The final term we introduce in this article is theory. They describe a set of systematically interrelated concepts or hypothesis that purports to explain and predict phenomenon. In OB, theories are also frequently referred to as models. We use the two terms interchangeably. There are no shortages of theories in OB. For instance we have theories to describe what motivates people, the most effective leadership styles and the best way to resolve conflicts and how people acquire power. In some cases, we have half a dozen or more separate theories that purport to explain and predict a given phenomenon. In such a case, is one right and the others wrong? No! They tend to reflect science at work – researchers testing pervious theories, modifying them, and when appropriate proposing new models that may prove to have higher explanatory and predictive powers. Multiple theories attempting to explain common phenomenon merely attest that OB is an active discipline still growing and evolving.
A researcher might survey a group of employees to determine the satisfaction of each of them with his or her job. Then, using company absenteeism reports, the researchers could correlate the job satisfaction scores against individual attendance records to determine whether employees who are more satisfied with their jobs have better attendance records than their counterparts who indicated lower job satisfaction. Let us suppose the researcher found a correlation coefficient of +0.50 between satisfaction and attendance. Would that be a strong association? There is, unfortunately no precise numerical cutoff separating strong and weak relationship. A standard statistical test would need to be applied to determine whether the relationship was a significant one.
A point needs to be made before we move on: A correlation coefficient measures only the strength of association between two variables. A high value does not imply causality. The length of women’s skirts and stock market prices, for instance have long been noted to be highly correlated but one should be careful not to infer that a causal relationship between the two exists. In this instance, the high correlation is more happenstance than predictive.
As a potential consumer of behavioral science you should follow the dictum of caveat “let buyer beware”! In evaluating any research study, you need to ask three questions.
Is the study actually measuring what it claims to be measuring? A number of psychological tests have been discarded by employers in recent years because they have not been found to be valid measures of the applicants ability to do a given job successfully. But the validity issue is relevant to all research studies. So, if you find a study that links cohesive work teams with higher productivity you want to know how each of these variables was measured and whether it is actually measuring what it is supposed to be measuring.
Reliability refers to consistency of measurement. If you were to have height measured every day with a wooden yardstick, you would get highly reliable results. On the other hand if you were measured each day by an elastic tape measure, there would probably be considerable disparity between your height measurements from one day to the next. Your height, of course does not change from day to day. The variability is due to the unreliability of the measuring device. So if a company asked a group of its employees to complete a reliable job satisfaction questionnaire six months later, we would expect the results be very similar provided nothing changed in the interim that might significantly affect employee satisfaction.
Are the results of the research study generalizable to groups of individuals other than those who participated in the original study? Be aware, for example, of the limitations that might exist in research that uses college students as subjects. Are the findings in such studies generalization to full-time employees in real jobs?