Abstract
Most scale development in advertising research is based on classic test theory. However, classic test theory is appropriate only when the researcher is interested in differentiating people. Much of advertising research is actually trying to differentiate advertisements rather than people. The present paper argues that advertising researchers should explicitly consider whether their research is intended to differentiate people or advertisements. If advertisements are being differentiated, then using classic test theory to evaluate the advertisement may mislead the researcher. The current paper discusses the use of generalizability theory in evaluating stimulus-centered scales and presents an empirical illustration where generalizability theory was far superior to classic test theory in determining whether a specific scale was appropriate.