A clear view of the salary differences in the 2025 developer survey, with uncertainty and survey limitations kept in view.
Remote respondents report higher pay than hybrid respondents, and average pay differs across the three education groups. These are survey associations—use them to frame questions for your hiring policy, not as evidence that changing work mode or credentials will cause a pay change.
Three exploratory views show where salary and satisfaction vary. Open any chart for a closer look.
Most title averages cluster together. Groups with fewer than 30 valid ratings are flagged in red, so small samples are not overread.
Bars show mean salary; diamonds show the median. The gap reveals how exceptionally high salaries can pull a mean upward.
Median salary rises from $25,573 at ages 18–24 to about $110,000 at ages 55–64. The oldest group is much smaller, and every age group contains a wide spread of salaries.
Hybrid and remote responses are compared using the same annual USD salary measure throughout.
Median $77,730
Median $90,527.50
↗ Remote respondents average $16,893.50 more than hybrid respondents. Both medians point in the same direction.
| Work mode | n | Mean | Median | Std. dev. | 25th pct. | 75th pct. | 99th pct. | Maximum |
|---|---|---|---|---|---|---|---|---|
| Hybrid | 7,803 | $101,790 | $77,730 | $347,265 | $47,566 | $116,015 | $400,000 | $18,387,548 |
| Remote | 6,934 | $118,684 | $90,527.50 | $442,810 | $47,650 | $150,000 | $500,000 | $33,552,715 |
Large maxima and standard deviations signal heavy right tails. The main analysis keeps all positive salaries; excluding salaries above the pooled 99th percentile ($450,000) as a sensitivity check still leaves remote ahead by $14,582.13.
Manual pooled calculation and SciPy's built-in test agree to numerical precision. With 14,735 degrees of freedom, both give p = 0.00958682.
Both intervals describe hybrid minus remote. Negative values mean the remote group reports more. Their endpoints differ because the bootstrap reflects the observed skew.
Each resample preserves its group's size.
Mood's median test compares counts above and below the pooled median ($81,912). Its p-value is 1.396 × 10⁻²⁵. A centered bootstrap test of the median difference gives p = 0.000100; its 95% interval is −$16,242 to −$10,000 for hybrid minus remote.
The two median p-values are not numerically close because the tests use different statistics and the 10,000-replication bootstrap cannot resolve p below about 0.0001. Both support a difference. A small p-value is evidence against equal group means or medians under the test assumptions; it is not the probability that the null hypothesis is true.
This comparison is limited to survey participants in the United States and Canada.
Median $140,000
Median $157,000
Median $160,000
One-way ANOVA: F(2, 4360), p = 2.70 × 10⁻⁸. This rejects the hypothesis that all three population means are equal, but does not identify every differing pair.
A null-centered, 10,000-replication bootstrap ANOVA also rejects equality at the 5% level. Unequal salary spreads and extreme values warrant care, but Welch ANOVA and a 99th-percentile sensitivity check reach the same overall conclusion.
Each education group is resampled at its original size.
| Mean difference | Observed difference | Bootstrap 95% interval | Includes zero? |
|---|---|---|---|
| Bachelor − Master | −$21,232.87 | −$32,381 to −$11,464 | No |
| Bachelor − Professional | −$36,713.36 | −$61,362 to −$15,862 | No |
| Master − Professional | −$15,480.48 | −$41,228 to $7,307 | Yes |
These percentile intervals are exploratory and not adjusted for multiple comparisons. A formal claim about specific pairs would need a multiple-comparison procedure.
All findings use the supplied cleaned Stack Overflow 2025 survey file. Figures and numbers mirror the accompanying analysis notebook. No individual-level dataset is published here.
Large reported salaries create long right tails. The primary analysis retains positive values; 99th-percentile exclusions are sensitivity checks only.
The tests assume independent survey responses. The survey file does not establish a randomized sample or prove independence, and unequal variances are indicated in both comparisons.
Country, occupation, experience and other factors may affect salary. The analyses compare reported groups; they cannot establish the effect of changing work policy or education.