The answer is very simple: the programmers had an average of 7.7 years of total programming experience, not 7.7 years of programming experience in Java.
Prechelt reports that "On average, the Java programmers had only half as much programming experience in Java as the C programmers had in C or the C++ programmers had in C++." How is this possible?
I'm not sure where this statement comes from. The raw data from the study provided to me by Prechelt did not contain any langauge-specific experience data. The only langauge-specific data was the number of lines of code written in the langauge in question, but this is not necessarily a valid measure of experience because in general a Java program will be shorter than an equivalent C or C++ program (and a Lisp program will be shorter still). The only way to really answer this question is to ask Prechelt. If someone does, please let me know what he said.