Results
The average size of a block group from
our sample was 0.04 square miles, with an
average population of 507. The population of
the block groups was predominately Black.
The block groups tended to have high rates
of household poverty and low educational
levels. Table 1 describes each of the variables
and lists their correlations with gonorrhea
rates, the broken windows index, and
the poverty index. Table 2 lists descriptive
information for variables used to create the
poverty and broken windows indexes; the
Cronbach α’s for these are 0.64 and 0.81,
respectively. With the exception of percentage
of females and percentage of population
younger than 25 years, all of the variables
were associated with both the poverty index
and the broken windows index.
Least squares regression indicated that
broken windows index shows a much stronger
relationship with gonorrhea rates than does
poverty (R2 = 0.241 for poverty and gonorrhea
vs R2 = 0.424 for broken windows and
gonorrhea; P = .000 for both correlations).
The results of the multiple regression
are shown in Table 3. After all the variables
were entered, only broken windows remained
significantly related to gonorrhea rates. Marital
status and off-sale alcohol outlet density
had a marginal but independent effect on
gonorrhea rates. Figure 1 illustrates how the
variables are associated in a hypothetical
model.
ANOVA indicated that gonorrhea rates
were significantly higher in neighborhoods
with both high broken window indexes and
high poverty indexes (46.6 per 1000 vs 25.8
per 1000 in neighborhoods with low broken
window and high poverty indexes [P <
.001]). When block groups with low broken
windows indexes and high poverty indexes
were compared with those with low broken
windows indexes and low poverty indexes,
there were no significant differences in rates