Scrolling up, the same sample put NPM use at 65.17%, so I feel pretty comfortable calling it sample bias.

The data is no doubt biased, but given just how far ahead vscode is, it would need to be a very heavy bias to see vim/nvim come out ahead among all developers (however we define that). It gives us some information, but I agree doesn't tell us what the proportion is among all developers.