Abstract Recent findings indicate that macroeconomic survey forecasts are anchoring biased and therefore are inefficient. However, despite highly significant test coefficients, a bias adjustment does not improve forecasts’ quality. We find that the cognitive bias is a statistical artifact because the anchoring test is biased itself. In particular, it produces misleading results if macroeconomic analysts use more comprehensive information than assumed by the test. Our results have important implications for a wide range of empirical research relying on survey data to capture market participants’ expectations, for example, studies analyzing the impact of macroeconomic conditions on asset prices, equity risk premiums, or market liquidity.