The New Year is four days away. Today, and every day this week, we bring you, two each day, and in no particular order, 10 of the most noteworthy stories of the year that has passed, and the statistics that made them possible. Today, Part 2 - Teachers and Politics.
3. Measuring a teacher's value. Tables were turned for the teachers of 470 elementary schools in Los Angeles when the LA Times published a searchable report card based on six years of data from the California Standards Tests for English and Math. Teachers courageous enough to google their grades would find the hallmark alphabetic system replaced for a color-graded scale--much like the now defunct Homeland Security Threat Monitor or litmus paper---showing their place along a spectrum from least to most effective. How much a teacher measured up was not determined by opinion but by a statistic: the estimated value that teacher added to the test performance of his or her students. It is called value-added modeling.
All kinds of things contribute to a student's performance in tests - the quality of his teacher, but also his own ability, how hard he works, his home life, his physical health, the quality of the school as a whole. To statisticians, what is now the standard approach to value-added modeling (VAM) is a mixed effects linear model for student test score data over several years. By including effects for a student's past test performance, as well as classroom and school factors, analysts attempt to isolate the contribution instructors have on student performance. The value a teacher adds (or subtracts) from a student's performance in a given year is the difference between the student's actual test score and his or her expected score, estimated from the mixed model.
As the Obama Administration has called for US school districts to develop more effective ways to evaluate teachers whilst giving greater incentive to implement innovate techniques through the Race to the Top grant competition, the use of VAM as an evaluative tool in education has become a national phenomena. And, as the response of the LA times shows, some teachers are already feeling the impact. In the summer of 2010, 25 teachers in the nation's capital were fired by the no-nonsense chancellor of schools, Michelle A. Rhee, based, in part, on their VAM litmus test results.
Developers of VAM continue to dispute the precision of the teacher rankings it produces--is the 41st VAM teacher really any different from the 33rd? Yet most agree that more evidence-based evaluation tools like VAM are needed in order for real gains in the classroom to be achieved. Education Secretary Arne Duncan has already given cautious approval of the actions of the LA times. It might not be long before VAM is a fixture of public education in America.
4. Political rhetoric finds a helpmeet in statistics. 2010 presented White House top economic adviser Lawrence Summers with a semantic quandary. Though the recession begun in December 2007 officially ended in June 2009, the US economy, in the words of the National Bureau of Economic Research, was "not operating at normal capacity". By the beginning of 2010, signs of recovery were inconsistent. In January, the President's Council of Economic Advisers reported that the 2009 $787 billion stimulus program had added nearly 2% points to GDP. Indicators of consumer expectation, stock prices, demand of capital goods and the money supply were also improving. Joblessness, however, had reached an unexpected high of 10%, 1.5% higher than the unemployment peak Obama's economic team had predicted.
What could be said about an economy whose rebound was not yet producing new jobs? The language that Summers settled on--"a statistical recovery and a human recession"--became the economic catchphrase of the year. From NPR to the World Economic Forum in Davos, Summers relied on this phrase to deliver a complex message of optimism and caution: the wounded US Economy was on the mend but needed more time to heal. By qualifying "recovery" with "statistically", Summers avoided a false declaration of victory while still conveying confidence that the US economy would eventually survive the decade's downturn unscarred.
But note: statistics are not inhuman, despite the contrast that Summers' phrase may seem to imply. It was statistics after all that measured the 'human recession' part of the story and told us that unemployment, and the misery that goes with it, was still high.
Tomorrow: Part 3: Marine life, and speech recognition.