Using interpolation to create a modeled surface entails approximating values of certain characteristics at all locations between known observations. In this case, our observations were the number of students per block, and that number was spatially located at the center of each block. Between the centers of each census block, we applied a mathematical formula that models the likely value at each point, taking into consideration the values of student counts in nearby blocks. The interpolation process is based on the premise that observations that are close in space are likely to have similar value, so in-between values are estimated based on nearby actual observations.
The resulting model can’t be used to determine actual counts at any one location. For example, the model erroneously shows students residing on Belle Isle, which is the result of Belle Isle being a large census block. In addition, the model could look considerably different depending on the assumptions made (the size of the grid or the weighting, for instance). Instead of regarding these values as actual observations, the model should be regarded as showing a spatial trend. While not precise, this smoothed spatial trend allows us to see areas of generalized higher concentration.
In the map above, orange and red indicate the highest numbers of students who attend public Detroit schools per area. As you can see, concentrations of these students exist around Lafayette Park, within the apartment complexes along 1-75 between Mack and Warren, and in Southwest Detroit, with smaller concentrations in Warrendale and around East English Village, on the far east side.
Conversely, large sections of the interior of Detroit, especially east and southeast of Hamtramck, show the fewest students. This situation isn’t completely surprising, given that this area of the city has high vacancy rates, large numbers of vacant lots and some of the worst building conditions in the city, according to the Motor City Mapping survey. The map below shows numbers of existing structures as a percentage of parcels per block group. The lighter colors represent areas of the city that have a higher percentage of vacant lots. While not perfectly accurate, these areas line up fairly well with the areas in the previous map showing the lowest concentrations of students — with the exception of Southwest Detroit. Southwest Detroit, with its large Hispanic population, is known to have higher birth rates than other parts of the city. These birth rates would explain the higher concentrations of students despite a higher number of vacant lots.
Generally speaking, this model illustrates the areas of the city where a relatively higher school capacity (higher number of student “seats”) is needed and areas where less capacity may be warranted. Of course, there are other criteria to evaluate in addition to proximity to students. Factors such as the number and quality of existing nearby schools, physical barriers such as highways, and societal barriers such as language and ethnicity are all important considerations as well.
While these data and maps are not prescriptive by themselves, they represent an important tool for policymakers working toward improving the climate for education in Detroit. Further layers of data would need to be added to this analysis to paint a more precise picture, but, optimistically speaking, the data and tools are becoming available to make data-driven decisions on education.