Has Teacher Quality Declined Over Time?

** Also posted here on "Valerie Strauss' Answer Sheet" in the Washington Post

One of the common assumptions lurking in the background of our education debates is that “quality” of the teaching workforce has declined a great deal over the past few decades (see here, here, here and here [slide 16]). There is a very plausible storyline supporting this assertion: Prior to the dramatic rise in female labor force participation since the 1960s, professional women were concentrated in a handful of female-dominated occupations, chief among them teaching. Since then, women’s options have changed, and many have moved into professions such as law and medicine instead of the classroom.

The result of this dynamic, so the story goes, is that the pool of candidates to the teaching profession has been “watered down." This in turn has generated a decline in the aggregate “quality” of U.S. teachers, and, it follows, a stagnation of student achievement growth. This portrayal is often used as a set-up for a preferred set of solutions – e.g., remaking teaching in the image of the other professions into which women are moving, largely by increasing risk and rewards.

Although the argument that “teacher quality” has declined substantially is sometimes taken for granted, its empirical backing is actually quite thin, and not as clear-cut as some might believe.

It is, of course, well-established that labor market opportunities for women have expanded a great deal over the past half-century, and that they are, as a consequence, less likely than before to choose occupations like teaching. This is true even when you account for the concurrent rise in female educational attainment (more women going to college).

But what kind of effect has this had on the “quality” of the teacher workforce? Needless to say, that’s a tough question to address, in large part due to the lack of data.

One way to do so is to see if incoming teachers’ measurable characteristics have changed over time. And, indeed, there is evidence of a decrease in the teacher applicant pool coming from more selective colleges and universities, as well as from the higher ranks of high school graduates.

But  there is one problem: None of these and most other “paper qualifications” has been shown to do a very good job of indicating who will be a great teacher, at least insofar as effectiveness is adequately gauged by test-based measures of teacher productivity (e.g., value-added and other growth models). It follows, then, that the trends above may not be the strongest evidence of an overall decrease in teacher quality.

One exception seems to be cognitive ability, such as reasoning and problem-solving, which does to some degree maintain an association with test-based effectiveness (see here, here and here).

An important 2004 longitudinal analysis of the trend in teachers' cognitive ability (as measured by math/reading tests) suggests that the proportion of female students who both chose teaching and scored in the top ten percent on these tests was around 15-17 percent in the late 1950s, compared with roughly 7-8 percent in the early 1990s (also see here). There were similar, though less pronounced, patterns among female students scoring in the top 70-90 percent. This suggests, in other words, that the highest-achieving (at least as measured by these aptitude tests) young women are more likely to choose other professions than they used to be.

(The results also suggest that the likelihood of choosing teaching may have actually increased slightly among top-scoring men, though samples were too small to get very precise estimates.)

It is, however, very important to note that, while there was this change in the likelihood of teaching at the top of the “cognitive ability” distribution, there was only a negligible decline in the scores of the average new female teacher over this same time period. This suggests that the decrease in the likelihood of teaching among the highest-scoring female students may have been offset by a decrease in the likelihood of choosing teaching among the lowest-scoring students. Other analyses have reached similar conclusions.*

That said, given the historic changes in the structure and degree of female labor force participation that have occurred over the past few decades, it’s certainly safe to say that female-dominated occupations like teaching have undergone changes in terms of who applies and why (both in the U.S. and other nations).

It’s likely – and tentatively supported by the empirical evidence discussed above - that the expanded opportunities to enter other professions may have compelled many women who would have otherwise been effective teachers to pursue other jobs, even if the extent to which this phenomenon has affected overall "teacher quality" remains an open question.**

It therefore makes sense for policymakers to consider how these “top candidates” might be attracted to the classroom, although the manners in which they are typically conceptualized (e.g., the "top third" of college graduates) don't quite square with the evidence, and the means of attracting them that get the most attention – such as performance pay – have, as yet, little evidentiary basis (see here for a different view).***

Nevertheless, overall, the evidence supporting the conventional wisdom that there has been a substantial decline in the “quality” of the average teacher over time is much less clear than is sometimes assumed. This of course does not change anything as far as the desirability of improving current and future teachers, but it does suggest we might be more careful about making blanket statements regarding the trend in "quality" among a huge workforce like teachers, and about offering simplistic explanations for incredibly complicated trends in educational outcomes.

- Matt Di Carlo

*****

* Whether or not this small change in the average might reflect a shift in "teacher quality" (or have had an appreciable effect on overall achievement) depends on many factors, including the relationship between “cognitive ability” and actual teacher effect sizes.

** One very interesting and relevant question about the size of the possible decline is the degree to which the skills that are “truly” associated with being a great teacher are also rewarded in the non-teacher labor market (also see here). In other words, do non-teacher jobs reward (e.g., via higher compensation) the skills that are widely shared by effective teachers?

*** For further research on why teacher “quality” may or may not have declined over time, see these papers examining the role of salary, unions, demographics of baby boomers and policies pertaining to childbearing.