Correlation Is Not Causation
Interpreting Social Data: Delight, don’t aggravate consumers with your ‘insights’
Posted by: Pete Holley, TBG’s Head of Research
As the data revolution takes hold, marketers are getting more interested in concepts like correlation in order to interpret consumer buying behaviour. However, as Pete Holley, TBG’s Head of Research, explains, correlation and causality is not the same thing. Marketers need to be careful when drawing inferences from Social Data.
Correlation is a measure of the association between two (or more) things. It’s quantified using correlation co-efficients. A correlation of 1 means there’s a perfect straight-line relationship between your two variables, so that if one increases, the other increases by the same amount:
“Ultimately, the only way to predict future events is to study how known past events manifested. The grey area between correlation and causation may not matter as much if you can prove that event A is usually followed by event B. You will never be right 100 percent of the time, but then again, what predictions are? The problem with Social Data is also a problem for “big data”: A lack of context can easily skew results. Thoughtful analysis can tell us a lot, but only if context asks the right questions. Brands that do this well will delight consumers, but those that get it wrong will aggravate them. For everyone’s sake, it’s critical to stay on the right side of the Delight-Aggravate continuum”