Teachers seeking better pay or smaller classes might not be advised to look at unions as the answer, according to researcher Michael Lovenheim. He argues in a careful paper that unionization had no discernible effects on average teacher pay or per-student district expenditures in Iowa, Minnesota or Indiana from 1972 to 1991. He did find that unionization increased teacher employment by 5 percent, but because that increase was associated with increases in district enrollment, the number of teachers per student stayed about the same.

Lovenheim's work challenges the more generally held view that unionization raises teacher salaries and lowers class size. He argues that his results are different because he has been able to pinpoint more exactly when local unions were created. Earlier research relied on rougher estimates of the timing, throwing off the calculation of union effects.

The solid line is the trend of salary changes before and after the union formed at the year marked 0. The dotted lines mark the 95 percent confidence interval for the salary changes. Most of the estimated percent changes in salaries are small; for example, the percent change in salary stays below 2.5 percent through the eighth year after unionization. Moreover, since a zero percent change is within the dotted lines for all years after unionization, the changes are statistically indistinguishable from zero.

Lovenheim is cautious generalizing his results. And of course, even in his sample states, unions could have had an effect on a host of other matters that teachers care about such as working conditions, benefits and the rules that govern layoffs.