IOO
Municipal Models
The first point that must be highlighted for this test is that the relatively small
size of the sample (N=98) is likely to be slightly erratic. This does not imply that results
will be biased, but it must be highlighted that it is not fair to expect extreme robustness
from the data. Regarding hypothesis 7.2, results are mixed. While the covariate
measuring legislative performance as a percentage of local bills is statistically
insignificant; increases in the net number of bills appear to be harmful for an electoral
victory. Contrary to the expectations, this barely significant finding casts some doubts
about the true role of bill submission over municipal performance. Predicted
probabilities show ridiculous and unrealistic percentages, forecasting an 88% of chances
of victory for a legislator who submitted zero bills. Given the mentioned relative scarcity
of data and the .10 significance (beyond the tolerable boundary of many researchers), I
will not consider this coefficient to be indicator of a substantive effect, but as evidence of
no effect. Such a finding forces a réévaluation of the scenario depicted in the theoretical
discussion. Should bill submission really affect general elections' results, or should it just
be influential at the stage of pursuit of the candidacy? This reasoning will be expanded
in the discussion section.
Contrary to the evidence found in the gubernatorial model, hypothesis 8.3 is not
supported by the municipal estimations. Following these results, party incumbency
should not affect a legislator's success in her municipal race. Is a higher rate of party
alternation a valid explanation? Cao's (2000) piece confirms that there is a huge variation
in party reelection across time and provinces. This dimension is not a part of the
purposes of this piece, but it is a substantive topic for further research.