We examine an asymptotic analysis of differentiable econometric models using the distance and direction (DD) method introduced by Cho and White (2012), in which the conventional analysis for the quasi-maximum likelihood estimation and inference can be treated as a special case. We extend their approach and revisit the conventional quasi-likelihood ratio,Wald, and Lagrange multiplier test statistics through a different perspective. This new perspective is further analyzed in a unified framework, and we exploit this to introduce new classes of test statistics.