Overall, there is little evidence to suggest a strong connection between NGOs and voter
turnout. However, as the next section will argue, that does not necessarily show that NGOs are
not having an effect on political participation and mobilization.
5. Protest Results
The results of the model estimating the effect of changes in numbers of NGOs on protest
are more startling. The variable for changes in NGOs is positive and significant, using robust
standard errors, regardless of the control measures used for social indicators. On average, an
increase of one NGO in a municipality is associated with an increase in .15 protests. Or, more
intuitively, an increase of 20 NGOs would predict an average increase of around 3 protests in a
year.
Of the control variables, the base number of NGOs is positive and significant, which
lends further support to the idea that NGOs are related to protest activities. Municipalities with
high total numbers of NGOs to start with are more likely to experience protest.
Three different measures of social indicators and level of development yield remarkably
similar results. Whether measured as percent of houses with electricity (Level of Development),
level of education, or percent unemployment, the main results of the model hold. (see models 3-5
in table one for a comparison). It is somewhat surprising that level of development has a negative
coefficient, suggesting that less developed municipalities are less likely to experience protest.
However, this is most likely a function of the fact that protest in very small municipalities might
be missing from the dataset (because it does not make national news), or because the decision to
take to the streets in protest might be a very different one in the smallest, least developed cities.
This possibility is supported by the large and significant negative coefficient for population size;
the very smallest cities do not have recorded experiences of protest.
19
Convention