The study of types is generally done as part of mathematics, not sciences. There are some attempts to study type system impact in real world programs, and ergonomics, which would fall under science, but these are a very small part of PLT.

However, the point of the article was different - the article was claiming that when you are writing the actual code for your program, you are normally using engineering-style reasoning (abduction), whereas when you are defining the hierarchy of types for your system, you are mainly applying scientific-style reasoning (induction).

Personally, I think that the article is a gross over-simplification of every single term - of deduction, abduction, and induction; of maths, science, and engineering; and of what programmers actually do. It is in fact so over-simplified that I don't see any valuable insight into the article at all. It almlst reads like a description of programming from a TV show that is trying to sound brainy.