One Woman Is Not Parity: The Metric That Hides the Gap
"At least one woman inventor" makes the patent gender gap look better than it is. The share-weighted numbers tell a different story.
21.9% of US patents have at least one woman inventor. The headline sounds like meaningful progress.
It is not the same as 21.9% of inventors being women. The share-weighted women inventor rate is 12.8%. And only 4% of US patents have an all-women inventor team. The gap between these three numbers is the gap between optical and actual progress.
This page walks the three metrics, explains why the "at least one woman" version is so widely cited despite overstating progress, and connects the metric choice to what it reveals about team-composition dynamics in US patenting.
This article presents published research data for educational purposes. It is not legal, policy, or investment advice. Patent gender statistics depend on name-based inference and methodology choices: see How we measure.
The three metrics, side by side
| Metric | US value (most recent) | What it measures |
|---|---|---|
| "At least one woman" | 21.9% (2019) | Patents with one or more women on the inventor team |
| Share-weighted WIR | 12.8% (2019) | Individual inventor share who are women (fractional count) |
| All-women teams | ~4% | Patents with no men on the inventor team |
Sources: USPTO Progress and Potential: 2020 Update; analysis by IPWatchdog of USPTO data.
The 21.9% headline gets quoted most often, including in press coverage and corporate diversity disclosures. The 12.8% share-weighted rate is the standard academic measure. The 4% all-women rate is the most-rarely cited and the most diagnostic of how the gap actually works.
Why "at least one woman" inflates apparent progress
Modern US patent applications often name multiple inventors (median around 2-3, with some industrial patents naming 5-10+). The "at least one woman" metric is binary: a patent with 1 woman and 4 men counts identically to a patent with 5 women and 0 men.
The asymmetry is amplified by a property of the metric itself: as patent teams grow, the probability that at least one woman appears can rise even if women remain a small minority of inventor positions. A field with 12% women in its R&D workforce and an average team size of 4 would mathematically produce "at least one woman" rates closer to 40% even with no underlying change in the share of women inventors. Growing team sizes inflate the headline metric independently of any progress on parity.
Empirically, the post-2000 growth in women's participation in patenting came almost entirely from women joining male-led mixed teams, not from all-women teams growing as a share of total filings:
- Share of US patents with multiple female inventors rose from 2.0% (1998) to 4.5% (2017).
- Share of US patents with all-women inventor teams has been roughly 4% for over a decade and has barely moved.
- 80% of all-women inventor teams are two-person teams.
- Among single-woman teams (one woman + men), 36% are single-woman + single-man pairs.
- Teams with one woman average 2.4 men per team.
Sources: USPTO 2019 Progress and Potential report; IAM Media analysis of USPTO data.
Translated into plain English: when corporate diversity disclosures and headlines say "30% of our patents have at least one woman inventor," what they often mean in practice is that 30% of patents have a male-led team with one woman on it. That is a meaningful form of progress; it is not the same as parity.
Mixed-gender teams produce more disruptive science
Yang, Tian, Woodruff, Jones, and Uzzi (PNAS, 2022) analyzed 6.6 million medical-science papers published since 2000 and found that gender-diverse teams produce substantially more novel and higher-impact research than same-gender teams of equivalent size. The effect held across fields and team sizes.
That finding is good news for the rising "at least one woman" metric: more mixed teams should mean more breakthrough work. The corollary is that the stubbornly flat all-women team rate represents a missing category of innovation, not just a missing category of inventors.
Subramani and Saksena (Nature Biotechnology, 2025) extended the team-composition analysis to citation outcomes for granted US patents. They found that majority-female-inventor patents receive fewer forward citations than comparable majority-male patents, with the citation gap reaching up to 22% in some specifications. The same team composition that produces more disruptive science is being commercially under-rewarded.
What the metric choice reveals
For corporate disclosures
Companies signing the USIPA Innovation Diversity Pledge (50+ signatories as of 2025, including Meta, Microsoft, Lenovo, Adobe, AT&T, HP, Cisco) increasingly disclose women inventor rates. Best practice in those disclosures is to report both the "at least one" rate and the share-weighted WIR side by side. Meta has reported a 2023 inventorship rate of 17.6% and a fractional inventorship rate of 16.8%, both above the US baseline.
A pledgee reporting only "at least one" data, without share-weighted comparison, is not being dishonest, but is presenting the most-flattering view of the same underlying performance.
For policy
Setting a national or institutional target on "at least one woman" is largely meeting the current curve. Setting a target on share-weighted WIR is harder. Setting a target on all-women team rate is hardest of all — and the most diagnostic of whether women-led invention pipelines are actually growing.
For the academic literature
Most peer-reviewed analyses (USPTO Office of the Chief Economist, EPO, WIPO, NBER) use share-weighted rates as the primary measure. Press coverage and corporate communications often quote "at least one." When you see a single big number cited in news media, check which one it is. The gap between the two is roughly 9 percentage points.
The all-women team plateau is the most stubborn metric in the field
Of the three metrics, the all-women team share is the one that has moved least. Two decades of progress on the "at least one" metric have not translated into corresponding growth in all-women teams.
Three plausible explanations, none mutually exclusive:
- Pipeline composition. If women remain a minority in most R&D-intensive fields, mathematics alone limits how often any randomly-composed team would be all-women. Software at 7% women translates into a vanishingly small expected rate of all-women software teams under random assortment.
- Network and institutional effects. Inventor teams form through professional networks that are not random. Women in male-dominated fields tend to work on mixed teams because the pool of female colleagues is small. The same dynamic that creates "at least one" growth suppresses all-women team formation.
- Funding and prosecution support. All-women-led patent applications may face higher prosecution costs and lower grant rates (see Where women's patents disappear). If true, all-women applications are also more likely to be abandoned mid-prosecution, suppressing the granted-patent count.
Disentangling which explanation matters most would require team-formation data we do not currently have. What we do have is the observation that the metric has been stuck for two decades, and that movement on the "at least one" metric is not, by itself, a leading indicator of all-women team growth.
Why this matters for the headlines you read
When a major outlet reports that "women now hold 25% of US patents," it is almost always citing the "at least one woman" version. When a peer-reviewed paper reports the gap as "12.8 percentage points off parity," it is almost always citing the share-weighted version. Both are technically correct. Both produce a different mental model of how close the US patent system is to gender parity.
Reading patent gender gap coverage is largely an exercise in noticing which number is being quoted and against what comparator. The share-weighted number is the harder one to move, the slower one to grow, and the more honest representation of where the US patent system actually is.