24
Table 1
The Impact of Voice and Influence on Policy Satisfaction
B
Robust
Standard Error
Significance
Non-Influential Quiet Voice
-0.537
0.219
0.014
Non-Influential Loud Voice
-0.449
0.282
0.111
Influential Quiet Voice
-0.374
0.268
0.163
Neutrality 0.669
0.172
0.000
Efficiency 0.302
0.085
0.000
Honesty 0.747
0.204
0.000
Competence 0.432
0.169
0.010
Length of Residence
0.007
0.004
0.096
Constant -3.274
0.424
0.000
Number of Cases
892
Log Likelihood
-529.32
Pseudo R-Square
0.140
Chi-Squared (Probability < 0.000)
362.70
Notes: Dependent variable is Policy Satisfaction. Results are from logistic regression with standard errors corrected for
clustering on city of residence. Bolded entries are significant at the .05 level; entries in italics are significant at the .10
level (all significance tests are two-tailed). See Appendix A for descriptions and coding of all variables.
Figure 3
Predicted Probability of Policy Satisfaction
0.60
0.51
0.49
0.47
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
Non-Influential Quiet Voice
(p<0.05)
Non-Influential Loud Voice
(n.s.)
Influential Quiet Voice
(n.s.)
Influential Loud Voice
(Baseline)
Notes: Probabilities calculated using the logistic regression results from Table 1. Control variables were set to their
mean values (modal values for dichotomous variables), and the Voice-Influence variables were changed from 0 to 1
individually. The “Influential Loud Voice” category serves as the baseline, excluded category. Only the “Non-
Influential Quiet Voice” variable has a statistically significant regression coefficient.
Convention