Last week we looked at the concept of reliability, or accuracy of measurement. Closely associated with reliability is the concept of validity of measurement.
The ways in which validity can be assessed have been described in the article about testing. Briefly, a measurement is valid if it agrees with other measures known to be valid (that is, it has concurrent validity), if it predicts future performance (predictive validity), or if it behaves in ways predicted by theory (construct validity). An unreliable test cannot of course be valid.
The point I would like to make here is that validity of measurement is often ignored, especially in testing. For example, one of the difficulties in assessing the effectiveness of programs of adult basic education has been the lack of a reliable or valid testing instrument. Ehringhaus (1991), for example, found that unreliable tests were used by almost all of the 427 teachers of adult basic education she surveyed.
When you ignore validity, you are asking for trouble. If you are using tests to assess the effectiveness of a program, for example, using an invalid test will tell you nothing about the success of the program. You might as well estimate the temperature with a sundial. If you use an invalid test for placement, you will be no more successful in matching people to jobs or training programs than if you had picked names out of a hat.
The same reservations apply to measures we develop ourselves. If you are using a questionnaire to assess employee morale, for example, you will have a much better idea of the utility of the questionnaire if you see how well its measures of morale correlate with things like staff turnover or absenteeism. If it doesn't correlate with them, you have no reason to assume it is a measure of morale.
On the other hand, if it does correlate, you now have a better measure of morale than those other measures. An account of how a measure like this can be validated is provided in another article.
Ehringhaus, Carolyn (1991). Testing in adult basic education. Adult Basic Education, 1 (1), 12-26.
Validity © 2001, John FitzGerald
Home page | Decisionmakers' index | E-mail