Why Your Quant Metrics Are Lying To You

This morning I fired up my browser and there on the Nielsen internal home page was a gleaming new post declaring Most Households Read Food Labels by our own Todd Hale, SVP Consumer and Shopper Insights.  For those who know me, they would anticipate my immediate reaction to such a statement: an almost visceral desire to holler, “That’s BS!”  Sometimes I accidentally say these things out loud and then I have to sheepishly apologize to those around me for my explosive behavior.

Please don’t misunderstand, I respect Todd highly and he offers an accurate analysis of what Nielsen’s Homescan panelists (125,000 demographically diverse households spread across the United States), answered in response to a 2008 survey asking the primary shopper about their tendencies for reading labels on food and beverage packages.

Label Reader

But if you relied solely on the data that in summary says 61% of households agree completely or agree somewhat that they read product labels you would be drawing very faulty conclusions as to how readily consumers are responding to what is printed on the product labels.  And such conclusions can lead to bad policy decisions among government and health regulators, bad ingredient decisions by food scientists, and bad marketing decisions by brand managers.

Well why is that you might ask?  Two reasons come to mind:

Quantitative data offers only a narrow slice of understanding consumers’ actions, attitudes, and beliefs

Quantitative data is useful in defining and tracking trends in behavior.  But you cannot get to the truth of what someone actually does unless you actually go out and observe them in the environment where they engage in this activity.  You see, it’s not even good enough to sit down with consumers in a focus group or one on one interview and ask them what they do.  In the case of consumers reading labels, you must go out and visit grocery stores and observe from afar to determine whether people are actually reading labels.  Then if you approach them you can find out what they are looking for on the label and why they read it.  The ethnographic approach complements quantitative data by providing definitive proof of behavior and rounds it out with cultural and personal drivers.

 A combination of confirmation bias and the Heisenberg Uncertainty Principle impacts truthful self reporting

Maybe you realize it, but the truth is, we lie all of the time when our behaviors do not match what we know “the experts” or “society” or even we ourselves have come to believe is the “appropriate behavior.”  If you don’t floss on a regular basis and the dentist confronts you with the results of this by saying “your gums are a little inflamed, are you flossing regularly?” how likely are you to admit, no, I never floss even though I know it’s good for me?  So think about those individuals who reported that they agree completely or agree somewhat that they read product labels.  Some of them probably absolutely do read the labels, especially if they or someone in the home have a restricted diet due to food allergies, health conditions, or less likely, are on a diet.

However, my experiences as a new product development consultant where I engaged with hundreds of consumers and visited dozens of stores demonstrated that most consumers move through the store in a mindless mode of filling their carts with the products and brands they know and only rarely stop to examine the label.  Nielsen’s own (warning PDF) DeltaQual research (warning PDF) shows that once these habits or Omega Rules are formed, only a certain type of trigger might cause a consumer to stop and scrutinize the brand or label more closely.  This differs of course by category where shopper modality adjusts accordingly but the average shopping list from week to week is probably 80% identical and probably 99% identical across an 8 week period.  I described one such delta moment scenario in my last post concerning the not-from-concentrate orange juice “lie” that ruptured the Breillatt household’s perception of what constitutes “fresh squeezed orange juice in a box.”

In reality, consumers may not even realize the lie to the “truth” that they claim defines their behavior.  It is not unusual when visiting a home and asking one member of the family questions about their behavior to have a son or daughter or spouse pipe in with, “No way, that’s not what you REALLY do.” You see, as the Heisenberg Uncertainty Principle (as misused by social scientists) states, “the very act of observing a phenomenon inevitably alters that phenomenon in some way.”  Consumers may believe they’re checking labels because they’ve heard how important it is to be aware of what they’re eating what with the food pyramid from the USDA, the constantly improving Nutrition Facts, and dietary recommendations from pediatricians, Dr. Oz, Dr. Gupta, and every other guru they encounter through media. They remember that of course they checked out the label of many of their foods when the big uproar came up around trans fats or high fructose corn syrup or whatever the latest nutritional crisis might be.  Of course, they forget that they did that once during three weeks’ worth of shopping 2 years ago until they were caught up again by the latest scandal to grip TMZ and the gossip rags at the checkout line.

Allow me to expand on that thought with a recent and relevant example – especially for you social media researchers who are trying to figure out how to leverage the data such sites like Twitter and Facebook offer.  Just a few days ago, the Nielsen funded Council for Research Excellence released the results of the Video Consumer Mapping Study which identified that:

The amount of time Americans report they spend watching online video has been, on average, grossly overstated by conventional forms of media research and audience measurement …conversely, traditional TV viewing has been pretty drastically under-reported” by research that asks people how they consume video.  The reason why, is that research based on how people perceive they consume media isn’t nearly as accurate as research that actually observes how they use it.

The ad industry historically has known about such “halo effects” and the fact that it is considered socially unpopular for people to report that they watch as much TV as they actually do. On the other hand, people tend to over-report their online and mobile video consumption, because “it is new and cool.”

As Arsenio Hall used to say, “Things that make you go hmm.”  Social biases have a real impact on how survey participants perceive themselves and therefore how they report their activities.

So my recommendation is, if you’re going to rely upon a statistic as the foundation for taking an important action to your product, make sure you have a good interpretative lens based on true consumer / user behavior to provide the full color for interpreting the data.