as well as in public policy circles, NGOs, and democratic social movements

around the world.

111

State Repression and Domestic Democratic Peace

Suffrage

0.08

Change in Probability of Repression

0.06

0.04

0.02

0

1 2 3 4 5 6 7 8 9

-0.02

-0.04

Categories of Repression

Figure 4.1 The In¬‚uence of Suffrage on Repressive Lethality

Suffrage

In order to understand how Suffrage in¬‚uences repression, I introduce the

Bollen and Paxton indicator into the basic model discussed in Chapter 3.

All of these results are presented in Table 1, column 1 in Appendix 2. To

focus the discussion and present the statistical ¬ndings parsimoniously, in

the next two chapters I do not provide the ordered probit results for the

variables in the basic model nor do I provide graphic representations of

predicted probabilities for the base model in the equations estimated.1

What does the analysis reveal? Mere statistical tables are not suf¬cient

for understanding causal in¬‚uences on ordinal variables, I provide in Fig-

ure 4.1 the predicted change in probabilities (i.e., the probability of being

in different categories of repression as Suffrage moves from its minimum

to its maximum).2

Recall from the last chapter that one reads these ¬gures by simply looking

at a speci¬c category of repression (i.e., 1“9) “ the x-Axis, with reference

to the independent variable of interest (Suffrage in this case). One then

1 Those interested in how the base model varies across equations can view the general simi-

larities of the results by observing the tables in Appendix 1.

2 A likelihood ratio test shows that the model including Suffrage is “better” than the basic

model (that is, the one that excludes Suffrage). This aspect of Voice thus diminishes state

repressive behavior.

112

Democratic Paci¬cation

identi¬es the corresponding change in the probability of achieving this

category derived from moving the relevant explanatory variable from its

minimum to its maximum “ the y-axis. If the value is positive, then, given

an increase in the relevant independent variable, the likelihood of a govern-

ment using a particular strategy of repression is increased. Correspondingly,

if the value is negative, then, given an increase in the independent variable,

the likelihood of a government using a particular strategy of repression

decreases.

When Figure 4.1 is considered, the empirical ¬ndings are clear. Support-

ing Hypothesis 1 that Voice encourages the likelihood of lower level repres-

sion while discouraging the likelihood of higher level repressive activity,

the results suggest that at higher levels of Suffrage, a country has a higher

probability of being in a lower repressive category, all other variables held

constant. The magnitude of these in¬‚uences, however, is quite small. Com-

paratively, for example, the in¬‚uence of GNP per capita on achieving the

lowest three categories was 5 percent, 25 percent, and 1 percent, respec-

tively. Overall, these ¬ndings offer weak support for the peace proposition.

The in¬‚uence of Suffrage across values of the dependent variable is

important to identify, for it is somewhat more complex than my hypothesis

suggests. According to the earlier discussion, democracy should increase the

probability of achieving low-level repression (category 1 and/or 2 on the

dependent variable), which decreases the probability of achieving repressive

behavior that was more lethal (3“9). From the statistical results, the prob-

ability of achieving the least repressive category (1) and the probability of

achieving the next category (2) are increased by Suffrage, but the in¬‚u-

ence of this variable is not particularly strong (yielding a 1 percent chance

of achieving category 1 and a 6 percent chance of achieving category 2).

Against expectations, Suffrage also increases the probability of achieving

repression where violence is low but restrictions are high (category 3); the

magnitude of in¬‚uence is lower than that exhibited in categories 1 and 2

(approximately less than half a percent). The ¬rst two ¬ndings are directly

consistent with the paci¬cation argument provided above (Hypothesis 1),

but the third is not. Indeed, the latter ¬nding provides preliminary sup-

port for Hypothesis 5, which suggests that state violence is more likely

than political restrictions to be reduced by democratic characteristics of

the political system.

The model provides no insight into repression where violence is mod-

erate and restrictions are limited because of the low number of cases (cat-

egory 4) “ this is similar to category 7. The results are largely consistent

113

State Repression and Domestic Democratic Peace

Table 4.1. The In¬‚uence of Suffrage on the Most Likely Current Value

of Repression

1 2 3

Most Likely Value of Most Likely Value of

Current Repression Current Repression

When Suffrage = 0 When Suffrage = 100

Lag Repression

1 1 1

2 2 2

’

3 2 2

’

4 2 2

2’

5 5

’

6 5 5

’ 2’

7 5

’

8 6 6

8’

9 9

= change; ’ = negative change; + = positive change.

Legend:

above this category, however, in terms of the expectations of Hypothesis

1: the most lethal forms of repression are consistently reduced by Suffrage.

Again, these in¬‚uences are relatively marginal “ albeit somewhat variable.

For example, when the level of state violence and restrictions is moderate

(category 5), the in¬‚uence of Suffrage is somewhat weak at approximately

2 percent as compared where the level of state violence is moderate but

where restrictions are high at approximately 3 percent (category 6). Above

this category, the in¬‚uence of Suffrage decreases. Thus, where violence

is high and restrictions are moderate (in category 8), the probability of

attaining this value of repression is decreased to below 2 percent when Suf-

frage is increased from its minimum to its maximum, whereas in category

9 (where both violence and restrictions are high), the negative in¬‚uence is

even weaker.

These results also convey additional information. For example, the pre-

vious discussion assesses the in¬‚uence of Suffrage on repressive lethality as

the former increases from its minimum to its maximum, but this does not

consider the fact that governments start at different levels of repression.

Some governments might be at level 2 (low political violence but moder-