The assumption that the relationship between market structure variables and profitability is the same across industries is critically examined. Using observations on food, tobacco and cosmetic firms, the null hypothesis of equal coefficient vectors is rejected for these industries. This implies that the observations are from different populations and cannot be considered to constitute one sample. It is well known that the least squares estimator is biased if the observations represent different populations and the independent variables are correlated. This casts doubt upon the results obtained in previous studies when no allowances are made for differences in the relationship between industries. In this study observations are grouped into five homogeneous subsets. Weighted and Generalized Least Squares estimates show that the coefficient of the advertising variable differs substantially across subsets. Advertising has a significant and positive impact in only one of the five subsets. These results provide only limited support for the hypothesis that advertising serves as a product differentiation barrier-to-entry in a market.