Guy Brandenburg flags an incredibly interesting and revealing study in the June 2010 issue of Journal of Political Economy, "Does Professor Quality Matter? Evidence from Random Assignment of Students to Professors." From the press release:
Highly credentialed and experienced professors are better at preparing students for long-term academic success than their less-experienced counterparts, but that ability isn’t necessarily reflected in their students’ teaching evaluations. That’s according to research by a pair of economists published in this month’s Journal of Political Economy.
The study’s authors, Scott Carrell of U.C. Davis and James West of the U.S. Air Force Academy, say their results raise questions about the value of student evaluations as measures of instructor quality...
Their data come from Calculus I and follow-on classes at the U.S. Air Force Academy. All Air Force Academy students are required to take Calculus I, Calculus II, and nine math-based technical courses regardless of their majors, and professors in all sections of classes use an identical syllabus and give identical exams. That gives the researchers a chance to compare instructors on a relatively even playing field.
The study found that students’ achievement in follow-on coursework was strongly influenced by their Calculus I instructor. Students who had a seasoned Calculus I professor with a Ph.D. tended to do better in follow-on coursework than students who had less-experienced and less-credentialed Calculus I instructors. This happened despite the fact that students of seasoned professors tended to have lower grades in Calculus I. The results, the researchers say, suggest that less experienced instructors have a tendency to “teach to the test,” while more experienced teachers produce “deep learning” of the subject matter that helps students down the road. (emphasis added)
For an educational study, they've got a huge and clean dataset. It is a shame that it isn't a K-12 study, but I can't think of a reason that it isn't applicable to younger students as well.
If this phenomenon exists in K-12 as well, think of the implications for the entire teacher evaluation, data driven instruction, etc. regime.
Worse, if this phenomenon exists in K-12 as well, imagine how long it would take to prove it.