Unions and ACT/SAT Scores

I saw this "fact" posted on FB:

Only 5 states don’t have collective bargaining for educators. Their ACT/SAT rankings: SC-50th/NC-49th/GA-48th/TX-47th/VA-44th.

I wanted to investigate, so I tried to dig into the data (SAT here, ACT here) but noticed instantly that Wisconsin ranks highly for the SAT, but has only 6% participation. That is suspiciously low to infer too much from.

Then I ran into this post. I thought I would share my comment here along with the paper I found.

Its a difficult thing to determine from simple descriptive stats. States with low participation in a particular test are likely to suffer from a selection bias. In Wisconsin you take the SAT if you are planning to go out of state. You go out of state if you are smarter, richer, etc. Which is why our SAT scores are high. You need to combine ACT, and SAT (through some conversion) then adjust the state's data for race, income, education, and percent going to college, along with union penetration. That will give you a better handle on the union’s effect. Unions definitely have very bad aspects too them, but they also have good aspects to them as well. The question is in part about the net effects. I would think their ability to get higher pay for union members, and better benefits relative to the non-union setting results in them attracting better teachers (on average) and results in greater stability. Again, we all know situations where this does not work well, but on average it might make for better teachers. That's not to say you couldn’t create the same environment without unions. Or to say you couldn't work with the unions to minimize their negative effects. When I graduated from Madison I knew people with teaching degrees who went to Texas, because of the lower standards and the ease of getting a teaching job there. The good ones eventually moved back to Wisconsin, and the bad ones left teaching. Wisconsin skims the cream as it were. It lets Texas do the work of separating the good from the bad teachers. Texas students lose, Wisconsin students win.

This paper looks like it might do a better job of identifying these issues:

Title: Do Teacher Unions Hinder Educational Performance? Lessons Learned from State SAT and ACT Scores.
Authors: Steelman, Lala Carr; Powell, Brian; Carini, Robert M.

From a quick read I think it does a reasonable job addressing the issues I highlighted above. Their conclusion, with a very thoughtful discussion of its implications is below:

They find a significant and positive relationship: that is, the presence of teacher unions appears to be linked to stronger state performance on these exams. These findings challenge the position that teacher unions depress student academic performance, and in so doing invite further empirical scholarship on this topic from a range of academic disciplines.

Our finding that teacher unions are positively linked to state average SAT and ACT scores prompts the question of why. Clearly, our study challenges the "rent-seeking" view outlined earlier, which envisions teacher unions at odds with what parents desire from schooling, namely, the educational advancement of their children. The zero-sum orientation that permeates much research on unions and assumes that worker gains inevitably result in production losses appears misguided, at least with respect to teacher unions. Still, our data cannot distinguish among the previously outlined explanations for the positive relationship between unions and state-level SAT and ACT scores. However, in supplementary analyses (available from the authors), we were able to test one possibility: that teacher unions are positively related to lower average class size (i.e., student-teacher ratios), higher per ca¬pita expenditures on education (adjusting for interstate variation in the cost of living), and higher salary (also adjusting for cost of living) . Although these variables are linked to state SAT and ACT scores, their inclusion in our models did not significantly reduce the effect of teacher unionization. Other mechanism (s) (i.e., better working conditions; greater worker autonomy, security, and dignity; improved administration; better training of teachers; greater levels of faculty professionalism) must be at work here. Future scholarship should be directed at unraveling why teacher unions appear to favor¬ably influence academic outcomes.

Finally, this study cannot tell us if there is an overall net benefit of teacher unions, at least with respect to cost effectiveness. Because we examined the link between teacher unions and productivity but not costs, we cannot gauge whether the higher test scores are enough to offset the purportedly higher costs of unionization. Whether there is a net benefit of teacher unions hinges not only on the impact of teacher unions on economic (and non¬economic) costs, but also on the specific costs deemed acceptable (by the public, policymakers, or academe) for a unit increase in educational productivity — an assessment for which consensus may be difficult to reach. Moreover, even if, through some mechanism, unionization raises test scores, teacher unions may be a relatively inefficient vehicle of educational reform: for example, states might raise scores more with an identical investment in school infrastructure, additional teacher training, or special programs.