Squeezing statistics to make them fit pre-fab politics
by Craig Westover
When one is a pot, one hesitates to call the kettle black. Nonetheless, when think tanks use research the way a drunk uses a lamppost, for support rather than illumination, the public is ill served, misinformation becomes widely distributed, policymakers are misled and bad policy follows.
The finding in a study released by the progressive think tank Growth & Justice that increased teacher salaries are linked to higher graduation rates is a case in point.
With support from Growth & Justice, Columbia University professor Henry M. Levin and Queens College assistant professor Clive R. Belfield investigated investments in K-12 education with a positive rate of return. Minnesota 20/20 (another progressive think tank) distilled the paper into bullet points and proliferated the idea that “increasing teacher salaries by 10 percent raises high school graduation rates 5 percent.”
According to Minnesota 20/20, the chair of the Minnesota House K-12 Education Finance Committee, Mindy Greiling, DFL-Roseville, said the Levin/Belfield paper contributes to a strong case for more school funding. She is advocating a $1 billion annual increase in the education budget.
Methinks Greiling et al are leaning on the lamppost of research for support of state-run education rather than using it to illuminate ways to actually improve children’s education. That is too bad because Levin and Belfield’s study (and other writing by the authors) illuminates several significant relationships between teacher salaries, teacher quality and improved student outcomes that are vital to meaningful discussion of education reform.
There are several implications of the Levin/Belfield research that Growth & Justice and Minnesota 20/20 miss and policymakers ought to consider.
The first is the unspoken observation that the current quality of public school teachers is not good enough; schools, say the authors, need to tap “an enhanced talent pool.”
“There is wide recognition that the quality of the teacher in the classroom is the most important single influence on the quality of education,” Levin writes. He notes higher teacher salaries “are not a magic elixir” but they “are capable of drawing a larger pool of talent into teaching.”
(The irony is that criticizing teacher quality from outside the system is verboten; criticizing teacher quality from inside the system is reason for more funding.)
The authors also contend the benefit of higher teacher salaries is realized through turnover and new hires. But if wages are increased, turnover ought to diminish. Doesn’t that mean a somewhat slow migration to the enhanced talent pool?
Indeed, the regression study on which Levin and Belfield base their conclusions provides a “time lag” of 10 years to allow for differences in salaries to work their way through to student performance. That cannot be a comforting thought to parents sending their children to first grade with the promise of higher quality teachers by the time their children reach high school.
Another implication of regression analysis is that the relationship between teacher salaries and graduation rates is a mathematically predictive number, not direct cause and effect. Holding “all other things equal,” increasing teacher salaries increases graduation rates; but we know in the real world all other things are not equal. How teacher salaries are implemented is an important piece of the puzzle.
Skeptical as I am about spending more money to prop up a state-run monopoly education system, I can accept the notion that teacher salaries are linked to student outcomes, but there is obviously something missing in the Minnesota 20/20 bullet point interpretation. More money alone doesn’t make a better teacher. There must be some intervening change in teacher behavior related to increased salary that leads to higher graduation rates.
“It is not only the overall level of teachers’ salaries that should be considered, but also the structure of the salary scales themselves,” writes Levin. “Fully capitalizing on higher teacher salaries may require very large changes in teacher recruitment, selection, professional development, and evaluation. … With a larger talent pool, only the best teachers should be retained. This suggests a superior system of teacher evaluation … (with) heavy dependence on evaluative data for awarding tenure.” (My emphasis added.)
That teacher salaries are linked to improved graduation rates is a data point, not a policy or a reform plan. Think tanks and policymakers primarily concerned with tweaking the education funding formula and securing more education funding are putting the cart before the horse. Waterboarding good research to make it confess conclusions not supported by the data may be modus operandi for politicians, but should be anathema to think tanks. To motivate a valuable discussion of education reform, some think tanks need to let go of the lamppost and follow the light.
This commentary originally appeared in the Nov. 15 St. Paul Pioneer Press. For the Growth & Justice view of it’s Education Summit, see Angela Eilers, “Homing in on smart school investments that pay off twice, also in the Nov. 15 Pioneer Press.