In meta-analysis multiple regression models, the dependent variable is a parameter estimate, and the independent variables are study descriptors. The error term in such a multiple regression model arises from two sources. First, the parameter estimate (dependent variable) differs from the underlying true parameter value because of estimation error. Second, there is model (or specification) error arising from the fact that the study descriptors (independent variables) do not fully capture all of the variation in the true parameter. We provide an iterative procedure to estimate the meta-analysis model in such a context, assuming that the standard errors (or t-statistics) of the estimated parameters are available for the individual studies included in the meta-analysis. Compared to the classical ordinary least squares method, the proposed method produces more efficient estimates of meta-analysis parameters, yields valid statistical significance tests, and produces a more appropriate (and less pessimistic) estimate of the goodness of the meta-analysis model. An application of the proposed method in the context of a meta-analysis of diffusion models of new consumer durables indicates that price, rate of price decline, year of product introduction, population size, and the nature of the product category significantly affect the diffusion model parameters.