In a previous blog post, I wrote about the common misconception that a statistically significant correlation between variables means that one of those variables is causing the other. For example, having a best friend at work correlates highly with an overall measure of employee engagement (according to Gallup Inc.), but that doesn’t mean that if you have a best friend at work, you will feel engaged in your work. "Level 5 Leadership" correlates highly with successful companies (according to Jim Collins), but that doesn’t mean that if you have a "Level 5" leader, you will have a successful company. This doesn’t take away from the value of these studies, it just cautions against making a causal leap in logic.
I am reminded of another correlation that has serious repercussions for many college-bound teenagers. This is the finding that students who attend elite colleges are more likely to have highly successful careers than students who attend other colleges. Parents make the mental leap from this finding to the belief that their children must attend one of these prestigious institutions (e.g. Ivy League) in order to be successful. This is absurd. Many other factors could explain success (graduate school enrollment, income, job satisfaction, etc.) of a high percentage of students who attend these so-called “elite” schools. Maybe they are successful because they are highly motivated with a great deal of family support. Maybe they would be successful no matter where they went to school. And many students who attend these elite institutions do not achieve success, as defined by the investigators, while many others attend less prestigious institutions and become very successful in their fields. So, what might be true statistically for a large group might not be true for an individual. Some individuals will do well and be quite successful at less prestigious colleges and universities…and at a small fraction of the cost. (Full disclosure: I serve on the Board of a community college.)
Are you bothered by other correlations without causation? What are they?
