Examine racial inequities in unsolved murder cases

Roughly 50 percent of murders in American cities do not lead to an arrest, according to an analysis by the Washington Post (alternative article). This number does not account for race or geography, and in some neighborhoods, homicide arrest rates are much lower (alternative article). If you live in one of these neighborhoods, you may know the fear, depression, frustration, prolonged grief, and anger associated with unsolved homicide.

In this tutorial, you'll use statistical analysis in ArcGIS Pro to identify American cities where unsolved homicides are most prevalent. Then, you'll assess which segments of the population are most impacted and determine whether there is any racial or demographic inequity in regards to unsolved homicides. This workflow doesn't address why homicides remain unsolved; the reasons are complex and varied. But as you work through the tutorial, you'll have opportunities to think about possible causes and implications.

This tutorial is for GIS analysts interested in learning about strategies for analyzing social and racial equity. You'll develop skills in summarizing and mapping central tendency, analyzing categorical data, and creating regions from points.

This tutorial was last tested on August 2, 2024, using ArcGIS Pro 3.3. If you're using a different version of ArcGIS Pro, you may encounter different functionality and results.

Requirements

• ArcGIS Pro (see options for software access)
• 1 credit

Calculate homicide arrest rates

Some cities, such as Baltimore and Chicago, solve so few homicides that vast areas stretching for miles experience hundreds of homicides with virtually no arrests. In other places, such as Atlanta, police manage to make arrests in a majority of homicides—even those that occur in the city's most violent areas.

―"Where Killings Go Unsolved," the Washington Post, June 2018 (alternative article)

To understand which cities in the United States have the best and the worst record for making arrests after someone is murdered, you'll analyze homicide data for the past 10 years. In particular, you'll summarize individual homicides by city and symbolize the result by unsolved homicide rate. Then, you'll create a bar chart that ranks each city.

Along the way, you'll view articles about the potential causes of unsolved homicides. You'll also have the option to explore homicide arrest rates by race and ethnicity.

Note:

This tutorial includes links to several articles that provide context, background, or additional information to extend your learning. Some sites where these articles are hosted may ask you to sign up for a subscription to continue viewing their content. If you don't want to subscribe, alternative articles are provided where possible.

Explore the data

First, you'll download an ArcGIS Pro project package and become familiar with the homicide data.

1. Download the Unsolved Homicides project.

   The Unsolved_Homicides file is downloaded to your computer. This file is an ArcGIS Pro project package file (.ppkx), which contains the map and data you'll use in this tutorial.

2. Locate the Unsolved_Homicides file on your computer and double-click it. If prompted, sign in to ArcGIS Pro using your ArcGIS organizational account.

   Note:

   If you don't have access to ArcGIS Pro or an ArcGIS organizational account, see options for software access.

   ArcGIS Pro opens and displays a map and data.

   

   The map shows homicide data for 50 cities across the United States. (This data is originally from the Washington Post and is hosted on ArcGIS Online.) Blue points represent solved homicide cases. A case is solved or closed if an arrest is made or if an arrest isn't possible, such as when the perpetrator dies. Red points represent open homicide cases, which remain unsolved.

   Note:

   This data was projected using the USA Contiguous Albers Equal Area Conic projection, appropriate for the full study area. If you plan to focus your analysis on a single city, you'll want to use the projection associated with that city. For more information about projections, read this blog post or complete this Learn ArcGIS tutorial.

3. In the Contents pane, right-click the Homicides layer and choose Attribute Table.

   

   The layer's attribute table appears. It contains more information about each homicide, including the date it occurred and the victim's race, age, and sex.

   The Disposition field indicates whether the case is open or closed, and whether it was closed by arrest or without arrest. The Unsolved=1, Closed=0 field simplifies the disposition into one of two numbers. Later, you'll use this field to calculate the rate of unsolved homicides by city. The Category field contains the same information, but in text form.

   Note:

   Crime data is almost always imperfect. Accuracy and consistency can be affected by reporting and recording errors, variations in police practices and police resources, and longstanding institutional bias associated with race, income, and gender. While these issues are beyond the scope of this tutorial, it's important to think critically about them when drawing conclusions.

   Race and ethnicity are intersectional and fluid. While people may associate with multiple races, they are generally categorized as a single ethnicity. The homicide data only records the victim race field as one of Asian, Black, Hispanic, White, Unknown, or Other. These groupings may not be appropriate for all data analyses. It's unclear if the Hispanic category includes Blacks or African Americans, Asians, or Whites who associate with the Hispanic ethnicity.

4. Close the Homicides table.

   Next, you'll create charts of homicide counts and the rates at which homicides are closed.

5. In the Contents pane, right-click the Homicides layer, point to Create Chart, and choose Bar Chart.

   

   The Homicides - Chart of Homicides view and Chart Properties pane appear. First, you'll chart the number of homicides in each city.

6. In the Chart Properties pane, for Category or Date, choose City. For Sort, choose Y-axis Descending.

   

   The chart view updates.

   

   Chicago has the largest number of homicides. Other cities with high homicide counts are Philadelphia, Houston, Baltimore, and Detroit.

   Tip:

   Depending on the size of your chart view, not all city names may be visible. Point to a bar on the chart to see the city name.

   Next, you'll split the homicides into solved and unsolved categories.

7. In the Chart Properties pane, click the General tab, and for Chart title type Comparison of data counts by City.

   

8. In the Chart Properties pane, on the Data tab, for Split by, choose Category.

   

   The chart updates.

   

   In this chart, every city has two bars, one for solved homicides and one for unsolved homicides. Several cities have more unsolved than solved homicides. The difference between solved and unsolved homicides does not appear to be correlated to the total number of homicides. Even some cities with relatively few homicides have more unsolved than solved.

   To better show the ratio of solved to unsolved homicides by city, you'll change the chart to visualize proportions.

9. In the Chart Properties pane, click the Series tab. For Display multiple series as, click 100% Stacked.

   

   The chart updates, with the bars for each category stacked on top of one another.

   

   In this chart, bars depict the percentage of total homicides that are solved and unsolved, instead of the total numbers of solved and unsolved homicides. Some cities, such as Chicago, have solved as few as 33 percent of their total number of homicides. Other cities, such as Richmond, have solved 78 percent.

10. Close the chart view and the Chart Properties pane.

    From your initial exploration of the data, you've learned that the problem of unsolved homicides varies widely across the United States. Using statistical analysis, you'll gain even more insight.

Summarize homicide data

The homicides appear on the map as individual points. But you're less interested in individual homicides than citywide trends. You'll summarize the homicide data using the Mean Center tool, which identifies the geographic center of a cluster of points and creates summary statistics for each cluster.

Once you aggregate the data, you'll map citywide trends in unsolved homicides using symbology.

1. On the ribbon, click the Analysis tab. In the Geoprocessing group, click Tools.

   

   The Geoprocessing pane appears.

2. In the Geoprocessing pane, search for Mean Center. In the list of results, click the Mean Center tool.

   

   Running this tool creates points at the geographic center of each city's reported homicides. It'll also find an average value for each cluster of a dimension field that you choose.

   You'll find the average value of the Unsolved=1, Closed=0 field. In this field, all values are either 0 or 1. Taking the average of these values will give you a fraction between 0 and 1. This average is equal to the unsolved homicide rate. By summarizing homicides using this field, you'll calculate the homicide rate for each city.

3. In the Mean Center tool, for Input Feature Class, choose Homicides. For Output Feature Class, type City_Point_Rates.

   

   Next, you'll choose the case field, which determines how the points are clustered, and the dimension field.

4. For Case Field, choose City. For Dimension Field, choose Unsolved=1, Closed=0.

   

5. Click Run.

   The tool runs and creates the City_Point_Rates layer, which is added to the map.

   

   Every city has one point to represent it. You'll symbolize these points using the unsolved homicide rate that the tool calculated, which is stored in the new Unsolved field.

6. In the Contents pane, right-click City_Point_Rates and choose Symbology.

   The Symbology pane appears.

7. In the Symbology pane, for Primary symbology, choose Graduated Colors. For Field, choose Unsolved.

   

   This symbology style divides the symbols into five categories based on the unsolved homicide rate. The breaks for each category have a large number of decimals, so you'll adjust them.

8. For Upper value, double-click each value and change them (going from the first value to the last) to 0.25, 0.35, 0.45, 0.55, and 0.68.

   

   You'll change the labels to match. Because the values are a rate ranging from 0 to 1, by multiplying them by 100, you get a percentage. You'll use the labels to communicate the values as percentages.

9. For Label, change the labels (from the first label to the last) to 21 to 25%, 25 to 35%, 35 to 45%, 45 to 55%, and 55 to 68%.

   

   Next, you'll format the symbols for all categories to be larger.

10. Click More and choose Format all symbols.

    

    A gallery of symbols appears.

11. Click Circle 3.

    

    The symbols are updated on the map, but all of them are the same red color. Next, you'll choose a color scheme.

12. At the top of the pane, click the Return to primary symbology pane button.

    

13. For Color scheme, click the drop-down menu, check Show names, and choose the Purple-Red (Continuous) color scheme.

    

14. Close the Symbology pane. In the Contents pane, uncheck the Homicides layer.

    

    The unsolved homicide rate is over 55 percent in six cities: Chicago, Buffalo, Baltimore, Detroit, New Orleans, and Stockton. For more than 50 percent of all murders in these cities, no one is arrested for the crime. The lowest unsolved homicide rates are located in Richmond, Charlotte, Tulsa, Albuquerque, and San Diego.

    Based on the map, there is no clear spatial distribution to high and low unsolved homicide rates. Cities near one another and cities in the same state may have significantly different rates. However, by summarizing the homicide data, you have a better nationwide understanding of the problem of unsolved homicides in the United States.

Rank cities by unsolved homicide rate

Your map shows the location of cities with high and low unsolved homicide rates. Next, you'll create a chart that ranks these cities from the highest to the lowest rate.

1. In the Contents pane, right-click City_Point_Rates, point to Create Chart, and choose Bar Chart.
2. In the Chart Properties pane, for Category or Date, choose city. For Aggregation, choose <none>.
3. For Numeric field(s), click Select and check Unsolved. Click Apply.

   A chart that shows the unsolved murder rate for each city is created. However, the chart organizes the cities alphabetically, not by the rate. You'll change how the chart is sorted.

4. For Sort, choose Y-axis Descending.

   

   Now, the bars in the chart are organized by the y-axis value (the unsolved murder rate). Before you examine the chart, you'll change the chart title and the axis titles so the chart is more legible.

5. In the Chart Properties pane, click the General tab. For Chart title, type Unsolved Homicide Rates by City.
6. For X axis title, type City. For Y axis title, type Unsolved Homicide Rate.

   

   The chart and axis titles are updated in the chart.

   

   The cities are ranked from highest to lowest homicide rates, and the color of each bar matches the color of the city's symbol on the map. As you previously learned, Chicago has the highest unsolved homicide rate in the country, but now you can better visualize which cities are higher and lower at a glance.

7. Close the chart and the Chart Properties pane.

   Tip:

   To reopen any chart you've created, double-click the chart name in the Contents pane.

8. On the Quick Access Toolbar, click the Save button. Click Yes if prompted to save to the current version of ArcGIS Pro.

   

   The project is saved.

Analyze rates by race and ethnicity

You've explored the homicide data, created charts, and performed statistical analysis to determine the unsolved homicide rate in each city. Your analysis has focused on where homicides are unsolved, but it may also be important to ask what factors play a role in whether a homicide is solved. The following resources provide more insight into the issue of unsolved homicides:

• "Why are unsolved murders more common in certain communities?" by PBS
• "Black murders accounted for all America's clearance decline" by Murder Accountability Project
• "An Unequal Justice" by the Washington Post
• "Shoot Someone In a Major U.S. City, and Odds Are You'll Get Away With It" by The Trace

Many of these resources show that unsolved murder rates have increased over the past 50 years. Additionally, these resources indicate demographic or racial discrepancies between areas where murders are solved more or less often.

Next, you'll repeat your mean center analysis and create a new chart ranking unsolved homicide rates, but this time take into account the race of the victim. Is a murder less likely to be solved if the victim is of a certain race?

1. In the Contents pane, right-click the Homicides layer, point to Create Chart, and choose Bar Chart.
2. In the Chart Properties pane, for Category or Date, choose Victim Race.

   The chart updates to show the total number of homicides by victim race. Black victims make up the majority of homicides, solved and unsolved. Many of the resources you looked at in the section introduction claimed that the unsolved homicide rate was higher for Black victims than victims of other races. You'll run the Mean Center tool again on only homicides with a Black victim to see if your data supports these claims.

3. In the chart, click the Black bar.

   

   By selecting the bar, you also select all homicides in the Homicides layer where the victim's race is Black. Geoprocessing tools, including the Mean Center tool, will only run on selected features if there are any.

4. Close the chart and the Chart Properties pane. If necessary, return to the Geoprocessing pane.

   The Geoprocessing pane shows the parameters for the last tool you ran, which is the Mean Center tool. You'll change some parameters and run the tool again.

   Tip:

   If you closed the Geoprocessing pane or it doesn't show the Mean Center tool, you can reopen any tool you previously ran in the project with the parameters you used. On the ribbon, in the Analysis tab, click History. In the History pane, on the Geoprocessing tab, double-click Mean Center to reopen the tool.

5. In the Mean Center tool, for Output Feature Class, type City_Black_Unsolved_Rates. Confirm that Input Feature Class is set to Homicides, Case Field is set to City, and Dimension Field is set to Unsolved=1, Closed=0.

   

6. Click Run.

   The tool runs and the City_Black_Unsolved_Rates layer is added to the map. You'll import the symbology from the City_Point_Rates layer to this layer.

7. In the Contents pane, confirm that City_Black_Unsolved_Rates is selected.
8. On the ribbon, click the Feature Layer tab. In the Drawing group, click Import.

   

   The Import Symbology window appears.

9. In the Import Symbology window, for Symbology Layer, choose City_Point_Rates.

   

   By default, the Unsolved field is chosen as the symbology field. This default is what you want, because you used the Unsolved field to determine the graduated symbology of the City_Point_Rates field.

10. Click OK. In the Contents pane, uncheck City_Point_Rates.

    

    Note:

    The cities of Dallas, Kansas City, and Phoenix do not record victim race for homicides, so none of these cities have data on the map.

    When looking only at Black victims, there are more cities with unsolved homicide rates over 55 percent. Only one city (Charlotte) has an unsolved homicide rate under 25 percent. If you repeated this analysis for White victims, you would find no cities with an unsolved homicide rate above 55 percent.

    Your analysis suggests that race is a factor in whether homicides are solved, as the resources you looked at in the section introduction indicated. A failure to solve Black homicides may lead to mistrust of law enforcement in Black communities, and subsequently reduced cooperation and even fewer arrests.

    Next, you'll create a chart to compare the unsolved homicide rates across all races and ethnicities.

12. In the Contents pane, right-click Homicides, point to Create Chart, and choose Bar Chart.
13. In the Chart Properties pane, for Category or Date, choose Victim Race. For Split by, choose Category.
14. Click the Series tab. For Display multiple series as, choose 100% Stacked. In the table, drag the Unsolved category above the Solved category.

    

15. On the General tab, for Chart title, type Solved and Unsolved Homicide Rates by Race/Ethnicity. For X axis title, type Victim Race/Ethnicity.

    The chart updates. The bar showing Black homicide rates is still selected, so you'll clear the selection.

16. On the ribbon, click the Map tab. In the Selection group, click Clear.

    

    The selection is cleared.

    

    The chart confirms that unsolved homicide rates are highest when the victim is Black. A homicide with a Black victim remains unsolved about 49 percent of the time, compared to about 28 percent for a homicide with a White victim.

17. Close the chart and the Chart Properties pane. Save the project.

You've summarized more than 52,000 homicides and calculated the unsolved homicide rate for each city. You also analyzed unsolved homicide rates by race and ethnicity, particularly focusing on arrest rates when the victim is Black. Your findings support the claims in the resources you read that race plays a role in the rate at which homicides go unsolved.

This part of the analysis was focused on the national level. Next, you'll examine a single city with a high unsolved homicide rate and consider some of the consequences of living in a community where murders do not lead to arrests.

Evaluate demographics for low arrest rate areas

Nearly a year after Aice Jackman was gunned down in the street, his mother and 5-year-old brother walked into a Dunkin' Donuts, where the boy spotted a pit bull puppy and dashed over to pet it. Kaiesha Skinner's gaze followed her young son and then settled on the man holding the leash. Their eyes met. She froze: It was the same man who she believes killed Jackman. She grabbed her youngest son's hand, yanking him away from the man and back to their car. "We all know who shot my son," Skinner said later. "They just haven't arrested him."

―"An Unequal Justice," the Washington Post, July 2018 (alternative article)

Previously, you learned that a homicide in the United States is more likely to go unsolved if the victim's race is Black. To better understand the racial and social inequity of unsolved homicides, you'll compare the characteristics of the people who live in neighborhoods where homicides remain unsolved to the characteristics of people in all homicide areas. Rather than victims or perpetrators, this part of the workflow is about communities. Unsolved homicides impact the communities where they occur, creating fear, apprehension, and grief.

You'll focus on a single city: Buffalo, New York. This city has one of the highest unsolved homicide rates in the country. (If you want, you can repeat the following analysis for any city where the Washington Post collected data.)

Create a layer for Buffalo homicides

First, you'll create a layer containing only homicides in Buffalo.

1. If necessary, open your Unsolved Homicides project in ArcGIS Pro.
2. In the Contents pane, uncheck City_Black_Unsolved_Rates and check Homicides.

   Tip:

   By this point in the analysis, your Contents pane has a large number of layers. To reduce the amount of space they take up in the pane, you can collapse their symbology by clicking the arrow next to the layer name.

3. Right-click Homicides, point to Data, and choose Export Features.

   The Export Features window appears. The Homicides layer is already chosen for Input Features.

4. In the Export Features window, for Output Name, type Buffalo_Homicides.

   You can create an expression to export only a subset of all homicides. You'll create an expression to ensure your exported layer contains only homicides that occurred in Buffalo.

5. Expand the Filter tab, and under Expression, build an expression that reads Where City is equal to Buffalo.

   

6. Click OK.

   The tool runs and the Buffalo_Homicides layer is added to the map. This layer only contains homicides that occurred in Buffalo.

7. If necessary, close the Feature Class to Feature Class (Conversion Tools) window.
8. In the Contents pane, right-click Buffalo_Homicides and choose Zoom To Layer.

   The map navigates to the extent of the layer, which is the same as the location of Buffalo, New York, located near the border between the United States and Canada.

   

9. Turn off the Homicides layer.

Find areas where unsolved homicides predominate

Your goal is to analyze the demographics of neighborhoods in Buffalo where there are a large number of unsolved murders. To locate those neighborhoods, you'll use colocation analysis. Colocation analysis assesses the composition of homicides surrounding each unsolved homicide and compares it to the typical composition found across the city.

For example, if homicides in Buffalo are 50 percent solved and 50 percent unsolved, colocation analysis expects 50 percent of the homicides near any given homicide to be 50 percent solved as well. An area where unsolved homicides are surrounded by the expected or higher-than-expected percent of solved homicides is an area where unsolved homicides are collocated with solved homicides. More important to your analysis are areas of isolation, where unsolved homicides are surrounded by a higher-than-expected percent of other unsolved homicides. These are communities and neighborhoods where murders occur, but arrests aren't made.

1. If necessary, open the Geoprocessing pane. Search for and open the Colocation Analysis tool.

   By default, the Input Type is set to Single dataset. This input type is used when the two variables in your analysis (in this case, solved and unsolved) are in the same layer.

2. For Input Features of Interest, choose Buffalo_Homicides. For Field of Interest, choose Category.

   The Category field has two values, Solved and Unsolved, which will be the basis of your colocation analysis. You want to analyze unsolved murders and their relationship to neighboring solved murders.

3. For Category of Interest, choose Unsolved. For Neighboring Category, choose Solved.

   Next, you'll choose the distance at which murders are considered to be neighboring. You'll use one mile as the distance, as this is a size appropriate for local neighborhoods and communities.

   Note:

   The distance you choose affects your results. It should be appropriate to the scale of your analysis and the area you're examining. One mile was selected to reflect a distance where homicides might have an impact on local residents. Someone who lives in Buffalo or someone who has a better understanding of neighborhood dynamics in Buffalo (such as how far fear, feelings of safety, and other emotions reverberate), might justify a different scale of analysis. To model more complex relationships (such as travel time), you could provide a spatial weights matrix.

4. For Neighborhood Type, choose Distance band. For Distance Band, type 1 and choose Statute Miles.

   Last, you'll choose the number of permutations, which determines the statistical significance of the relationships between solved and unsolved homicides. A higher number provides more accurate results, but it may cause the tool to take longer to run. You can choose 99, 199, 499, 999, or 9999 permutations. Unless performance is an issue, you should usually choose 999 or 9999 permutations for the best accuracy.

5. For Number of Permutations, choose 999.
6. For Output Features, type Colocation_Relationships_Buffalo.

   

7. Click Run.

   The tool runs and adds the Colocation_Relationships_Buffalo layer to the map. A warning appears at the bottom of the Geoprocessing pane. This warning states that some features in the dataset had no neighbors. It's okay to ignore this warning, as homicides with no neighbors are neither areas of colocation nor areas of isolation.

8. In the Contents pane, uncheck the Buffalo_Homicides layer.

   

   The layer symbology has four categories. You're interested in the Isolated - Significant category (dark green), which is where unsolved homicides are predominately surrounded by other unsolved homicides and the pattern is statistically significant. These areas, where murders are likely to go unsolved, are located primarily in the eastern part of the city.

Enrich areas where unsolved homicides predominate

You've found where unsolved homicides predominate. Next, you'll learn more about these neighborhoods and communities. In particular, you'll examine their racial composition and demographic characteristics, which you'll later compare to other areas in the city.

First, you'll create a buffer around the areas where murders remain unsolved. Then, you'll enrich that buffered area with demographic data.

1. On the ribbon, on the Map tab, in the Selection group, click Select By Attributes.

   

2. For Input Rows, confirm that Colocation_Relationships_Buffalo is selected.
3. Under Expression, create an expression that reads Where LCLQ Type is equal to Isolated - Significant.

   

4. Click OK.

   Points where unsolved homicides predominate are selected.

   

5. In the Geoprocessing pane, click the Back button. Search for and open the Pairwise Buffer tool.

   You'll create a buffer of one mile, which matches the distance you used to determine neighbors in your colocation analysis. You're using the Pairwise Buffer tool instead of the regular Buffer tool because it is better suited for creating overlapping buffers from numerous points.

6. In the Pairwise Buffer tool, for Input Features, choose Colocation_Relationships_Buffalo. For Output Feature Class, type Unsolved_Homicide_Areas_Buffalo.
7. For Distance, type 1 and choose Statute Miles.

   By default, a buffer will be created for every input feature. You'll dissolve all buffers into a single feature, which will make subsequent analysis more convenient.

8. For Dissolve Type, choose Dissolve all output features into a single feature.

   

9. Click Run.

   The tool runs and creates a one-mile buffer around the selected points.

   

   Note:

   Because the permutations done by the Colocation Analysis tool add a random component to the analysis, there's a chance your results will be slightly different from those shown in the example images. These minor differences will not affect your conclusions.

10. On the ribbon, in the Map tab, in the Selection group, click Clear to clear the selection.
11. In the Geoprocessing pane, click the Back button. Search for and open the Enrich tool.
12. For Input Features, choose Unsolved_Homicide_Areas_Buffalo. For Output feature class, type Unsolved_Homicide_Area_Data.
13. Next to Variables, click the add button.

    The Data Browser window appears. You can use this browser to search for and add the demographic data variables that the tool will use to enrich the layer. The more variables you add, the more credits the analysis costs. You'll add six variables covering race, age, and income.

14. In the Data Browser window, double-click the Race category.

    

15. Double-click Population by Race. For the following three variables, select the percent sign, unselect the number sign, and check the box:
    • 2024 White Population
    • 2024 Black Population
    • 2024 Asian Population

    Note:

    There are many other distinctive racial groups that can be included in an analysis of this type. You're welcome to add more racial groups, but for the purposes of this tutorial, these will be the only groups used.

    

    Note:

    Data is periodically updated. Use the latest data available.

16. Click the back button and double click Hispanic Origin. For 2024 Hispanic Population, select the percent sign, unselect the number sign, and check the box

    

17. Click the back button or click Categories to return to the list of categories. In the search box, type median age and press Enter.
18. Check the 2024 Median Age variable.

    

19. Search for and check the 2024 Wealth Index variable.

    

    You've selected six variables total.

20. In the Data Browser window, click OK.

    

    The variables are added to the Enrich tool parameters. Notice that there is a message indicating that this tool requires credits, which are the cost of using the demographic variables in your analysis. The credit consumption for running this tool is low (less than 1 credit), so you will proceed with running it. You also can get an estimate of the credits that will be used, so you know before you run the tool.

    Note:

    Learn more about credits.

21. At the top of the tool, click estimate credits.

    

    The estimated credit consumption is only .06 credits.

22. Click Run.
23. Close the Geoprocessing pane. In the Contents pane, right-click UnsolvedHomicideAreasData and choose Attribute Table.

    Note:

    You may need to scroll down the Contents pane to find the layer.

    The table has fields for each of the variables you selected.

    

    In Buffalo communities where homicides remain unsolved, 64 percent of the residents are Black. The median age is 33.6 and the wealth index is 32. The national average wealth index is 100, so a wealth index of 32 indicates that the neighborhood is predominately poor.

    While these statistics suggest that neighborhoods with unsolved homicides are primarily Black and below the national average in terms of wealth, you can't make any claims about racial inequity until you compare this area to all homicide areas.

Obtain statistics for all homicide areas

It's possible that all areas where homicides occur in Buffalo have a similar demographic profile, regardless of whether homicides are solved or unsolved. You'll need to compare the demographics of the unsolved homicide area to the demographics of the other homicide areas across the country.

1. Close the table.

   To avoid repetition and credit expenditure, the buffer area for other homicide areas across the United States has already been created and enriched using the same steps you followed for Buffalo.

2. In the Contents pane, right-click Homicide Area Data and choose Attribute Table.
3. In the table, click the row number for Buffalo to select it.

   

   The entire row is highlighted.

4. Scroll to the right until you reach the 2024 variables.

   Based on the table, the area where all homicides occur in Buffalo has a Black population of 31 percent, compared to 75 percent in the unsolved homicide area. Additionally, this area has a wealth index of 53, compared to 29 in the unsolved homicide area. These findings suggest that homicides tend to go unsolved in areas that are poorer and have a larger Black population.

   The story is similar for other cities where unsolved homicide rates are high. For instance, if you repeated this analysis for Chicago, you would find that the population of areas where homicides tend to remain unsolved is 62 percent Black, while the population of all homicide areas is 30 percent Black. The wealth index is 45 in unsolved homicide areas and 79 in all homicide areas. In Baltimore, Detroit, and New Orleans, the other cities with the highest unsolved homicide rate, areas where homicides remain unsolved have a higher percentage of Black residents and a lower wealth index compared to all homicide areas.

   The following image compares the ethnic and racial differences between unsolved homicide areas (red) and solved homicide areas (blue) in Chicago, Baltimore, Detroit, and New Orleans:

   

   Note:

   The Hispanic field created by the Enrich tool represents ethnicity (not race) and includes Asians, Blacks or African Americans, and Whites who identify as Hispanic.

   These findings are evidence of racial inequity. They demonstrate that Black communities experience a disproportionate burden with regard to unsolved homicides compared to other communities.

5. On the ribbon, on the Map tab, in the Selection group, click Clear.

   You've compared the demographic data for Buffalo and some of the other cities with high rates of unsolved murders. Next, you'll extend your analysis to the entire country. Rather than look at every city individually, you'll summarize the demographic characteristics for all cities with homicide data. Specifically, you'll calculate averages for the race, age, and wealth fields, weighted by each city's population.

   If the results from this analysis reveal that unsolved homicide areas do not demonstrate similar demographic characteristics to all homicide areas, you might be uncovering potential evidence of social and racial inequity.

   First, you'll multiply each demographic field by the population field. By weighting the fields by population, you ensure that cities with larger populations have a larger influence on the average. Doing so prevents a city with a small population, which may have an extremely high percent of one demographic, from skewing the results.

6. In the Geoprocessing pane, search for and open the Calculate Field tool.

   This tool calculates a new field based on an equation that you create. You'll run this tool six times, once for each of the demographic fields. Each time you run the tool, you'll use an equation that multiplies one of the demographic fields by the population field.

7. In the Calculate Field tool, enter the following parameters:
   • For Input Table, choose Homicide Area Data.
   • For Field Name (Existing of New), type wAsian (the w stands for weighted).
   • For Field Type, choose Double (64-bit floating point).
   • For wAsian =, type !populationtotals_totpop_cy_1! * !raceandhispanicorigin_asian_cy_p_1!.

   Tip:

   Alternatively, you can build the expression by double-clicking the 2024 Total Population field, clicking the multiplication sign (*), and double-clicking the 2024 Asian Population: Percent field.

   

   Note:

   The field names appear in the expression, not the alias.

8. Click Run.

   The tool runs and a new field named wAsian is added to the end of the table. It contains the percentage of each city's population that is Asian, weighted by population.

9. Run the Calculate Field tool five more times, changing the Field Name and the second half of the Expression parameters based on the following list (keep the Field Type parameter set to Double and keep !populationtotals_totpop_cy! at the start of every expression ):

   Field Name	Expression
   wBlack	!populationtotals_totpop_cy! * !raceandhispanicorigin_black_cy_p_1!
   wHispanic	!populationtotals_totpop_cy! * !raceandhispanicorigin_hisppop_cy_p!
   wWhite	!populationtotals_totpop_cy! * !raceandhispanicorigin_white_cy_p_1!
   wMedianAge	!populationtotals_totpop_cy! * !gender_medage_cy!
   wWealthIndex	!populationtotals_totpop_cy! * !wealth_wlthindxcy_1!

   The calculated values populate the Homicide Area Data attribute table.

   

   Next, you'll create summary statistics that add together all of the values for each field. When you divide the summed values by the summed total population, you'll find the average percent for each demographic variable across the country.

10. In the Geoprocessing pane, search for and open the Summary Statistics tool.

    This tool calculates statistics for all values in a field or fields.

11. In the Summary Statistics tool, set the following parameters:
    • For Input Table, choose Homicide Area Data.
    • For Output Table, type Sums.
    • For Statistics Field(s), for Field, choose wAsian, wBlack, wHispanic, wWhite, wMedianAge, and wWealthIndex.
    • If necessary, for Statistic Type, choose Sum for each field.

    

12. Click Run.

    The tool runs and creates the Sums table, which is added to the bottom of the Contents pane.

13. Close the Homicide Area Data table. In the Contents pane, under the Standalone Tables section, right-click the Sums table and choose Open.

    

    To calculate the averages, you'll divide the weighted sums by the total population sum (50,119,283). You can perform these calculations using the Calculate Field tool, but for the purposes of this tutorial, the calculations are provided.

    • Average percent Asian population in all homicide areas: 472639288.4 / 50119283 = 9.4 percent
    • Average percent Black population in all homicide areas: 1175732428.04/ 50119283 = 23.45 percent
    • Average percent Hispanic population in all homicide areas: 1525332275.05 / 50119283 = 30.43 percent
    • Average percent White population in all homicide areas: 1881577324.59 / 50119283 = 37.54 percent
    • Average median age in all homicide areas: 1808939077.2/ 50119283 = 356.1
    • Average wealth index in all homicide areas: 3924682224 / 50119283 = 78.3

    The following table compares the statistics for unsolved homicide areas in Buffalo, all homicide areas in Buffalo, and all homicide areas for the 50 cities in the United States where homicide data was collected. A homicide area is defined as a location within one mile of any homicide.

    Field	Unsolved homicide areas in Buffalo	All homicide areas in Buffalo	All homicide areas in 50 cities
    Percent Asian	2.5	4.1	8.9
    Percent Black	75.1	32.3	25.2
    Percent Hispanic	4.4	11.7	30.3
    Percent White	17.9	54.9	47.7
    Median Age	36.4	36.8	35.4
    Wealth Index	29.0	49.0	78.4

    The unsolved homicide area you analyzed in Buffalo has a higher percentage of Black residents and a lower wealth index than the other homicide areas across the United States. While you only analyzed Buffalo in this workflow, this trend is also true for Baltimore, Chicago, Detroit, and New Orleans, the other communities most impacted by unsolved homicides.

    Why compare the demographic data for unsolved homicide areas to all homicide areas, rather than comparing it to all areas where solved homicides predominate or to all other homicide areas (all but the significant isolated unsolved homicides)? All of these comparisons are valid, but your approach is the most conservative option because the demographic data for all homicide areas includes the demographic data for unsolved homicide areas.

14. Close the table. Save the project.

    Your analysis provides data-driven evidence of racial and social injustice. It doesn't identify why there are areas where homicides remain unresolved, but it does show that a particular segment of the population is unfairly bearing the tragic consequences.

    What would it feel like to live in a community where murders aren't resolved? If you do live in one of these communities, what are the broader impacts? The following articles might provide insight:

    • "Sick and Grieving: The Toll of Unsolved Murders" by The Colorado Trust
    • "No One Has Been Arrested" by the Washington Post
    • "Crime and Violence" by the Office of Disease Prevention and Health Promotion
    • "Neighborhoods and Violent Crime" by the United States Department of Housing and Urban Development

    In this tutorial, you identified regions where unsolved homicides predominate in Buffalo, New York. Then, you obtained demographic data for these areas and compared it to the demographic data for all homicide areas. If the demographic data was similar, you could conclude that everyone who lives in a homicide area deals with the tragic consequences of unsolved homicides equally. Instead, you found stark differences. Areas where homicides remain unsolved have a higher proportion of Black and low-income residents, suggesting racial, social, or judicial inequities.

Next steps

What actions can you take to promote social equity in your community, school, or workplace, or among the people, organizations, and institutions with which you engage? You can repeat the workflow for your own city or a city near you or use ArcGIS StoryMaps to create a story that shares your results. Alternatively, you could adapt the workflow to understand if there are racial or social inequities associated with access to parks or high-quality schools; proximity to environmental hazards; or the availability of jobs, affordable housing, or healthcare services.

Additionally, you can explore the following resources:

• Murder Accountability Project, a nonprofit corporation established in 2015 to improve America's accounting of unsolved homicides, to assist law enforcement in clearing cold cases, and to inform the public about the growing problem of unsolved murder
• GIS Equity & Social Justice, an Esri user community
• Racial Equity: GIS for Racial Justice, GIS workflows and resources to map and better understand racial inequity
• Racial Equity GIS Hub, an ongoing, continuously expanding resource hub to assist organizations working to address racial inequities
• Using GIS to Operationalize Racial Equity & Social Justice, a recorded video session on how organizations use GIS to assess the impact of their choices and develop equitable policies and programs

To learn more about the complexities and limitations of crime data and crime clearance rates, explore the following resources:

• "Case Closed? How 'Solved' Murder Stats Are Misleading" by governing.com
• "Why big-data analysis of police activity is inherently biased" by The Conversation
• "Crime clearance rate in the United States in 2019, by type" by Statista

To learn more about why crimes aren't solved, explore the following resources:

• "Open Cases: Why One-Third of Murders in America Go Unresolved" by NPR
• "Police solve just 2% of all major crimes" by The Conversation
• "Killings of black people lead to arrests less often than when victims are white" by Macomb Daily

To learn more about combating systemic racism by supporting investment in Black communities, explore the following resources:

• "'A desperate need': Black homicide epidemic demands prevention, justice" by The Hill
• "Make Black lives matter by investing in Black communities - Part 1: The deadly impact of structural racism" by the American Federation of Teachers
• "Make Black lives matter by investing in Black communities - Part 2: Increased policing does not create safer schools and communities" by the American Federation of Teachers
• "Make Black lives matter by investing in Black communities - Part 3: Investing in the matter of Black lives" by the American Federation of Teachers

You can find more tutorials in the tutorial gallery.

Acknowledgements

• The Washington Post data used in this workflow was obtained from ArcGIS Living Atlas of the World. Victim names were removed. The data covers 10 years but is incomplete for some cities and time periods.
• Special thanks to Ted Mellnik from The Washington Post for encouraging this work.
• Special thanks to Kevin Butler, Lynne Buie, Andy Mitchell, Hannes Ziegler, and David Wynne, who participated in an analysis challenge using this data and provided inspirational ideas.
• Special thanks to Margot Bordne, Clinton Johnson, John Beck, and Khalid Duri for their helpful feedback.
• Gallery image from Pexels and shared under the Pexels License.
• Light Gray Canvas map sources: Esri, HERE, Garmin, FAO, NOAA, USGS, OpenStreetMap contributors, and the GIS User Community