It's time we sit down and have a serious chat.
Almost all your stats and claims are bullshit.
Yes, it's rant time.
Yes, I'm using the patronizing filter, sorry.
My patient filter is all out of batteries.
They're not destroying companies at a horrifying rate, or any rate we can calculate.
Businesses are going to fail because there are already 100 other cupcake shops in the same neighborhood.
We're talking about a just a handful of individuals responsible for all these "statistics".
csoonline.com/article/315370…
- Cite your sources
- Publish source data whenever possible, or at least make it available upon request
- Put data in the correct context
- State your sample size. If you don't want to because it's n=12, maybe you shouldn't be using it.
- Make sure you've interpreted the stats correctly! When in doubt, ask an expert - many are willing to check your work for free.
- Most importantly, validate before sharing or publishing! There's no CTRL+Z for the Internet. I've been guilty of this myself - watching my own bullshit tweet go viral, as if just to mock my shitty mistake.
Don't do it for me, do it for our data scientists. Make it so that they don't have to turn to drink whenever they see bad stats. /END