This paper reports the findings of a longitudinal study of the Australian womens' clothing market. Using MANOVA to analyze results from a field experiment allows variation in respondents' beliefs and preferences to be partitioned between (a) factors of theoretical and/or managerial interest (market segments and time), (b) factors representing secondary aspects of the research design (presentation and context of clothes) and (c) traditional measurement threats (repeated measurement). The results obtained for this illustrative case suggest that the variance generated by design choices and measurement effects may be equal to, or greater than, that associated with our theoretical constructs. If such a conclusion were more generally true, it would indicate a pressing need for the improved reporting of research designs and greater attention to the study of methodological artifacts.