Is there such thing as a fuzzy version of prop.test? Let's say I have the output from cmeans, i.e. set of clusters. Every object belongs to each cluster "in a fuzzy way". object cluster1 group dependentvar obj1 0.21 A 1 obj2 0.84 A 0 obj3 0.90 B 1 obj4 0.02 B 0 Having data in above form, is it possible to perform (something like) a prop.test against groups A and B, inside cluster1? Since every object belongs to cluster1 "to some degree", every observation should apply to the test "to some degree". Is there already a test in R that does that? Can you recommend a relevant reading? -- Maciej Blizi??ski <m.blizinski at wit.edu.pl> http://automaciej.blogspot.com/