Particularly if you have some issues with the method that TIOBE uses.
Not sure they address the problem as far as PowerBuilder is concerned though. For example, the first method measures data from StackOverlow and GitHub. The first makes sense. I doubt there are many if any PowerBuilder projects on GitHub though, which doesn't mean the language isn't used, it just means it isn't used with GitHub.
Similar issues with one of the sources that one of the commenters on the article makes. They recommend looking at data from itjobswatch.co.uk, at least as far as popularity within the UK.
You won't see PowerBuilder on the list, but if you search for PowerBuilder separately you'll get hits. Apparently the site doesn't categorize PowerBuilder as a language. That's actually fair, it's a tool, not a language. But it demonstrates the problem using most of these statistics to try to determine where PowerBuilder fits in.