>I'm not sure I follow. Roach's main quantitative point is that the
>workweek is a lot more than 35.5 hours, especially in the professional and
>management sectors that make up 35% of the workforce. If I understand
>Doug's elaborations correctly, this is not an artifact of averaging, but
>rather results from a set of clearly wrong assumptions, like that computer
>programmers work 35 hour weeks. That seems not only clearly wrong, but
>clearly wrong by 50% or more even if you only counted the hours such
>people are actually in the office, never mind working at home or on the
>train. Surely there's nothing in principle stopping us from researching
>this question quantatively? Surveying in more detail how many hours
>people in the BLA's basket of professions actually work? And seeing if
>that number's wrong? I don't even see why the work-at-home hours should
>be beyond the reach of survey estimation.
The BLS extrapolates (salaried) white-collar hours from hours logged by (hourly) production workers. There's room for error in the production worker stats alone - they're based on what employers report they paid their toilers for, not how many hours they actually spent on the job. But there's even more room for error in the extrapolation to managerial hours, since most of the research on which the formulas are based come from the 1970s. They're working on time-use studies now, but the results aren't likely to come in for another two years.
Doug