I examine differences between effective tax rates (ETRs) and book-tax differences (BTDs) as alternative measures of corporate tax avoidance or tax aggressiveness. When BTDs are scaled by pretax income, the scaled BTD is statistically equivalent to the ETR. When BTDs are scaled by assets the resulting BTD measure is equivalent to the ETR multiplied by the firm's pretax return on assets (ROA), potentially adding measurement error unless ROA differences are part of the research design. Adding a control variable for ROA doesn't solve this problem. Any theoretical reason to use BTDs rather than ETRs in empirical research should be based on the effect of ROA. Simulated regression results demonstrate that, when ETR is the correct theoretical measure, BTDs don't provide any information beyond that provided in ETRs, and scaling by assets results in lower regression coefficients, potentially reflecting measurement error.