family income behave the same way in Model 2 and in Model 1, except that the impact of median family
income is smaller in 2001 in the second model.
These models contain the racial and Hispanic origin measures. In 2001, higher proportions of
non-white residents were associated with higher turnout, though the standardized coefficients suggest the
effect is weaker than the demographic and economic measures. The situation is different in 2005. Black
and Hispanic voters were strongly mobilized in the 2005 election, especially the proportion of Hispanic
origin. The Asian population was also mobilized but not as heavily. The R
2
increases from Model 1 to
Model 2. from .258 to .333 in 2001 and from .160 to .241 in 2005. Compared to the New York models, we
are still explaining less of the variation in turnout in Los Angeles, even with measures for racial and ethnic
composition.
Model 3 uses specific racial and ethnic group variables. In 2001, native born blacks in Los
Angeles were mobilized. While they remained positively mobilized in 2005, the effect was weaker. The
proportion of Iranian and Armenian residents in were negatively associated and significant in both
elections. While Russians were mobilized in 2001, they were not in 2005. The coefficients show that
turnout was positively associated with the proportion who are Chinese; in 2001, the effect was significant,
but not so in 2005. The same pattern is visible for Koreans, with weak significance in 2001 and no
statistical significance in 2005. It appears that immigrants from the Middle East and Asia have been less
consistent over the elections.
The last groups we consider are Hispanics in Los Angeles. The largest Hispanic population in
the city are Mexicans. Mexicans are well mobilized in both elections; looking at the standardized
coefficients we see that the positive effects are the strongest in the models. Salvadorans and
Guatemalans are also mobilized into Los Angeles politics and the effects appear strong. The R
2
for
Model 3 increases to .354 for 2001 and .273 in 2005. The increase is only a slight improvement over the
second model.
We conclude from this analysis that, while underlying features of social structure pertinent to each
setting, but roughly similar across the settings, explain the continuities of electoral engagement across the
Convention