Causal Analysis – Does X cause Y

Causal Analysis

As a market researcher, I am always curious about the impact of marketing activities and purchase behavior. In short I wonder if our activities cause people to purchase our products and services. The notion of causal analysis goes back some time in the area of social, economic and business research. Our colleagues in the physical and life sciences have a much easier time asserting that X causes Y. Such is not the case in the business world.

With that said the advent of web analytics and marketing automation has moved us closer to being able to say, with the conviction that spending a dollar on one activity causes consumers to respond in a predictable way. Yet as my early mentor Dr. Nabil Razzouk would say, “at best we can predict some of the behavior some of the time.”

Most survey projects in consumer or B2B marketing research do not involve experimental design, although there is ample room to do so. Instead, we are often faced with the exercise of assessing the degree of association between variables. This is done through statistical significance testing. For example, the chart below illustrates the relationship between job satisfaction and job hunting. It is highly significant (p = .000) and would appear that satisfaction (or lack thereof) in one’s job may lead to a job search.

Causal Analysis

Yet, we are unable to say with certainty that low levels of job satisfaction cause a person to seek new employment. To make this definitive statement we would need to conduct experiments where we altered people’s job satisfaction levels and assessed changes in job seeking behavior. This would be a very complicated process.

However, what we can say is that there is a significant association between levels of employee satisfaction and an employee’s willingness to search for new employment. Expanding on this, we can say that employees who are dissatisfied with their work are more than twice as likely to be conducting a job search (80% vs. 36%) than those who are presently satisfied.

Remember association is not causation. However, do not let this fact stop you from developing and implementing marketing programs designed to encourage specific behaviors. Causal analysis requires an experimental mindset. If we can employ this technique effectively only then can we make specific assertions about the impact of our marketing programs on key behaviors.