Using the scientific method to debug your test case failures

Step	Explanation
Observe a phenomenon	program crashed or produced output that wasn't what you expected
Ask a question	why did the program crash
Gather information & apply expert knowledge	get information regarding the inner workings of the program could rely on a domain expert's input
Form hypothesis	I think that the program did this when the incorrect output was produced
Test hypothesis	Is it testable? actionable? If your hypothesis was incorrect, go back to "Gather information"