Basic Marketing: A Global Managerial Approach

Perreault−McCarthy: Basic

Marketing: A

Global−Managerial

Approach, 14/e

8. Improving Decisions
   with Marketing
   Information

Text © The McGraw−Hill

Companies, 2002

240 Chapter 8

Much marketing research is based on nonrandom sampling because of the high

cost and difficulty of obtaining a truly random sample. Sometimes nonrandom

samples give very good results—especially in industrial markets where the number

of customers may be relatively small and fairly similar. But results from nonrandom

samples must be interpreted, and used, with care.

An estimate from a sample, even a representative one, usually varies somewhat

from the true value for a total population. Managers sometimes forget this. They

assume that survey results are exact. Instead, when interpreting sample estimates,

managers should think of them as suggestingthe approximate value.

If random selection is used to develop the sample, researchers can use various

methods to help determine the likely accuracy of the sample value. This is done in

terms of confidence intervals—the range on either side of an estimate that is likely

to contain the true value for the whole population. Some managers are surprised to

learn how wide that range can be.

Consider a wholesaler who has 1,000 retail customers and wants to learn how many

of these retailers carry a product from a competing supplier. If the wholesaler ran-

domly samples 100 retailers and 20 say yes, then the sample estimate is 20 percent.

But with that information the wholesaler can only be 95 percent confident that the

percentage of all retailers is in the confidence interval between 12 and 28 percent.^20

The larger the sample size, the greater the accuracy of estimates from a random

sample. With a larger sample, a few unusual responses are less likely to make a big

difference.

Even if the sampling is carefully planned, it is also important to evaluate the

quality of the research data itself.

Managers and researchers should be sure that research data really measures what

it is supposed to measure. Many of the variables marketing managers are interested

in are difficult to measure accurately. Questionnaires may let us assign numbers to

consumer responses, but that still doesn’t mean that the result is precise. An inter-

viewer might ask “How much did you spend on soft drinks last week?” A respondent

may be perfectly willing to cooperate—and be part of the representative sample—

but just not be able to remember.

Survey Sampling, Inc., and
Simmons Custom Research help
marketing researchers develop
samples that are really
representative of the target
market.

Validity problems can
destroy research

Research results are
not exact