One of the important steps towards constructing an appriopriate mathematical model for the real-life data is to determine the structure of dependence. A conventional way of gaining information concerning the dependence structure (in the second-order case) of a given set of observations is estimating the autocovariation or the autorcorrelation function (ACF) that can provide useful guidance in the choice of satisfactory model or family of models. As in some cases calculations of ACF for the real-life data may turn out to be insufficient to solve the model selection problem, we propose to consider the autocorrelation function of the squared series as well. Using this approach, in this paper we investigate the dependence structure for several cases of time series models.