Let’s face it: Gender and age have long been the hallmarks behind most marketing and selling approaches. Those core demographic pieces of information are thought to predict much about our consumer preference and choice, as are other areas such as parental status and income. After all, the term “mom jeans” came from somewhere.
The overwhelming majority of consumer research studies include questions about such primary demographics. And while those types of questions aren’t going anywhere any time soon, there’s no denying that the utility of primary demographics is limited if you are trying to learn as much as possible about the segments you care about.
For example, when you meet someone at a party, you don’t say, “Hi, I’m 34 years old. I have a bachelor’s degree and earn $77,000 per year. I’m married and own a home in a suburban neighborhood.” You might mention some of that, but you’d also tell people about your interests and hobbies, the TV shows you’ve you watched recently, where you’ve traveled, and so on.
If you’re writing a survey, you might not have the time or budget to ask people about a few dozen attributes (or more) that are seemingly unrelated to your topic of interest. But if you could ask people something beyond primary demographics such as gender and age that could tell you a great deal about all the other attributes that define a survey respondent, what should you ask? This is the question we set out to answer.
So how did we do it, and what did we learn?
At CivicScience, looking for meaningful associations between different subsets of our respondents is unquestionably one of the core objectives of our business. At any given moment, there may be several thousand questions being asked across the hundreds of websites we partner with, and our ability to tie all of the responses we collect together at an individual level gives us a unique opportunity to learn how different segments of the population differentiate from each other at a much more granular level than would be possible with traditional survey methods.
For this study, we began with 608 questions from our syndicated question library, chosen to reflect the mix of attributes most valued by CivicScience customers for marketing research purposes. For each one, we built 607 contingency tables showing how our respondents answered all possible two-way combinations of questions, and then identified which questions provided us with the most information about how our respondents answered all other questions in the set.
What we found is quite interesting… for example, we found that the following attributes have the strongest correlations with all other questions in the analysis:
- The type of car someone drives
- Second screen behaviors
- Whether someone adjusts their lifestyle to help the environment
- Ownership of home gym equipment
- How much they like shopping at Family Dollar, the Vitamin Shoppe, or eating food from Domino’s
- and more…
What does this mean? By mining the large amounts of survey data we’ve collected, we believe these findings show how we are able to help clients better determine what attributes are more meaningful to research and utilize for marketing strategy development beyond demographics. And because of the novel approach we have for collecting a wide-reaching set of attributes (well beyond what could be achieved using other survey methods), we are able to discover statistically meaningful insights that are uniquely advantageous.
The bottom line: it means there is opportunity to define new “stereotypes” to better serve marketing.
We also have a complete research paper on this subject matter, which was featured in a session at last month’s Advertising Research Foundation (ARF) Re:Think conference.