$W$ converges in distribution to $\chi^2_p$ where $p$ is the dimension of $\theta$.

$W$ converges to a standard normal distribution $N(0, 1)$.

$W$ is identical to the likelihood ratio test statistic.

$W$ is always smaller than the score statistic.