In a correspondence analysis one evaluates which variable's category corresponds with one category of the other variable. In essence, one obtains two (or more) dimensions (or factors) that determines the hidden structure in the relation between both variables. Each category of each variable is represented by one point in a Catersian plane where the axes represent each dimension (or factor).
So, how close should the euclidian distance between two categories be to consider correspondence effective?
I think that a good approximation is:
A very strong correspondence of 0.0 to 0.1
A strong correspondence of 0.1 to 0.2
A middle correspondence of 0.2 to 0.3
A weak correspondence of 0.3 to 0.4
A very weak correspondence of 0.4 to 0.5.
Because, for example, a distance of 0.500 implicates a minimal distance in one dimension of 0.354 and a maximal distance of 0.500.
And, a distance of 0.100 implicates a minimal distance in one dimension of 0.071 and a maximal distance of 0.100.
And, a weight score of 1.000 determines completely one dimension (or factor).