Explaining variations in the advertising & promotional costs/sales ratio: A reanalysis

The authors focus on their 1990 model (B&K; see Balasubramanian and Kumar 1990), which indicates that market share, market growth, and their interaction are important predictors of the ratio of advertising and promotional costs to sales (A&P/S). In sharp contrast, Ailawadi, Farris, and Parry's (1994; AFP) subsequent replication attempt asserts that market share and market growth are not good predictors of A&P/S. The authors' research documents serious problems in AFP's study, including erroneous model estimates caused by incorrect execution of the SAS TSCSREG procedure, pooling data for analysis in ways that cannot be justified statistically data preparation errors, and inappropriate operationalizations of market share and market growth. With regard to analyses of Profit Impact of Market Strategy and brand-level data reported by AFP, the authors again find that AFP have analyzed data in inappropriate ways, given the conceptual framework of the B&K model. Because these problems provide compelling evidence to disregard AFP's results, the authors conclude that the criticisms advanced by AFP against the B&K model are not valid.