29
Group-Level Analysis
My theory generated two hypotheses about the effects of group-level characteristics on the
likelihood of autonomy. I argued that increases in autonomy are most likely when groups are
unified or have many factions with distinct demands. I also argued that autonomy is most likely
when the self-determination group is organized through a coordinating group or institution.
To test these hypotheses, I use a logistic regression analysis on my random sample of self-
determination group/state dyads. In my ideal model, I include my two theorized independent
variables, number of factions making distinct demands and presence of a coordinating group or
institution. I also include any variables I think are theoretically related to both my independent and
dependent variables: whether there was an autonomy agreement in the last 5 years and the size of
the group.
In include a dummy variable in each year that indicates whether there was an autonomy
transfer in the previous five years. Although the analysis treats all yearly observations as
independent from each other, the political process that leads to a new autonomy arrangement
cannot be divorced from the past. Specifically, I expect that states/group dyads are less likely to
agree to a change in autonomy status when they have done so in the recent past. I also think that
transfers of autonomy can lead to further fractionalization of groups, and so the number of factions
with distinct demands will not be independent from previous transfers of autonomy.
I also include a variable for the size of the group. The size of the group may affect both the
likelihood of a new autonomy arrangement and the number of factions. Larger groups may have a
greater heterogeneity of demands just by the fact that they have more individuals with potentially
disparate preferences. Moreover, a group’s population and relative size may affect the likelihood of
a new autonomy arrangement. Relatively larger groups have hold more sway within the state and
Convention