This paper describes a stochastic homogenization analysis of a particle reinforced composite material using an approximation technique. In order to analyze the influence of a microscopic random variation of an elastic property of a component material on the homogenized elastic property of a particle reinforced composite material, the Monte-Carlo simulation is employed. Since the conventional Monte-Carlo simulation sometimes involves a higher computational cost, an approximate stochastic homogenization method using the Monte-Carlo simulation combined with a polynomial-based approximation technique is employed, and accuracy of the approximate Monte-Carlo simulation is investigated. In order to apply a lower order approximation to the approximate Monte-Carlo simulation effectively, the weighted least square method is proposed, and its effectiveness is discussed with the numerical results.