85
the multilevel estimation. Further research will try to unfold these subnational
mechanisms that explain so much variation in this model.
Table 6.10: Results of the Mayoral Models
|
Clustered SE |
Random Intercept | |
|
MayoralCandidate |
0.36*** |
0.30 *** |
|
(0.13) |
(0.05) | |
|
Committee Chair |
-0.35*** |
-0.32 *** |
|
(0.09) |
(0.03) | |
|
Distance to Median |
-0.43*** |
0.05 |
|
(0.09) |
(0.05) | |
|
Distance to Majority Median |
0.09 |
-0.32 *** |
|
(0.13) |
(0.04) | |
|
PJ member |
0.12* |
-0.00 |
|
(0.07) |
(0.02) | |
|
Provincial Party Member |
0.22 |
-0.06 |
|
(0.14) |
(0.06) | |
|
District Magnitude |
-0.00 |
0.03 *** |
|
(0.01) |
(0.00) | |
|
Constant |
-2.11*** |
-2.82 *** |
|
(0.09) |
(0.49)___________ | |
|
Observations |
101,533 |
101,533 |
|
Random Intercept |
5.636 | |
|
(2.374) | ||
|
Pseudo-R2 |
0.01 |
0.04 |
At the gubernatorial side, things also perform as expected. None of the models
shows statistical significance for the main covariate of interest. Thus, gubernatorial
expectations do not have any impact over expected behavior towards legislators'
districts. The only covariate that is robust across estimations, and also consistent with its
performance in the municipal model, is Committee chairmanship: the sign is always
negative and statistically significant. In more theoretical terms, we would say that