Case Study: Data Feminism in Practice — Challenging Missing Data

"What gets counted counts." — attributed to William Bruce Cameron (often misattributed to Albert Einstein — itself a lesson in how attribution shapes authority)

Overview

In 2017, a group of data scientists and journalists in Mexico realized something that was simultaneously obvious and invisible: there was no reliable, comprehensive dataset on femicide — the killing of women because of their gender — in their country. The Mexican government published homicide statistics, but the data did not distinguish femicide from other forms of homicide, did not consistently record victim gender, and did not capture the gendered dimensions of violence that made femicide a distinct phenomenon.

The absence of data was not an oversight. It was a structural choice — a choice not to count, not to categorize, not to make visible a pattern of violence that activists and families had been documenting for decades. The group, which became Data Civica, decided to build what the government would not: a comprehensive, community-validated femicide database.

This case study examines three instances where "missing data" about marginalized populations was challenged through counter-data practices: femicide tracking in Mexico, police violence documentation in the United States, and the community data practices that emerged from data feminism's insistence on making the invisible visible. It connects directly to Chapter 32's analysis of how the systematic absence of data perpetuates structural inequality — and how communities can challenge that absence through collective action.

Skills Applied: - Applying data feminism principles to real-world data gaps - Analyzing the politics of what is and is not counted - Evaluating counter-data practices as forms of data activism - Connecting missing data to policy failures and structural violence

The Situation

Case 1: Femicide in Mexico — Data Civica and the Counting of the Dead

Mexico has one of the highest rates of femicide in the world. The country's National Citizen Femicide Observatory estimated approximately 10 women and girls killed per day in 2023. But for years, this reality was obscured by the absence of official data.

What was missing. Mexican government homicide statistics recorded deaths but did not systematically distinguish femicide from other killings. When femicide was recorded, it was often classified under general homicide categories that obscured the gendered nature of the violence. Police reports frequently lacked information about the relationship between the victim and the perpetrator, the circumstances of the killing, and the victim's prior reports of domestic violence — all essential for identifying a homicide as femicide.

Why it was missing. The absence was not accidental. Accurately counting femicide would have required acknowledgment that gender-based violence was a systemic problem, not a collection of isolated incidents. It would have required police and prosecutors to investigate the gendered circumstances of killings, which meant additional training, resources, and political will. And it would have produced a number — a visible, citable, undeniable number — that would have created pressure for policy action. The absence of data served the interests of institutions that preferred the problem to remain unmeasured.

What Data Civica did. Data Civica, working with journalists, human rights organizations, and families of victims, built an alternative dataset. They collected information from news reports, court records, police reports obtained through freedom-of-information requests, and testimonies from families. Each case was cross-referenced against multiple sources and validated through community verification — a process that was labor-intensive but produced data of higher quality and specificity than official statistics.

The resulting dataset documented not just the number of femicides but their context: the age of the victim, the relationship to the perpetrator, whether previous reports of domestic violence had been filed (and ignored), the geographic location, and the legal outcome (whether the perpetrator was prosecuted, convicted, or remained free). This contextual data transformed abstract statistics into a structural analysis of institutional failure.

The impact. Data Civica's work contributed to the passage of Mexico's General Law on Women's Access to a Life Free of Violence and supported the expansion of "gender violence alerts" — official declarations that trigger enhanced resources and monitoring in regions with high femicide rates. The data did what official statistics had failed to do: it made the invisible visible, created accountability pressure, and provided the evidentiary foundation for policy change.

Case 2: Police Violence in the United States — Fatal Force, The Counted, and Mapping Police Violence

In the United States, no comprehensive federal database of police killings existed until journalists and activists built one.

What was missing. The FBI's Supplementary Homicide Reports (SHR) and the Bureau of Justice Statistics' Arrest-Related Deaths (ARD) program both collected data on police-involved deaths, but participation was voluntary, and many departments did not report. Independent analyses estimated that official statistics captured fewer than half of all police killings.

Why it was missing. The absence of data reflected a power dynamic: police departments, which would be the primary subjects of accountability pressure from accurate data, were the same entities responsible for reporting. Voluntary reporting gave departments an implicit veto over their own accountability. Departments in which police killings were disproportionately high had the strongest incentive not to report — and no mechanism existed to compel them.

What activists and journalists did. Three independent initiatives built what the government would not:

  • The Washington Post's Fatal Force database (launched 2015) tracks every fatal shooting by an on-duty police officer in the United States, compiled from news reports, public records, and law enforcement accounts.
  • The Guardian's The Counted (2015-2016) expanded the scope to include all police-involved deaths, not just shootings.
  • Mapping Police Violence (launched 2015, maintained by activist Samuel Sinyangwe and collaborators) combines multiple data sources and provides geospatial and demographic analysis.

These databases consistently documented that official FBI statistics undercounted police killings by approximately 50% — and that the killings were disproportionately concentrated in communities of color, particularly Black communities.

The impact. The independent databases transformed the policy debate. Before 2015, claims about the frequency and racial patterns of police killings were disputed as anecdotal. After 2015, the data was undeniable. The databases were cited in federal investigations, court cases, congressional hearings, and the platform demands of the Black Lives Matter movement. In 2021, the Department of Justice announced a new mandatory data collection program on police use of force — directly addressing the gap that community data had exposed.

Case 3: Trans and Non-Binary Population Data — Invisibility by Design

Most national statistical systems do not collect data on transgender and non-binary populations. Census forms in most countries offer only "male" and "female" options. Health surveys do not systematically record gender identity. Employment statistics do not track outcomes for trans workers.

What is missing. The specific needs, health outcomes, employment experiences, housing security, and safety of trans and non-binary people are invisible to evidence-based policymaking. You cannot design effective healthcare programs for a population whose health needs you have not measured. You cannot enforce anti-discrimination protections when the discrimination is not documented.

Why it is missing. The absence reflects multiple structural factors: data classification systems that impose binary gender categories (data feminism's principle of "rethink binaries and hierarchies"), political resistance to acknowledging transgender identities in official statistics, and the genuine difficulty of collecting sensitive demographic data from a population that faces discrimination and may be reluctant to disclose.

What advocates have done. Trans-led organizations have built their own data systems. The National Center for Transgender Equality's US Transgender Survey (USTS) — conducted in 2015 and again in 2022, with nearly 93,000 respondents in the latter wave — provides the most comprehensive data on the experiences of trans people in the United States. The survey documents rates of discrimination, healthcare access, homelessness, employment outcomes, and interactions with law enforcement that are not captured by any government data system.

The Williams Institute at UCLA School of Law has developed population estimates for LGBTQ+ communities using statistical modeling techniques applied to existing surveys — a methodological innovation that produces usable data from incomplete sources.

The impact. Community-generated data has been cited in legal challenges to discriminatory policies, used to design healthcare programs, and referenced in policy debates about non-discrimination protections. The 2020 US Census included a question about same-sex household relationships for the first time — a small step toward visibility, driven in part by decades of community data advocacy.

Analysis Through Data Feminism Principles

Principle 1: Examine Power

In each case, the absence of data served the interests of powerful institutions. Government officials who benefit from an unmeasured femicide crisis, police departments that benefit from untracked uses of force, and political actors who benefit from the invisibility of trans populations all had structural reasons to maintain data gaps. The first step in challenging missing data is asking: Who benefits from this absence?

Principle 2: Challenge Power

Counter-data practices do not merely fill information gaps — they shift power. When Data Civica published femicide statistics that the government refused to collect, it created accountability pressure that had not previously existed. When the Washington Post documented police killings that the FBI undercounted, it changed the terms of a national debate. Counter-data is a form of power redistribution: it gives communities the evidence they need to demand change.

Principle 5: Embrace Pluralism

Each counter-data initiative incorporated multiple forms of knowledge. Data Civica combined quantitative analysis with family testimonies. Mapping Police Violence combined database analysis with geospatial visualization and community reporting. The US Transgender Survey combined survey methodology with participatory design — trans community members shaped the questions, ensuring that the survey captured experiences that non-trans researchers might not think to ask about.

Principle 6: Consider Context

Counter-data practices insist on context. Official statistics strip context: a homicide is a homicide, a death is a death. Data Civica's femicide database documented the relationships, the prior reports, the institutional failures that transformed isolated deaths into a systemic pattern. The police violence databases documented the race, the circumstances, the weapons (or absence of weapons) that revealed patterns invisible in aggregate numbers. Context is what transforms data from a number into evidence.

Principle 7: Make Labor Visible

The labor of counter-data production is often invisible. Data Civica's researchers spent months cross-referencing court records and news reports. The families who provided testimonies relived their trauma to contribute to a database. The trans community members who completed a 92-page survey shared intimate details of their lives. This labor — emotional, intellectual, and organizational — is the foundation of counter-data practices, and it is overwhelmingly performed by the communities who are already most burdened by the consequences of missing data.

Connection to Chapter Themes

The Consent Fiction operates in reverse with missing data. Normally, we analyze cases where data is collected without meaningful consent. With missing data, the harm comes from data that is not collected — without the consent of the communities who need it. No one asked the families of femicide victims whether they consented to their loved ones' deaths being uncounted. No one asked Black communities whether they consented to police killings going untracked. The absence of data is itself a governance decision — one made without the input of those most affected.

The Accountability Gap

Missing data creates the ultimate Accountability Gap: you cannot hold institutions accountable for problems that have not been measured. If femicide is not counted, femicide prevention programs cannot be evaluated. If police killings are not tracked, police departments cannot be compared. If trans health needs are not documented, healthcare systems cannot be held responsible for failing to meet them. Counter-data practices close this gap by producing the evidence necessary for accountability.

The Power Asymmetry

The power asymmetry in missing data is structural: the institutions that could collect the data (governments, police departments, healthcare systems) are often the same institutions whose behavior the data would scrutinize. This creates a conflict of interest that voluntary data collection cannot resolve. Counter-data practices bypass this conflict by shifting the data collection function to communities and independent organizations — but this shift imposes costs on communities that are already resource-constrained.

Discussion Questions

  1. The politics of counting. Dr. Adeyemi argues that "silence in data is not just an absence — it is a statement about whose experiences are worth measuring." Apply this argument to a data gap you have observed in your own community or area of study. What is not being counted? Who benefits from the absence? What would change if the data existed?

  2. Counter-data sustainability. Counter-data practices are labor-intensive and often depend on volunteer effort, grant funding, or journalistic resources. The Washington Post's Fatal Force database has been maintained continuously since 2015, but its continuation depends on editorial commitment at a single news organization. How can counter-data practices be made sustainable? Should governments be required to collect the data that communities have demonstrated is necessary?

  3. The burden of visibility. Trans and non-binary data advocates face a tension: collecting data on gender identity can enable better services and policy advocacy, but it can also expose individuals to surveillance, discrimination, and targeting. How should data advocates navigate this tension between the benefits of visibility and the risks of exposure? Does the (in)visibility pillar of Taylor's data justice framework provide useful guidance?

  4. From counter-data to official data. In the police violence case, community-generated data eventually prompted the DOJ to establish a mandatory collection program. Is this the goal of counter-data — to pressure governments into collecting the data themselves? Or is there value in maintaining community-controlled alternatives even when official data exists? Consider the differences in who controls the data, what questions are asked, and whose interests the data serves.

Your Turn: Mini-Project

Option A: Missing Data Audit. Identify a data gap in an area you care about — education, healthcare, housing, criminal justice, environment, or another domain. Document: (1) what data is missing, (2) who would benefit if it existed, (3) who benefits from its absence, (4) what counter-data practices (if any) have attempted to fill the gap, and (5) what it would take to make the data collection official. Write a two-page analysis.

Option B: Counter-Data Case Study. Research a specific counter-data initiative not covered in this case study (examples: the Anti-Eviction Mapping Project, Feminicidio.net in Spain, the Missing and Murdered Indigenous Women database, citizen air quality monitoring networks, community-led COVID-19 data collection). Write a case analysis (800-1,200 words) covering: (a) what data was missing, (b) who created the counter-data, (c) what methodology was used, (d) what impact the data has had, and (e) what data feminism principles the initiative exemplifies.

Option C: Data Collection Design. Design a data collection instrument (a survey, an intake form, or a reporting tool) that would address one of the data gaps described in this case study (or one you have identified). Explain: (1) what data you would collect, (2) what categories and classifications you would use (and why), (3) how you would ensure the data serves the affected community rather than extracting from it, and (4) what ethical safeguards you would implement. Apply at least three data feminism principles in your design.

References

  • D'Ignazio, Catherine, and Lauren F. Klein. Data Feminism. Cambridge, MA: MIT Press, 2020.

  • Data Civica. "Claves Para Entender y Prevenir los Asesinatos de Mujeres en Mexico." Mexico City, 2020.

  • James, Sandy E., et al. "The Report of the 2015 U.S. Transgender Survey." National Center for Transgender Equality, 2016.

  • Lartey, Jamiles, and Simone Weichselbaum. "The Uncounted." The Marshall Project, 2017.

  • Mapping Police Violence. "2023 Police Violence Report." https://mappingpoliceviolence.org.

  • Segato, Rita Laura. "Territorio, Soberania y Crimenes de Segundo Estado: La Escritura en el Cuerpo de las Mujeres Asesinadas en Ciudad Juarez." Ciudad Juarez: De Este Lado del Puente, 2004.

  • Taylor, Linnet. "What Is Data Justice? The Case for Connecting Digital Rights and Freedoms Globally." Big Data & Society 4, no. 2 (2017).

  • Washington Post. "Fatal Force." Database, 2015-present. https://www.washingtonpost.com/graphics/investigations/police-shootings-database/.

  • Williams Institute, UCLA School of Law. "How Many Adults Identify as Transgender in the United States?" 2022.