Perhaps nothing separates qualitative from quantitative research more than this single word – validity. But as researchers know, there are many forms of validity just as there are ‘many forms of truth’ (John Stuart Mill, philosopher).

Clients will often ask, ‘Are these research results valid?’ There is no single answer to that question since validity of results can be assessed in a number of different ways. Some levels of validity are easier to achieve and others are more complex and difficult to satisfy. The following are the types of validity that are most relevant for marketing research.

Predictive Validity

Also known as internal validity, this is the extent to which a set of attributes measured are responsible for changes in the dependent variable they are intended to describe.

Statistical Validity

Relates to the correct use of statistics and should include examination of the following important questions. Are the measures being used statistically reliable? Are small sample sizes responsible for the inability to conclude that a relationship exists? In significance testing, is the correct power being used? Has the causal direction of the relationship been correctly identified? Have interaction effects or non-linear effects been taken into account?

Construct Validity

Involves assessing the degree to which the items being used to measure a construct actually do measure that construct. Both convergent and discriminant validity are types of construct validity. Convergent validity relates to the expectation that the items being used to measure a given construct, should be correlated amongst themselves. Discriminant validity relates to the expectation that the items measuring different constructs, should not be highly correlated with each other. In marketing research, a commonly used rule-of-thumb here is that the correlations measuring convergent validity should be higher than the correlations measuring discriminant validity.

Content Validity

This is also known as external validity and has to do with the requirement that the information gained from the sample data is able to be generalized back to, and be representative of, the population of interest.

Face Validity

Relates to the expectation that the research findings make common sense and are intuitive from a business perspective. It is this type of validity that management is most familiar with and concerned about. Yet from an academic viewpoint, its importance to management is grossly under-estimated. Managers are rarely able to judge predictive validity and most often assess the value of the research solely on its face validity.

Therefore, as practitioners we should strive to routinely validate results using the higher order forms of validity discussed herein, whilst ensuring at the same time, that face validity is achieved.