In creating a survey, a good survey question is essential. In addition, the following steps are useful.

1. Identifying the sample population and choosing a sampling strategy.
2. Choosing a method to conduct the survey: face-to-face, phone, paper survey, or internet survey.
3. Deciding on a number of questions to ask, the order, and the wording of these questions. The wording must be unbiased.

A good survey should be answerable with the results of the survey. It should not be too narrow or too general. For instance, the effect of tax rates on public expenditures is too general to be addressed in a survey. On the other hand, the influence of a tax deduction as an incentive for homeschooling children could be directly assessed with a survey.

Also, a survey question must address the variability in the data. A survey question can be phrased to comply with this. For instance, a question that asks whether high school students prefer art or music as extracurricular activities is broad. A question that addresses the variation in this preference for first-year, second-year, third-year, and fourth-year students would account for variability and generate more comprehensive data.

After the survey is conducted and the results are collected, inferences may be made on the data. The reliability of these inferences is highly dependent on the sampling strategies used in determining with whom to conduct the survey study.