This preview shows page 1. Sign up to view the full content.
Unformatted text preview: within a market area (Herﬁndahl
1950). The index is calculated by squaring the individual shares, that is, the share
of each ﬁrm competing in a market, and summing the results. In the current
research, the Herﬁndahl Index was adopted to measure the concentration of
persons within professed religious groups using U.S. counties as the basic units
of analysis.
The general formula for the Herﬁndahl Index is:
Hj = ∑ S2 ij
S represents the number in each denomination within a county divided by
the total number of church adherents in the county, i represents the index of
summation over all religious denominations in county j. H represents the probability
that any two persons, selected at random, within a county will be adherents of
the same organized religious group (Iannacone 1991). For example, in the simplest
case assume one county has ﬁve discrete religious group afﬁliations, each with
an equal 20 percent market share of adherents. The indexed concentration for the
county is H = .202 + .202 + .202 + .202 + .202, or .20. Thus, if two persons were
selected at random from that county the odds are one in ﬁve, or 20 percent, that
they will be adherents of the same recognized denominational group.
A second county has three discrete religious afﬁliations with 50 percent,
30 percent, and 20 percent market shares, respectively. The index then would
be H = .502 + .302 + .202, or .38. The odds are 38 percent that two people selected
at random will have the same religious afﬁliation.
Still a third county has 25 discrete religious denominational afﬁliations
each with the following percentages in each denominational afﬁliation: one with THE IMPACT OF RELIGIOUS HOMOGENEITY 345 40 percent, one with 20 percent, one with 10 percent, and 22 with 3.15 percent
each. The index score is H = .402 + .202 + .102 + 22 (.03152) = .23. In this case,
the odds of selecting two persons at random with the same denominational
afﬁliations are 23 percent.
The index shows the probability that two persons selected at random would
share the same denominational afﬁliation, relative to the concentration of adherents
in a fewer or greater number of religious groups. When more adherents are in
fewer religious afﬁliations the index score is higher. Conversely, when adherents
are spread over a greater number of afﬁliations the index score is lower. The
index theoretically could range from 0.00, when there are no adherents with any
afﬁliations, to 1.00, when all adherents have a single afﬁliation.
Covariates
In order to examine more completely the effects of religious homogeneity
on divorce, the effects of other potential inﬂuences on the divorce rate must be
held constant. Divorce rates have been found to be associated with higher levels
of geographic mobility and, by extension, lower levels of community involvement and integration (Breault and Kposowa 1987; Glenn and Shelton 1985).
Hence, percent population change from 1980 to 1990 is used as an indicator
of population instability. It also has been generally established that a higher
concentration of young adults contributes to a higher divorce rate (Martin and
Bumpass 1989); thus we include the percentage of the population 15 to 34 years
of age. Likewise, race and ethnicity have been shown to have an impact on the
divorce rate (U.S. Bureau of the Census 1992b). Hence, we use percentage of
whites, percentage of Native Americans, percentage of Asian/Paciﬁc Islanders,
and percentage of Hispanics/Latinos as measures of race and ethnicity. The
percentage of African Americans is not included because of its high correlation,
r = .82, with percentage white. Previous research has shown that the relative
concentration of males and females within a population impacts the divorce rate
(Guttentag and Secord 1983; Trent and South 1989). Here, we use the percentage
of females in the population as an indicator of gender concentration. Also, given
evidence that the level of economic instability is positively associated with the
divorce rate (South 1985), we include four censusbased measures that are
assumed to represent different dimensions of this variable: percentage of the
civilian labor force employed in manufacturing, percentage unemployed,
percentage urban, and the median family income.
Because general area of residence apparently has an effect on the likelihood of divorce in the United States, we also include measures of region. Thus,
the four major regions that have been historically utilized by the Census Bureau
in reporting population data were selected for input: Northeast, South, West, and
Midwest, each coded as 1 = Yes and 0 = No. 346 LARRY C. MULLINS ET AL. Descriptive Results
Descriptive statistics (means and standard deviations) are shown in Table 1.
An examination of the intercorrelations indicates that these v...
View
Full
Document
This note was uploaded on 01/27/2014 for the course SOCI 3040 taught by Professor Lauramckinney during the Fall '13 term at Tulane.
 Fall '13
 LauraMcKinney

Click to edit the document details