A Level: What's the difference between 'reliability' and 'validity'?

This question slips up a lot of students as the two are easily mistaken for one another. We talk about Reliability and Validity when we are critiquing a research method. We consider them when looking at how a psychologist has created a test and evaluating how good the test is. In short, Reliability refers to consistency and Validity refers to accuracy. 'Consistency' and 'accuracy' are trigger words for examiners, use them correctly and your marker will be confident that you've understood.

When we discuss consistency we are looking at how well the test functions. If we are putting in the same input over and over again, we should get the same result. A good example is the Meyers-Briggs Personality Test. In case you are unfamiliar with the test, it claims to tell you what Personality Type you are by asking questions about your reactions, motivations, beliefs and behaviors. Your personality type is static, meaning it shouldn't change. If you took the test in the morning and again in the evening, your outcome should be the same. The test should be resilient to certain external factors like your mood or fatigue. If your outcomes differed, then we would say that the test is 'unreliable' because your results were inconsistent.

Another example is an IQ test. Traditionally, Psychologists have thought your IQ is static (though this is now widely debated). If you took an IQ test whilst stressed or distracted and then took the same test again whilst calm and focussed, your results should be the same or 'consistent'. If they are, then it's a reliable test!

Validity is actually the more straightforward of the two. Validity is asking; Does this test work? Does it measure what it set out to measure? We often talk about validity when we look at new Diagnostic tools. For example, if you're studying OCD you will probably have learnt about the Yale-Brown scale, a series of questions that determine whether or not the client has OCD. A simple way to test the validity of this scale is to test it on people you know do not have OCD and people who already have a diagnosis of OCD. Those without the diagnosis should return negative results. Those with diagnoses of OCD should return postive results. We allow small error margins but on the whole the test should be able to identify the majority correctly. If it does, then it's a valid test, it worked.