In a clinical trial, a researcher tests the efficacy of a drug using the Wald statistic $W = \frac{(\hat{\beta} - 0)^2}{SE(\hat{\beta})^2}$. When the likelihood surface is highly asymmetric due to a small sample size, which phenomenon describes the failure of the Wald test to reject $H_0$ as the estimate $\hat{\beta}$ moves away from the null?