Adopting a Nested Logit model implies that the simple MNL model does not clearly represent the data set. To test if MNL model is preferable over a Nested Logit model, a t-statistics for the hypothesis that the logsum parameters are equal to one. The logsum parameter is a function of the underlying correlation between the unobserved factors for pairs of alternatives in the same nest. If the logsum parameter is equal or greater than 1, it means that the correlation in that nest is zero. Therefore, the Nested Logit model then collapses to the MNL model.