Many organizations use surveys to gather information and data in order to learn more about their customers. While it is important to ensure that surveys are targeted to the right audience, confidence in the development and implementation of a survey is even more vital. To prepare a successful survey, practitioners should follow nine basic steps; the first three steps are explored here.
Step 1: Establish the Goals
Survey design has three main goals:
- Decide what the organization wants to learn.
- Determine whom to ask.
- Determine what to ask them.
The first goal, deciding what the organization wants to learn from the survey, must be clearly stated before the design process can continue. Some common survey objectives are learning more about the potential market for a new service or product and collecting ratings of current services or products. In addition, organizations may just need to determine customer or associate satisfaction levels. Satisfaction levels have a direct impact on corporate image, which is even more important in the current economic times.
Step 2: Determine the Sample
Identifying the target population for the survey will help reduce stray answers and bias. From the population, practitioners need to determine the sample size. Utilizing basic sampling strategies, they must take into account the response rate, or “hit rate,” based on the type of media used to deliver the survey. For example, if a calculated sample size is 1,000 and the media used typically has a 10 percent hit rate, then practitioners must send out 10,000 surveys.
From a practical viewpoint, organizations must also consider time and money when selecting a sample. They want to learn about a population, but they do not want the process to take forever, or to be too expensive. Some budget considerations include:
- Staff time for survey planning, execution and analysis
- Labor and material costs for pre-testing
- Telephone fees
- Mailing costs, including envelopes and stamps
- Incentives to encourage responses
- Printing costs
In order to determine sample size effectively, practitioners may need some additional information, such as:
- Best estimate of the population size
- Best estimate of the rate of the survey characteristic
- The maximum acceptable margin of error
- The confidence level in saying that the true population rate falls within the margin of error
With this information, practitioners can calculate the proper sample size. Although statistical analysis software is helpful in these situations, it is also possible to use the following formulas.
When dealing with large populations where the sample will be less than 10 percent of the total population, use this formula:
- n is the sample size
- p is the rate of the survey characteristic
- D is the margin of error, or Delta
- z is the z-score for setting the confidence level using
- a, which is the amount of risk
When dealing with small populations where the sample will be more than 10 percent of the total, use this formula:
- n, p, D and z are the same
- N is the population size
- is the fininte population correction
Step 3: Choosing the Approach
Probability sampling is the most appropriate sampling method. Within this method, practitioners can choose from several approaches:
- Simple random sampling: This ensures that each member of the population has an equal chance of being selected to take the survey. For example, if there are x items in the population, each item has a 1/x chance of being selected. However, this “luck of the draw” approach can exclude some subgroups.
- Stratified random sampling: This approach divides the population into logical strata based on shared characteristics. From there, practitioners use random sampling within each stratum to ensure that each stratum is represented in the sample.
- Systematic random sampling: This is useful if the population is very large and individuals cannot be numbered for random sampling. Instead, practitioners systematically choose the sample by selecting individuals at fixed intervals (e.g., every 10th person on a population list).
- Cluster sampling: This approach, sometimes referred to as multi-stage sampling, is used when the population has many strata. In this case, practitioners randomly select clusters from the strata and then use random sampling within the clusters.
Before Moving On
The first phase in creating a successful survey sets the stage for each following phase. Practitioners must have a clear understanding of what they want to learn and from whom they want to learn it, and they must choose the right approach for gaining that knowledge. Once these key steps are defined, practitioners are ready to move to on to how and what to ask in the survey.