It is always interesting to see great minds taking an interest in new technology, then dictating a strategy or coming to a conclusion that we all previously accepted with older technology. Perhaps it is a testament to the fact that repetition helps improve mastery. Or, we continue to ignore our past learnings, so we repeat what we all should have known before.
There could be more reasons.
A report done by scientists from McGill University and Carnegie Mellon University suggests that marketers, brands, and social scientists that rely on social media to pull data in order to examine human behavior online and offline must be sure that they are not misleading the people they work for.
Why? Much of the data pulled from certain sites could be unrepresentative of the population, leading to poor conclusions and woeful recommendations.
You know, an element in market research called bias. An elementary element that those conducting these research studies should know.
But the thing is, the studies these scientists have seen did not pass the bias test; in fact, several of them failed miserably. The synopsis of the study tells of marketers using reports from Pinterest, which is dominated by 25–34 year-old women. Another example they used is that they found the accuracy of using Twitter for some political research to be at 65% rather than the 90% some reports are touting.
Scary, right? Want to go back through your reports to make sure the sites you cited had more information? Yes, us too.
The best research is done when every potential bias is removed or taken into account. The best decisions come from research that was well-prepared and vetted. The best campaigns are grounded by good business decisions, and customers come to buy products and services because the marketing behind the interaction was focused, targeted, and accurate.
But you all know this by now. Decades ago, researchers relied on catching people outside their homes because people didn't like being bothered at home. But the bias became that researchers would only get the people with busy schedules, so the homebody or driver missed out. Even in modern times, data gathered on telephone polls has been inaccurate because of the Do Not Call list, and the prevalence of having a cell phone and no landline.
Why wouldn't the same bias test be given to social media?
Easier said than done, we guess.
Though it is unfortunate that a study like this needed to be written, we are sure glad it was.
Dwayne W. Waite Jr. is partner and principal at JDW: The Charlotte Agency, a marketing and advertising shop in Charlotte, NC. He enjoys consumer behavior, economics, and football.