Because some countries in which conflicts were observed were newly emerging
countries, and the origins of their crises lay in colonial conditions prior to their independence, we
found we could not reliably connect such conflicts to conditions two years prior to event onset.
Even in cases where instability clearly began after formal independence, the country-year
structure of most social science data sets does not allow us to reliably observe conditions prior to
the onset of many crises in newly independent countries. Thus we only included in our samples
for model estimation those cases of instability onset where it was possible to measure conditions
in that same independent country two years prior to the onset of conflict.
9
Similarly, because we
did not collect data for years prior to 1955, we did not have data regarding conditions prior to
onset for conflicts that were either ongoing or began just after our observation period. So events
that were evident after 1955, but whose onset occurred in 1956 or earlier, were excluded from
our analysis. We thus excluded a total of 24 of our consolidated episodes on the grounds that we
could not obtain data that was clearly antecedent to the event onset, either because the country
was not independent or because the episode of instability began before our period of
observation.
10
In addition, we should note that for many nations and time periods, we must rely on
estimated, smoothed, or interpolated data. The Task Force has invested heavily in screening
data, cross-checking data from various sources, and subjecting data to expert review, in order to
obtain the most complete and reliable country-year data that it is possible to obtain.
Acknowledging these issues, we have tried to identify models that are parsimonious, robust, and
highly significant.
9
A sensitivity analysis using samples including and excluding these new-country cases suggests that this decision
has little effect on the mix of variables and magnitude of coefficients in our models.
10
Regrettably, these exclusions include a number of cases of great historical interest, including the Cuban
Revolution (start date 1956), and the Algerian Revolution (colonial event).
Convention