ask a question with this: think of the problem this way: What is the probability of observing a transition to democracy given that a country is always observed as a dictatorship? The answer is zero. What is the probability of observing a transition to dictatorship given that a country is always democracy? Again, the answer is obviously zero. We appear to "learn" nothing from these cases, so they drop out of the analysis. But do we really learn nothing from these cases