The Standard of Evidence for Change
Note: Dan Goldhaber, an economist and professor at the University of Washington, is guest-posting this week.
When Rick asked that I guest blog, I wasn't too sure that I could be pithy, coherent, and interesting--I'll let you judge. Either way, this is a terrific forum for making points that I've little chance to make in economics journals or through econometric analyses.
I am continually struck by the fact that policy debates over a whole variety of issues focus almost entirely on the downside risks of reform, while massively ignoring the costs or downsides of business as usual. This comes up around numerous issues ranging from school autonomy and funding to vouchers and privatization. For example, back in 2005, Governor Arnold Schwarzenegger--citing the enormous growth in pension costs for California's public employees--sought to change the state's defined benefit plans into 401(k)-style defined contribution plans. Union backlash was furious, with multiple teachers unions forming a group called "Seriously, Saving California" in bitter opposition despite the fact that pension obligations had soared to $2.6 billion. And, we are even warned that focusing too keenly on another important issue--teacher evaluation--may be a distraction from addressing broader social issues (for example, see here). Is focusing too much attention on teacher quality really a problem?
A particularly telling example is the current debate over using value-added assessments to help judge teacher effectiveness. Not surprisingly, unions have raised objections. In response to Florida's S.B. 6 value-added measure from earlier this year, for example, Robert Dow, the President of the Palm Beach County Classroom Teachers Association, warned that "Education is falling apart. If this Senate Bill 6 is signed into law, I think it will be the death knell of education." And it's entirely true that there are good reasons to be wary of using value added information to make high-stakes decisions. For one thing, student assessments may not be well-aligned to standards or designed to measure student growth. For another, policies that reward teachers for student test-based achievement may lead some to focus narrowly on tested items or even to cheat. Most recently, rigorous reports like this illustrate that value-added measures of teacher effectiveness will lead to some misclassification of teachers.
When it comes to making changes, there are clearly compelling reasons to proceed with caution. But the oft-ignored flip side is that current teacher policies don't look so hot either. We know teacher licensure, compensation, evaluation, and professional development systems are only weakly (at best) related to teacher effectiveness. For instance, Dominic Brewer and I found back in 2000 that whether a teacher is fully licensed by his or her state does not, in general, predict that teacher's effectiveness. And a wide variety of research shows that, collectively, teacher qualifications explain less than 5 percent of what makes a teacher effective or ineffective (for example, see here or here).
Shouldn't we consider the problems with the status quo when setting the bar for change? This is the argument that I make, along with colleagues Steve Glazerman, Susanna Loeb, Doug Staiger, and Russ Whitehurst, in a recent Brookings Institution report. In particular, we note that much of the debate about using value-added for teacher evaluation is framed around the potential consequences for teachers rather than focusing on the known or potential consequences for students. We act as though change should be clearly good for all involved, when in fact there are likely to be winners and losers. And sometimes the best interests of incumbent adults in a school system may not be perfectly (or even closely) congruent to the interests of students.
The general point is that the counterfactual matters and we are not comparing reform to an existing nirvana. As a related and ironic aside, the only reason that we can so clearly pin down the downsides of using value-added is that it is perfectly clear what it is intended to measure. This isn't true for the teacher credentials that are used now. For example, getting a Master's degree may help teachers more effectively teach students citizenship. It may also make them more likely to collaborate with their peers. We just don't know because it is not entirely clear how getting a Master's degree is supposed to affect teachers. We do know, however, that it doesn't generally make them more effective in teaching students reading and mathematics.
I pick on the Master's degree because the finding that it doesn't seem to predict teacher effectiveness is not a new one. And while I could be wrong, my guess is that most of the public still believes that teachers with Master's degrees are more effective, and that getting a degree makes teachers better at their craft. But education policymakers should, and I suspect do, know better.
Maybe I'm being too impatient, and too hard on the policymakers--but I don't think so. The problem is that the politics and culture in schools are such that the education system has an extreme case of risk aversion. What this means is that the barriers to trying to really rethink troubling practices or remake mediocre schools are remarkably high.
Things could be changing. The economic downturn could force reform. And Rick rightly promoted an extraordinary speech by Education Secretary Duncan in which he called on school systems to seriously question the status quo and treat the financial crisis as an opportunity to innovate and improve the productivity of the American education system. Some of the actions Duncan recommends entail taking on sacred cows like the Master's pay premium. He deserves praise for this. My complaint is that we have known pretty definitively for well-over a decade that paying for a Master's degree is unlikely to be the most productive use of a school dollar, yet it appears to take an economic disaster to get even high-level cover to talk about change, much less to take the action itself. It probably shouldn't come as a surprise that we set the standard of evidence for change higher than that applied to the status quo, but it sure is disappointing to see exactly how much higher it is.