The predictive validity of an assessment tool refers to that tool's ability to predict future behaviors or situations based on the results of the tool. For example, researchers have found high predictive validity for IQ tests when applied to future salary. Children with high IQs are more likely to be high earners when they become adults. Test creators that design an assessment with the ability to predict future situations in mind often focus mainly on designing a test with high predictive validity.
Convergent validity describes an assessment tool's relationship to similar assessments. If an assessment has a high convergent validity, then an individual scoring high on that assessment will likely often score high on assessments that evaluate a similar trait. For example, the psychologists Alfred Binet and John Raven separately designed tests of IQ. But even though their tests are structured differently, they both have high convergent validity. That is to say, if you score high on one of the tests, you will likely score high on the other. Likewise, if you score low on one, you will most likely score low on the other.
Surface validity is perhaps the easiest to assess. The name describes the test's appearance. An assessment with high surface validity will contain questions that appear to be testing the trait of interest. For example, in IQ testing, tests with high surface validity should contain questions related to logical reasoning or verbal skill but not questions related to personal preferences or political standing.
External validity is the hardest form of validity to analyze. It is the true ability of an assessment to measure what it is meant to measure. In many cases, external validity can only be estimated, not truly calculated. For example, IQ tests intend to measure an individual's ability to solve novel problems, but some argue that tests of verbal skill and logical reasoning are not equivalent to tests of the ability to solve novel problems. While most psychologists agree that IQ tests have high external validity, it is virtually impossible to establish this as fact. Thus, an IQ test can never have perfect external validity.