19
5. Results and discussion.
I run a total of two
models with both presented in Table 2. Model 1 in Table 2 is the
analysis of riots, while model 2 is the analysis of demonstrations. As can be seen from
these models, both hypotheses are supported. Even controlling for the different factors,
the evidence suggests that as the number of effective parties increases the number of both
riots and demonstrations are reduced. In both models the effective party variable is
significant and in the hypothesized direction.
The results on the other variables in the model are less straight forward, and in
some cases less expected. In both models Per Capita performs as expected, as a
significant and negative relationship exists. ELF is significant and in the expected
direction when used in the model on demonstrations. However, in the riots model ELF is
insignificant. The dichotomous sixties variable is also significant and in the expected
direction in both models.
The results are more unexpected in regards to the total population and the energy
consumption variables. Energy is significant in both models (though barely in regard to
demonstrations) yet it is in an unexpected direction. With greater energy consumption
comes an increase in the number of riots and demonstrations. This is opposite to much of
the literature on development and violence.
The results on the total population variable are equally puzzling. The variable is
insignificant in regard to the number of demonstrations. However, it is significant and
negative in regards to riots indicating that as the population increases the number of riots
will decrease. This is contrary to what Powell (1982) and Krain (1998) found in regard
Convention