Market Research Tutorial: Market Research Samples





Market research samples are a critical part of the underlying validity of a project. For quantitative studies, the statistical foundation of the results rest upon an assumed randomness of respondents in the survey.

Here, alas, the market research tutorial must talk statistics, at least briefly. If you remember back to any intro course on stats, recall that after a population (the total pool of customers or potential customers or competitors' customers or the like) reaches a certain size (that is, number of people), taking a sample of the population and simply profiling them through a survey provides an accurate picture of the underlying population. The certainty of that picture being correct is expressed as a "level of confidence."

In a standard market research report, (or even in a news article, say, on a presidential poll) that the results of the study are reported as valid with a level of confidence and margin of error, for example a 95% level of confidence with a +/- 3% margin of error. This essentially means that if the study was replicated 100 times (a random sample was drawn out of the population and surveyed), 95 times out of 100 the responses would not vary by more than 3 percentage points on any given yes/no type question. So you may see that candidate X is favored by 45% of voters and candidate Y is favored by 36%. The "true" percentages for candidate X would range from 42% to 48% and for candidate Y from 33% to 39%. That is, in 95 surveys out of 100.

Without going into a statistics lesson here (and maybe we have, sorry!), for our purposes in this guide, the critical elementin market research samples is the assumption of the randomness in drawing the sample out of the population.

The long and short of it is that to have a statistically-valid survey, you must ensure that you take a random sample fromthe population you are studying.

Let's say that population is your organization's customers. You would take the entire customer database, filtering for, if appropriate, activity within a certain time frame. You would then take that pool of customers and draw a random sample from it.

For telephone and online surveying, you will need to pull approximately 15 customers for every completed survey you need. So if you need 400 completed surveys, you would draw a random sample of 6000 customers.

But how to do determine how many completed surveys you will need? Onward to the next section....................


Forward to How to Determine Sample Size

Back to Choosing the Population


Top of Market Research Tutorial Market Research Samples