Analyzing Quality of Food Across Economic and Racial Groups in Chicago

Team NKMG

Gabriel Lima, Kenyu Kobayashi, Murat Topak, & Niko Masiero

Applied Data Analysis, EPFL, Autumn 2019

Racism in the United States

Racism is arguably institutionalized in American society. Result of almost 100 years of slavery and consequent lack of opportunity for the black population to get back on their feet has created various barriers to this community. This situation has led to the exclusion of the black population from health, educational, social, and economic resources. Data suggests that black Americans are more likely to suffer from mental illness, can live up to 20 years less than an average white American, and do not have the same higher level education opportunities.

Inequality in Food Distribution

Another important problem in the current world landscape is the precarious food distribution across different regions. For instance, diabetes has hit minorities the hardest and many layers of society worldwide suffer of poverty and hunger. However, there has not been any analysis on the quality of food provided to different social and racial groups in a certain location. Following the trend of minorities receiving lower quality services and products, we hypothesize that the same would follow in terms of the quality of the food available to the black and poorest population. Therefore, we ask the following research question:

Does the quality of food available to different racial and economic groups differ significantly?

Methodology

In order to answer our question, we utilized a dataset of food inspections in the city of Chicago. We use the results of the inspections as a proxy for a measure of food quality in a certain location. If a certain neighborhood has a higher failure rate, we can infer that the food available there is of lower quality. As a form of measuring economic and racial variables in the city of Chicago, we also obtained information about the racial distribution across neighborhoods, the city of Chicago’s measurement of social and economic hardship , and data regarding crimes committed in the city. With these datasets, we divide our main research question in three different sub-questions:

Research Questions

RQ1) Do neighborhoods with a higher concentration of black people are subjected to lower quality food?

RQ2) Does economic and social hardship correlate with food quality in a certain neighborhood?

RQ3) Do neighborhoods with a higher crime rate present lower food quality?

Food Inspections

The dataset provided by the city government identifies in which restaurant the inspection was conducted, the final result, violations, risk, location, and date. With this information, we calculate 6 measures of food quality and food inspections for each neighborhood.

Inspections per capita: Number of inspections conducted in a certain neighborhood divided by the neighborhood's population.
Rate of failure: Percentage of inspections that received "Fail" in a certain neighborhood.
Average risk: The average risk given by inspectors of all establishments in a certain neighborhood.
Violations per capita: Average number of food violations per citizens of a neighborhood.
Inspections per establishment: Average number of inspections divided by the number of establishments of a neighborhood.
Violations per establishment: Average number of food violations divided by the number of establishments of a neighborhood.

We aimed to analyze whether there is some correlation between inspection results and the composition of a neighborhood in terms of race and economy. Note that 1) and 5) do not measure food quality, but how often inspections are conducted. We also aim to understand whether there is a bias on where inspections are conducted.

Census Data

Among all the variables obtained in the US nationwide census, we only used two factors for our project. First, we obtained each neighborhood's population so we can normalize our measurements according to the population size of a neighborhood, and its racial distribution, as a measure of how many citizens belong to racial minorities.

Hardship Index

The city of Chicago also makes available the economic hardship of all its community areas. The economic hardship index is a function of:

Unemployment (of citizens over the age of 16 years)
Education (of citizens over 25 years of age without a high school diploma)
Per capita income level
Poverty (below the federal poverty level)
Crowded housing (housing units with more than one person per room)
Dependency (population under 18 or over 64 years of age)

We used the economic hardship index as a measurement of economic conditions of each community area in Chicago. A higher index indicates poorer conditions. We scaled the hardship index to a range between 0 and 1 for presentation purposes.

Crime Data

This dataset was also provided by Chicago's government. It reflects reported incidents of crime of various types. We have chosen to remove crimes that do not directly correlate with the level of violence of a neighborhood. For instance, we remove non-violent crimes and records of car violations as the vehicle might not belong to the neighborhood where the crime was reported. We present the number of crimes per capita in the map on the right.

Results

Choose the food inspection index you would like to present in the scatter plots.
Hovering over points show the reported values and their significance level.

Conclusion

Our results indicate that there does not exist a strong correlation between the food quality and our tackled socio-economic variables. With the exception of a small non-linear correlation between the percentage of African Americans in a community and the number of inspections, both Pearson and Spearman correlation indexes are not statistically significant.

RQ1: Do neighborhoods with a higher concentration of black people are subjected to lower quality food?

A higher concentration of African Americans in a certain neighborhood is not correlated to a decrease in food quality, measured by the results of food inspections. Interestingly, our results indicate that these neighborhoods are subject to inspections more frequently than others. We hypothesize that this is partly caused by a prejudice in the sanitary surveillance agency, which believes that establishments in certain black communities are less safe and will result in higher failure rate. Therefore, surveillance agencies often conduct inspections in these neighborhoods.

RQ2: Does economic and social hardship correlate with food quality in a certain neighborhood?

There is no significant correlation between economic hardship and food quality according to our results. Even though citizens in neighborhoods with high hardship indexes might not be able to acquire similar quantities of food, they have food of similar quality available.

RQ3: Do neighborhoods with a higher crime rate present lower food quality?

Higher criminality does not significantly correlate with the results of food inspections. This result indicates that more violent communities do not have any disadvantage in terms of the quality of the food available.

Limitations

Note that the results of food inspections might not be exclusively dependent on the quality of the food available, but also on the establishment where it is served. Nevertheless, our results indicate that there is no significant difference between the food available at different neighborhoods based on their racial distribution, economic hardship, and criminality. We believe that this is partially caused by the fact that even though Chicago is a big city, most of its food comes from similar producers and companies due to location proximity. Results might differ if we analyze larger scale datasets, such as in the US or the world.

Remarks

We by no means argue that there is no prejudice in how and what food has been distributed to different communities in the city of Chicago. Our work does not cover what is actually available to be bought and consumed by citizens, but only uses the results and frequency of food inspections as a proxy for the quality of the food available to different communities. Nevertheless, our results indicate that this discrepancy, if found in studies using other measures of food quality, is expected to be, if not null, not excessive.