Defect Clustering And Pesticide Paradox In Software Testing

When we discuss principle of testing, Defect clustering & Pesticide Paradox are two important principles of testing which often called together.

Defect Clustering

In an application small number of modules contains maximum number of bugs or shows most operational failures

Once defect clustering is identified, testers will focus on these known areas that may also be called as Hotspots, same information is helpful for development team also as they can emphasis more on major affected modules.

Defect clustering cases are often include in Smoke testing which ensures major complex conditions are verified. Defect clustering analysis is often done after first round on test by Project manager or Test manager with the help of defect sheet. Defect sheet should have information of Module for all reported defect. Manager can take pivot of Module Vs Defect count and easily analysis occurrence of Defect clustering

Moreover it is based on the Pareto principle, also known as the 80-20 rule, where 80% of issues are from 20% of modules. 

Importance of Defect Clustering 

• QC can focus on the same modules/functionality in order to find more number of available bugs
• Time is saved thus cost for finding defects

Below is the sample example which shows identified modules (highlighted) of Defect clustering.

Module	Count of issues	Percentage %
Master 1	4	2.07253886
Master 2	8	4.14507772
Master 3	28	14.50777202
Master 4	5	2.590673575
Master 5	9	4.663212435
Transac 1	24	12.43523316
Transac 2	25	12.95336788
Transac 3	6	3.10880829
Transac 4	33	17.0984456
Report 1	22	11.39896373
Report 2	8	4.14507772
Report 3	21	10.88082902
Total	193

Pesticide Paradox

Pesticide paradox refers to when same set of test is executed again & again then eventually the same test case would no longer find new bugs.

After 2 or 3 rounds of testing, count of new defects start dropping as most of the bugs got fixed by cleaning ‘Hot Spot’ area. Development team put more focus & become extra careful in those module where testers found more defect and might overlook in other places 

Thus by executing same set of test cases would no longer gives more defect. The test case should be updated for new & different type of test in such a way that it will produce more defects

Methods to Avoid pesticide paradox

• Preparing new set of test cases & updating it tot already available test case
• Developing new set of tests case to perform test on different function/module of software 

By using above methods testers can find more effective defects in the modules which are not been focused by testers earlier or modules which are not taken extra care as testers are focusing in defect clustering modules

The first method is time consuming, as number of cases will be too large thus increase the testing time and in turn will increase the cost of testing.

To manage this testers should identify & remove test cases which are of less importance & are of low priority.

Testers can identify & remove useless test cases without compromising with the quality as all the test cases covering important areas of the software will be retained.

Bottom Line:

Defect cluster is useful during initial round of testing but it is always good to use one of the Pesticide Paradox method

Although created test case has maximum coverage and average rate of finding defects is high but we still need to keep updating, reviewing test cases on regularly basis.

About Author:

Vinayak Jadhav is a consultant in Systems Plus Pvt. Ltd. Within Systems Plus, he actively contributes to the areas of Technology and Information Security. He can be contacted at:  vinayak.j@spluspl.com