Supporting Statement PART B_01-29-14_FINAL

SUPPORTING STATEMENT – PART B

PART B: COLLECTIONS OF INFORMATION EMPLOYING STATISTICAL METHODS

1. Description of Activity

The following table provides information on the sampling procedures for each of the Annual Customer Satisfaction surveys that would fall under this generic license. All of the surveys will be administered via a web-based survey tool¹.

Target sample sizes are determined by calculating the number of surveys that must be sent, given a specified response rate and non-deliverable rate, to achieve the desired margin of error calculated using the standard binary formula, the Finite Population Correction, and the 95% confidence level.

Table 1: Proposed Samples and Sampling Method by Annual Recurring Survey

Survey	Estimated Population	Potential Sample	Sampling Method	Anticipated Response Rate	Anticipated # of Responses
Annual Customer Satisfaction Surveys (General Public)
Contractor Pay	13,000	13,000	Census of contractors with whom DFAS worked	25%	3,250
Retiree and Annuitant	630,000	5,000 Retirees 5,000 Annuitants	Random selection	35%	3,500
Vendor Pay	65,000	25,100	Selected randomly based on database of DFAS Vendors	15%	3,765
Total	708,000	48,100		20%	10,515

Estimated Responses by Ad Hoc/Ongoing Survey

Unlike the Annual Customer Satisfaction surveys, DFAS MyPay sends a survey invitation out following each interaction with the MyPay system or interaction with a MyPay representative, thus, is a sample of convenience of all MyPay users. Data from recent years shows approximately 9,027,000 MyPay interactions from members of the general public, resulting in fewer than 18,000 responses.

Similarly, the DFAS Customer Care Center offers an Interactive Voice Response survey after each telephone interaction, thus is a sample of convenience of DFAS Customers. Data from recent years shows approximately 400,000 customers are members of the general public, but fewer than 50,000 call the Care Center each year and approximately 15% complete the survey.

2.  Procedures for the Collection of Information

a. Type of sampling: See Table 1

b. Variables used to stratify the sampling frame: For populations below 50,000, a census survey is used. For larger populations, a random sample is used based upon the transactions that took place within the previous calendar year. DFAS receives the population from each mission area, and provides the list to OPM to generate the random samples. See Table 1

1. Surveys:

   1. Vendor Pay Survey: This list includes all vendors/contractors paid by DFAS (All sites). A randomly selected sample is surveyed.

   2. Retired and Annuitant Survey: This list includes all customers that have had at least one event activity (Address Change, Allotment start/stop, etc.) within the previous calendar year. A randomly selected sample is surveyed.

c. Proposed sample size overall: See Table 1

d. Proposed sample size per cell or subgroup: N/A

e. How will potential participants be recruited: For participants in the Customer Satisfaction Survey, invitations and reminders will be sent via e-mail. For MyPay, an invitation will be proffered on the exit page of the MyPay system after each customer interaction. For the Customer Care Center, an invitation is proffered following a telephone interaction with a representative.

f. Are any potential participants Active Duty military: Yes

g. Are any potentially vulnerable populations being targeted: No

h. What anticipated response rate was used to derive the cell sizes for the sample: N/A

i. What criteria were used to determine who will receive a survey: As noted in Table 1, for some populations of the Customer Satisfaction Survey, there will be a census survey invitation and for others a random sample will be used. For MyPay surveys, an invitation is proffered after each customer interaction. For the Customer Care Center, an invitation is proffered following a telephone interaction with a representative.

k. What database is the source of contact information for the sample: The following databases are used: Defense Joint Military Pay System (DJMS); Defense Civilian Pay System (DCPS); and the Defense Travel System (DTS).

l. Estimation procedures: Weighting will not be used

m. Degree of accuracy needed: The target maximum margin of error is ± 2.5 percentage points.

n. Unusual problems requiring specialized sampling procedures: None

o. Use of periodic or cyclical data collections to reduce respondent burden: Burden should be minimal because the Customer Satisfaction Surveys are conducted annually, most with random selection from a very large population. Also, the MyPay and Customer Care Center interaction surveys are very short and are offered as after each interaction with the system.

3.  Maximization of Response Rates, Non-response, and Reliability

Response rates are maximized by: (a) informing respondents that a third-party, OPM, is conducting the surveys on behalf of DFAS; (b) keeping the survey lengths to a minimum; (c) sending up to two reminder notices; and (d) providing respondents with save-and-return functionality. The sampling plan, when possible, takes anticipated response rates into account for purposes of meeting target margins of error. Estimates of population demographics are generally not known, so sample demographics cannot be compared to them to estimate generalizability.

4.  Tests of Procedure

Approximately the same administration procedures that have been used for the past nine years will be used in ensuing years, and any lessons learned will be incorporated. Some of the changes are due to DoD policy and others are improvements to obtain the most accurate contact/email address for each survey distribution list. We no longer send OPM SSNs in order to pull samples. All distribution lists are requested in February of the survey year. Distribution lists are asked to be pulled based upon the previous calendar year transactions/communications.

5. Statistical Consultation and Information Analysis Steve Burnkrant, Ph.D., Phone: 202-553-1279, e-mail : [email protected]

1 A physical copy of the Retiree and Annuitant Survey is available to respondents by request only.

3