Validity

Validity refers to the extent to which something does what it is intended to do. From this broad perspective, validity equally applies to an object designed to perform certain tasks, to a program targeted to certain goals, or to an instrument intended to measure a given concept or construct. Therefore, it is not limited to the problem of measurement, although it is in this context that we will use it throughout this entry. The concept of validity applies to all measurement situations, but is particularly crucial to social epidemiology, where most research work has to deal with constructs that have to be operationalized. This entry contains a brief review of the concept with emphasis on its empirical and theoretical implications.

The constructs are variables that cannot be directly observed or measured. Quality of life, motivation, satisfaction, and socioeconomic status are typical [Page 1071]examples of constructs. Whenever a construct has to be operationalized, that is, defined in terms of concrete data that can be gathered or behaviors that can be observed, one is inevitably confronted with the problem of validity. Measuring a construct by means of an instrument, such as a questionnaire, always poses a problem of validity; for instance, we may try to measure someone's quality of life by asking him or her a series of questions about limitations, pain, and so on, but we must bear in mind that the answers to these questions are not a direct measure of their quality of life but at best an approximation of it. Another frequent example of a validity problem is statistical inference, where we have to estimate some parameters of a finite or infinite population by examining a sample of it. A sample is valid if it adequately reproduces the characteristics of the population that we want to study. Valid samples are usually said to be representative.

An instrument used to measure a construct can be a single indicator (e.g., income as a measurement of socioceonomic status) or can consist of a set of items (e.g., questionnaires designed to measure quality of life). Validity can be assessed on two grounds: theoretical and empirical.

The theoretical validation of an instrument implies a thorough examination of its contents with the purpose of verifying whether it reflects the meaning we have attached to the construct it is intended to measure. On the other hand, the empirical validation entails a careful testing of the properties that should correspond in practice to that meaning. Accordingly, two main dimensions are involved in the process of validation: the ontological and the methodological dimensions.

To theoretically verify that an instrument reflects the meaning of the construct, we must explicitly state what the construct is or what it means for us. We thereby assume an ontological position that is in practice equivalent to making a contextual definition. For instance, an instrument designed to measure quality of life can be validated neither on theoretical nor on practical grounds if we have not previously defined what quality of life is, or what it means for us. Obviously, the definition may vary from one context to another. For example, it may not be the same when applied to patients with cancer as when applied to healthy people, and it may also differ from one cultural setting to another. People with different views of quality of life will surely not agree on the pertinence of an instrument designed to measure it.

...