Why the Testing Method Developers Prefer Is Rarely Ever the One That Finds the Most Bugs

A replicated controlled study confirms that developers’ perceptions, preferences, and opinions about software testing techniques do not reliably predict actual defect detection effectiveness. While results largely align with the original experiment, the replication reveals persistent bias toward preferred techniques—particularly those perceived as easier to apply—despite weaker performance. The findings reinforce a core lesson of empirical software engineering: decisions about testing techniques should be guided by evidence, not intuition.