1

General definitions

On one hand, I've been working in embedded systems, specially in aeronautics (were DO178 applies to software) and i'm used to these definitions:

  • validation: check with test, inspection, analysis, demonstration, that the final product/service, fulfill the customer requirements. "Are we building the product right?"
  • verification: check that each process is done correctly. e.g.: all the customer requirements are covered by a test, inspection, analysis or demonstration. Your product could be faulty, but your documentation is correct. "Are we building the right product?"

Is my interpretation correct?

On the other hand, I've seen in the application world (like in ISTQB certification) that:

  • validation has the still the same meaning (but restrictive to a complete product)
  • verification is also used to describe a validation process, but not for a full product/system: for sub-systems. Is my understanding correct? Is it a common practice?

In the context

In our company we are making appliances, and in our team/department, we are making sub-system for those appliance (GUI). Some colleagues of mine are using the "verification test" term to describe the activity of functional test on our sub-system, considering we are testing a part of the appliance. They say the validation tests cannot be done as our sub-system is not the final product.

In other words, they are saying that, because the customer ask for an appliance, they are not considering themselves and the other members of the department as customers for the sub-system. Therefore, the activity of functional testing is not a activity of validation, because does not correspond to the definition "Checking that the product does what the clients asked for".

Here comes my question: Is there any "TRUE" standard definition of verification and validation? How can I make myself understood if everybody use a same word to describe 2 distinct activities?

Note: a first question was asked here but was not referring to any documentation and thus subject to opinion-based answers

Guillaume D
  • 2,202
  • 2
  • 10
  • 37

2 Answers2

2

Slightly opinion based, still, but my familiarity with it is as follows:

Verification - is about demonstrating that you have correctly and completely done what you set out to do or agreed (in the contract usually). So for example, correctly transposing & applying design values, correctly applying design methods etc. It can be objective & quantitative, e.g based on:

  • testing by field trials, pilot schemes etc
  • independent (3rd party) verification of the design, calculations & assumptions etc
  • design reviews
  • repetition using alternate methods to increase confidence of results
  • comparison with similar proven designs

Validation - is about demonstrating that what you set out to do is or was reasonable, valid & appropriate, in the context of end-use. Fitness for purpose. Its more subjective, but can be argued in court by pointing to:

  • usage, custom & long-term satisfaction
  • technical judgement based on analysis
  • testing in real world applications

The above is from BS 7000 : Part 2 (1997) concerning manufactured products.

FrugalTPH
  • 533
  • 2
  • 6
  • 25
1

"Verification" and "validation" are similar sounding words which means the same thing in common parlance. To minimize ambiguity and confusion, it is best to look for clarification when listening or reading, and never to rely on a difference between the words when speaking or writing yourself. Talk about "testing against requirements" versus "end-user testing", so nobody mixes them up.

To make matters worse, the two terms have similar but different distinctions in the world of scientific computing. One of them means "auditing code against math/theory" while the other means "testing against experimental data". Again, just say what you mean, and you'll sound less fancy but far clearer.

Obviously, it looks a little silly if you ask "do you really mean validation/verification and not the other?", though I wish it were that easy. To save face, just discuss the specifics, and you'll get far more nuance than trying to differentiate those two V words ever will.

hegel5000
  • 876
  • 1
  • 7
  • 13