Our robust data set and scientific tools make clear that we have a big public health problem on our hands. But what does science tell us about our ability to band together and reverse course, or to create something entirely new?

Often, our response to public health and other socioeconomic problems is a new law or policy to steer resources in a particular direction. Such measures are needed, and important. But what can be done at the local level, to complement broader action and direct and manage change?

Imagine that you have the potential to improve food access and public health for one of the 50,000 blocks that we studied. Maybe you’re a developer, bringing a full-service grocer to that block, or a community resident, working with a convenience store owner to increase the quality and selection of fresh fruits and vegetables. Given the magnitude of food imbalance in Metro Detroit and especially the city itself, and the fact that there are thousands of blocks and thousands of people in affected areas, it would be human nature to wonder if the one project you might work on could make a meaningful difference.

Psychologically, the answer sometimes feels like no – that the problem is too large to tackle – but mathematically, the answer is always yes, at least to some degree. Illustrating the first of many steps in our process of assessing food balance and public health brings this point to life: we measure the distance from every single food venue that we know to exist to the center of every single block in our study area to identify the shortest distance to each type of food venue from each block. This involves millions of individual calculations to compute the final scores for each block.  For example, just to calculate the first set of distance scores for USDA Food Stamp retailers, the program must execute over 50 million computations. Our study is a static picture: one moment in time in the history of Detroit. Moving forward, if we are to track the impact of different types of stores opening and closing block-by-block, we must recalculate the distance and food balance scores all over again for all 50,000 blocks each time, as the scores are only meaningful in relation to one another. It’s all relative. Therefore, all blocks have the potential to be affected in some way.

Clearly, the blocks closest to the one where our hypothetical project might take place – the new full-line grocer or the upgraded convenience store – have the greatest potential to improve, but only by recalculating the universe of all 50,000 blocks do we know the true level of impact. And if we intervene with a new project or program on 5 strategically placed blocks in our continuum, instead of just one random block, the ripple effect along the continuum is amplified.

Clearly, the blocks closest to the one where our hypothetical project might take place – the new full-line grocer or the upgraded convenience store – have the greatest potential to improve, but only by recalculating the universe of all 50,000 blocks do we know the true level of impact. And if we intervene with a new project or program on 5 strategically placed blocks in our continuum, instead of just one random block, the ripple effect along the continuum is amplified.

There are 200 tracts in Detroit that have Food Balance Scores of 2 or more, meaning that they are out-of-balance in terms of food access (the distance to the closest mainstream food venue is at least twice as far as the distance to the closest fringe food venue). If we want to focus on out-of-balance tracts in Detroit, instead of blocks, but built up from the block each time, we can pair the Food Balance Score with a wide range of variables, such as age, race, household size, car ownership, etc. Our food venue data is most valuable the lower we go down in geography: not only the block but the exact location on the block. The value comes from the data being highly accurate and not extrapolated from a more distant source. Demographic data, such as from the Census or from sources that attribute characteristics based on modeling techniques, are usually not accurate or available at the block level. By going up to the tract or at least the block group we are able to round out some of these errors.

This was a short excerpt. Read the Appendix today to see the full text and illustrations. Because the appendix is such a large document, it is also broken up into sections in the event you have trouble opening the full version.

Download Attachments