Differences in which teachers get cut under a seniority-based layoff system compared with one based on teacher effectiveness are large and persistent, a new analysis of the New York City teaching corps finds.
But unless the number of layoffs are considerable, the two approaches don't seem to differ all that much on how they affect class sizes, the analysis concludes.
In general, critics of seniority-based policies say they require more teachers to be cut than under an effectiveness-based system, thereby raising class sizes. They also disproportionately affect low-income, high-minority schools that tend to have more novice teachers, such critics say. (If you're just coming to the debate, read a summary of the contours of the debate here, or my longer feature on the issue here.)
The researchers ran several analyses on data on 4th and 5th grade teachers in Gotham. They modeled two layoff scenarios to respond to a (fictional) budget shortfall equivalent to a 5 percent reduction in total salaries paid to those teachers. They used seniority-based layoff procedures for one analysis and an effectiveness-based procedure for the other, using "value added" achievement-growth measures attributable to individual teachers.
Four scholars in all conducted the analysis: Donald Boyd and Hamilton Lankford at the University of Albany in New York; Susanna Loeb at Stanford University in California; and James Wyckoff, at the University of Virginia in Charlottesville. These researchers have studied New York City over the past 10 years and have conducted a number of studies on the teaching force there.
The value-added analysis comes with the usual caveat that value-added measures don't incorporate all aspects of high-quality teaching, and that measures over just one year tend to be unstable. The researchers used multiple years of data where possible to minimize the latter problem.
Among the researchers' findings:
• Seven percent of teachers would be cut under a seniority-based layoff policy, compared with 5 percent under a policy based on value-added effectiveness;
• The population of laid-off teachers varies considerably based on the layoff approach with only 13 percent of teachers identified for separation under both scenarios;
• Most schools would lose relatively few teachers under the simulations, but the "effectiveness" scenario does a slightly better job of mitigating the concentration of layoffs in specific schools. Twelve percent of schools lose more than 20 percent of teachers under the seniority model, compared with 8 percent of such schools under the effectiveness model;
• The typical teacher laid off under the effectiveness scenario is, unsurprisingly, much less effective than one laid off under seniority—a difference in effectiveness equivalent to 26 percent of a standard deviation of student achievement. That's more than twice the difference between a first- and fifth-year teacher's average effectiveness;
• To address concerns that value-added effectiveness measures "fade out" over time, the researchers simulated how layoffs would have been made in summer 2007, rather than 2009, and then tracked those teachers an additional two years. They found that, while there is an overall decline in the difference between the seniority- and effectiveness-based scenarios over time, the difference is persistent, equal to 12 percent of a standard deviation of student achievement.
• However, because of the relatively small percentage of layoffs, an effectiveness-based layoff doesn't have much effect on the 4th and 5th grade workforce as a whole, equivalent to about 1/10th of a standard deviation in teacher value-added.
The bottom line, the authors write, is that informing teacher layoffs with information about effectiveness can improve the quality of instruction in some classrooms. Given the limitations with value-added, effectiveness measures should incorporate other approaches, such as classroom observations.