Growth modeling has been of interest in many assessment programs, including both highstakes and low-states tests. Growth could be modeled using different approaches. This study models growth with an Item Response Theory (IRT) based approach that utilizes item response data. It investigates the impact of complex student clustering structure where students are cross-classified by two grouping variables on student growth modeling. Specifically, this study focuses on modeling student linear growth between two time points, along with school clustering effect and father's occupation clustering effect, using a cross-classified multilevel dichotomous IRT Model. To examine the performance of the proposed cross-classified multilevel IRT model for growth modeling, a simulation study is carried out. The magnitudes of the clustering effects for both the school and the father occupation grouping variables and the variance of the growth parameters are manipulated in this simulation study. The performance of the proposed model is evaluated in terms of parameter recovery under different simulation conditions. The impact of ignoring the cross-classified clustering structure on students' growth modeling and other model parameter estimation is also investigated by comparing the proposed model with two three-level multilevel IRT models in terms of parameter recovery and model-data fit. Result has showed that the proposed model outperforms the three level IRT models in terms of model fit and model parameter estimation precision.