Actually, the study found that static languages were worse. They tested students on two problems. In one there was no significant difference, and in the other the static language performed worse.
Here is that study: "An Experiment About Static and Dynamic Type Systems", by Stefan Hanenberg, University of Duisberg-Essen, 2010.
However, they didn't use existing languages; they created a brand new language and IDE that had a typed and untyped variant. Whether this confounds or strengthens the case is unclear to me.
In any case, I think any survey of static type systems that doesn't include a language with type inference is very flawed. Java's type system is an annoyance in the small, and is a mixed bag in the large. Haskell's type system is a totally different story.
Here is that study: "An Experiment About Static and Dynamic Type Systems", by Stefan Hanenberg, University of Duisberg-Essen, 2010.
http://www.cs.washington.edu/education/courses/cse590n/10au/...
However, they didn't use existing languages; they created a brand new language and IDE that had a typed and untyped variant. Whether this confounds or strengthens the case is unclear to me.
In any case, I think any survey of static type systems that doesn't include a language with type inference is very flawed. Java's type system is an annoyance in the small, and is a mixed bag in the large. Haskell's type system is a totally different story.