Data Integrity & Evidence-Based Policy

Evidence-based policy rests on one foundational requirement: accurate data. Before governments can address a problem, they must be able to see it clearly — to count who is affected, who is responsible, and whether interventions are working.

For decades, sex-disaggregated data has been one of the most important tools in women's policy. It revealed the wage gap. It exposed the under-diagnosis of heart disease in women. It documented male violence against women as a systemic pattern. It showed girls' underrepresentation in STEM fields and women's absence from corporate leadership. Without this data, the problems remained invisible — and invisible problems cannot be solved.

That data is now being systematically corrupted.

Across Canada, government agencies, Statistics Canada, police services, health authorities, and educational institutions are replacing biological sex with self-declared gender identity in official records. The consequences for women's policy — and for women's safety — are profound.

Crime Statistics: Hiding Who Commits Violence Against Women

In January 2019, Statistics Canada and the Canadian Association of Chiefs of Police announced that the Uniform Crime Reporting Survey — the primary national database on crime — would replace the variable "sex" with self-declared "gender." Police services across Canada were directed to record the gender identity of offenders and victims as they self-report, not as biologically observed.

The consequences are not theoretical. When a male offender who identifies as a woman commits sexual assault, he is recorded as a female perpetrator. When his victim — a woman — reports to police, the crime is recorded as a woman-on-woman assault, or an assault by a "gender diverse" person on a woman.

The pattern this creates in aggregate data is predictable and alarming: apparent increases in violent and sexual crime committed by "women," decreases in male perpetration of violence against women, and statistical distortion that makes the dominant pattern of male violence against women progressively harder to see.

This matters because public policy addressing violence against women depends on accurate identification of who commits it. Programs targeting male perpetrators of domestic violence, sentencing guidelines calibrated to male recidivism patterns, risk assessment tools designed for male sex offenders — all depend on knowing the biological sex of offenders. When that information is corrupted at the point of data collection, every downstream intervention is compromised.

Women who are victimized by males recorded as female have their experiences distorted before they reach a dataset. Their male rapist becomes, statistically, a woman. The pattern of their victimization — male violence against females — is erased.

Health Data: When Biological Reality Disappears from Medicine

Sex matters in medicine. Female and male bodies differ in ways that are clinically significant across virtually every area of healthcare: cardiovascular disease presents differently in women than men and was historically under-diagnosed as a result; autoimmune conditions affect women at far higher rates; certain cancers are sex-specific; drug metabolism varies by sex; and responses to treatment differ between female and male patients.

Sex-disaggregated health data is the foundation for understanding these differences and developing appropriate clinical responses. When that data is replaced with self-declared gender identity, the foundation cracks.

What data corruption looks like in practice:

• A male who identifies as a woman and retains a prostate may be recorded in women's health datasets, distorting statistics on conditions that affect only biological females

• A female who identifies as a man and takes testosterone may experience testosterone-specific health effects recorded in men's health data, obscuring female-specific risks

• Clinical trials that stratify by "gender" rather than sex may fail to detect sex-specific drug interactions or treatment responses

• Cancer screening programs calibrated by gender identity rather than biological sex may miss cancers in bodies that have the relevant anatomy but are recorded under a different gender

The history of medicine is already marked by the consequences of ignoring female biology. Women were excluded from clinical trials for decades. Diagnostic criteria developed on male populations were applied to female patients with harmful results. Sex-specific medicine, developed painstakingly over the past thirty years, has begun to correct these errors. Replacing sex with gender identity in health data risks repeating these mistakes in a new form — not through explicit exclusion, but through the quieter mechanism of data that no longer reflects biological reality.

The Census: Canada's Foundational Demographic Tool

The Canadian Census is the primary instrument by which governments plan services, allocate resources, and track demographic change over time. Its value depends on longitudinal consistency — on the ability to compare data collected in 2021 with data collected in 1971, 1981, 1991, and beyond.

In the 2021 Census, Statistics Canada replaced the question about sex with a question about gender. Respondents were asked to select from male, female, or "please specify" — a free-text field accommodating non-binary identities. The stated rationale was inclusivity. The practical consequence is that the 2021 Census data on sex is no longer directly comparable with all previous Census data on sex, because the question is no longer asking the same thing.

Sex and gender identity are different variables. Gender identity is a self-reported, subjective, and potentially changeable characteristic. Biological sex is an observable, stable, biological fact. Conflating them in foundational demographic data does not make the data more inclusive — it makes it less reliable for the purposes it was collected to serve.

Governments use Census data to plan healthcare services, housing, education, and social programs. When those plans require understanding the needs of female populations — obstetric services, gynaecological care, refuge from male violence, access to female-specific programs — accurate counts of biological females are necessary. Gender identity data does not provide this.

Employment and Education: When Equity Metrics Lose Their Meaning

Employment equity legislation was designed to remedy specific, documented patterns of discrimination: women's under-representation in leadership, male-dominated professions, and high-paying fields; girls' historical exclusion from STEM education; women's disproportionate burden of unpaid care work and its economic consequences.

The data collected to track progress on these equity goals — women's representation on corporate boards, STEM graduation rates, the wage gap — depends on knowing who is biologically female. When males who identify as women are counted in the "women" column of equity statistics, the data stops measuring what it was designed to measure.

A company that hires biological males who identify as women can record those hires as progress toward gender equity targets while the representation of actual biological females in its workforce remains unchanged. A university that admits males who identify as women to women-in-STEM programs can report improved gender parity while biological females remain as underrepresented as before.

The problem is not merely statistical. Equity policy built on corrupted data will consistently appear to be succeeding while actually failing the population it was designed to serve.

Research: The Downstream Effect of Data Corruption

Policy depends on research. Research depends on data. When foundational data is corrupted, the entire chain from observation to intervention breaks down.

Researchers studying violence against women cannot analyze patterns of male perpetration if offender sex has been replaced by gender identity in criminal justice data. Epidemiologists studying sex-specific disease cannot draw meaningful conclusions from health datasets that conflate biological sex with self-declared gender. Economists studying the gender wage gap cannot produce reliable analysis from employment data in which sex has been replaced by gender identity.

Sex-disaggregated data is required for research across an enormous range of fields: criminology, epidemiology, labour economics, sociology, psychology, and public health. When that variable is corrupted or eliminated, entire lines of inquiry become impossible — and the problems they were designed to illuminate become statistically invisible.

The Historical Trend Problem

An additional consequence is the corruption of historical trend analysis. Research relies on comparisons over time. When the definition of key variables changes midway through a time series, comparisons become unreliable. We can no longer ask "are things getting better or worse for women?" because "women" no longer means the same thing in 2024 data as it did in 2004 data. Decades of baseline data — painstakingly collected to track female disadvantage — is now incompatible with the data being collected today.

The Transparency Problem

The shift from sex to gender identity in official data has not been publicly announced, debated in Parliament, or subject to formal policy consultation. It has occurred through administrative changes at Statistics Canada, through guidance issued to police services by the Canadian Association of Chiefs of Police, and through quiet amendments to data collection methodologies across government departments.

Most Canadians do not know that this change has occurred. They assume that when crime statistics report on "female" offenders, those are biological females. They assume that when equity statistics report on women's representation, those figures count biological women. They assume that when health data reports on women's outcomes, it is describing the health of biological females.

These assumptions are no longer warranted — but the change that invalidated them was never publicly explained.

Democratic accountability requires that citizens be informed about changes to foundational institutions. When the basis of national crime statistics changes, Parliament and the public are entitled to know. When Census methodology changes in ways that affect comparability with decades of historical data, there should be a public explanation of the consequences. The quiet administrative substitution of gender for sex in official data collection is a matter of significant public interest that has received almost no public deliberation.

What Accurate Data Requires

Restoring data integrity does not require ignoring gender identity or eliminating accommodations for people who identify outside the male/female binary. It requires maintaining biological sex as a distinct, recorded variable in datasets where it matters — which is most of them.

1. Reinstate sex as a primary variable in crime statistics. Offender and victim sex should be recorded as biological sex, not self-declared gender identity. Gender identity can be recorded as an additional variable where relevant. These are not mutually exclusive, but biological sex must be preserved.

2. Restore sex to the Census. The 2021 change that replaced a sex question with a gender question must be reversed or supplemented. A separate question on biological sex should be reinstated to maintain comparability with historical data and to enable accurate planning for sex-specific services.

3. Require sex-disaggregated reporting in health data. Health datasets must record biological sex as a mandatory variable. Gender identity can be collected additionally where clinically relevant. Clinical trials must stratify by biological sex, not gender identity.

4. Protect sex-based equity metrics. Employment equity legislation must specify that "women" in equity targets refers to biological females. Equity statistics must track biological female representation, not gender identity.

5. Require parliamentary approval for major methodological changes. Changes to foundational data collection methodologies — particularly those affecting national census data, crime statistics, or health surveillance — should require parliamentary debate and approval, not administrative decision.

6. Publish clear methodology documentation. Any dataset that conflates sex and gender identity should prominently disclose this fact, the date the change was made, and the implications for comparability with historical data. Users of official statistics are entitled to know what they are measuring.

Why This Is a Women's Rights Issue

Data corruption may seem like a dry, technical matter — a question for statisticians rather than advocates. It is not.

Every major advance in women's rights over the past century has been driven by data that made women's disadvantage visible. The wage gap required data to document. The prevalence of domestic violence required data to quantify. The under-diagnosis of heart disease in women required data to reveal. The exclusion of girls from educational opportunities required data to prove.

When the data is corrupted, the problems it revealed begin to disappear — not because they have been solved, but because they have been made statistically invisible. A pattern of male violence against women becomes, in corrupted data, violence by perpetrators of indeterminate or mixed sex. The wage gap, tracked against a "women" category that includes biological males, produces numbers that no longer reflect female economic disadvantage. Health disparities obscured by data that can't distinguish biological sex can no longer drive sex-specific medical research.

Replacing sex with gender identity in official data is not a neutral technical choice. It is a policy decision with direct consequences for the visibility of women's disadvantage and the accountability of systems designed to address it. Women's rights advocates have always relied on data to make the case for change. We cannot afford to lose it.

Conclusion

Accurate, sex-disaggregated data is infrastructure. It is as foundational to women's policy as roads are to economic development — invisible when it works, catastrophic when it doesn't.

The quiet administrative replacement of biological sex with self-declared gender identity across Canada's official statistics is dismantling that infrastructure without public debate, without parliamentary approval, and without acknowledgment of the consequences.

Evidence-based policy requires evidence. Evidence requires data. Data requires that the variables it measures actually correspond to the reality it claims to describe.

When crime data records male perpetrators as female, we cannot see male violence clearly. When health data conflates sex and gender, we cannot research sex-specific disease accurately. When equity metrics count males as women, we cannot track female underrepresentation honestly. When the Census no longer counts biological females, we cannot plan services for them effectively.

Restoring data integrity is not a cultural war. It is a precondition for evidence-based governance. Women's lives depend on policy that works — and policy that works requires data that is true.