Who Should Receive the Survey
After developing the survey form, the next step is to conduct the survey. Ideally, you would survey all persons within the FOM, which may be possible for smaller groups. However, for larger groups, and to control pre-chartering costs, it is likely more reasonable to survey a random sample of the potential membership. Surveys can be delivered electronically, by mail, or in person.
How to Perform a Random Survey
There are two types of sampling for a random survey: a statistically valid sample and a targeted sample.
A statistically valid random sample contains two key components:
- Every individual within the FOM must have an equal chance of being selected; and
- The number of responses must be a fair representation of the entire FOM.
How to Ensure Equal Chance for Selection
A simple statistically valid random sample involves surveying every nth person from a list of potential members. However, such a selection method is not required. You may survey the membership using other forms of random sampling. Examples of other forms of statistically valid random samples are illustrated as follows:
- If the FOM consists of members of five religious entities, offering the survey form to all attendees at all five religious entities on a given day would be considered statistically valid random sampling. If you only survey the members of one of the five religious entities or limit the survey to a selected demographic group (for example, those under the age of 25 or those with children), it would not be considered a statistically valid random sample.
- If the FOM consists of a community charter, conducting the survey at several different locations throughout the community would be considered statistically valid random sampling. However, if only one segment of the community or one geographic area (such as one city in a county) was surveyed, and not all segments of the community or various geographic areas had an opportunity to receive the survey, the distribution would not be deemed a statistically valid random sample.
A targeted random sample does not contain the two key components found in a statistically valid sample and a specific number of responses is not required.
Due to these shortcomings, the results from a targeted random sample cannot be extrapolated and applied to the entire FOM population. For this reason, in most cases, the assumptions in the business plan and pro forma financial projections are better supported using data from the statistically valid random sample method.
The following example illustrates the differences in applying the results of a statistically valid random sample survey compared to a targeted random sample.
Regardless of whether a survey is conducted based on a statistically valid random sample or a targeted random sample, it is inappropriate to survey only those individuals you assume are interested in joining the credit union while excluding others within the FOM. That is not a random sample and the survey results and all data gathered from the survey would be unreliable and considered a weak foundation for developing a business plan and financial projections.
EXAMPLE: Assume the FOM population is 5,000; surveys are sent to 1,000 random individuals; and 500 individuals responded to the survey. Out of the 500 responses, 150, or 30 percent, expressed an interest in joining the credit union within its first two years of operation.
If the results of the 150 were based on a statistically valid sample, you could extrapolate the 30 percent favorable response rate (150/500) and apply it to the entire FOM population. You can assume approximately 1,500, or 30 percent of the 5,000 individuals in the population, would be interested in joining the credit union within its first two years of operation. The two-year annual growth projections in the pro forma financial projections could then be developed with this in mind. For example, the projected new members for year one and year two could be 500 and 1,000, respectively, resulting in total new members after two years of 1,500.
Before performing a survey, contact the NCUA if you have questions about whether your random sampling method is proper, particularly if it does not include every nth individual in the potential membership. All random sampling methods must be explained in detail in the PFCU application, and if the NCUA determines a random sampling method improper, we will require a new survey.
How to Determine the Number of Responses Required
The next factor to consider when conducting a statistically valid random sample of the potential membership is the number of survey responses you must receive to adequately predict the preferences of the potential membership. The number of required survey responses will vary depending on the size of the population. You may use Table 1 to determine the number of responses needed based on the number of total potential members, also known as the population size. The table uses a 95 percent confidence level1 and a 5 percent confidence interval.2
For example, if the PFCU’s potential membership is 5,000 members, you must receive 357 responses to the membership survey to adequately predict the preferences of the entire potential membership.
Table 1: Number of Responses Needed
|Number of Responses Needed
|Up to 1,000
|1,001 to 3,000
|3,001 to 5,000
|5,001 to 10,000
The PFCU should use this table when completing either a statistically valid random sample or a targeted random sample.
How to Determine the Number of Surveys to Distribute
How will you know how many surveys to send out to receive the required number of responses? It is highly unlikely that you will receive a 100-percent response rate to the membership survey. Thus, the final factor to consider when conducting a random sample of the potential membership is the anticipated response rate from the sample of members. The organizing group must use judgment in estimating the survey response rate, as it will differ for each potential field of membership.
In the previous example, if you estimate 40 percent of those who receive the survey will respond, you must send out 893 surveys in order to receive 357 survey responses (357 being 40 percent of 893).3 If a low response rate is anticipated, consider surveying a larger number of potential members to ensure you get an adequate number of responses.
PFCUs should be aware a low response rate may be an indication the potential membership is not supportive of a new federal credit union. If a low actual response rate is experienced, the PFCU should be able to provide reason(s) why the potential membership is not being responsive to the survey.
1 The confidence level tells you how sure you can be of the results. It is expressed as a percentage and represents how often the population would pick an answer that lies within the confidence interval. The 95 percent confidence level means you can be 95 percent certain.
2 The confidence interval (also called the margin of error) is the plus-or-minus figure usually reported in newspaper or television opinion poll results. For example, if you use a confidence interval of 5, and 47 percent of your sample picks an answer, you can be relatively certain that for the entire relevant population between 42 percent (47 minus 5) and 52 percent (47 plus 5) would have picked that answer.
3 The number of surveys to send to achieve the required number of survey responses is calculated by dividing the number of required responses by the estimated response rate.