Methodology

ESTIMATED GRADUATION RATES
The data used to estimate graduation rates came from three sources: data published by the U.S. Department of Education’s National Center for Education Statistics, and that contained in previous Schott Foundation 50 State Reports on Public Education and Black Males (published in 2004 through 2012). Moving averages were applied to these data to calculate the estimated graduation rates. Where a state or locality made the data readily available, those data points were used.

Several different forecasting methods were tested in order to calculate the most reliable estimated graduation data. “Moving averages” were ultimately used to estimate more recent graduation data as well as missing, unavailable or unreported graduation data at the state level and for specific school districts.

A “moving average” is one of the most basic forecasting techniques. A moving average uses a defined number of data points over a specific span of time or period, and averages them, creating a predicted value. It is called “moving” because the period for which the average is created moves forward. For example, in a simple moving average, if there were five data points, the first period might generate an average from the first and second data points, the second period would generate an average from the second and third data point, the third period would generate an average from the third and fourth data point, and so on. The idea behind the use of moving averages in forecasting each next period will not be much different than the past few periods. Therefore, while it is not recommended for long-term forecasting, it produces reliable short-term predictions. In the case of this report, the predictions computed using the moving average method represent estimates for past graduation rates that have not yet been published

STATE LEVEL METHODOLOGY
At the state level, moving averages were used to create two different sets of estimated average freshman graduation rates (AFGR) for the 2012-13 school year. These different methods correspond to the data used in the estimation. In the first estimate of state-level average freshman graduation rates for the 2012-13 school year, moving averages were applied to the published average freshman graduation rate data for Black, Latino and White males from 2002 through 2010. In the second estimate of state-level average freshman graduation rates for the 2012-13 school year, moving averages were applied to the estimated diploma counts of Black, Latino and White males from 2002 through 2010. These initial estimated diploma counts were generated using the published average freshman graduation rate data for Black, Latino and White males from 2002 through 2010, and the enrollment bases used in the calculation of the graduation rates. These enrollment bases were calculated using U.S. Department of Education enrollment data. By multiplying the enrollment bases by their corresponding graduation rates, estimated diploma counts were computed. The estimated diploma counts for the 2009-10 school year were checked against the published data for that year (the only year for which diploma counts were available by race and gender); the estimated diploma counts matched closely to the published data. Using the estimated diploma counts generated by way of the application of moving averages and the enrollment bases from the U.S. Department of Education data, the estimated state-level average freshman graduation rates for the 2012-13 school year were calculated.

LARGE URBAN DISTRICT METHODOLOGY
At the district level, moving averages were used to estimate the 2011-12 graduation rates of Black and White males. Moving averages were applied to district-level graduation data, which was compiled through previous Schott Foundation reports, covering 2001-2, 2003-4, 2007-8 and 2009-10. The data in these previous reports represent the graduation rate of Black and White males in 2001-2, 2003-4, 2007-8 and 2009-10, and may differ from locally reported data due to variations in how school districts and states calculate and report graduation data. In all of the above estimations, missing, unreported or unavailable data were all estimated. It should be noted that the number of data points used and the extent to which the data fluctuates influences the accuracy of the estimations.