Research vs. Conventional Wisdom I: Teacher Attrition

Research findings about teachers and teacher labor markets sometimes seem to defy conventional wisdom. Dan Goldhaber, director of CALDER at the American Institutes for Research and the Center for Education Data & Research at the University of Washington, explores teacher attrition in this first of three Education Week guest blog posts.

Most of the research at least tacitly confirms the conventional wisdom about what matters in education (and I'd guess other areas of social science research too). This doesn't make the research unimportant. An example: most people can recall the great influence of specific teachers on their lives, but America's teacher policy didn't hone in on the importance of individual teacher effectiveness until research showed the differential effects of teachers on student test growth or long-term economic well-being.

This week, however, I'm going to take advantage of Rick's kind offer (and dubious decision) to turn over Straight Up to focus on some cases where research findings about teachers, and teacher labor markets, appear to overturn the conventional wisdom. We know that even the most rigorous countervalent research seldom unseats common perceptions, but I'm going to give it a shot.

The conventional wisdom on attrition is that we lose about half of new teachers within their first five years of teacher teaching. As near as I can tell, this finding stems largely from Richard Ingersoll's work using data from the Schools and Staffing and Teacher Follow Up surveys.

Now, this 50 percent finding is subject to many caveats related to the data's limitations. For instance, the data cover a sample of teachers, not the whole population, so the survey isn't and wasn't intended to be representative of teachers new to the profession. Also, the follow-up survey used to assess attrition was conducted a year later, so it misses the many teachers who return to the classroom after a year or two of absence. There are other caveats about the survey and attrition research, but I'll skip these as they are well-documented in Ingersoll's report and by others, like Matt Di Carlo. But two points are worth making. The first: shortcomings with the data, not the analysis, put the finding on rather shaky ground. The second: the entrenched lore that we lose half of our teachers after five years is repeated as fact again and again. Indeed, if you Google the phrase "half of new teachers leave," you get nearly 10,000 hits.

But this 50 percent attrition figure is flat out wrong! How do we know? A recent National Center for Education Statistics (NCES) report, based on a survey that is focused on teachers new to the profession, provides lots of interesting findings about what kinds of teachers are leaving which kinds of schools (see NCTQ's recent Teacher Quality Bulletin). The headliner is that 17 percent of new teachers leave the profession after 5 years—nothing like half! This attrition rate differs little from what we see for new employees in other professions.

Why is this important? I can think of at least three reasons. First, the chasm between the two figures illustrates the need to be far more circumspect about easily characterized research findings. It's hard to shake a meme that has taken hold, and this one has been around for more a decade, but a desire to be accurate should trump catchy but disproven factoids.

Second, the "half of new teachers leave" meme has been repeatedly used to buttress another claim that rests on shaky empirical ground: good people are being hounded out of the teaching profession by new accountability pressures and a dumbing down of teaching. True, some good teachers are leaving, but it sure doesn't look like they are decamping in droves--even when accountability pressure for individual teachers is ramped up. In fact, if anything, research is showing that more academically capable individuals have been entering teaching in recent years (see here and here). This is another case where research contrasts with the conventional narrative.

Finally, an emerging policy concern is that impending teacher retirements will lead to a shortage of skilled teachers. How much of this fear reflects an incorrect belief about teacher attrition's magnitude? I have no idea, but surely our level of concern about the "looming teacher shortage" (again, Google it to see the many stories out there on this) ought to cool down. In fact, if you share the number-loving gene and think concern can be measured, it should be about two-fifths of what it was before the NCES report was released.