The Ecological-Evolutionary Typology of Human Societies and the Evolution of Social Inequality

3.4Specialization

In contrast to the number of hierarchical levels, which represent the vertical complexity of a society, the degree of specialization may be viewed as representing horizontal complexity. Table 3.4 shows how Type of Society relates to the level of specialization. The specialization scale has 5 levels corresponding to (1) no specialization, (2) pottery only, (3) loom weaving only, (4) metalwork only, and (5) specialization in metalwork, weaving, and pottery. Note that the categories of this variable are not unquestionably ordinal (see the original discussion of this variable in Murdock and Provost 1971). Since the specialization scale in only available in the SS, percentages are estimated from fewer observations.
------ Table 3.4 about here ------
With respect to the sequence of four types from HG to AG the monotonic pattern is once again in evidence. For HG societies the modal pattern is one of no specialization (55.6%) while for AG the modal pattern is complete specialization of pottery, weaving, and metalwork (77.4%). SH and AH societies in between the HG and AG extremes. Note how SH societies range over a wide range in degree of specialization, from none (20%) to complete (2.9%). FI are in between HG and SH with respect to specialization, while HE resemble AH in this respect (although the comparisons for FI and HE involve small numbers of cases).

3.5Explanatory Power of Ecological-Evolutionary Typology

In order to secure a more quantitative estimate of the strength of the relationship between indicators of social complexity and the ecological-evolutionary typology Table 3.5 presents the results of regression analyses of the number of jurisdictional levels beyond the local community, and of the degree of specialization, each variable treated as if it were an interval variable. Since even the ordinal nature of the variables, especially degree of specialization, may be reasonably questioned this is certainly a highly conservative approach, since any deviation from the equal intervals assumption would attenuate the fit of the regression model and thereby lower the estimated explanatory power of the typology.
------ Table 3.5 about here ------
Results for hierarchy are on the left of Table 3.5. The ecological-evolutionary typology is represented by a set of indicators with values (0,1) for each type of society. HG is omitted as the reference category. The coefficients can therefore be interpreted as deviations of the dependent variable from the HG average. Note that all regression coefficients are significant at better then the .05 level, except for FI. The coefficients for SH (.374), AH (1.119) and AG (2.136) exhibit a pattern of increasing magnitude reflecting the monotonic sequence of Table 3.3. The degree of hierarchy for FI is not significantly different than for HG. The coefficient for HE is significant and positive, however, with a value intermediate between that of AH and AG. The fit of the model (R² = .407, adjusted R² = .403) is nothing short of astonishing given the random noise introduced in the dependent variable by deviations from the assumption of an interval scale and perhaps even from that of an ordinal scale.

The high degree of fit is interesting in a larger perspective, since the scale of hierarchy is quite similar to the classical typologies based on the nature of the political system such as the ones of Service (1962, 1975), Fried (1967), or Diamond(1998:268-269). It is theoretically significant that the ecological-evolutionary typology, constructed as it is entirely from variables related to subsistence and technology, should be able to reproduce so well other typologies based on entirely different variables related to levels of political organization.

Results for the specialization scale are shown on the right of Table 3.5. The pattern of coefficients is qualitatively very similar to that for hierarchy: there is a monotonic trend in the positive coefficients for SH, AH, and AG; the coefficient for FI is not significantly different from 0 (i.e., from that for HG); the coefficient of HE is positive and significant, with a value intermediate (this time) between that for SH and AH. The fit of the model (R² = .578, adjusted R² = .565) is even stronger than for hierarchy, despite the fact that the ordinal assumption is perhaps even more questionable for the specialization scale than it is for hierarchy.

4Social Rigidity, Inequality & Freedom

Section 3 examined features of society that could be grouped under the general heading of social complexity. Data analysis revealed a general pattern of increasing complexity along the HG, SH, AH, AG sequence, less complexity for FI, and a higher level of complexity for HE, closer to the one for AG. Other dimensions of stratification systems relate more explicitly to the invidious nature of inequality, being measures of social rigidity (manifested in inheritance of the position of headman, and inheritance of real property), presence of the practice of slavery, and two composite measures of the degree of stratification and the presence of despotism, respectively. We explore these dimensions in this section.

4.1Inheritance of Leadership Position

------ Table 4.1 about here ------

Table 4.1 shows how Type of Society is related to inheritance of the office of local headman (0 indicates no inheritance, 1 indicates that the office is inherited by the patrilineal or matrilineal heir). Looking down the rightmost column we find a pattern different from before. There is the beginning of a monotonic upward trend in inheritance of headman position in the sequence HG (52.6%), SH (53%), and AH (65.6%). But then the monotonic increase is reversed, down to 29.6% inheritance for AG. The figures for FI (55.8%) and HE (61.7%) are both higher than for AG. So instead of a monotonic progression with a peak for AG, we now have a dip for the latter! This pattern is unlikely to be due to random variation, as the AG percentage is supported by 98 cases. Furthermore, the dip in the likelihood of inheritance of headman position is not due (as one might surmise) to the fact that AG societies choose their leaders by a method even more invidious than inheritance. In fact, a tabulation (not shown) of the original variable (EA V72) on which Table 4.1 is based reveals that the modal method of succession in AG (37.8%, more than for any other type) is "election or other formal consensus, nonhereditary", which clearly represents a more "democratic" method than inheritance. We now look at the other variables to see if this new pattern is found for other dimensions.