data_descriptions.Rmd

---
title: "The State of Gun Violence in Charlottesville & Albemarle"
author: Samantha Toet & Michele Claibourn, [The Equity Center](https://www.virginiaequitycenter.org/)
output: 
  html_document:
    template: gun-violence-template.html
    css: mystyle.css
    toc: true
    toc_float: true
---

![](https://virginiaequitycenter.github.io/cville-equity-atlas/assets/img/equity-center-logo.png){width="350"}

------------------------------------------------------------------------

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = FALSE, message = FALSE, warning = FALSE)
library(ggmap)
library(ggrepel)
library(janitor)
library(leaflet)
library(lubridate)
library(magrittr)
library(reactable)
library(reactablefmtr)
library(readxl)
library(scales)
library(sf)
library(tidycensus)
library(tidyverse)

```

# Introduction

Every day, more than 120 people in America are killed with guns.[^1] The gun homicide rate in the U.S. is 26 times higher than that of other high-income countries.[^2] In 2024, the US Surgeon General's Advisory declared gun violence as a public health crisis.[^3]

[^1]: Everytown - Beyond Measure: Gun Violence Trauma - <https://everytownresearch.org/report/gun-violence-trauma/.Published> May 17, 2023.

[^2]: Grinshteyn, E., & Hemenway, D. "Violent death rates in the US compared to those of the other high‑income countries, 2015." Preventive Medicine 123 (2019), 20‑26. <https://doi.org/10.1016/j.ypmed.2019.02.026>

[^3]: US Surgeon General's Office - The U.S. Surgeon General’s Advisory on Firearm Violence: A Public Health Crisis in America - <https://www.hhs.gov/sites/default/files/firearm-violence-advisory.pdf>. Published July 2, 2024.

Charlottesville and Albemarle are not immune to the epidemic of gun violence. The rate of violent crime in our community increased by 30% from 2021 to 2022, and in the early months of 2023, police had already investigated five homicide cases compared to zero cases in 2021.[^4] On November 13, 2022, three UVA football players were shot and killed on campus. Devin Chandler, Lavel Davis Jr. and D’Sean Perry lost their lives to an act of gun violence. While incidents of gun violence have decreased in the first half of 2024,[^5] the number of bullets recovered at each scene has increased, pointing to the ever-changing nature of the issue.

[^4]: UVA President's Council - Community Safety Working Group Report - <https://prescouncil.president.virginia.edu/sites/g/files/jsddwu616/files/2024-01/Final%20Community%20Safety%20Working%20Group%20Report_24.pdf>. Published September 22, 2023.

[^5]: Cville Weekly - CPD Chief says crime rates lower overall despite recent spike in gun violence - <https://www.c-ville.com/cpd-chief-says-crime-rates-lower-overall-despite-recent-spike-in-gun-violence> - Published July 3, 2024.

The [Gun Violence Solutions Project](https://provost.virginia.edu/subsite/gun-violence-solutions-project) is a University-sponsored effort to work with community partners on assessing, developing, and implementing actionable solutions to reduce gun violence in mid-sized cities and regions, like Charlottesville and Albemarle. One strand of the Gun Violence Solutions Project is focused on researching solutions that could be implemented locally, while exploring law and policy alongside the historical, cultural, and structural factors that shape gun violence.

This document is meant to support the regional work of the Gun Violence Solutions Project to help our community understand the larger picture of gun violence and [gun violence trauma](#gun-violence-trauma) in our region by bringing together multiple sources of data. Each section includes an overview of the data source presented, a brief discussion of notable trends, and limitations that may complicate or constrain conclusions.  

The numbers presented here each represent part of a story. By bringing them together, and in conversation with our community, we hope to collectively tell a fuller story. The community -- residents of Charlottesville and Albemarle  -- is the central stakeholder. The folks who live and work here **are** the data. Given the wealth of lived experience and the range of perspectives within our community, these voices must be centered. Together we can develop a shared understanding and greater clarity about the local problems of gun violence, identify knowledge and data gaps, and promote wider community engagement in conversations about solutions.

Ultimately, we hope this work will be part of:

1.  Understanding the full impact of gun violence on our community and how different people experience this impact;

2.  Working with residents, advocates, and decision-makers to identify approaches to prevention and intervention for all types of gun violence; and

3.  Evaluating the effectiveness of current and future efforts to eliminate gun violence.

<details>

<summary>**Data Notes**</summary>

The data provided below represent information that is, for the most part, publicly available. This document is meant to serve as an overview of the existing resources so that we can begin representing the nature, scope, and dimensions of the problem locally.

Much of this data relies on information created when people interact with representatives of public services. While reduced to counts and numbers, the experiences and processes the data represent are deeply human. People's decisions select, record, and shape the data. Though no personally identifiable information is contained in this report, each data point represents a moment in the lives of members of our community.

The code and (most of the) data are available at our [GitHub repository](https://github.com/virginiaequitycenter/cville-alb-gun-trauma). We encourage anyone interested to review these analyses or extend them by contributing their own data-driven insights.

Because the data below is drawn from various sources -- news outlets, the Virginia Department of Health, and local and national law enforcement agencies -- we provide a glossary to clarify terminology.

Please refer to the [Terminology] section of the Appendix for more information and a list of definitions.

</details>

------------------------------------------------------------------------

# Firearm Injuries & Deaths 

The leading cause of death by [gun violence](#gun-violence) in the Blue Ridge Health District is suicide, which made up 76% of all firearm-related deaths from 2018 to 2022. This mirrors national patterns, where more than 60% of gun deaths are by suicide. The chart below includes the number and rate of firearm-related deaths among Blue Ridge Health District residents. Homicide -- community violence, domestic violence, mass shootings, and other intentional violence -- accounts for 22% of all deaths. 

The Blue Ridge Health District includes Charlottesville City, and the counties of Albemarle, Greene, Louisa, Fluvanna, and Nelson. 

```{r vdh_intent, echo=FALSE, message=FALSE}

vdh_intent <- read_csv("data/vdh_intent.csv") %>%
  rename(intent = intent_of_injury) %>%
  mutate(intent = case_when(intent == "Undetermined/legal/war" ~ "Undetermined or Police Intervention",
                            TRUE ~ intent))

vdh_intent %>%
  select(-patient_health_district) %>% 
  arrange(desc(firearm_deaths)) %>%
  reactable(
    defaultColDef = colDef(
      header = function(value) 
        str_to_title(gsub("_", " ", value, fixed = TRUE)),
      minWidth = 70,
      headerStyle = list(background = "#f7f7f8")
    ),
    columns = list(
      rate_of_firearm_deaths_per_100k_pop = colDef(style = color_scales(vdh_intent, colors = c("#FFFFFF", "#FA8C00")))
    ),
    bordered = TRUE,
    highlight = TRUE)

```

------------------------------------------------------------------------

## Injuries in Cville/Albemarle

The [Virginia Department of Health](https://www.vdh.virginia.gov/surveillance-and-investigation/syndromic-surveillance/firearm-injury-surveillance/) also reports data on emergency department visits for firearm injuries. The table below shows the rate of [firearm injuries](#firearm-injury) per 10,000 emergency department visits in Charlottesville and Albemarle from 2015 - 2023. 

```{r vdh_injuries}

vdh_injuries <- read_csv("data/vdh_injuries.csv")

albemarle <- vdh_injuries %>%
  filter(str_detect(patient_city_county, "Albemarle")) 

names(albemarle) <- gsub("_", " ", names(albemarle)) %>%
  str_to_title() %>%
  gsub("Ed", "ED", .)

albemarle %>%
  arrange(albemarle$Year) %>%
  select(-"Patient City County") %>%
  reactable(
    defaultColDef = colDef(
      align = "center",
      #defaultSortOrder = "desc",
      headerStyle = list(background = "#f7f7f8")
    ),
    columns = list(
      "Total ED Visits" = colDef(format = colFormat(separators = TRUE)),
      "Rate Of Firearm Injuries Per 10k ED Visits" = colDef(style = color_scales(
        albemarle, colors = c("#FFFFFF", "#FA8C00")
      ))
    ),
    bordered = TRUE,
    highlight = TRUE
  )

```


<style>
p.comment {
background-color: #d8e6ef;
padding: 10px;
margin-left: 25px;
border-radius: 5px;
}
</style>

<p class="comment">
<b> Notable Trends</b> 
<br> 
<br> 
In 2020, the rate of firearm injuries peaked at 8.3, with 29 firearm injuries among 34,875 emergency department visits. However, 2020 also had the lowest number of emergency department visits which aligns with national trends of decreased emergency department use during the COVID-19 pandemic.[^6] After falling in 2021, the firearm injury rate began trending upward.
</p>

[^6]: Melnick, Glenn et al. “COVID-19 driven decline in emergency visits: Has it continued, is it permanent, and what does it mean for emergency physicians?.” The American journal of emergency medicine vol. 61 (2022): 64-67. <doi:10.1016/j.ajem.2022.08.031>

<details>
<summary><b>Data Notes</b></summary>

City/county localities are assigned using the patient's residential zip code for Virginia residents. Patients with non-Virginia or unknown zip codes are grouped as 'Out of State.'

Firearm injury visits are identified using key terms in the chief complaint (reason for visit) and discharge diagnosis codes. Key terms include: gun with wound, GSW, gunshot, buckshot, revolver, rifle, shotgun, firearm, pistol, handgun, been shot, I was shot, I got shot, combination of hit, ricochet, graze with bullet. ICD-10 diagnosis codes: W32.0, W32.1, W33.0, W33.1, W34.0, W34.1, X72, X73, X74, X93, X94, X95, Y22, Y23, Y24, Y35.0, Y38.4 SNOMED diagnosis codes: 41430008, 56768003, 63409001, 69861004, 77301004, 86122002, 111050005, 219257002, 283545005, 218081007, 218086002, 218082000, 218087006, 218088001, 269796009, 242869008, 219199009, 219200007, 219201006, 219204003, 219205002, 219203009, 219198001, 219142001, 219143006, 219144000, 219145004, 219146003, 287184008, 287193009.

This excludes follow-up visits, visits involving other types of guns (e.g., staple gun), and visits where firearm was used as a weapon but not fired (e.g., pistol whip).

</details>

------------------------------------------------------------------------

## Deaths in the Blue Ridge

The [Virginia Department of Health](https://www.vdh.virginia.gov/firearm-injury/firearm-related-deaths/) tracks causes of death among Virginia residents using death certificates and reports on all firearm-related fatalities. Below we show the distribution of gun deaths by age in our region between 2018 and 2022.

```{r vdh_age}

vdh_age <- read_csv("data/vdh_age.csv")

vdh_age %>%
  select(-health_district) %>%
  arrange(age_group) %>%
  reactable(
    defaultColDef = colDef(
      header = function(value) str_to_title(gsub("_", " ", value, fixed = TRUE)),
      align = "center",
      defaultSortOrder = "desc",
      headerStyle = list(background = "#f7f7f8")),
    columns = list(
      rate_of_firearm_deaths_per_100k_pop = colDef(style = color_scales(vdh_age, colors = c("#FFFFFF", "#FA8C00"))
        )),
    bordered = TRUE,
    highlight = TRUE,
    defaultPageSize = 11) 

```

<p class="comment">
<b> Notable Trends</b> 
<br> 
<br> 
The table above shows that the rate of firearm death is highest for individuals aged 18 and 19 years old, with a rate of 18.3 based on 8 confirmed firearm-related deaths between 2018 and 2022 in the Blue Ridge Health District. A rate of 18.3 means that for every 100,000 people aged 18-19, 18 individuals would have lost their lives to gun violence. While there are fewer than 100,000 residents aged 18-19, the rate helps compare risk of gun-related deaths across age ranges. The rate above 14 per 100k individuals for all ages between 18 and 44 years old. The age group with the second-highest percentage of gun-related deaths are individuals above the age of 75, with 16 deaths per 100,000 people. Given that suicide is the number one cause of firearm-related death in our region, we assume that a majority of the deaths for individuals 75 and older were suicide.
</p>

<details>

<summary><b>Data Notes</b></summary>

Deaths are classified as firearm-related using the definition from the Centers for Disease Control.[^7]

[^7]: Centers for Disease Control - About Firearm Injury and Death: Fast Facts - <https://www.cdc.gov/firearm-violence/about/?CDC_AAref_Val=https://www.cdc.gov/violenceprevention/firearms/fastfact.html>. Accessed March 15, 2024.

VDH reports the number and rate of firearm-related deaths among Virginia residents by patient health district and age from 2018 through 2022. Virginia Department of Health districts are assigned based on the residence of the patient at the time of death, not where the death occurred. Data includes Virginia residents only, regardless of whether they died in Virginia.

</details>

------------------------------------------------------------------------

# Incidents of Gun Violence {.tabset}

The [Gun Violence Archive](https://www.gunviolencearchive.org/) curates a database of incidents of gun violence and gun crime, sourced from local and state police, media, data aggregates, and government. The database includes information about 201 incidents of gun violence affecting 306 participants in the Charlottesville/Albemarle region from May 2014 - April 2024. An individual may be involved in more than one incident (see [incidents](#incident) vs. [crimes](#crime)). Incidents range from shots fired reports, to unlawful gun ownership, to acts of violence such as assault and homicide. The Gun Violence Archive does not include information about suicides.

The data includes two datasets: one has information about each incident and the associated outcomes (for example, the number of people injured and/or killed), and the other lists [participant](#participant)-level information (for example, the age range and gender of both victims and suspects). Both include the location and date of each incident along with characteristics such the number of injuries, and the nature of the gun incident.

## Incident-level data

#### Victims of Gun Violence in the Charlottesville/Albemarle Region

```{r gva_incidents}

gva_incidents <- read_csv("data/gva_incidents.csv") 

gva_incidents %>%
  group_by(yr = year(incident_date)) %>%
  summarise(total_injured = sum(victims_injured + suspects_injured),
            total_killed = sum(victims_killed + suspects_killed)) %>%
  ungroup() %>%
  pivot_longer(matches("total")) %>% 
  ggplot(aes(x = yr, y = value, fill = name)) +
  geom_col() +
  labs(x = "Year",
       y = "Number of Victims",
       caption = "Total incidents: 201") +
  scale_fill_manual(labels = c("Injured", "Killed"),
                    values = c("#F8BE3D", "#007BAB"),
                    guide = guide_legend(title = "Victim Status")) +
  scale_x_continuous(breaks = scales::pretty_breaks(n = 11),
                     guide = guide_axis(angle = 35)) +
  theme_minimal()
```

<p class="comment">
<b> Notable Trends</b> 
<br> 
<br> 
The bar graph above shows that the number of victims both injured and killed in incidents of gun violence have increased tenfold from 3 in 2014 to 33 in 2022. 2022 had the highest number of victims injured, with 24 individuals reported being shot, while 2023 was the deadliest year, with 12 individuals losing their lives to gun violence. While 2024 only includes data through April, so far the numbers are trending lower than in previous years.
</p>

------------------------------------------------------------------------

## Individual-level data

#### Ages of Participants

```{r gva_participants}

gva_participants <- read_csv("data/gva_participants.csv") 

gva_participants %>%
  drop_na(age) %>%
  mutate(yr = year(incident_date),
         role = str_to_title(role)) %>%
  ggplot(aes(age, fill = role)) +
  geom_histogram(position = 'identity') +
  facet_grid(~role) +
  scale_fill_manual(labels = c("Suspect", "Victim"),
                    values = c("#F8BE3D", "#007BAB"),
                    guide = guide_legend(title = "Participant Role")) +
  labs(x = "Age",
       y = "Number of People", 
       caption = "Total particpants: 306") +
  scale_x_continuous(breaks = scales::pretty_breaks(n = 10),
                     guide = guide_axis(angle = 35)) +
  theme_bw()
```

<p class="comment">
<b> Notable Trends</b> 
<br> 
<br> 
The bar graphs above show the age ranges of suspects and victims of gun violence in Charlottesville and Albemarle from May 2014 to April 2024. The age of suspects peaks between the ages of 18 and 23; the age of victims peaks around the ages of 19-22 and again around ages 33-33.
</p>

<details>
<summary><b>Data Notes</b></summary>

The data does not include any identifiable information about the participants, like name, race, ethnicity, socioeconomic status, or mental health status.

**Voluntary Participation**

Because the data is sourced largely from public news websites, if an act of violence didn't receive local news attention it was not included in the Gun Violence Archive database. That means this data underrepresents some information, especially shots fired numbers. Notably, it looks like **shots fired incidents haven't been added to the database since 2021.**

**Suicides**

Suicides are not part of the Gun Violence Archive's normal data collection procedures. Individual suicides are collected through the CDC's Annual Report[^8] and, because of privacy and CDC policy, they are only available as an aggregate number, without detail. Suicides associated with Officer-Involved incidents and Murder-Suicides are included in Gun Violence Archive tallies because they are considered incidents of community violence.

[^8]: Centers For Disease Control - Provisional Suicide Deaths in the United States, 2022 - <https://www.cdc.gov/media/releases/2023/s0810-US-Suicide-Deaths-2022.html>. Published August 10, 2023.

</details>

------------------------------------------------------------------------

# Crimes Involving Firearms {#crimes}

Crime data for Virginia are pulled from Virginia's [Uniform Crime Reporting - Incident Based Reporting](https://vsp.virginia.gov/sections-units-bureaus/bass/criminal-justice-information-services/uniform-crime-reporting/#UCR-IBR) system.  The state system follows the FBI's National Incident-Based Reporting standards, and the state mandates participation by all publicly-funded law enforcement agencies. The data contains summaries of all [crimes](#crime) where a firearm was used in Virginia, Albemarle County, and Charlottesville City between 2016 - 2023, separated by whether the individuals believed to have committed the crime was an adult or [youth](#youth). For this source, youth are considered to be under the age of 18.

#### Adult Compared to Youth Firearm Rates

```{r nibrs_compare}

nibrs_crime <- read_csv("data/nibrs_crime.csv")
age_labs <- c("Under 18", "Above 18")
names(age_labs) <- c("n_juvenile", "n_adult")

nibrs_crime %>%
  mutate_at(vars(matches('n_')), ~ . / est_pop_district * 1e4 ) %>%
  pivot_longer(cols = c(n_juvenile, n_adult)) -> tmp_plt_dat

ggplot(tmp_plt_dat, aes(year, value, color = district, group = district)) +
  geom_line(size = 1.5) +
  facet_wrap(~name,
             labeller = labeller(name = age_labs),
             scales = "free_y") +
  labs(x = "Year",
       y = "Crime Rate per 100k Population") +
  scale_color_manual(values = c("#232D4B", "#F8BE3D", "#007BAB"), 
                    name = "Region",
                    breaks = c("virginia", "charlottesville", "albemarle"),
                    labels= c("Virginia", "Charlottesville", "Albemarle")) +
    geom_label(data = filter(tmp_plt_dat, year == 2023),
             aes(label = paste0(round(value, 1))),
             show.legend = F, hjust = .75, alpha = 0.75, fontface = "bold") +
  scale_x_continuous(breaks = scales::pretty_breaks(n = 8),
                     guide = guide_axis(angle = 35)) +
  theme_bw()
  
```

<p class="comment">

<b> Notable Trends</b> 
<br> 
<br> 
The figure above shows the rate of gun violence crimes per 100k population over time.  In Virginia, the crime rate has been generally increasing from 19.2 in 2016 to 24.4 in 2023, with a peak of 25.6 in 2022. That means on average in Virginia in 2023, for every 100,000 people, there were 24 crimes involving firearms. The rate in Charlottesville has been much more variable, peaking above state levels from 2020 to 2022 and reaching a maximum of 28 in 2021. Albemarle has had a relatively steady rate during this period, ranging from a minimum of 7.4 in 2021 to a maximum of 11 in 2022.
In general, following the COVID-19 pandemic, crimes involving firearms increased for all age groups however most notably among youth under 18. The Charlottesville youth firearm crime rate surpassed state levels in 2022, however that number has been slowly decreasing. 
</p>
<details>
<summary><b>Data Notes</b></summary>

**Participation**

The Virginia Incident-Based Reporting system is managed by the Virginia State Police Data Analysis and Reporting Team. Although participation is mandatory, there is the potential for some data loss as each law enforcement agency finalizes their data submission process and data is updated. 

</details>

------------------------------------------------------------------------

# Police Reporting

This dataset includes an anonymized collection of 890 gun-related incidents recorded by the Charlottesville Police Department (CPD), the Albemarle County Police Department (ACPD), and the UVA Police Department (UPD) between January 1, 2019 - May 18, 2024. The data was retrieved from their SQL database by the ACPD Sr. Crime Analyst, de-identified, and then shared with the Equity Center in June 2024. To preserve anonymity while also providing geographic information, approximate block numbers are used instead of specific addresses. This data is used further below to compare incidents of gun violence to underlying population characteristics.

Each observation represents the initial information that is provided by individuals calling for police assistance. Some incidents are listed as "Unverified" which means that although a call was made to the police, there was not sufficient evidence to confirm an incident of gun violence actually occurred (see *Data Notes* section below.)

As a result, this is not a definitive collection of public safety data for Charlottesville and Albemarle. These reports likely miss some acts of gun violence as not all incidents are reported to the police. Both personal and structural histories impact individuals' likelihood of calling the police. In addition, while verified reports indicate evidence of shots fired, they do not mean that an arrest was made.

#### Incidents of Gun Violence in Charlottesville & Albemarle

```{r police_incidents}

regional_gv <- read_csv("data/regional_gv.csv") %>%
  mutate(locality = word(locality, 1),
         yr = year(reported_date))

shots_fired <- regional_gv %>%
  filter(description == "Shots Fired",
         verified == "VERIFIED") %>%
  group_by(yr, locality) %>%
  count() %>%
  rename(shots_fired = n)

other <- regional_gv %>%
  filter(description != "Shots Fired",
         verified != "UNVERIFIED") %>%
  group_by(yr, locality) %>%
  count() %>%
  rename(other = n)

years <- shots_fired %>%
  left_join(other, relationship = "many-to-many") %>%
  pivot_longer(cols = c("shots_fired", "other"),
               names_to = c("type"),
               values_to = "count")
  
ggplot(years, aes(yr, count, fill = type)) +
  geom_col() +
  facet_wrap(~locality) +
  scale_fill_manual("Incident Type",
                    labels = c("Aggravated Assault & Homicide", "Shots Fired"),
                    values = c("#F8BE3D", "#007BAB")) +
  labs(x = "Year",
       y = "Number of Verified Incidents",
       caption = "Note: the 2024 data only includes incidents up to May 18.") +
  scale_x_continuous(breaks = scales::pretty_breaks(n = 7),
                     guide = guide_axis(angle = 35)) +
  theme_bw()

```

<p class="comment">
<b>Notable Trends</b> 
<br> 
<br> 
The stacked bar plots above show the types of firearm-related incidents in Charlottesville and Albemarle between 2019 and early 2024. While the number of shots-fired reports is higher in Albemarle County, the number of violent incidents such as aggravated assault and homicide is higher in Charlottesville. The distinct geographies contained in each locality likely accounts for some of this difference. For example, shots fired in a rural area of Albemarle County may be a result of wildlife hunting season, while shots fired in an urban area of Charlottesville is typically not associated with legal wildlife hunting.
</p>

----------------------------------------------------------------------

#### Age & Race of Victims of Gun Violence

The table below shows the race and age breakdown of the 537 total victims of gun violence (injury and death) reported to the
police between January 2019 to May 2014.

```{r victim_demos}
race_county <- read_csv("data/race_county.csv")

demos <- read_excel("data/raw/Regional GV Data - 2019-2024 YTD.xlsx", sheet = "Demographic Data for Cases") %>%
  clean_names() %>%
  distinct() %>%
  mutate(reported_date = ymd_hms(reported_date),
         yr = year(reported_date),
         subject_type = str_remove_all(subject_type, "/Offender"),
         sex = case_when(
           sex == "Unknown" ~ NA,
           TRUE ~ sex),
         race = case_when(
           race == "African American" ~ "Black",
           race == "Unknown" |
             race == "New World Text" |
             race == "Microsoft Word" ~ NA,
           TRUE ~ race)) %>%
  replace_na(list(sex = "Not Provided"))

age_groups <- demos %>%
  mutate(age_grp = case_when(
    age <= 9 ~ "9andunder",
    age >= 10 & age <= 14 ~ "10-14",
    age >= 15 & age <= 17 ~ "15-17",
    age >= 18 & age <= 19 ~ "18-19",
    age >= 20 & age <= 24 ~ "20-24",
    age >= 25 & age <= 34 ~ "25-34",
    age >= 35 & age <= 44 ~ "35-44",
    age >= 45 & age <= 54 ~ "45-54",
    age >= 55 & age <= 64 ~ "55-64",
    age >= 65 & age <= 74 ~ "65-74",
    age >= 75 ~ "75andup"
  ))

total_victims <- demos %>% filter(subject_type == "Victim") %>% nrow()
total_pop_cvalb <- sum(race_county$pop_est)
n_bl_cvalb <- sum(race_county$black)
pct_bl_cvalb <- percent(n_bl_cvalb/total_pop_cvalb)

black_victims <- demos %>% filter(subject_type == "Victim", race == "Black") %>% nrow()
pct_bl_vic <- percent(black_victims/total_victims)

victims <- age_groups %>%
  filter(subject_type == "Victim") %>%
  group_by(race, age_grp) %>%
  summarise(n = n(),
            pct = round((n() / 537) * 100, digits = 1)) #537 total victims 

orange_pal <- c("#fff2e6", "#ffb54d")

victims %>%
  select(-n) %>%
  pivot_wider(names_from = age_grp, values_from = pct) %>%
  select(`9andunder`, `10-14`, `15-17`, `18-19`, `20-24`, `25-34`, `35-44`,
         `45-54`, `55-64`, `65-74`, `75andup`) %>%
  ungroup() %>% 
  reactable(
    defaultColDef = colDef(
      align = "center",
      minWidth = 50,
      format = colFormat(suffix = "%"),
      na = "–",
      headerStyle = list(background = "#f7f7f8"),
      style = color_scales(data = ., colors = orange_pal, span=2:12)
    ),
    columns = list(
      race = colDef(name = "", minWidth = 100, format = colFormat(suffix = ""), na = "NA"),
      `9andunder` = colDef(name = "0-10"),
      `75andup` = colDef(name = "75+")
    ),
    bordered = TRUE,
    highlight = TRUE
  )
```

Some elements of the police data are also available through [Citizen Connect](https://albemarlecountyva-uvaecc-transparency.connect.socrata.com/), an interface that allows users to view a variety of policing interactions across the region.

<p class="comment">
<b>Notable Trends</b> <br> <br> According to the table above, Black individuals make up `r pct_bl_vic` of the victims of gun violence, but only `r pct_bl_cvalb` of the total population in Charlottesville and Albemarle. In particular, Black individuals aged 25-34 make up 16.9% of the victims.   
</p>

<details>
<summary><b>Data Notes</b></summary>

**Verification**

Each incident is listed as being verified, unverified, or as a case. An incident is considered verified if it can be proved by an officer to have happened or there is reasonable evidence of its occurrence. For example, a shots fired incident would be verified if there were shell casings found at the scene.

An incident would be considered unverified if there was no proof after the fact that it happened. For example, someone may call the police to report hearing loud noises, however when the officers arrived they find either no evidence of a shooting, or evidence to prove otherwise, such as used fireworks supplies.

An incident would be marked as a case if it required significant follow up, or was otherwise recorded as a crime to be investigated. Typically more violent incidents such as aggravated assault or homicide become cases.

**Data Sharing**

Because this data may include information that is actively being used in court, it is the only data source in this report that is not fully open-sourced, or publicly available. This decision was made so that the individuals being represented can maintain their anonymity throughout the duration of their trial and subsequent followup. A similar, but more limited, report can be generated based on data availble in the Charlottesville Open Data Portal, and instructions on how to do so are outlined in [scripts.R](https://github.com/virginiaequitycenter/cville-alb-gun-trauma/blob/main/scripts/data_prep.R).

**Collection Mechanisms, Specifically for Race and Ethnicity **

We do not know the process for collecting and recording demographic information. Since we do not know how the demographic information is recorded (For example, is it self-reported? Is it assumed and reported by an officer? A bystander?) we recognize that this information is not all-inclusive or complete, and we should be cautious when making any inferences based on this data. Ethnicity was not recorded. 

</details>



---------------------------------------------------------------------

## Charlottesville Open Data Portal 

Another way of exploring local gun violence data is through the publicly available [Charlottesville Open Data Portal](https://opendata.charlottesville.org/datasets/charlottesville::crime-data/about). This database includes 526 gun-related crime reports, and 390 gun-related arrests in City of Charlottesville from March 2019 through April 2024.

This data differs from the above data, shared directly by the police, in the following ways:

1.  It includes only reports in the Charlottesville district;
2.  It includes all calls to police, not just gun-related calls;
3.  It does not include additional incident information such as verification or if it was assigned to a detective as a case.

The dataset is made up of two dataframes. The first includes information about initial crime report. This data represent the first step when gun-related crimes are recorded; the police-provided data represents the observations after subsequent steps -- officer investigation -- have occurred. The second data set includes arrest data. Both datasets have been anonymized to remove names and other identifiable information such as house numbers.

```{r odp_crimemap, fig.show="hold", fig.width=5, out.width= "50%"}

odp_crimes <- read_csv("data/odp_crimes.csv") 
odp_arrests <- read_csv("data/odp_arrests.csv")
load(file = "data/cville_map.RData")

# Geographic spread
ggmap(cville_map) +
  stat_density2d(data = odp_crimes, aes(fill = ..level.., alpha = 0.1), # removes 55 values
                 geom = "polygon") +
  theme(legend.position="none") +
  scale_fill_viridis_c(direction = -1) +
  theme_void() +
  theme(legend.position = "none") +
  labs(title = "Gun-Related Crime Reports",
       caption = "Total reports: 526")

ggmap(cville_map) +
  stat_density2d(data = odp_arrests, aes(fill = ..level.., alpha = 0.1), # removes 96 values
                 geom = "polygon") +
  theme(legend.position="none") +
  scale_fill_viridis_c(direction = -1) +
  theme_void() +
  theme(legend.position = "none") +
  labs(title = "Gun-Related Arrests",
       caption = "Total arrests: 390")

```

<p class="comment">
<b> Notable Trends</b> 
<br> 
<br> 
The two maps above show the density of gun-related crime reports and arrests in the Charlottesville region. The map on the left demonstrates that most gun-related crime reports between March 2019 and April 2024 were concentrated in the West Main Street area, specifically near the 10th St intersection. There are also smaller concentrations of reports in Fifeville and on the east end of the Downtown Mall, near the Pavilion. 
<br> 
<br> 
The map on the right demonstrates that most gun-related arrests are concentrated on the east end of the Downtown Mall, the Roosevelt-Brown Boulevard region, and the south side of Charlottesville. The location with the most arrests is the Charlottesville Police Station, located on the East End of the Downtown Mall. When we asked the police department for clarification, they explained that most arrest warrants are written and filed at the police station, and that when individuals receive warrants, they turn themselves in at the police station. Most incidents of gun violence do not have an arrest at the scene of the crime, but have an arrest days later from a warrant issued, which is issued at the police station.
</p>

------------------------------------------------------------------------

### Arrest Reports

From the Charlottesville Open Data Portal:

> "Arrest data contains information on people taken into custody by City of Charlottesville police officers. More serious crimes such as felony offenses are more likely to result in an arrest. However, arrests can occur as a result of other offenses, such as parole violations or a failure to appear for trial."

Below we use the arrest data to understand whether a lot of people are committing a few crimes or a few people are committing a lot of crimes. The figure shows the distribution individuals arrested among the 390 [arrests](#arrests) for firearm-related charges made between May 2019 and April 2024. The data has been anonymized so that identifiable information has been removed.

#### Number of Firearm-Related Arrests per Person

```{r odp_arrests}

# Arrest numbers per person 
odp_arrests %>%
  group_by(id) %>%
  summarise(charges = n()) %>%
  ungroup() %>%
  ggplot(aes(y = charges)) +
  geom_bar(fill = "#007BAB") +
  geom_text(aes(label =..count..), stat = "count", hjust = -0.25, size = 3) +
  labs(x = "Number of People",
       y = "Number of Charges per Person",
       caption = "Total number of people arrested: 233") +
  scale_y_continuous(breaks = c(1:15)) +
  geom_label(aes(90, 10, label = "86% of people arrested \nfor gun-related crimes \nhave 2 or fewer charges")) +
  theme_minimal()
```

<p class="comment">
<b> Notable Trends</b> 
<br> 
<br> 
Of the 233 individuals arrested, 147 have only one gun-related charge during this five-year period, 53 have two charges, and one person has 15 charges. In other words, 86% of people arrested for gun-related crimes have 2 or fewer charges. This suggests a small group of people are committing a significant proportion of gun-related crimes in Charlottesville.
</p>

<details>
<summary><b>Data Notes</b></summary>

**Jurisdiction Limits**

These datasets only contain information reported by City of Charlottesville Police. It does not contain information about incidents that solely involve other police departments operating within the city (for example, University of Virginia police or Albemarle County police).

**Filtering**

The data has been filtered to only include crime reports that involve firearms, which required some subjective wrangling. Specifically, only offences labeled *Shots Fired/Illegal Hunting*, *Robbery - Armed*, and *Weapons Violations* are used. As a result, this data may significantly underrepresent the number of firearm incidents. For example, crime reports that have *Assault* as the primary offense are not included because we cannot be certain the assault included firearm.

</details>

-----------------------------------------------------------------------

# Officer-Involved Shootings

Virginia Law defines an officer-involved shooting as the discharge of a firearm by a law-enforcement officer that results in the death or serious bodily injury of another. [^9]

The table below shows an anonymized version of the 10 victims of officer-involved shootings in Charlottesville and Albemarle from 2016 to today.  

[^9]: Virginia Law Code - § 52-28.2. Reporting of officer-involved shootings; inclusion in annual Crime in Virginia report - <https://law.lis.virginia.gov/vacode/title52/chapter6/section52-28.2/>. Accessed November 15, 2024.

```{r ois}

police_shootings <- read_csv("data/gva_officer.csv") %>%
  rename("Date of Shooting" = incident_date,
         "Outcome" = outcome,
         "Victim Age" = victim_age,
         "Victim Gender" = gender, 
         "Victim Race" = race,
         "Did the victim have a weapon?" = weapon)

police_shootings %>%
  reactable(
    defaultColDef = colDef(
    minWidth = 70,
    align = "left",
    headerStyle = list(background = "#f7f7f8")),
    bordered = TRUE,
    highlight = TRUE
  )

```

<p class="comment">
<b>Notable Trends</b> <br> <br> 
The table above shows the demographic information about the 10 victims of officer-involved shootings in Charlottesville and Albemarle. Of the victims, 40% died as a result of the incident, 50% were identified as Black, and the median age was 37. All of the victims were males that were described as being in possession of a firearm. 
</p>


<details>
<summary><b>Data Notes</b></summary>
<br>
<b>Multiple Data Sources</b>
<br>
<br>
This data is downloaded from the Gun Violence Archive and then manually augmented by reviewing associated news articles for additional information such as age, race, and if the victim had a gun. 
<br>
<br>
<b>Suicide at Standoff</b>
<br>
<br>
Some of the deaths indicated above were described as self-inflicted, such that they occurred during a police chase or standoff where shots had been fired. These are still considered as officer-involved deaths, because they are still incidents of gun violence. The overall goal is to provide information on most types of gun violence, and gun crime, no matter the root cause. 

</details>

----------------------------------------------------------------------

# Residential Characteristics

The data shown above is focused on describing key outcomes. To better understand the underlying contexts of gun trauma in our community we compare the location of these outcomes to the composition of key residential characteristics that may influence antisocial or violent behavior.

The maps below show clusters of gun violence incidents reported by our local police jurisdictions between 2018 and 2024 overlaid on top of community population characteristics. These characteristics are from the 5-year [2022 American Community Survey](https://www.census.gov/programs-surveys/acs/news/data-releases.2022.html#list-tab-1133175109). The maps show the distribution of population characteristics in our region, such as youth populations, childhood poverty rates, median earnings, and unemployment rates.

Next to each map is a tab with a scatterplot comparing the presence of a population characteristic to rates of gun violence within each census tract. The population characteristic being examined is represented along the bottom and the proportion of total gun violence incidents from 2018 - 2024 is shown along the side. Each circle represents a different census tract, and the size of the circle is proportional to the population of the tract.

## Youth Populations

### Are incidents of gun violence happening where there are more youth populations? {.tabset}

The map below shows the percentages of young people, defined both as the population that are under 18 and 25 years old, for each census tracts. With over 16,700 undergraduate students coming to the region each semester to study at UVA,[^10], a portion of the 18 to 24 population of Charlottesville and Albemarle is transient. The JPA - Fontaine, Carr's Hill - McCormick, and Darden tracts house a large number of students, with nearly all first year students living in dorms in the McCormick region. As an alternative, and to focus in on our local youth population, we also show the population under 18.

[^10]: UVA Statistics, Fall 2022 Enrollment - <https://admission.virginia.edu/admission/statistics>) - Accessed May 19, 2024.


#### Map

<details>
<summary><b>Map Instructions</b></summary>
The map below shows the percentages of individuals under 18 and 25 years old, respectively. Each grey circle represents a cluster of gun violence incidents. Click on a circle to zoom into the cluster and explore the location and date of each incident. Click on a census tract to see population measurements. Zoom in to see specific areas and incidents more closely. Toggle between the population percentages using the selector in the top right. 
</details>

```{r youth_pops}

# Get police-provided gun violence data
gv <- read_csv("data/regional_gv.csv")

# Get census shape data
dat <- readRDS("data/census.RDS")

# Convert lat/lon coords to sf points
gv_pts <- gv %>%
  st_as_sf(coords = c("lon", "lat"), crs = 4326)

# Count incidents per tract
gv_sf_summary <- dat %>% 
  mutate(counts = lengths(st_intersects(., gv_pts))) %>%
  group_by(tract) %>%
  mutate(incidents_pop = (counts / pop_est),
         percent_incidents = (counts / 887) * 100,
         locality = str_to_title(locality)) 

pal_age <- colorNumeric(palette = "viridis", 
                    domain = 0:100, 
                    reverse = TRUE)

gv_sf_summary %>%
  filter(tract_name != "JPA - Fontaine") %>%
  leaflet() %>%
  addProviderTiles(providers$CartoDB.Positron) %>% 
  addPolygons(group = "Under 18",
              stroke = TRUE, 
              weight = 0.5,
              opacity = 1,
              color = "black", 
              fillColor = ~ pal_age(percent_under18),
              fillOpacity = 0.5,
              popup = paste0("Percent of Population Under 18: ", 
                             round(gv_sf_summary$percent_under18, digits = 2), "%", "<br>",
                             "Number of People Under 18: ", gv_sf_summary$total_under18, "<br>",
                             "Total Population: ", gv_sf_summary$pop_est, "<br>",
                             "Tract: ", gv_sf_summary$tract_name, ", ", gv_sf_summary$locality),
              highlightOptions = highlightOptions(
                fillOpacity = 1,
                bringToFront = FALSE
              )) %>%
  addPolygons(group = "Under 25",
              stroke = TRUE, 
              weight = 0.5,
              opacity = 1,
              color = "black", 
              fillColor = ~ pal_age(percent_under25),
              fillOpacity = 0.5,
              popup = paste0("Percent of Population Under 25: ", 
                             round(gv_sf_summary$percent_under25, digits = 2),  "%", "<br>",
                             "Number of People Under 25: ", gv_sf_summary$total_under25, "<br>",
                             "Total Population: ", gv_sf_summary$pop_est, "<br>", 
                             "Tract: ", gv_sf_summary$tract_name, ", ", gv_sf_summary$locality),
              highlightOptions = highlightOptions(
                fillOpacity = 1,
                bringToFront = FALSE
              )) %>%
  addLegend("bottomright",
            pal = pal_age,
            values = ~ percent_under25, 
            title = paste("Percentage of the Population"), 
            labFormat = labelFormat(suffix = "%"),
            opacity = 1) %>%
  addLayersControl(baseGroups = c("Under 18","Under 25"),
                   options = layersControlOptions(collapsed = FALSE)) %>%
  addMarkers(data = gv, 
             lng = gv$lon,
             lat = gv$lat,
             popup = paste0("Description: ", gv$description, "<br>",
                            "Date: ", gv$reported_date, "<br>",
                            "Status: ", gv$verified),
             clusterOptions = markerClusterOptions(
               showCoverageOnHover = FALSE,
               iconCreateFunction=JS("function (cluster) {    
    var childCount = cluster.getChildCount();  
    if (childCount < 100) {  
      c = 'rgba(211,211,211);'
    } else if (childCount < 1000) {  
      c = 'rgba(211,211,211);'  
    } else { 
      c = 'rgb(211,211,211);'  
    }    
    return new L.DivIcon({ html: '<div style=\"background-color:'+c+'\"><span>' + childCount + '</span></div>', className: 'marker-cluster', iconSize: new L.Point(40, 40) });

  }")
             ))
  
```

------------------------------------------------------------------------

#### Scatterplot

##### Under 18 Population Percentages Compared to Incidents of Gun Violence

```{r age_scatter}

scatter_text <- gv_sf_summary %>%
  filter(percent_incidents > 4)

gv_sf_summary %>%
  filter(tract_name != "JPA - Fontaine") %>%
  ggplot(aes(x = percent_under18, y = percent_incidents, size = pop_est, color = locality))  +
  geom_point(alpha = 0.8) +
  scale_size(range = c(.1, 12)) +
  geom_label(data = scatter_text, 
            aes(x = percent_under18, y = percent_incidents, label = tract_name), 
            size = 3, hjust = 1,
            inherit.aes = FALSE) +
  scale_color_manual(values = c("#007BAB", "#F8BE3D"), 
                    name = "Region") +
  scale_y_continuous(labels = function(x) paste0(x, "%")) +
  scale_x_continuous(labels = function(x) paste0(round(x), "%")) +
  labs(x = "Percentage of the Population Under 18 Years Old",
       y = "Percentage of Total Gun Violence Incidents",
       #title = "Under 18 Population Percentages Compared to Incidents of Gun Violence",
       subtitle = "Each circle represents a different census tract",
       caption = "Years: 2018 - 2024",
       size = "Estimated Population") +
  guides(size = "none") +
  theme_minimal()

```

<p class="comment">
<b>Notable Trends</b> 
<br> 
<br> 
According to the scatterplot above, there is not a strong relationship between the percentage of youth living in a region and the occurrence of gun violence. The regions with the highest proportion of gun violence are Fifeville - Cherry Avenue, 10th & Page - Venable, Ridge Street, Downtown, and Belmont. These tracts are all near the median value of youth proportions, which is approximately 20%.
</p>

<details>
<summary><b>Data Notes</b></summary>

**ACS 5 Year Summaries**

The American Community Survey (ACS) 5-Year Estimates are summaries for geographic areas that are based on data collected over a five-year period. The ACS data represented here is from 2018-2022. The 5-year estimates are considered more accurate than the 1-year estimates because they use a larger sample size, which results in smaller margins of error.

**JPA/Fontaine**

The ACS indicates that there are only 14 individuals under the age of 18, all of which are female, living in the JPA - Fontaine census tract. Because of the margins of error for the estimate in this tract, we have removed JPA - Fontaine from the scatterplot.

</details>

------------------------------------------------------------------------

## Poverty

### Are incidents of gun violence happening where there is more poverty? {.tabset}

The Census Bureau uses a set of income thresholds that vary by family size and composition to determine who is below poverty. If a family's total income is less than the threshold for that family's composition, then every individual in that family is considered to be living below poverty. We use both the percent of the total population below poverty and the percent of children below poverty. For tracts with heavy university student populations, the poverty rate is exaggerated. The child poverty rate excludes the transient low-income status of college students.

#### Map

<details>
<summary><b>Map Instructions</b></summary>
The map below shows the rates of children and adults living in poverty in our region. Each grey circle represents a cluster of gun violence incidents. Click on a circle to zoom into the cluster and explore the location, date, and description of each incident. Click on a census tract to see population measurements. Zoom in to see specific areas and incidents more closely. Toggle between the Overall Poverty Rates and Child Poverty Rates using the selector in the top right. 
</details>

```{r pov_map}

pal_pov <- colorNumeric(palette = "viridis",
                        domain = NULL,
                        reverse = TRUE)

gv_sf_summary %>%
  leaflet() %>%
  addProviderTiles(providers$CartoDB.Positron) %>% 
  addPolygons(group = "Overall Poverty",
              stroke = TRUE, 
              weight = 0.5,
              opacity = 1,
              color = "black", 
              fillColor = ~ pal_pov(poverty_est),
              fillOpacity = 0.5,
              popup = paste0("Poverty Rate: ", gv_sf_summary$poverty_est, "%", "<br>",
                             "Childhood Poverty Rate: ", gv_sf_summary$cpov_est, "%", "<br>",
                             "Population: ", gv_sf_summary$pop_est, "<br>",
                             "Tract: ", gv_sf_summary$tract_name, ", ", gv_sf_summary$locality),
              highlightOptions = highlightOptions(
                fillOpacity = 1,
                bringToFront = FALSE
              )) %>%
  addPolygons(group = "Child Poverty",
              stroke = TRUE, 
              weight = 0.5,
              opacity = 1,
              color = "black", 
              fillColor = ~ pal_pov(cpov_est),
              fillOpacity = 0.5,
              popup = paste0("Poverty Rate: ", gv_sf_summary$poverty_est, "%", "<br>",
                             "Child Poverty Rate: ", gv_sf_summary$cpov_est, "%", "<br>",
                             "Population: ", gv_sf_summary$pop_est, "<br>",
                             "Tract: ", gv_sf_summary$tract_name, ", ", gv_sf_summary$locality),
              highlightOptions = highlightOptions(
                fillOpacity = 1,
                bringToFront = FALSE
              )) %>%
  addLegend("bottomright",
            pal = pal_pov,
            values = ~ poverty_est, 
            title = "Estimated Poverty Rates",
            labFormat = labelFormat(suffix = "%"), 
            opacity = 1) %>%
  addLayersControl(baseGroups = c("Overall Poverty", "Child Poverty"),
                   options = layersControlOptions(collapsed = FALSE)) %>%
  addMarkers(data = gv,
             lng = gv$lon,
             lat = gv$lat,
             popup = paste0("Description: ", gv$description, "<br>",
                            "Date: ", gv$reported_date, "<br>",
                            "Status: ", gv$verified),
             clusterOptions = markerClusterOptions(
               showCoverageOnHover = FALSE,
               iconCreateFunction=JS("function (cluster) {    
    var childCount = cluster.getChildCount();  
    if (childCount < 100) {  
      c = 'rgba(211,211,211);'
    } else if (childCount < 1000) {  
      c = 'rgba(211,211,211);'  
    } else { 
      c = 'rgb(211,211,211);'  
    }    
    return new L.DivIcon({ html: '<div style=\"background-color:'+c+'\"><span>' + childCount + '</span></div>', className: 'marker-cluster', iconSize: new L.Point(40, 40) });

  }")
             ))
```

------------------------------------------------------------------------

#### Scatterplot

##### Childhood Poverty Rates Compared to Incidents of Gun Violence

```{r pov_scatter}

scatter_text2 <- gv_sf_summary %>%
  filter(cpov_est > 31)

gv_sf_summary %>%
  filter(tract_name != "JPA - Fontaine") %>%
  ggplot(aes(x = cpov_est, y = percent_incidents, size = pop_est, color = locality)) +
  geom_point(alpha = 0.8) +
  scale_size(range = c(.1, 12)) +
  geom_label_repel(data = scatter_text2, 
                   aes(x = cpov_est, y = percent_incidents, label = tract_name), size = 3, 
                   min.segment.length = unit(0, 'lines'),
                   inherit.aes = FALSE) +
  scale_color_manual(values = c("#007BAB", "#F8BE3D"), 
                   name = "Region") +
  scale_y_continuous(labels = function(x) paste0(x, "%")) +
  scale_x_continuous(labels = function(x) paste0(round(x), "%")) +
  labs(x = "Percentage of Children Under 18 Living in Poverty",
       y = "Percentage of Total Gun Violence Incidents",
       #title = "Childhood Poverty Rates Compared to Incidents of Gun Violence",
       subtitle = "Each circle represents a different census tract",
       caption = "Years: 2018 - 2024",
       size = "Estimated Population") +
  guides(size = "none") +
  theme_minimal()

```

<p class="comment">
<b>Notable Trends</b> 
<br> 
<br> 
According to the scatterplot above, there is a positive relationship between the percentage of youth living in poverty and the occurrence of gun violence. The census tracts with the highest percentages of youth living in poverty are 10th & Page - Venable, Fifeville - Cherry Avenue, Oak Hill - Southwood, and Ridge Street. Of those four tracts, Oak Hill - Southwood is the only region where the proportion of gun violence is below 7.5%.
</p>

<details>
<summary><b>Data Notes</b></summary>

**ACS 5 Year Summaries**

The American Community Survey (ACS) 5-Year Estimates are summaries for geographic areas that are based on data collected over a five-year period. The ACS data represented here are from 2018-2022. The 5-year estimates are considered more accurate than the 1-year estimates because they use a larger sample size, which results in smaller margins of error.

Because the data represents summaries from 2018-2022, the plots below do not show changes in poverty rates by individual year but are rather an average of the past 5 years. This information can still be used to estimate approximate poverty rates for 2023 and 2024, which haven't been released in the ACS 5-Year Estimates yet.

**Student Housing**

A significant portion of the JPA-Fontaine Census tract is made up of UVA students. Many of these students are not earning an income and therefore skew overall poverty levels for that area.

</details>

------------------------------------------------------------------------

## Median Earnings by Education

### Are incidents of gun violence related to low personal earnings? {.tabset}

Earnings consist of wages and salary from a job, and are the primary source of direct income for most individuals. Other sources of income include Social Security payments, pensions, child support, public assistance, annuities, money derived from rental properties, interest and dividends. Median personal earnings provide a clearer picture of the economic opportunities afforded in an area, as it emphasizes employment income over assets. Lower median earnings are primarily a result of jobs not paying sufficient wages. Median earnings of all employees, of employees with a Bachelor's degree or more, and of employees with a high school degree only are provided.

#### Map

<details>
<summary><b>Map Instructions</b></summary>
The map below shows the median earnings of each individual by education level. Each grey circle represents a cluster of gun violence incidents. Click on a circle to zoom into the cluster and explore the location, date, and description of each incident. Click on a census tract to see population measurements. Zoom in to see specific areas and incidents more closely. Toggle between the different education levels selector in the top right. 
</details>

```{r earnings_map}

pal_earn <- colorNumeric(palette = "viridis",
                        domain = c(15000:130000),
                        reverse = FALSE)

gv_sf_summary %>%
  leaflet() %>%
  addProviderTiles(providers$CartoDB.Positron) %>% 
  addPolygons(group = "All Education Levels",
              stroke = TRUE, 
              weight = 0.5,
              opacity = 1,
              color = "black", 
              fillColor = ~ pal_earn(med_earn_25),
              fillOpacity = 0.5,
              popup = paste0("Median Earnings: ", scales::dollar(gv_sf_summary$med_earn_25), "<br>",
                             "Poverty Rate: ", gv_sf_summary$poverty_est, "%", "<br>",
                             "Population: ", gv_sf_summary$pop_est, "<br>",
                             "Tract: ", gv_sf_summary$tract_name, ", ", gv_sf_summary$locality),
              highlightOptions = highlightOptions(
                fillOpacity = 1,
                bringToFront = FALSE
              )) %>%
  addPolygons(group = "High School Degree or Equivalent",
              stroke = TRUE, 
              weight = 0.5,
              opacity = 1,
              color = "black", 
              fillColor = ~ pal_earn(med_earn_hs),
              fillOpacity = 0.5,
              popup = paste0("Median Earnings: ", scales::dollar(gv_sf_summary$med_earn_hs), "<br>",
                             "Poverty Rate: ", gv_sf_summary$poverty_est, "%", "<br>",
                             "Population: ", gv_sf_summary$pop_est, "<br>",
                             "Tract: ", gv_sf_summary$tract_name, ", ", gv_sf_summary$locality),
              highlightOptions = highlightOptions(
                fillOpacity = 1,
                bringToFront = FALSE
              )) %>%
  addPolygons(group = "Bachelors Degree or Equivalent",
              stroke = TRUE, 
              weight = 0.5,
              opacity = 1,
              color = "black", 
              fillColor = ~ pal_earn(med_earn_bd),
              fillOpacity = 0.5,
              popup = paste0("Median Earnings: ", scales::dollar(gv_sf_summary$med_earn_bd), "<br>",
                             "Poverty Rate: ", gv_sf_summary$poverty_est, "%", "<br>",
                             "Population: ", gv_sf_summary$pop_est, "<br>",
                             "Tract: ", gv_sf_summary$tract_name, ", ", gv_sf_summary$locality),
              highlightOptions = highlightOptions(
                fillOpacity = 1,
                bringToFront = FALSE
              )) %>%
  addLegend("bottomright",
            pal = pal_earn,
            values = c(15000:130000), 
            title = "Median Earnings",
            labFormat = labelFormat(prefix = "$"), 
            opacity = 1) %>%
  addLayersControl(baseGroups = c("All Education Levels",
                                  "High School Degree or Equivalent",
                                  "Bachelors Degree or Equivalent"),
                   options = layersControlOptions(collapsed = FALSE)) %>%
  addMarkers(data = gv, 
             lng = gv$lon,
             lat = gv$lat,
             popup = paste0("Description: ", gv$description, "<br>",
                            "Date: ", gv$reported_date, "<br>",
                            "Status: ", gv$verified),
             clusterOptions = markerClusterOptions(
               showCoverageOnHover = FALSE,
               iconCreateFunction=JS("function (cluster) {    
    var childCount = cluster.getChildCount();  
    if (childCount < 100) {  
      c = 'rgba(211,211,211);'
    } else if (childCount < 1000) {  
      c = 'rgba(211,211,211);'  
    } else { 
      c = 'rgb(211,211,211);'  
    }    
    return new L.DivIcon({ html: '<div style=\"background-color:'+c+'\"><span>' + childCount + '</span></div>', className: 'marker-cluster', iconSize: new L.Point(40, 40) });

  }")
             ))

```

------------------------------------------------------------------------

#### Scatterplot

##### Median Annual Earnings Compared to Incidents of Gun Violence

```{r earnings_scatter}

scatter_text3 <- gv_sf_summary %>%
  filter(percent_incidents > 5 |
           med_earn_25 < 40000)

gv_sf_summary %>%
  ggplot(aes(x = med_earn_25, y = percent_incidents, size = pop_est, color = locality)) +
  geom_point(alpha = 0.8) +
  scale_size(range = c(.1, 12)) +
  geom_label_repel(data = scatter_text3,
                   aes(x = med_earn_25, y = percent_incidents, label = tract_name), size = 3,
                   min.segment.length = unit(0, 'lines'),
                   inherit.aes = FALSE) +
  scale_color_manual(values = c("#007BAB", "#F8BE3D"), 
                   name = "Region") +
  scale_y_continuous(labels = function(x) paste0(x, "%")) +
  scale_x_continuous(labels = scales::label_dollar()) +
  labs(x = "Median Annual Earnings (25+ years old)",
       y = "Percentage of Total Gun Violence Incidents",
       #title = "Median Annual Earnings Compared to Incidents of Gun Violence",
       subtitle = "Each circle represents a different census tract",
       caption = "Years: 2018 - 2024",
       size = "Estimated Population") +
  guides(size = "none") +
  theme_minimal()

```

<p class="comment">
<b>Notable Trends</b> 
<br> 
<br> 
According to the scatterplot above, the three regions with the most gun violence (Fifeville - Cherry Avenue, 10th & Page - Venable, and Ridge Street) all have median annual earnings around \$40,000. The only other regions with median annual earnings below \$40,000 are Rose Hill - Harris - Venable, Carr's Hill McCormick Road, JPA - Fontaine, and Old Ivy - Darden School. These blocks also house a significant portion of UVA students, who do not have any reportable annual earnings.
</p>

<details>

<summary><b>Data Notes</b></summary>

**ACS 5 Year Summaries**

The American Community Survey (ACS) 5-Year Estimates for median earnings are summaries for geographic areas that are based on data collected over a five-year period. The ACS data represented here is from 2018-2022. The data represents the combined responses across all five years.


**Missing Data**

There are a few tracts where median earnings by specific education level are not available. This generally happens when there are large margins of error in the measurements or very small sample sizes for the tract. As a result the map displays these areas in grey.

</details>

-----------------------------------------------------------------------

## Unemployment Rates

### Are incidents of gun violence related to unemployment? {.tabset}

Unemployment also speaks to the economic opportunities available to residents. Below we map the rate of unemployement among residents of each neighborhood.

#### Map

<details>
<summary><b>Map Instructions</b></summary>
The map below shows unemployment rates in our region. Each grey circle represents a cluster of gun violence incidents. Click on a circle to zoom into the cluster and explore the location, date, and description of each incident. Click on a census tract to see population measurements. Zoom in to see specific areas and incidents more closely. 
</details>

```{r unemployment_map}

gv_sf_summary %>%
  leaflet() %>%
  addProviderTiles(providers$CartoDB.Positron) %>% 
  addPolygons(group = "Unemployment Rate",
              stroke = TRUE, 
              weight = 0.5,
              opacity = 1,
              color = "black", 
              fillColor = ~ pal_pov(unemployment_rate),
              fillOpacity = 0.5,
              popup = paste0("Unemployment Rate: ", gv_sf_summary$unemployment_rate, "%", "<br>",
                             "Population: ", gv_sf_summary$pop_est, "<br>",
                             "Tract: ", gv_sf_summary$tract_name, ", ", gv_sf_summary$locality),
              highlightOptions = highlightOptions(
                fillOpacity = 1,
                bringToFront = FALSE
              )) %>%
  addLegend("bottomright",
            pal = pal_pov,
            values = ~ unemployment_rate, 
            title = "Unemployment Rates",
            labFormat = labelFormat(suffix = "%"), 
            opacity = 1) %>%
  addMarkers(data = gv, 
             lng = gv$lon,
             lat = gv$lat,
             popup = paste0("Description: ", gv$description, "<br>",
                            "Date: ", gv$reported_date, "<br>",
                            "Status: ", gv$verified),
             clusterOptions = markerClusterOptions(
               showCoverageOnHover = FALSE,
               iconCreateFunction=JS("function (cluster) {    
    var childCount = cluster.getChildCount();  
    if (childCount < 100) {  
      c = 'rgba(211,211,211);'
    } else if (childCount < 1000) {  
      c = 'rgba(211,211,211);'  
    } else { 
      c = 'rgb(211,211,211);'  
    }    
    return new L.DivIcon({ html: '<div style=\"background-color:'+c+'\"><span>' + childCount + '</span></div>', className: 'marker-cluster', iconSize: new L.Point(40, 40) });

  }")
             ))

```

------------------------------------------------------------------------

#### Scatterplot

##### Unemployment Rates Compared to Incidents of Gun Violence

```{r employment_scatter}

scatter_text4 <- gv_sf_summary %>%
  filter(percent_incidents > 5 |
          unemployment_rate > 7)

gv_sf_summary %>%
  ggplot(aes(x = unemployment_rate, y = percent_incidents, size = pop_est, color = locality)) +
  geom_point(alpha = 0.8) +
  scale_size(range = c(.1, 12)) +
  geom_label_repel(data = scatter_text4,
                   aes(x = unemployment_rate, y = percent_incidents, label = tract_name), size = 3,
                   min.segment.length = unit(0, 'lines'),
                   inherit.aes = FALSE) +
  scale_color_manual(values = c("#007BAB", "#F8BE3D"), 
                   name = "Region") +
  scale_y_continuous(labels = function(x) paste0(x, "%")) +
  scale_x_continuous(labels = function(x) paste0(x, "%")) +
  labs(x = "Unemployment Rate",
       y = "Percentage of Total Gun Violence Incidents",
       #title = "Unemployment Rate Compared to Incidents of Gun Violence",
       subtitle = "Each circle represents a different census tract",
       caption = "Years: 2018 - 2024",
       size = "Estimated Population") +
  guides(size = "none") +
  theme_minimal()

```

<p class="comment">
<b> Notable Trends</b> 
<br> 
<br> 
According to the scatterplot above, there is no strong relationship between unemployment rates and proportions of gun violence for each census tract. The two regions with the highest rates of unemployment, 10th & Page - Venable and Ridge St. - also have significantly high proportions of gun violence. Fifeville - Cherry Avenue, is a notable outlier: although it has some of the highest proportion of gun violence, the unemployment rate is only 1.3, below the regional median of 3.4. Commonwealth - Townwood - Berkmar and Carr's Hill McCormick Road also have higher rates of unemployment, but lower rates of gun violence.
</p>

------------------------------------------------------------------------

# Access to Firearms

Nearly a third of Americans own a gun[^11] providing easy access to firearms for many people. Someone can legally purchase a gun from their local big-box store, their neighborhood gun dealer, or a specialty weapons trade show. The minimum age in Virginia to purchase a handgun is 21, however individuals as young as 18 can legally purchase rifles and shotguns[^12]. Residents can open carry their weapons, except in areas where firearms are prohibited by statute (such as UVA Grounds[^13]).

[^11]: PEW Research - Key facts about Americans and guns - <https://www.pewresearch.org/short-reads/2023/09/13/key-facts-about-americans-and-guns/> - Published September 13, 2023.

[^12]: Virginia State Police - Firearms/Concealed Weapons FAQ - <https://vsp.virginia.gov/services/firearms/faq> - Accessed April 30, 2024.

[^13]: UVA Policy - SEC-030: Regulation of Weapons, Fireworks, Explosives, and Other Prohibited Items - <https://uvapolicy.virginia.edu/policy/sec-030> - Published 2022.

People can also obtain guns illegally: from manufacturing their own, to under-the-table sales, to trade and barters, and theft.

This section explores two means of accessing guns in Charlottesville and Albemarle: from the many licensed dealers and from thefts, specifically from vehicles.

## Firearms Dealers {.tabset}

The [Bureau of Alcohol, Tobacco, Firearms, and Explosives (ATF)](https://www.atf.gov/firearms) maintains a record of all federally licensed firearms dealers. Some of these dealers are widely frequented businesses, such as the Walmart on Route 29, while others are individuals that operate out of residential neighborhoods. The list and map below show the 22 authorized dealers in Charlottesville and Albemarle county as of April 2024.

### Map

#### Locations of Firearms Dealers in Charlottesville & Albemarle

```{r dealer_map, out.width="90%"}

# Census shape data 
alb_sf <- dat %>%
  st_union()

atf_dealers <- read_csv("data/atf_dealers.csv")
load(file = "data/alb_map.Rdata")

ggmap(alb_map) +
  geom_sf(data = alb_sf, inherit.aes = FALSE, alpha = 0, linewidth = 0.75) +
  theme_void() +
  geom_point(atf_dealers,
             mapping = aes(x = lon, y = lat, col = business_type, shape = business_type),
             size = 3) +
  scale_color_manual(name = "Business Type", values = c("#F8BE3D", "#007BAB")) +
  scale_shape_manual(name = "Business Type", values = c(17, 19)) +
  theme_void() +
  labs(caption = "Total licensed dealers: 22")
  
```

<p class="comment">
<b> Notable Trends</b>
<br> 
<br> 
The map above shows that a majority of licensed firearms dealers operate out of businesses in Charlottesville, while there are a handful of residential entities that are licensed to deal firearms in Albemarle county.
</p>

------------------------------------------------------------------------

### Table

The type of license issued indicates if the seller is a dealer or manufacturer of firearms and/or other destructive devices like explosives. The sellers in Charlottesville and Albemarle have the following license types:

-   **Type 01** - Dealer in Firearms Other Than Destructive Devices

-   **Type 02** - Pawnbroker in Firearms Other Than Destructive Devices

-   **Type 07** - Manufacturer of Firearms Other Than Destructive Devices

-   **Type 09** - Dealer in Destructive Devices

-   **Type 10** - Manufacturer of Destructive Devices, Ammunition for Destructive Devices or Armor Piercing Ammunition

```{r dealer_list}

atf_dealers %>%
  select(license_name, business_name, premise_street, license_type) %>%
  reactable(
    defaultColDef = colDef(
      header = function(value)
        str_to_title(gsub("_", " ", value, fixed = TRUE)),
      align = "left",
      headerStyle = list(background = "#f7f7f8")
    ),
    columns = list(
      license_type = colDef(minWidth = 50)
    ),
    bordered = TRUE,
    highlight = TRUE
  ) %>%
  reactablefmtr::add_source("List of all Federal Firearm License types: https://www.atf.gov/resource-center/fact-sheet/fact-sheet-federal-firearms-and-explosives-licenses-types", align = "right", font_size = 10, font_color = "grey")

```

------------------------------------------------------------------------

## Theft from Vehicles

Thefts from cars are the largest source of stolen guns and have increased triplefold compared to the last decade.[^14]

[^14]: Jay Szkola, Megan J. O’Toole, and Sarah Burd-Sharps - Gun Thefts from Cars: The Largest Source of Stolen Guns - <https://everytownresearch.org/report/gun-thefts-from-cars-the-largest-source-of-stolen-guns-2/> - Published May 9, 2024.

According to the Virginia Uniform Crime Reporting Database, in 2023 \$1,577,221 dollars worth of guns were reported stolen from vehicles in Virginia. Specifically, in Albemarle the amount reported was \$20,857, and in Charlottesville, \$11,940.

The plot below shows the number of guns stolen from vehicles in Charlottesville City and Albemarle County from 2016 to 2023.

#### Theft of Firearms from Vehicles

```{r nibrs_theft}

nibrs_theft <- read_csv("data/nibrs_theft.csv") %>%
  filter(district != "Virginia")

ggplot(nibrs_theft, aes(year, n_stolen, colour = district)) +
  geom_line(linewidth = 1.5) +
  labs(x = "Year",
       y = "Number of Firearms Stolen") +
  scale_color_manual(values = c("#007BAB", "#F8BE3D"), 
                    name = "Region") +
  geom_label(data = nibrs_theft, 
             aes(label = n_stolen), 
             show.legend = F, 
             alpha = 0.75,
             fontface = "bold") +
    scale_x_continuous(breaks = scales::pretty_breaks(n = 7),
                     guide = guide_axis(angle = 35)) +
  theme_minimal()

```

<p class="comment">
<b> Notable Trends</b> 
<br> 
<br> 
In general, the rate of firearms being stolen from vehicles has increased, with 48 guns being stolen from vehicles in 2023 in Charlottesville and Albemarle.
</p>

<details>
<summary><b>Data Notes</b></summary>
Theft data is pulled Virginia's [Uniform Crime Reporting - Incident Based Reporting](https://vsp.virginia.gov/sections-units-bureaus/bass/criminal-justice-information-services/uniform-crime-reporting/#UCR-IBR) system. See more about collection methodology and limitations in the [Crimes Involving Firearms](#crimes) section. 
</details>

------------------------------------------------------------------------

# Next Steps

This work has focused on understanding the larger picture of gun violence in our community -- the nature, scope, frequency, and targets -- by pulling data from multiple sources, each of which tells only a partial story. As this work continues and we continue to expand on this report, we will work with community partners to identify additional data that can help us better understand the underlying causes and larger impacts and to strategize on and evaluate solutions. This might include information on community factors such as regional history, adverse childhood experiences, access to resources, and other social structures that are in play.

## Project Repository

The work supporting this report, including our data collection documentation and the corresponding data, is publicly available in the [Virginia Equity Center GitHub Charlottesville and Albemarle Gun Violence Repository](https://github.com/virginiaequitycenter/cville-alb-gun-trauma).

## Acknowledgements

We'd like to thank Misty Graves, the Director of Human Services for the City of Charlottesville, and Kaki Dimock, Chief Human Services Officer for Albemarle County for their feedback on this report and participation in the [Community Safety Implementation Group](https://prescouncil.president.virginia.edu/community-safety). We'd like to thank Andrew Friedman, Senior Crime Analyst with the Albemarle County Police Department for supporting us with anonymized local data. We'd like to also thank the Equity Center's Community Research team, specifically Dr. Sherica Jones-Lewis, Jess Harris, Sydney Schuler, and DeVon Wade for their continued support, feedback, and dissemination. 

## Citation

Samantha Toet, Michele Claibourn. “The State of Gun Violence in Charlottesville and Albemarle” Published November 22, 2024. https://virginiaequitycenter.github.io/cville-alb-gun-trauma/data_descriptions.

This work is licensed under a [Creative Commons Attribution 4.0 International License](https://creativecommons.org/licenses/by/4.0/).

------------------------------------------------------------------------

# Appendix

## Terminology

#### *Firearm Injury* {#firearm-injury}

The [CDC](https://www.cdc.gov/violenceprevention/firearms/fastfact.html) defines a firearm injury as:

> "A gunshot wound or penetrating injury from a weapon that uses a powder charge to fire a projectile. Weapons that use a powder charge include handguns, rifles, and shotguns. Injuries from air- and gas-powered guns, BB guns, and pellet guns are not considered firearm injuries as these types of guns do not use a powder charge to fire a projectile."

Types of firearm injuries include:

-   Intentionally self-inflicted
-   Unintentional (accidental injuries that happen while someone is cleaning or playing with a firearm)
-   Interpersonal violence (homicide or assault)
-   Legal intervention (violence inflicted by law enforcement on duty)
-   Undetermined intent (not enough information to determine)

#### *Gun Violence* {#gun-violence}

Gun violence is an offense committed with firearms, such as handguns, shotguns, or semi-automatic rifles. It can include homicide, violent crime, attempted suicide, suicide, and unintentional death and injury. The national [Gun Violence Archive](https://www.gunviolencearchive.org/) describes gun violence as:

> "all incidents of death or injury or threat with firearms... a shooting of a victim by a subject/suspect is considered gun violence as is a defensive use or an officer involved shooting. The act itself, no matter the reason is violent in nature."

#### *Gun Violence Trauma* {#gun-violence-trauma}

The [American Sociological Association](https://www.asanet.org/footnotes-article/society-survivors-survivor-society-examining-gun-trauma-us/) describes gun violence trauma as the psychological and emotional impact that gun violence can have on individuals and communities. It can result from direct or indirect exposure to gun violence and has long-term negative effects on wellbeing. This information is difficult to objectively measure, so while this analysis focuses on numbers of direct exposure to gun violence for individuals it is important to remember the lasting indirect trauma gun violence has on our community.

#### *Incident* {#incident}

An incident of gun violence can be defined as a time-boxed event where individual acts of violence took place. An incident may include multiple offenses and individuals involved. For example, if an individual went on a shooting spree and injured three people, that would be considered as one incident with three offenses.

#### *Crime* {#crime}

The [FBI](https://cde-prd-data.s3.us-gov-east-1.amazonaws.com/nibrs/_all/NIBRS_DataDictionary.pdf?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Content-Sha256=UNSIGNED-PAYLOAD&X-Amz-Credential=ASIAQC732REKXFJPIH3N%2F20240603%2Fus-gov-east-1%2Fs3%2Faws4_request&X-Amz-Date=20240603T163942Z&X-Amz-Expires=900&X-Amz-Security-Token=FwoDYXdzEI7%2F%2F%2F%2F%2F%2F%2F%2F%2F%2FwEaDO5feniVpfL1KaEfuSKhAgc9X%2Bc%2FW5%2FwpIAXVxPTxAseiEEljen2PEvNlATTpWS0PVPhu5gaKiro8w%2FJwucRn7cwt8LLmhXwGCrQ0yB%2FnhKPJvD4dbt92qV8HrSOg3FoRTONZuuJl9AhOYWgPQK7m1l5tXLVnQn1PUDuR27DIqBPnf68tH48uxZfV2PWZE2spjxAuCLhB7irYW1ajEvAmwypISpfFhGeYW%2FGcUin533xKw6pcY9tJvFx1B%2BL8MJghNnFPVWlnAmWjNP96sygayq1wbUkRO68c1hMJ%2FzVpE9jztEOR64I30TYQ%2Bmh%2FzKqNuvG4AHuEzOTmgnlxBNx2vARAUYR9G9V5cv7%2BZnfx4RwQLbeqdm%2B0bgi8Vk1Zsgk9E5s0191P%2FLg1NCrgcgxO4co%2B7r3sgYyLckfeOKTl8QnBNYYGAjQajWhazqXvVCFGE8oOza7mqX7vICsDO%2B5%2FBdoDriF%2FQ%3D%3D&X-Amz-Signature=901dc6929b34f9da36734967c546cd02236700b5e8e192ac449783f770dfc684&X-Amz-SignedHeaders=host&x-id=GetObject) defines gun-related crimes as the number of illegal offenses using a firearm committed during an incident of gun violence. Gun violence is not synonymous with gun crimes. Not all crimes are violent in nature. For example, a person may be charged with the crime of illegally possessing a firearm, despite not brandishing or using the weapon.

#### *Arrest* {#arrest}

An arrest can be defined as the physical taking or seizing of a person by a police officer. It is an act that indicates the intention to take that person into custody. An arrest can happen after an incident of gun violence, or it can happen for nonviolent issues such as parole violations or a failure to appear for trial. Anyone arrested is considered innocent until proven guilty in a court of law.

#### *Participant* {#participant}

A participant is defined as an individual involved in an act of gun violence. Participants can be victims, those injured or killed by firearms, or offenders, those using firearms. In cases of suicide, participants are considered victims. It is important to note that while the definition is limited to individuals that are physically affected by gun violence, the spectrum of victimization often extends much further to the community.

#### *Youth* {#youth}

For this analysis, youth are considered to be juveniles under the age of 18. Some visualizations include individuals up to 24 years old as noted in the description.

## References