Developer Testing in the IDE: Patterns, Beliefs, and Behavior

While most of this fits roughly with my own observations, the following caught my attention:

> only a quarter of test cases is responsible for three quarters of all test failures

This rate is totally off my personal experience.

25% of test cases lead to 75% of test failures?

I'm used to have < 5% of the tests leading to > 95% of the test failures. That is, only few of the tests are really important in hindsight. The only problem is that nobody know which ones are the 5%, so better be safe than sorry.