The #MeToo era produced a generation of corporate policies and a relatively small change in measured harassment. The Equal Employment Opportunity Commission has tracked sexual-harassment charges every year since the late 1990s; the absolute number rose modestly in 2018, then settled back near the long-run baseline of 6,000–7,000 charges annually. That is not because harassment fell — most credible surveys of women in the workforce continue to find that roughly 40–60% report having experienced workplace sexual harassment at some point in their careers — but because reporting is, and has always been, a thin slice of the underlying behavior. The EEOC's own Select Task Force on the Study of Harassment in the Workplace estimated in 2016 that roughly three out of four workers who experience harassment never formally report it.
That reporting gap is the policy problem. Almost everything the field has done since 2017 — mandatory training videos, online reporting portals, third-party hotlines — has tried to make the reporting step easier without changing the incentives that make workers refuse to use it. The result is a system that documents the harassment willing to report itself and ignores most of the rest.
Training doesn't work the way HR thinks it does
The most honest body of evidence comes from Frank Dobbin (Harvard) and Alexandra Kalev (Tel Aviv University), whose work on what corporate diversity programs actually move — synthesized in Getting to Diversity: What Works and What Doesn't (Harvard, 2022) — has been replicated across decades of EEO-1 data. Their finding for harassment training is consistent and depressing: stand-alone compliance training has, at best, null effects on subsequent harassment claims and, in some studies, a small negative effect on women's representation in management afterward, possibly because such training activates resentment among the men it targets.
What does move the needle, in the same body of research, is a narrow set of structural changes: investigations performed by trained external parties rather than line HR; managers' inclusion records as a factor in their own performance reviews; and explicit climate surveys whose results are reported up to the board. A 2020 RAND Corporation study of harassment in the U.S. military, the largest organization where these interventions have been tested at scale, found that command-climate metrics tied to commander evaluations produced the most measurable reductions in incident reports.
The legal regime moved in 2022, and almost nobody noticed
The Ending Forced Arbitration of Sexual Assault and Sexual Harassment Act, signed into law in March 2022, voids pre-dispute arbitration clauses and class-action waivers for sexual harassment and assault claims. That single change converted a substantial fraction of harassment disputes from confidential arbitration — where settlement data never reaches the public — into open court, where reputational risk to employers is real. The Speak Out Act, passed later in 2022, voids pre-dispute nondisclosure agreements for the same categories. These were not headline laws, but they are the most significant federal interventions on harassment since the 1991 Civil Rights Act.
The empirical question now is whether these statutes produce a measurable change in employer behavior. Early evidence from researchers at the University of Iowa law school and the Berkeley Center on Comparative Equality and Anti-Discrimination Law suggests they have begun to shift settlement practices and the structure of internal investigations — but the data is thin. Most general counsels are quietly redrafting handbooks and saying as little as possible publicly.
The hidden categories: power, race, and the gig workforce
The largest categorical gap in harassment policy is the workers it doesn't reach. Independent contractors are not covered by Title VII. Gig platform workers face customer-facing harassment with almost no formal recourse — a problem documented at length in Mary Gray and Siddharth Suri's Ghost Work (2019), and which the EU Platform Work Directive of 2024 begins to address abroad but the U.S. has not. Domestic workers are still partially excluded from federal labor protections. Roughly a quarter of the U.S. workforce now sits, at least part-time, in one of these legally thinned categories.
Race compounds. The Center for WorkLife Law's 2023 survey of women of color in STEM and academic medicine found that Black and Latina women reported the highest rates of being mistaken for staff, having their expertise questioned, and being excluded from informal networks — the day-to-day pattern that doesn't trigger a Title VII charge but predicts attrition. McKinsey and LeanIn.Org's Women in the Workplace 2023 report found that Black women were 50% more likely than white women to say they had experienced microaggressions in the past year, and twice as likely to say they had to hide aspects of themselves at work. Those are not separable from harassment; they are its baseline atmosphere.
Bystander programs are the most promising single intervention
The intervention with the strongest causal evidence is bystander training that teaches specific scripts ("I noticed that comment, can we step back?") rather than abstract values. The CDC, drawing on the same research base used in campus sexual-violence prevention, has identified bystander programs as one of the few primary-prevention strategies with consistent evidence in workplace and community settings. Bringing In The Bystander, originally developed at the University of New Hampshire, has been adapted for corporate use with documented reductions in measured rape-supportive attitudes and increases in intervention behavior. The reason these programs work is also why they are unpopular: they put responsibility on the room rather than on the target.
What an honest harassment policy looks like in 2026
An employer serious about this problem stops measuring training-completion rates and starts measuring four things instead. First, the climate-survey delta — how the proportion of employees who say they would feel safe reporting moves year over year. Second, time-to-resolution and complainant retention rates after investigations, because losing the complainant is the strongest signal that the system retaliated. Third, manager-level patterns — do certain managers' teams report at higher rates, and what happens to their careers. Fourth, the rate at which harassment incidents resolve through internal channels versus arbitration or litigation, post-2022.
None of those metrics get put in an annual report yet. The companies that publish them first will gain a real competitive advantage in recruiting, and will also be the first to find out what their workplaces actually look like. For more on the broader workforce-policy frame this connects to, see the New Labor Movement → pillar.
The 2022 federal end of forced arbitration in harassment cases was the biggest change to workplace harassment law in a generation, and almost no one noticed because it didn't come with a hashtag.
Updated May 21, 2026. This piece was substantively rewritten as part of NWLB's 2026 editorial refresh.
