Even though classifications need to be relevant, there are no firm guide lines for relevance. One important caution is that the classification should clearly flow from the analyst,s views of the causes of the problem. This will allow for transparency of assumptions and avoid narrow definitions of prob lems. For example, different classifications of poverty problems can be derived from causal assumptions that it is a problem of inadequate income, cultural deprivation, or psychological motivation. Should the analyst focus on only one type of classification, contrary assumptions are ignored, and that can lead to the formulation of flawed policies. Ignoring relevant data may also lead to faulty explanations of poverty rates (e.g., among the elderly and disabled versus families with children) and a corresponding inability to explain groups differential access to program benefits with cuts in federal Medicaid expenditures (Johnson 1997, 15).Failure to classify data relevantly and to make under lying assumptions explicit can produce erroneous policies.