Average/mean is disproportionately influenced by a small number of highly paid employees.
In this case, the article is clearly demonstrating that despite Amazon's commanding market share, incredible wealth, and the largess bestowed upon their executives, some of their workforce is so poorly paid that they may not even make a living wage.
I would disagree with that. Imagine this dataset of salaries:
[1, 1, 1, 1, 1, 10, 10, 10, 10, 10, 10]
You have around half employees being paid like shit (1) and half being paid well (10), but the median will show a good salary (10).
>In Seattle, Amazon’s more than 45,000 employees are paid an average of more than $110,000, according to an analysis of individual worker data posted to job review site Glassdoor. The median, or midpoint, of Amazon’s 566,000 individual employee salaries worldwide stood far lower.
Granted, we are using average data from Glassdoor, whereas Amazon is self-reporting the median salary information. Plus, Amazon has part-time workers whose wages it annualizes to include the dataset. This makes me wonder though, are they counting the benefits that those workers are not receiving due to their lower number of hours?