8
scores less than 40 as "1" and the other cases as "0". The inclusion of these two dummy
variables will allow for a comparison of the effects of ideology on the more conservative
Democrats and less conservative Democrats.
Findings
How does support for Hispanic issues vary across the ethnicity of the House member
(measured by Black caucus and Hispanic member) in comparison to NHNB Democrats while
also controlling for percent Hispanic of the district? An ordinary least squares regression
analysis of these variables on Hispanic support is presented in the Model 1 column of Table 1.
[TABLE 1 ABOUT HERE]
Surprisingly, the analysis indicates that members of the Black caucus are more likely to
support Hispanic issues than Hispanic members of Congress. Furthermore, consistent with the
literature, Hispanics and NHNB Democrats support Hispanic issues at approximately the same
levels. Interestingly, the percent Hispanic of the district does not have an independent effect.
The constant reveals the low level of support from Republicans (excluding the 3 Hispanic
Republicans) for Hispanic issues.
If party has such an important impact, what will happen if the Hispanic variable is
recoded so that it only measures Hispanic Democrats? Perhaps descriptive representation is
important, but some are better at providing it than others (Dovi 2002)? In other words, perhaps
Hispanic Democrats provide better direct substantive representation than Hispanic Republicans.
An analysis with the Hispanic variable recoded to include only Hispanic Democrats is presented
in the Model 2 column of Table 1.
Convention