The use of sampling weights in the analysis of the 1998 Workplace Employee Relations Survey