Font size:
More than 60 variables were used in the calculation of the 2012 TTB rankings.[26] Though its methodology is complex, the outcome is clear: the TTB list is strongly correlated with student poverty levels. Plotting each school’s TTB score against the percentage of students eligible for a free lunch through the National School Lunch Program shows a strong, negative relationship (see Graphic 2).[*] TTB rankings and student poverty are so closely related that if the state had ranked schools simply on the percentage of students eligible for free lunch in the 2012-2013 school year, nearly half would have seen their ranking change by less than 10 percentage points.[27]
Graphic 2: TTB Ranking and Percentage of Students
Eligible for a Free Lunch, 2012-2013
For the 2012-2013 school year, 55 percent of the variation in TTB scores for schools could be explained by the percentage of students eligible for free lunch at each school.[†] This has been consistent across years — this same figure was 56 percent the previous school year. Interestingly, the correlation between a school’s TTB rank and student poverty is similar in value to the correlation between raw standardized test scores and student poverty.[28] In other words, the formula used for the TTB list does not appear to mitigate the fact that schools serving larger portions of low-income students will generate lower standardized test scores on average.
Of the variables used to calculate TTB rankings, those that measured student test scores were most strongly correlated to poverty. However, several of the variables used to measure learning growth (such as annual gains in math and reading test scores) were also similarly tied to school poverty rates.[‡] That is, poverty affects the level of student achievement as well as the rate of growth in learning as measured by MDE.
[*] Eligibility for this federal program is based on household income. For more information about the National School Lunch Program, see: "National School Lunch Program," (U.S. Department of Agriculture, 2012), http://goo.gl/9a3wC (accessed Sept. 17, 2013).
[†] This is figure is based on analysis assuming a linear relationship between school rankings and poverty. A nonlinear model was tested and did explain slightly more of the data variation. However, to keep the analysis consistent and simple, a linear model was used to analyze the relationship between school rankings and poverty levels in Michigan and other states.
[‡] Author’s calculations using "2011-12 Data Needed for Replication of the Top-to-Bottom Ranking," (Michigan Department of Education, 2012), http://goo.gl/E0kEI (accessed Sept. 17, 2013). This finding echoes one made by the Cleveland Plain Dealer in June 2013, when reporters discovered that the state’s “value-added” rankings were correlated to students’ family incomes. Patrick O'Donnell, "Teachers' 'value-added' ratings and relationship to student income levels questioned," Cleveland Plain Dealer, June 18, 2013, http://goo.gl/GliICF (accessed Sept. 17, 2013).
