Explaining military intervention
Much of the research that has been done on the military in Third World poli-
tics has been from a macro and quantitative standpoint. Here the literature is
based on statistical analysis of a large number of countries which are ranked
by variables that measure levels of instability or military intervention. These
are then correlated with socio-economic factors that seem likely to have
explanatory value. A good example of the genre is Wells’ study of 31 coun-
tries in Sub-Saharan Africa in 1970. Wells related, through multivariate analy-
sis, social and economic variables to military coupsin order to explain why
some African countries had experienced them and others had not. Population
size and growth rate, urbanization, literacy, mass-media availability, GNP per
capita, economic growth rate and measures of ‘centrality’ (the geographical
concentration of political and economic life) were taken as relevant indicators
of social and economic conditions that might be thought a priorito be related
to the likelihood of military intervention. Indicators of the size of the military
and police force, and of defence expenditure, were used to establish the sig-
nificance of organizational characteristics in military intervention. The level
of loans from the USA was taken as indicative of external influence on vul-
nerability to the coup d’état. The results were disappointingly inconclusive.
Even combining all the independent variables only explained 56 per cent of
the variation in coupactivity. The final explanatory model produced owed
more to case studies than statistical analysis (Wells, 1974).
This is fairly typical of the problems that are encountered when making
quantitative statistical analyses to explain and predict military coups
(Dowse, 1969, pp. 217–22; Jackman, 1986; Johnson et al., 1986; O’Kane,
1986). First there is the question of which states should be included in the
analysis. Were colonies ‘states’? Should they have been included in the pop-
ulations at risk from military intervention as they have been in some analy-
ses? Statistical explanations require the rule of ceteris paribusto be
satisfied, so the cases compared have to have things other than military
intervention equal. With nation-states ‘other things’ are hardly ever ‘equal’.
Grouping together ‘Black African states’, for example, could be said to pro-
duce heterogeneity rather than homogeneity. Aggregating a meaningfully
comparable collection of countries produces problems, the solutions to
which do not repay the effort put into finding them.
The second source of doubt about such analysis is how far the specific
measurable variables chosen represent dependent and independent factors
such as social mobilization, characteristics of the military itself, degrees
of political development, international economic dependence and so on.
Military Intervention in Politics 175