So what can the power of randomness mean for small businesses?
Warning – This article begins in a rather “wonkish” fashion to discuss what might seem an arcane point about survey research, but I hope you’ll find it interesting anyway, and I promise to bring it back to relevance to ordinary small business owners. I’ll skip the statistical stuff.
As a long-standing research professional, I am often shocked to encounter people who don’t understand the power that is inherent in the concept of randomness.
Just recently I came across a potential client who was absolutely outraged that I would suggest that sampling for a project should be random within given population. He wanted everything to be set in quotas, for everything to be set in advance. To him the word random was synonymous with sloppy.
Much of this prejudice comes from dictionaries that will define the concept in seemingly negative terms such as from Merriam-Webster:
Definition of Random:
A haphazard course
— at random:
Without definite aim, direction, rule, or method
Subjects chosen at random
Haphazard is such an ugly word. But in the right context it is fun – walking randomly around a city to get a sense of what it is like without a set plan can be marvelous: San Francisco, Paris and Manhattan come to mind.
But in the context of research, randomness is the goose that laid the golden egg. Because without being able to apply random sampling, the industry wouldn’t exist.
To Quota or Not to Quota:
But to go back to this client’s erroneous point, he misunderstood an important point. In research we often use quotas for specific purposes based on either known factors or important objectives of the research.
Applying quotas to targets can be crucial – you may want the study to be comprised of as many men as women because you know that as many men as women buy your product.
You may need to read the population by age because you have observed that your customers tend to be older and you want to better understand younger consumers. So you establish quotas to make sure that other factors don’t mess up your ability to derive from the analysis what you need --– men and younger people are generally harder to reach than women and older people (and business owners may be the toughest of all).
But your targets are strategic and generally based on known facts.
These known facts are often based on data that is collected randomly. If you look at how a company like Nielsen would establish quotas for specific demographic groups for placing their TV people meters, they do so based on very thorough enumeration studies that are based on random sampling.
What this specific client wanted to do was control a key variable that would skew the data.
There’s the rub. If you want to bias a study away from what is out there in the population you are studying towards what is inside your head, you can do that.
If you know what you want people to say, then you carefully control whom you ask.
However, if you want to know what is really going on out there in whatever sampling frame you have established for yourself such as your service area or people who are engaged in a category (such as new car purchase intenders, users of body lotion, families with children in diapers, etc.), you ultimately have to rely on random sampling.
Here is a formal definition of random sampling:
A simple random sample is a subset of a statistical population in which each member of the subset has an equal probability of being chosen. An example of a simple random sample would be the names of 25 employees being chosen out of a hat from a company of 250 employees.
This basic concept is how pollsters can talk to 1,200 people in the US out of 350 Million and be able to draw conclusions about people’s positions on policy, attitudes, beliefs and candidates.
Obviously there are complications – they often look for “likely voters” (which is a debate unto itself) and may establish regional quotas so they can be sure they cover the entire country.
Also, they may need to apply weighting to re-balance the sample against known variables. For example, if they collect 40% of their data from a group that they know is only 33% of the population, they would apply “weights” to scale down that data and scale up the data from the group that was under-represented.
Relevance to Small Businesses
So what can the power of randomness mean for small businesses?
It can help you develop a culture of research – that is, building a tool set that will help you run your business better much as a new accounting system or inventory control might do.
For example, evaluating your customer satisfaction to see what things you are doing well and what things could be improved. To do this, you can’t simply ask those who look nice and are smiling at you, you need to apply a random selector process whereby every fourth or seventh or twentieth customer gets asked to do a quick survey.
If you have a new product or service line and you want to see how people might react to it before making a big investment, then you would need to test the concept with your customers.
Even if not everyone will do it (they won’t), by applying a random selector you build in randomness into the design that would make the data more reliable than a cherry-picked sample. This could save you a lot of time and help you maximize your profit potential by using objective input you can trust.
This shouldn’t be about getting the answers you WANT to hear, it’s about getting the answers you NEED to hear.
Obviously the topic of professional consumer research is much more complex and detailed than what is listed above. I just wanted to touch on the importance of randomness so you can grasp its’ potential impact on your business.
I hope this helps.
About the author
Carlos E. Garcia, a Latin Business Today, Partner: Research, Trends, Insights was born to Mexican immigrant parents, grew up in East Los Angeles and attended Pomona College, UC Berkeley and National University (BA, MA and MBA respectively). He has over thirty years of experience in the field of US Hispanic consumer research, twenty one years at the helm of his own company, Garcia Research. Most recently SVP at GfK: Knowledge Networks, where he headed up their Hispanic research efforts. He's gone full circle and now back at the helm of Garcia Research, a Hispanic market and Multicultural-focused research firm.Website