It's not the most comprehensive data, but I would put more stock in this than in paid tweets from employees.
If they really wanted to address the issue in a transparent way, why not conduct some independent research and release results to the public? E.g. an anonymous survey of a representative sample of warehouse workers.
I agree that we should take paid tweets by employees with a large grain of salt, but I also want to point out that you likewise have essentially 0 evidence that they're not telling the truth, apart from random surveys and a few investigative reports, most of which are about warehouses halfway around the world.
I think it's import to remember that, like Amazon, journalists also like to appeal to a certain narrative (how often do they report about the amount of Good Amazon has done for the world?).
A pattern of a few (potentially based) data points is hardly a pattern.
https://www.theverge.com/2018/4/16/17243026/amazon-warehouse...
https://www.cnbc.com/2018/07/16/amazon-prime-day-threatened-...
https://www.theguardian.com/technology/2018/jul/30/accidents...
https://www.cbsnews.com/news/inside-an-amazon-warehouse-trea...