Experimental models for software diagnosis

Experimentation and data collection are becoming accepted practices within the software en-gineering community to determine the e ectiveness of various software development practices.
However, there is wide disagreement as to exactly what the term \experimentation" means in
this domain. It is important that we understand this concept and identify how we can best
collect data needed to validate software methods that are e ective. This understanding will
provide a basis for improved technical exchange of information between scientists and engineers
within the software community.