** Also posted here on “Valerie Strauss’ Answer Sheet” in the Washington Post.
Please check out the “sequel” to this post, which includes a multivariate analysis, and, more importantly, two other posts (here and here), which present summaries and discussions of the actual evidence on the relationship between unions and test scores.
For years, some people have been determined to blame teachers’ unions for all that ails public education in America. This issue has been around a long time (see here and here), but, given the tenor of the current debate, it seems to bear rehashing. According to this view, teachers unions negatively affect student achievement primarily through the mechanism of the collective bargaining agreement, or contract. These contracts are thought to include “harmful” provisions, such as seniority-based layoffs and unified salary schedules that give raises based on experience and education rather than performance.
But a fairly large proportion of public school teachers are not covered under legally-binding contracts. In fact, there are ten states in which there are no legally binding K-12 teacher contracts at all (AL, AZ, AR, GA, LA, MS, NC, SC, TX, and VA). Districts in a few of these states have entered into what are called “meet and confer” agreements about salary, benefits, and other working conditions, but administrators have the right to break these agreements at will. For all intents and purposes, these states are free of many of the alleged “negative union effects.”
Here’s a simple proposition: If teacher union contracts are the problem, then we should expect to see higher achievement outcomes in the ten states where there are no binding teacher contracts.
So, let’s take a quick look at how states with no contracts compare with the states that have them.
In states where there are binding contracts, there is some variation in coverage (the percentage of teachers covered under contracts). In most of them (34, including D.C.), districts are required to bargain with unionized teachers, and coverage in these states is very high. There are a few other states in which contracts are binding once they’re finished, but districts are not required to bargain (Louisiana also technically falls into this category, but since Katrina, there are no binding contracts in the state). The results for these states are virtually identical to those for the bargaining states.
In the table below, using data from the 2009 National Assessment of Educational Progress (NAEP), I present average scale scores for states that currently have binding teacher contracts and those that don’t. The averages are weighted by grade-level enrollment, and they include only public non-charter schools (since most charters in all states have no contracts).
As the table clearly shows, the states in which there are no teachers covered under binding agreements score lower than the states that have them. Moreover, even though they appear small, all but one of these (8th grade math) are rather large differences.
To give an idea of the size, I ranked each state (and D.C.) by order of its performance —its average score on each of the four NAEP exams – and then averaged the four ranks. The table below presents the average rank for the non-contract states.
Out of these ten states, only one (Virginia) has an average rank above the median, while four are in the bottom ten, and seven are in the bottom 15. These data make it very clear that states without binding teacher contracts are not doing better, and the majority are actually among the lowest performers in the nation.
In contrast, nine of the ten states with the highest average ranks are high coverage states, including Massachusetts, which has the highest average score on all four tests.
If anything, it seems that the presence of teacher contracts in a state has a rather large, positive effect on achievement.
Now, some may object to this conclusion. They might argue that I can’t possibly say that teacher contracts alone caused the higher scores in these states. That there are dozens of other factors besides contracts that influence achievement, such as lack of resources, income, parents’ education, and curriculum, and that these factors are at least partially responsible for the lower scores in the ten non-contract states.
My response: Exactly.
Note: After publishing this post, we learned that there is one binding contract in Louisiana (in Saint Tammany Parish), and two in Arkansas (Little Rock and Pulaski). We apologize for the errors, and we also thank Stuart Buck, a commenter who pointed out the Little Rock contract.