by the much smaller State House
districts and even show detailed geographic variation across
precincts within a State House district.
- A recent state legislative data collection project led by
Malcolm Jewell (1992) provided valuable district-level data, from
which scholars have learned an enormous amount. By continuing in
this tradition, precinct-level data will increase the resolution of
our knowledge of electoral politics substantially. In contrast to
data on the 50 States, 435 U.S. House Districts, 1,916 State Senate
Districts, 3,139 counties, and even the 4,675 districts of the lower
house of state legislatures, the approximately 170,000 precincts in
the U.S. provide considerably more detailed information. They
contain information about small, local communities, with much more
variation than the higher level aggregates.
- Scholars using electoral data recognize its geographical nature,
but they have only rarely been able to access geographical
information. As a result, the vast majority of published analyses,
even those on topics such as redistricting or political geography,
have necessarily ignored the geographic placement of districts.
Maps have not had a central place in the study of American politics
since V.O. Key was writing. The ROAD data enable scholars to study
the geographic nature of American politics and to draw maps easily.
That is, not only are precinct-level data available, but we provide
the data in geographic formats, when possible, providing information
on local context. In particular, scholars will be able to use
mapping software, such as ArcView or MapInfo, to analyze
geographical features of American politics and to merge them with
other types of geographical data.
- Scholars will be able to use these aggregate data to draw
inferences about individual behavior using newly available methods
of ecological inference (King, 1997) and associated public domain
software programs (available at
http://GKing.Harvard.Edu).
Survey research has taught us a great deal, but as data on random
collections of isolated individuals from unknown geographic locations,
they miss much that the ROAD project can provide. To put it
differently, if you were an ambitious graduate student in the late
1940s or before interested in the quantitative study of American
politics, you would probably be drawing maps, doing detailed studies
of local politics. If instead you (like almost everyone in the field
today) started any time during the second half of this century, after
Robinson's (1950) ecological fallacy article and following the advent
of modern survey research, you likely became a survey researcher.
Today, the literature is dominated by survey analyses, but with new
aggregate data and methods, we all have many new opportunities to
redress this imbalance.
- For the first time, scholars will be able to study data from
numerous offices at many different levels of aggregation -- from
precincts, to state assembly districts, to state senate districts,
to U.S. House districts, or to states. (Counties and other
aggregation levels are also possible.) Even without survey data,
this will make it possible to study how the same voter groups cast
their ballots across many different offices. ROAD data will enable
more detailed studies of split ticket voting and of the factors
leading to divided government at many levels, for any or all states.
- The ROAD data should make possible many new studies of
legislative redistricting, and associated analyses and forecasts of
political and racial fairness, compactness, the consequences of
equal population constraints on gerrymanderers, and related issues.
- Finally, this is the first data set to be generally available to
the academic community that is on par in terms of quality and
quantity with the data politicians and political strategists have
been using for decades to target campaign resources. As a result,
this data set could also produce a new, more detailed studies of
campaign strategy, but on a massive and comprehensive nationwide
scale.
In part because this data set is of such exceptional value, and in
part because it would take many researchers many lifetimes to exploit
it fully, we are releasing it prior to publishing much from it. The
data have been deposited in the ICPSR.
[Supported in part by NSF SBR-9321212]