Report-year research

How to compare nonprofit compensation trends from 2020 through 2026.

990Search offers report editions from 2020 through 2026. A defensible trend comparison uses the same position, geography, organization scope, charity category, and annual-expense band in every year—and reads each edition’s source and validation notes before interpreting a change.

A report-year change is not automatically a market trend

A later result can change because compensation changed, but it can also change because different organizations filed, an amended return replaced an original, the available source cohort changed, or title evidence supported a different role assignment. Treat the report as a repeated public-filing benchmark, not as a survey of the same employers in every year.

Hold the comparison group constant

Choose one standardized role and repeat the same nationwide, state, or county scope; private-foundation choice; charity category; and annual-expense band. If one year uses all organizations while another uses a narrow category or a different budget group, the difference describes the filters—not a clean trend.

Compare the whole distribution

Track observation count, average, median, and available percentiles together. A rising average with a stable median may indicate movement in the upper end of the distribution. A falling count may make a percentile less stable. A suppressed cell means the published threshold was not met; it is not a zero-dollar result.

Read the report edition and source-year statement

The year in a report title identifies the 990Search report edition. Form 990 data is historical and filed after the underlying fiscal period. The finished report describes the edition and available validation status so the word “current” is not mistaken for real-time payroll.

A repeatable trend workflow

  1. Choose one role and one defensible peer group.
  2. Run the same filters for each report edition from 2020 through 2026.
  3. Record count, average, median, and the same available percentiles.
  4. Flag changes that coincide with a large population shift or a suppressed cell.
  5. Use filing-level “View data” evidence to investigate surprising results.
  6. Document the edition, filters, and date of the analysis.
Do not smooth historical results to force an upward line.

A later year does not have to exceed the year before it. 990Search preserves the published evidence and methodology instead of changing valid observations to create a preferred trend.

Questions about historical comparison

Can nonprofit compensation be compared across report years?

Yes, if the comparison holds the role and peer-group filters constant and accounts for edition-specific source years and methodology notes. A change in filters is not a compensation trend.

Does a later report edition mean live payroll data?

No. Form 990 compensation is reported after the fact. Each report edition identifies the filing evidence and methodology used for that publication.

Should averages or medians be used for a trend?

Review both. Averages respond to every included value, while medians show the midpoint and are less affected by a small number of unusually high observations.