Ruth Göhlen: Fraternal interest groups and violent conflict 47
As a condition for the formation of fraternal interest groups Paige and Paige
mainly take into account economic criteria. As the term suggests, the indicator of fra-
ternal interest group strength is of the ordinal type. But even with the indicator of
Paige and Paige the effect of fraternal interest groups is weak and insignificant, as my
Statistical analysis with the Standard Cross-Cultural Sample (SCCS) shows (cf. Ap-
pendix, Tab. 3).
It moreover has to be considered that one of their defining elements of those
groups, also called ‘largest effective kinbased political subunit’, is the explicit decision
making of those groups whether or not to engage in or to wage war. For our purposes
It is necessary to omit this constituent part of the definition, because of its tautological
effect in relation to the hypothesis under consideration.
Control variables
According to most of the existing cross-cultural studies on the fraternal interest
group-hypothesis certain conditions can prevent these groups to wage war. The
authors of the first of these studies, Thoden Van Velzen and Van Wetering (1960),
argued that social stratification could have such a preventive effect, since in stratified
Cultures “the ruling groups can monopolize power” or there can be “differences bet-
Ween the behavior towards members of one’s own, and towards those of other castes
Or classes” (S. 189). Consequently, these authors selected only societies without social
Stratification for testing the hypothesis. And they go a rather high correlation value
(® = 0.73, p «0.001, N = 51, cf. Appendix, Tab. 1). Thus the failure of the tests so far
furthermore could be caused or conditioned by the inclusion of stratified societies.
If we omit stratified cultures, the indicator of Paige and Paige produces a strong
and significant effect (cf. Appendix, Tab. 3, Pearson's r = —0.494, p = 0.011, N = 36 /
Tesp. after dichotomizing: r = —0.587, p = 0.0001, N = 36).
: The effect of the residence-indicator on war is raised too, but remains weak and in-
Significant (cf. Appendix, Tab. 2).
_ Later cross-cultural studies have replaced social stratification with political centra-
lization (Otterbein and Otterbein 1965; Otterbein 1968; et al.). The argument is: cen-
tralized political systems prevent fraternal interest groups from waging war. This
Seems quite plausible. Surprisingly the political centralization-variable does not pro-
Uce the expected results (cf. Appendix, Tab. 1). The variables used as context vari-
ables in the tests consider only the presence of formal political institutions or similar
Aspects, but not its effectiveness of interest for intervention. The question is, whether
the neglect of these aspects in fact accounts for the weak effects.
Since this context variable is very convincing, further studies are needed, studies
that belong to the domain of political organization. At present my collegue Klaus Kra-
Jewski is constructing a variable, which in my opinion can be well applied to the hypo-
thesis discussed here.