Given how much more popular automated testing and TDD have become in the past 7 years since that book was published (and much longer for some of the sources cited for those data), I'd be very surprised if we haven't gotten much better at that particular form of testing.
Also I'm quite curious how you can actually count the number of bugs found from each of these methods. Test-after unit testing would even be preventative if you wrote tests immediately after writing the code. If doing that made you think about more edge cases and thus fix up the code then it seems like those bugs would not get counted.
Seems like many of these would have different amounts of a preventative effect that would be difficult to measure....especially if you wanted to isolate their effect from each other.
As I understand the paper, the researchers tried both asking programmers to use certain techniques to debug a program with known bugs, then measured the number of bugs they caught against the known total.
In addition they went to larger companies, checked which bug detection techniques they were using and how well each one detected bugs.
I have the gated PDF, email me if you want a copy.
Also I'm quite curious how you can actually count the number of bugs found from each of these methods. Test-after unit testing would even be preventative if you wrote tests immediately after writing the code. If doing that made you think about more edge cases and thus fix up the code then it seems like those bugs would not get counted.
Seems like many of these would have different amounts of a preventative effect that would be difficult to measure....especially if you wanted to isolate their effect from each other.