Probability Sampling -- Why Not Sample Everyone?
by M. K. Jeppeson

I have worked with people in business who collect reams of data -- about customer attitudes, sales, client demographics, changes the customer would like to see, and so on.

The scale of some of those databases is both compelling , and appalling! 

People hire statisticians as consultants to make sense out of their data (that's what we are for). And they analyze the data and come up with statistical results:
Frequencies, Correlations, Analyses of Variance, to name a few.

So what?

Well, when properly interpreted, the statistics provide an aggregate profile of the "average" person, place, or circumstance, for each category that is of interest to the business.
They then assist the business by describing the results in light of their particular business interests, and Voila - data has been translated into information. 

The above process is descriptive because we are looking no further than the existing customer data and providing a general description.

But what happens if we want to extrapolate? What if the business wants to expand and work with potential customers in, say, Scottsdale, or all of Maricopa County, or maybe the whole of the State of Arizona? What if they want to expand it to potential clients throughout the United States, Europe, Asia, or the World? 

How am I going to get information from all of those potential customers to find out how best to manage and expand my business? 
I would suggest that it probably can't be done effectively outside your local community - in fact, it probably should not be done that way at all.

The best way to get the information you need from potential customers is to make a statistically appropriate sample using tried and true probability techniques.
Sigma Research (or any top-notch statistical consultant) can show you how to cost-effectively sample for the best results. 

Why is probability sampling best?  
Because it gives a cost-effective and mathematically accurate representation of all potential customers in the target market. 

Each person, place, or thing that you want to gain information about, has an "equally likely or known" chance of being picked from the defined population. 
For example, probability sampling has definite advantages over direct census-taking. A census is a survey of all people, like a certain agency woefully attempts to conduct each decade in our great country.
By todays' standard, these are your tax dollars at work, but also Constitutionally bound to be performed in that fashion - a continued and heated debate! 

Some advantages of probability sampling: 
(1) It affords a true representation of the population, minimizing biased results. 

(2) You can select a small, feasible sample size and obtain valid information about the population at large.
Gallup Poll statisticians have proven this fact.

Sigma Research has worked with probability samples of from 50 to up to 4,000 people.
The sample size depends upon whether one is sampling Scottsdale Road or the World.
Notice sometime how small and accurate a well-drawn polling sample can be. 

(3) Probability sampling has actually been proven to be more accurate* than census-taking. 
(* Accuracy = low bias + high precision).
An actual head-count census actually creates an incredible amount of non-sampling error in surveying that includes clerical mistakes, administrative errors, inaccurate data entry, and so on -- just because of the enormity of the undertaking.
This guaranteed sort of problem can be avoided using a small, well-chosen probability sample. 

(4) With probability, you can make general statements about the responders to your survey 
(mailer, questionnaire, or research protocol), and you can draw valid conclusions about your target population.
This will help you explore and make on-target decisions to successfully expand your business into previously unknown territory.

There are several types of probability methods, like Random, Stratified and Cluster sampling. 
A professional statistician can help you select and use a method that's appropriate and suited to your business needs.
For more information on the subject, please contact us - we would be honored to model, develop or conduct your survey market research or guide you in successfully accomplishing it in-house!

Maureen Keryan Jeppeson, Ph.D.(Cand.), is Owner /Director of Sigma Research in Chandler, Arizona.
She has been a research and statistical consultant since 1987.