On using the jackknife to estimate quantile variance