Friday, September 13, 2013

The "denominator problem" in assessing teacher compensation

Do public school teachers receive compensation in line with the market value of their skills? To answer that question, we have to know how much teachers work relative to other professions. Andrew Biggs and I call this the "denominator problem" in pay comparisons. The hourly wage is total salary divided by work hours, with the numerator determined easily enough using teacher contracts or survey data. But what goes in the denominator?

How much time do teachers work?
Andrew and I have an upcoming academic article that addresses the denominator problem using some detailed time-use data. Without it, researchers are stuck with using either "contract hours," which tend to understate teacher work time, or self-report data, which overstates work time.

On the Corner yesterday, I wrote about how the OECD not only uses these less reliable data, but also mixes and matches methods when comparing countries:
OECD studies are sometimes just too ambitious. The truth is that large-scale international comparisons are almost inevitably plagued by inconsistent and unreliable data. The most recent example: The OECD’s Education at a Glance 2013 shows American teachers working more hours than teachers in every OECD country except Chile (see table D4.1 in the study). Sounds impressive!

But consider how the data are collected. For many countries, the OECD appears to measure teacher work time based on statutory or contractual work hours. For the U.S., however, the OECD relies on the federal government’s School and Staffing Survey, which directly asks American teachers how much time they work.

That’s an inconsistency likely to inflate U.S. hours relative to the rest of the world.
Read the whole thing here. And for a discussion of other methodological challenges to assessing teacher compensation, try this longer backgrounder.

1 comment: