This research aims at detecting the behavior of the parameter of first order Autoregressive model. The parameter takes various functions and formulations with the estimates of the parameter through the method of least square. The random errors of the model follow Non-Gaussian distribution (discrete, continuous) with the experiments taking various sample sizes through conducting simulation experiments. The results would be compared with absolute bias, mean square errors, mean absolute percentage errors and absolute standard bias. The most important results of the study: The values of MSE and MAPE will be decreased wherever the size of the sample is increased for all distributions, and the smallest value for MSE and MAPE is when n=150. This value is exclusively relevant to third model and for all distributions. Through simulation experiments no essential differences have appeared for the results between continuous and discrete distributions.