Classical regression analysis is usually used in many analyses of longitudinal data or repeated measurement data (the response of each individual is observed repeatedly) . It assumed that there is no correlation between pairs of observations. Wherea, multiple observations on the same object generally produce correlated outcomes. Ignoring correlation in regression analysis can lead to incorrect conclusions. In longitudinal data analysis is to regard the correlations between pairs of observations in this case the structure of covariance. For this reason, inference from longitudinal data analysis can make more correct conclusions. The objective of the study is to know the effectiveness and efficiency of longitudinal data analysis to describe the change of response over time comparing of the cross sectional data analysis. Simulation data was used to investigate the behavior of longitudinal data analysis. The result of this study showed that longitudinal data analysis was powerful to increase the information of the change of the response over time succesfully. If the correlation between two observations in same object is increasing, the cross sectional data analysis becomes inefficient to describe means response as a function of time.