Does Who Files Change the Outcome?
Partly, and less than the raw numbers suggest. Small entities are granted 13.2 points less often than large ones. More than half of that travels with technology mix and with who files the application. Hold field, examiner and filing account fixed and 5.8 points remain, and 12.4 for micro entities.
The raw difference
Taken alone this looks like a straightforward difference in treatment. Most of it is not.
Subtracting these rates gives 13.5 and 28.2 points, which is not quite what the adjustment table below reports. The two use different samples: these headline rates cover all 1,235,857 resolved applications, while the model uses the 1,223,976 with a complete art unit, filing year, examiner, account and entity record. Its unadjusted gaps are 13.2 and 27.5 points.
What the difference is made of
Small and large entities do not file the same applications into the same art units. Holding the comparison inside an art unit and filing year, then adding the examiner, then adding the filing account, removes more than half the gap.
| Held fixed | Small entity | Micro entity |
|---|---|---|
| Nothing, the raw gap | -13.2 | -27.5 |
| Art unit and filing year | -8.2 | -23.2 |
| Plus the examiner | -7.8 | -21.3 |
| Plus the filing account | -5.8 (0.14) | -12.4 (0.35) |
| Share of the raw gap left | 44% | 45% |
The examiner adds almost nothing: the gap moves by 0.4 points when the examiner is held fixed. Most of what the controls remove is technology and who files.
Examiners in the same art unit do differ
Within a single art unit, the gap between examiners is wider than the gap between entity sizes. Adjusting for the work in front of each examiner narrows it without closing it.
| Measure | Grant rate, of resolved |
|---|---|
| Median art unit grant rate | 85.2% |
| Raw gap between examiners, median art unit | 21.5 points |
| Adjusted for each examiner's docket | 13.7 points |
| Gap expected by chance alone | 4.9 points |
The last row is the one that keeps the finding honest. Some spread between examiners would appear even if every examiner behaved identically, simply because dockets differ and samples are finite. The adjusted gap is about three times that, so it is not noise, but it is not 21.5 points either.
What this does not say
- The controls cannot see the quality of the application, the budget behind it, or how hard it was pressed. Any of those could account for what remains.
- Entity status is current rather than as filed, so an applicant whose status changed is recorded under the later one.
- No examiner is named, here or in the underlying study. Two thresholds apply and they are not the same rule: the examiner comparison counts only examiners with 50 or more resolved applications, and no figure anywhere in this work is published from a cell smaller than 30 applications.
- A grant rate is not a measure of whether the right applications were granted.
Our own counts from the USPTO record. These describe what happened across past applications. They say nothing about any particular application, examiner or firm, and none of this is legal advice.
The figures on this page are available as a CSV: /data/does-who-files-change-the-outcome.csv. It is generated from the tables and charts above, so it carries the same numbers and cannot fall out of step with them.