Don’t Be Fooled by Correlations

The HR world loves correlations. Every conference I go to, and every cool HR tool I see shows me some awesome statistic that connects whatever they are hawking to dramatic results. For example, I saw this in an infographic put out, not surprisingly, by an employee engagement software company: “Companies with higher employee engagement achieved 2.5x the revenues of competitors with lower engagement.”

That seems pretty cool, right? Like you should double down on your efforts to improve engagement. Imagine what you could do with all that extra revenue!! One problem: correlation is not causation

Let’s give them the benefit of the doubt and assume (a) they did actual statistical analysis behind this statement and (b) their measure of engagement is decent. Even so, in the end, all they have is a correlation between higher revenues and higher engagement. But we have NO IDEA if the higher engagement caused the organizations to be more successful in sales or revenue generation.

For all we know, the higher engagement they measured is actually the result of the higher revenues. Maybe we should completely ignore engagement, and take the money we were going to spend on their software and hire more salespeople instead.

There’s nothing wrong with correlations, but they are most useful as a guide for further research and developing sharper hypotheses, NOT for generating conclusions about what you should do. Don’t be fooled.

Oh, and having looked at all this correlation data, Maddie and I have developed a new hypothesis about engagement. It’s not caused by sales, we don’t think, but it is a result of people being successful (and NOT from having a best friend at work, either). More on this in our new book coming out this fall.

Let's Talk About Workplace Culture
%d bloggers like this: