Pick any two things and compare what they pay. Every figure here comes from ministry leaders — and every cell too thin to trust says so instead of guessing.
Choose what to break out and what to compare it against. Cells with fewer than 12 paid respondents are suppressed — a number drawn from eight people isn't a finding.
Each tile is a state, sized equally so small states stay readable. Only states with at least 15 respondents show a figure. The panel starts on the national picture — click a tile to compare that state against it, and click the same tile again to return.
Pooling to four regions gives each gender — paid respondents — and even then no salary difference clears two standard errors. The access gap beneath each pair does. That contrast is the finding, not a shortcoming of the sample: see Who Gets Paid.
Pick a group to see how its pay is distributed across bands — not just the average, which can hide a lot.
Volunteers are never counted as $0. The survey asks compensation as a band, with “I am a volunteer” as one option. Averaging unpaid volunteers into a pay figure would manufacture gaps that are really differences in who gets paid at all. Every average here uses paid respondents only; unpaid rates are reported separately.
Averages are estimates. Because pay is banded, we assign each band its midpoint ($100,000+ is treated as $115,000) and average those. This is a reasonable approximation, not payroll data. Median bands are also shown, which are more robust for banded responses.
Suppression is enforced in the data, not the display. Cells under 12 paid respondents (15 for states) carry no figure at all, so no amount of clicking can surface a number built on a handful of people.
These are unadjusted comparisons. No controls for church size, tenure, role, or hours. Church size alone moves average pay from roughly to , so most apparent gaps between groups partly reflect which churches they work in.