Dodd is generally credited with providing clear empirical support for the proposition that, in the period after the Second World War, minimal winning coalitions in European party governments will tend to last longer in office than non-minimal winning coalitions. There has been a considerable body of research on this and related questions. Dodd, as well as most other authors treating cabinet coalition formation, has attempted to model features of cabinet formation such as cabinet duration or cabinet type (e.g. minimal winning v. minority government v. oversized coalitions) largely or entirely using data pooled from all cabinets in each of a number of different countries over some considerable time period. One difficulty with this method is that system-level variables (such as number of parties, or the presence of large anti-system parties), which might be able to explain aggregate-level between-county variations in cabinet type or cabinet durability, are not likely to be the same variables that are useful in explaining within-country differences. A second difficulty is that certain system-level characteristics such as effective number of parties or number of cleavage dimensions are highly correlated with both cabinet type and cabinet duration and, as a consequence, these variables are highly correlated with one another when pooled cross-national data are used. Thus, if the analyst is not very careful, results of pooled cross-national data may lead to mistakes about causal structure and a confusion of within-country and between-country effects.