level discrimination changes from 0 to 1.
government “some of the time.” But the likelihood of “never” trusting government changes
dramatically as self-identification and perceptions of discrimination change. It rises from 0.08 to
0.23 when a black respondent who identifies as American goes from perceiving a low level of
discrimination to perceiving a high level of discrimination, and then it diminishes to 0.13 when
that same respondent identifies primarily as black.
[Table 5 About Here]
Latinos and Asians share more similarities with whites than blacks do. Education and
being a Democrat lower trust while generalized trust raises trust in government. The model also
shows that Latinos are slightly more trusting of government than Asians. And as Michelson
found, acculturation seems to lower trust, as indicated by the negative coefficients on
complicated role. Here, panethnic discrimination diminishes trust rather than personal
discrimination. But as with blacks, this effect only applies to people who identify primarily as
American. Identifying as Latino or as Asian neutralizes the damaging effect of discrimination.
Predicted probabilities of each level of trust for Latinos and Asians are displayed in tables
6a and 6b.
increase the likelihood of trusting government “some of the time” (0.37 0.57, for Latinos) or
“never” (0.02 0.06) while decreasing the likelihood of trusting government “most of the time”
(0.45 0.32) or “just about always” (0.15 0.05). It is important to note as well that a
17
Predicted probabilities are calculated using CLARIFY, holding all other variables constant at their means for
blacks (see King, Tomz, and Wittenberg 2000).
18
Running models for Latinos and Asians separately indicates that the impact of generational status is driven
entirely by Latinos.
19
Predicted outcomes are calculated using CLARIFY, holding all other variables constant at their means for Latinos
and Asians, with “speaks English at home” held constant at 1, while changing panethnic discrimination from 0 to 1.
18