In an effort to make our client's jobs easier we have built a real time driver analytics engine in Murmur - this allows them to quickly see what aspects of organisational culture have the biggest impact on staff engagement. If you're looking to act on your culture, "driver analysis" is a crucial tool in selecting the significant and most cost effective means.
But what exactly is a driver analysis?
I like to think that I am OK at explaining things simply - having taught statistics at university for a number of years with some modicum of success (as far as student feedback goes anyway). However, in the context of an employee survey, there are a few common difficulties that make a short answer difficult. Perhaps if we sort out these points of confusion first we can arrive at a clearer conclusion.
Confusion Point 1: Driver analysis does not refer to one specific statistical technique or test
As I footnoted in a previous post, I think some consultant somewhere just made this term up. The term applies to a range of statistical methods for assessing relationships; in employee surveys this usually means the relationship various questions (or groups of questions) have with the engagement index being used in your survey. The idea being that this will tell you what 'drives' engagement.
Confusion Point 2: Different people and organisations use different types of driver analyses.
As mentioned above, since the term applies to different techniques, there are a range of techniques being used. You might hear terms such as regression (linear, logistic, non-linear, etc.), impact analysis, relative importance analysis, shapley values, and correlation (Pearson, Spearman's, etc.) - as well s some other colourful terms.
Confusion Point 3: Causality is not directly assessible via driver analysis
Driver analyses can tell you which things are most related to engagement. So if you look at the link between job satisfaction and engagement you will find a big link. But, this is not much help; telling an executive they need to improve people's job satisfaction in order to 'drive' engagement is silly right? Driver analysis has to be done using sensible inputs into the analysis. In Murmur we don't use "non-actionable" items in our driver analyses.
Engagement driver analysis refers to a range of statistical techniques that tell you which things are most related to engagement. If the top driver of engagement is a learning and development question, it simply means that people who respond most positively to that question are also likely to be the most engaged. If you act to improve responses to that question then you have an improved chance of making your employees more engaged.
However, that's not the end of the story. Murmur is designed to allow you to track the change that comes from your actions - ultimately this is the most important thing.
So what type of driver analysis should you use? More on that next time.