What is meant by “imbalanced?” Is this in terms of the balance of response/non-response on the dependent measure or balance of response/non-response on the predictor measures?
A. Either one. If the proporxxxtions of responses and non-responses differ greatly, it is one kind of imbalance in the data. The other kind of imbalance is that the distribution (univariate or multivariate) of predictors under response (or non-response) is highly asymmetrical. Exact methods are called for in either case.
