Which condition is necessary for the Wald test statistic $W = \frac{(\hat{\theta} - \theta_0)^2}{\widehat{Var}(\hat{\theta})}$ to converge in distribution to a $\chi^2_1$ random variable?