If you're dealing with national data and are not too concerned about 'underground' activity, I'd say you're right. Establishment data offers more reliable data on establishments, where most workers are employed, but part-timers are then counted twice if they work at two establishments and workers commuting from out-of-state are counted among the state's employed. Also just like household surveys, the sample used may become more biased over time. Currently here in Mississippi I don't find either alone sufficient--and they've just changed the formula they use for extrapolating from their household survey data, which caused some comparability problems. Marianne Hill [EMAIL PROTECTED] On Tue, 7 Feb 1995 07:22:01 -0800, Peter.Dorman writes: >Dear Pensters: > >I am about to say something in print that I recall being true, but I think I >should run it past you to make sure. In general, isn't it the case that the >establishment surveys are more reliable indicators of the level and >distribution of employment than the household surveys? I am combining the >establishment data on nonfarm employment with the household data on *farm* >employment, because the latter is not collected by establishment, and, for the >fixed-weight index I'm constructing, variation within industries over time is >more important than the right proportions between industries at any one time. >Does this make sense? > >Peter Dorman