Abstract
The possibility to assess the significance of individual explanatory variable in the regression model belongs among the basic diagnostic tools of data analysis. The paper studies the problem for the least weighted squares - a generalization of the (ordinary) least squares as well as of the least median of squares or the least trimmed squares.
The paper at first shows how to cope theoretically with the problem and then by the numerical simulations demonstrates that the corresponding quantiles converge with the increasing sample size to a limit value, which is nearly identical with the Student's quantiles for the respective sample sizes.