Science of Surveys

---

Step 4: Sample Selection

One step of the survey process often overlooked is the determination of the survey sample. Two important tasks are needed when deciding who will receive your survey. First, you must identify your target population.

A target population is the entire group of people from which you would like to obtain information. For many surveys within small companies or associations, you might be able to survey the entire population.

For other large-scale surveys, it is impossible to survey everyone. In those cases, it is vital that the sample you choose represents the larger population in every meaningful way. Without a representative sample, your results will not reflect those of the larger group and your survey will fail.

Many sophisticated techniques exist to generate survey samples including stratification, cross-sectional and quota sampling. However, the easiest and quickest procedure is random sampling. Random sampling works by allowing everyone in the population an equal chance of being included in the sample.

Statisticians commonly accept random sampling as an effective way to generate representative samples, given that the sample is sufficiently large.

Sample Bias
Short of sampling the entire population, it is nearly impossible to get a bias-free sample. However, identifying and correcting for specific signs of bias can be helpful when interpreting survey results. The following list describes sources of bias in samples and how they influence survey results.

Sample Probable Bias Reason

Survey Goal	Sample	Probable Bias	Reason
How do people perceive our business?	Current customers	Favorable	If they are still your customers, they will tend to be more favorable about your business than non-customers.
How do people perceive our business?	Ex-customers	Unfavorable	Since they decided to leave your business, they will tend to be more negative about your business than others
What is your opinion about?	People who "phone in" about their opinion	Extreme	Views Because it takes effort to call in, people who participate usually have either a very positive or negative opinion about the topic than the general population.
What is your opinion about?	Daytime phone interviews	Non-working or self-employed	Using a daytime phone survey will create a bias toward non-working or self-employed people because those are the people at home during the day to answer the phone.
What is your opinion about	Internet surveys	Technology savvy or younger generations	Although this is less the case now than in the past, there is still a bias toward younger people or those comfortable with technology in some of the more mature industries.

Source: http://ubmail.ubalt.edu/~harsham/stat-data/opre330Surveys.htm

1. Determine goals and specify objectives
2. Conduct question brainstorming and pre-testing
3. Prepare the questionnaire layout and data format
4. Sample Selection
5. Survey administration
6. Data compilation and analysis
7. Reporting of Results