The overall average hides 7 different numbers

country splits data-job-postings into 7 groups. Here is what salary_min_usd looks like inside each.

• US — mean 120,039, median 118,500 (1,579 rows)
• GB — mean 64,365, median 69,000 (436 rows)
• NL — mean 62,590, median 69,500 (245 rows)
• CA — mean 59,213, median 67,000 (247 rows)
• DE — mean 58,829, median 65,500 (333 rows)
• BR — mean 30,650, median 34,000 (243 rows)

Top to bottom that is 120,039 against 25,038, a spread of 379.4%. The pooled average is 104,106.

select country,
  count(*)                                                    as rows,
  round(avg(salary_min_usd)::numeric, 2)                      as mean,
  percentile_cont(0.5) within group (order by salary_min_usd) as median
from data_job_postings
group by 1
order by mean desc;

A spread that wide means the pooled number is not a summary, it is an artefact of the mix. Change the proportion of US rows and the overall average moves without any individual group changing at all — which is how a metric goes up while every segment goes down.

Notice the mean and median columns disagree only slightly here. Always compute both in the group-by. The comparison between them per segment is free and tells you whether you are looking at a level difference or a tail difference.

This is the setup for Simpson's paradox — the case where every segment moves one way and the total moves the other.