Annual Employee Survey Template Page 11
ADVERTISEMENT
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
299
2. Construct the Sample Frame. This is a list of individuals from which you will select your
sample based on the target population parameters identified in #1 above (e.g., full-time, part-
time, permanent, temporary).
3. Determine the Size of the Sample. Agencies may either conduct a census (i.e., survey all
employees in the target population) or select a sample of employees from the target population to
survey. At a minimum, the sample should be large enough to produce 400 completed surveys to
ensure reports accurately reflect the target population. The typical response rate for the FHCS
has been approximately 50 percent. Based on these findings, OPM recommends the selected
sample include at least 800 employees to obtain 400 completed surveys. Smaller agencies (i.e.,
smaller than 800 employees) will need to survey all of their employees to meet this requirement.
If you have evidence from past surveys in your agency that your response rate is typically higher
or lower than 50 percent, you should adjust your sample size accordingly.
In addition, agencies should consider the number and size of the subgroups for reporting and
statistical analyses. To meet the requirements of the AES regulation at subpart C of 5 CFR part
250, agencies only need to report AES results for the agency/department as a whole (e.g.,
Department of the Interior, Office of Personnel Management). It should be noted, though, for
large agencies, reporting at the department level masks the meaningful results at the
subcomponent level. It is recommended agencies evaluate results and focus action planning
efforts at the subcomponent level, as well.
If agencies sample at the subcomponent level, they only need to report the overall agency results
to OPM and on their Websites. However, as indicated in the “Survey Reporting” section below,
an agency will need to report information on the number of people surveyed in the various
subcomponents to demonstrate representativeness.
OPM recommends a sample size of at least 30 in each group to be compared. This should allow
for standard statistical tests for differences. Testing can be done for smaller sample sizes but this
requires the use of small sample methods such as the exact binomial distribution. In any case,
agencies should consult a statistician to assist with the analysis of the data.
Where feasible, a larger sample is always better. Whenever a survey is administered to a sample
rather than the entire target population, sampling error will occur. Sampling error is the
difference between the true population value and the population value estimated from the
sample. When one interprets sample data, there is a chance wrong conclusions will be drawn
because of sampling error. Surveys are also subject to nonsampling error, including nonresponse
error, measurement error, coverage error, and processing error. However, the extent of sampling
error can be estimated directly from the sample itself. That estimate is often referred to as the
margin of error, a statistical measure (confidence interval) that indicates the precision of a
sample estimate.
For example, assume the margin of error is plus or minus 3 percent for a 95 percent level of
confidence. If the percent favorable is 92 percent, a statement such as the following would be
U.S. Office of Personnel Management
November 2008
ADVERTISEMENT
0 votes
Related Articles
Related forms
Related Categories
Parent category: Life