Accounting research frequently uses R2, for example, to measure value relevance. We show analytically that scale effects present in levels regressions increase R2, and this effect increases in the scale factor's coefficient of variation. Thus, between-sample comparisons of R2 are invalid, unless one controls for differences in the scale factor's coefficient of variation. Applying our analysis to prior research, we show that the documented increase in value relevance of accounting is attributable to increases in the coefficient of variation of the scale factor. Controlling for this effect, there has been a decline in value relevance, as measured by R2.