Party Competition as a Function of Demographic Variables

The general hypothesis is that variation in party competition
can be accounted for in terms of socio-economic diversity. Since this
general hypothesis is not amenable to direct testing, eleven lowerlevel
hypotheses concerning the relationship between party competition
and variation in socio-economic structure were used. The county was
the unit of analysis. Party competition was measured for two levelsthe
presidential and the gubernatorial--for each county. Party competition
was defined at each level in teras of the percentage of the
vote which a county gave to the candidate who failed to carry the
county. For the presidential levelf an average was taken for the
1956 and 1960 elections. Por the gubernatorial level, an average was
taken for the elections between 1956 and 1960. The higher is the
average, the higher is party competition for the county.
Measured thusly, party competition was predicted to vary directly
with: (1) per cent urban; (2) median income for the county;
(3) median number of years of education of those twenty-five years of
age or older; (4) the degree of income variation for the county;
(5) the degree of variation in number of years of school completed by
those twenty-five years of age or older in the county; (6) the degree
of dispersion among the major occupational categories within the
county; (7) population density per square mile; (8) the per cent of the
labor force engaged in white-collar occupations; (9) the per cent Negro
of the county population. Party competition was predicted to vary inversely with (l) the extent to which urbanism in a county deviated
from fifty per cent ; and , (2) the extent to which the per cent employed
in white-collar occupations deviated from fifty per cent.
Multiple correlation and regression analysiB was used to teat
the hypotheses. A selected sample was used, consisting of the counties
of two sets of states. The Homogeneous Set consisted of the counties
of Maine, New Hampshire, Wyoming, and North Dakota. The Heterogeneous
Set consisted of the counties of New Jersey and Ohio. Also, these two
sets were combined to form a Combined Set.
The multiple correlation coefficients indicated that party competition
could be accounted for in terms of socio-economic diversity
to a significant degree at the presidential level but not at the gubernatorial
level for each of the sets. However, many of the relationships
were not in the predicted direction, indicating that party competition
will not necessarily increase with all types of socio-economic
diversity. A surprising finding was that the relationships between
party competition and the independent variables were often reversed for
the two ballot levels . It was suggested that perhaps two party systems
are operating in these states at the different ballot levels. The most consistent finding was the negative relationship-contrary
to the predicted positive relationship--between party competition
and the median educational level of a county. Several others of the
independent variables were found to have some tmportance, but there
were many inconsistencies in the findings between the ballot levels and,
to a lesser extent, between the sets of counties. The results auagest
that the simple linear model which vas used might need complicating; the
relationships may be more complex than can be accommodated by a linear
model.