2900-0680 justification Part B

B. COLLECTIONS OF INFORMATION EMPLOYING STATISTICAL METHODS

1. Provide a numerical estimate of the potential respondent universe and describe any sampling or other respondent selection method to be used. Data on the number of entities (e.g., households or persons) in the universe and the corresponding sample are to be provided in tabular format for the universe as a whole and for each stratum. Indicate expected response rates. If this has been conducted previously include actual response rates achieved.

The respondent universe for this study is guided by the study’s issues and research questions. Each survey has different target respondent populations and subpopulations, which are described below.

Population for Survey of Service Disabled Veterans

The target population for the survey is those veterans who have been rated by VA as service disabled. The data collection requires statistically representative samples that will support analysis of subgroups of interest to the Commission at the 95 percent confidence interval. To fully research the study questions, the respondent sampling frame for the study will be constructed from several VA administrative files that represent the different service eras of disabled veterans and contain the various types of disabilities and disability ratings received by veterans. A sample of individuals will be drawn that will allow the analysis to distinguish differences by disability rating, the primary body system that is the basis of the disability and specific categories of special monthly compensation (SMC).

To develop the sample design for the survey of service-disabled veterans, the process begins with a review of the population statistics for disabled veterans, grouped by primary disability and level of disability. The subgroups for veterans are shown in Table 3.

Table 3. Population of Veterans by Disability and Level

Body System of Primary Disability	Disability Rating
	0%	10%	20-40%	50-90%	100%
Musculoskeletal		366,900	513,498	209,306	14,344
Skin		69,033	33,332	12,363	1,381
Auditory		106,480	69,845	30,800	7,278
Neurological		28,287	46,264	45,583	7,451
PTSD		5,278	29,646	121,760	50,630
Mental (Excludes PTSD)		36,337	38,955	65,854	58,783
Digestive		37,866	40,268	15,118	4,714
Cardiovascular		52,415	51,646	51,089	12,685
Respiratory		33,694	38,248	35,459	8,625
Endocrine		11,259	74,284	24,445	2,821
Genitourinary		11,780	18,361	18,479	11,992
Visual		9,705	20,839	11,205	1,209
Gynecological		1,832	5,843	13,450	298
Systemic (Infectious, Immune…)		1,157	1,663	2,173	997
Dental		1,868	2,291	992	66
Hemic/Lymphatic		598	4,731	2,699	3,709
Subtotal	0	774,489	989,714	660,775	186,983
Cat 1: 0% w SMC K	4,101
Cat 2: 100% w SMC S,L,M,N, or O[1]					45,765
Cat 3: 100% w R1 or R2					7,827
Total	4,101	774,489	989,714	660,775	240,575

The cells in this table were modified to avoid cells that are too small to be meaningful for policy purposes:

Skin Collapses 50-100 percent ratings into one group

Visual Collapses 50-100 percent ratings into one group

Gynecological Collapses 50-100 percent ratings into one group and 10-40 percent into one group

Systemic Collapses 50-100 percent ratings into one group and 10-40 percent into one group

Dental Collapses all into one group

Hemic/Lymphatic Collapses 10-40 percent into one group

Projections can be made to cell population as defined above. After these adjustments, the population is arrayed in Table 4 and serves as the basis for sampling.

Table 4. Population of Veterans by Disability and Level—Modified

Body System of Primary Disability	0%	10%	20-40%	50-90%	100%
Musculoskeletal		366,900	513,498	209,306	14,344
Skin		69,033	33,332	13,744*
Auditory		106,480	69,845	30,800	7,278
Neurological		28,287	46,264	45,583	7,451
PTSD		5,278	29,646	121,760	50,630
Mental (Excludes PTSD)		36,337	38,955	65,854	58,783
Digestive		37,866	40,268	15,118	4,714
Cardiovascular		52,415	51,646	51,089	12,685
Respiratory		33,694	38,248	35,459	8,625
Endocrine		11,259	74,284	24,445	2,821
Genitourinary		11,780	18,361	18,479	11,992
Eye		9,705	20,839	12,414*
Gynecological			7,675*	13,748*
Systemic(Infectious, immune...)			2,820*	3,170*
Dental			5,217*
Hemic/Lymphatic			5,329*	2,699	3,709
Subtotal	0	769.034	996,227	663,668	183,032
Cat 1: 0% w SMC K	4,101
Cat 2: 100% w SMC S,L,M,N, or O[1]					45,765
Cat 3: 100% w R1 or R2					7,827
Total	4,101	769.034	996,227	663,668	236,624

*Combined with adjacent cell(s)

Sample Design and Allocation for Disabled Veterans Survey

The sample will be stratified to obtain estimates by cell. The criterion used in this design is that point estimates will be 95 percent certain for categorical variables in each cell, such that the estimates are within 5 percentage points of the population percentage. This is commonly referred to as 95/5 precision. For a normally distributed continuous variable, this criterion is equivalent to estimates being within 10 percent of a standard deviation. For all cells, the sample sizes should be sufficient to identify significant differences for continuous variables between cells when the differences are of half a standard deviation with p< .01 and power of .95. We would also be able to calculate differences in proportions of 20 percent or more with the same significance and power. Both of these effects are considered medium-sized effects (Cohen, 1988).

The formula for calculating sample size in a simple random sample is:

(p(1-p)/n)^½ = 0.0255

(.5(1-.5)/n)^½ = 0.0255 and thus

.25/.0255² ≈ 384

where p = the proportion being estimated (in this case .5)

and n = sample size

This formula results in a sample size of 384 veterans drawn from an infinitely large population to meet the 95/5 criterion.

However, we know that the population of each cell is not infinitely large. We may, therefore, reduce the required sample size by applying what is known as the “finite population correction.” The finite population correction, or fpc, is equal to:

fpc = 1- nf/N

where n is equal to the sample size calculated for an infinitely large population, N is equal to the known population size and nf is the size needed for the same precision for the given finite population. The ratio nf /N is referred to as the sampling fraction. To get the same precision that n would yield for an infinite population, one needs a sample of size nf where:

nf = n/(1+n/N)

Substituting n = 384 and N = 1,366 (the number in the systemic row at 10%) into the formula above and then multiplying n by the fpc we get:

nf = 384/(1+384/1,366) = 300.

Therefore, when a population is extremely large, the number needed to achieve 95/5 precision is 384. However, when the population is smaller, the sample size can be smaller, though the decrease is not proportional. The precision must be calculated for the worst case scenario, which happens when the point estimate is 50 percent.

The final respondent universe and the number of completed surveys needed for each cell is presented in Table 5.

Table 5. Sample Allocations by Cell for the Survey of Service-Disabled Veterans
(Sample Size Needed for 95/5 Precision)

Body System of Primary Disability	Disability Rating
	0%	10%	20-40%	50-90%	100%
Musculoskeletal		384	384	383	374
Skin		382	380	374	0
Auditory		383	382	379	365
Neurological		379	381	381	365
PTSD		358	379	383	381
Mental (Excludes PTSD)		380	380	382	382
Digestive		380	381	375	355
Cardiovascular		381	381	381	373
Respiratory		380	380	380	368
Endocrine		372	382	378	338
Genitourinary		372	376	376	372
Eye		370	377	373	0
Gynecological		0	366	374	0
SYSTEMIC (Infectious, Immune...)		0	338	343	0
Dental		0	358	0	0
Hemic/Lymphatic		0	358	336	348
Subtotal	0	4,519	5,982	5,596	4,021
Cat 1: 0% w SMC K	351
Cat 2: 100% w SMC S,L,M,N, or O[1]					381
Cat 3: 100% w R1 or R2					366
Total	351	4,519	5,982	5,596	4,021

The total sample size is 21,221 completes for veterans. In addition to samples in each cell, this design also produces a self-weighting subsample of approximately 1,600 that represents the entire disabled veteran population. The precision for the total population, using the self-weighting sample, yields a 95 percent confidence interval of 2.45 percent. The precision would be even greater if all 21,221 cases were weighted.

Population and Samples for Survey of Survivors of Service-Disabled Veterans

The target survey respondent population for the survey of survivors of service-disabled veterans is surviving spouses of veterans who died while on active duty or due to service-connected causes. All spouses receive DIC from VA, and the survey will assess their quality of life.

In addition to DIC, some spouses receive benefits from DoD’s Survivor Benefit Program (SBP). The circumstances of survivors who are eligible for SBP could differ significantly from those who do not. Thus there are four groups of interest in the survey. They are the following:

• Surviving spouses of veterans who died in the last 5 years and receive SBP

• Surviving spouses of veterans who died in the last 5 years and receive no SBP

• Surviving spouses of veterans who died more than 5 years ago and receive SBP

• Surviving spouses of veterans who died more than 5 years ago and receive no SBP

Table 6 presents the universe for the surviving spouse population and subpopulations. Table 7 presents the sample sizes, using the methodology described previously.

Table 6. Universe for the Survey of Surviving Spouses

Surviving Spouses	DIC w/ SBP offset	DIC w/o SBP offset
Veteran died >5 years	36,623	193,975
Veteran died ≤ 5 years	13,897	57,142
Total	50,520	251,117

Table 7. Sample Allocations by Cell for the
Survey of Survivors of Service-Disabled Veterans
(Sample Size Needed for 95/5 Precision)

Surviving Spouses	DIC w/ SBP offset	DIC w/o SBP offset
Veteran died > 5 years	380	383
Veteran died ≤ 5 years	374	381
Total	754	764

In addition the sample will target surviving spouses under 40 years of age. Table 8 shows the number of surviving spouses who are under 40.

Table 8. Universe for the Survey of Young Surviving Spouses

Surviving Spouses	DIC w/ SBP offset	DIC w/o SBP offset
Veteran died > 5 years	267	2,063
Veteran died ≤ 5 years	363	2,764
Total	5427

The sample of Young Surviving Spouses will augment the sample that will naturally occur in the sample of all surviving spouses. As shown in Table 9, a total of 35 Young Surviving Spouses are expected to be present in the samples of surviving spouses selected to obtain representation of those with and without SBP offset and those widowed five years or less, or more than 5 years. These 35 respondents will be overlapped—that is, they will be included in both the estimates for the 4 cells in the sample of all surviving spouses, as well as in the national estimates of Young Surviving Spouses.

Table 9. Overlap for the Survey of Young Surviving Spouses and All Spouses

Surviving Spouses	DIC w/ SBP offset	DIC w/o SBP offset
Veteran died > 5 years	3	3
Veteran died ≤ 5 years	8	21
Total	35

Table 10 shows the distribution of the 359 needed to obtain a nationally representative sample of Young Surviving Spouses at the 95/5 level of precision (including the overlapping sample).

Table 10. Sample Allocations by Cell for the
Survey of Young Surviving Spouses
(Sample Size Needed for 95/5 Precision for Total Population)

Surviving Spouses	DIC w/ SBP offset	DIC w/o SBP offset
Veteran died > 5 years	18	136
Veteran died ≤ 5 years	24	181
Total	359

Operationally, the allocations for the main survivor survey will be adjusted to include only the survivors over 40 (reducing the allocation of the cells by 3, 3, 8 and 21 respectively) and a corresponding number from the Young Survivor sample will be used in their place).

Population for Survey of National VSOs

The target population for this survey is approximately 2,000 National VSOs who are co-located at VA facilities and are actively assisting veterans and their survivors to prepare and present claims. A companion internet survey of approximately 2,000 Veterans Benefits Administration employees who rate disability claims will also be administered, but clearance is not sought for them because they are government employees. VSO respondents will be asked to provide insights into the challenges and issues associated with implementing the laws, regulations, and procedures for assisting service-disabled veterans and their survivors. We will survey all of the approximately 2,000 National VSOs co-located at VA facilities and all of the approximately 2,000 VBA raters. Because these populations are relatively small (2,000 each) and web-based surveys are relatively efficient means of data collection, we will not sample from them, but rather will survey both populations in their entirety.

EXPECTED OVERALL RESPONSE RATE

We expect an overall response rate of 80 percent for the survey of service-disabled veterans and the survey of survivors of service disabled veterans. The response rate will be impacted by the degree to which obsolete address and telephone contact information can be replaced with corrected information, as well as the degree to which proxy respondents for disabled veterans are able and willing to respond to the survey. Our a priori estimated response rate is based on previous studies of veterans and their survivors, where response rates in excess of 80 percent were achieved for respondents with valid contact information. These studies focused on whether the benefits that veterans and their survivors were receiving actually met their economic needs, and thus are similar in many ways to the proposed surveys.

Veterans and survivors who may be excluded from the survey are those whom we do not succeed in reaching by telephone after we have performed extensive efforts to locate and call them, and individuals who are physically or mentally unable to participate in a telephone interview and for whom the contractor does not succeed in getting a surrogate to answer for them. In some cases, an attorney or financial institution will have responsibility for the financial affairs of these individuals. Such trustees are not appropriate surrogates nor are they likely to disclose contact information for an appropriate surrogate. The contractor does intend to call and interview respondents who are institutionalized when it is possible to get past the institutional gatekeepers and the respondent is capable of completing the survey. Many institutionalized individuals are capable of completing a phone survey.

For the survey of National VSOs, we expect the response rate to be 90 percent or greater because the Commission will encourage responses to this survey, and because the Commission will enlist the cooperation of the National Veterans Service Organizations that employ the VSOs to endorse and legitimize the survey and also encourage responding to it.

2. Describe the procedures for the collection of information, including: Statistical methodology for stratification and sample selection; the estimation procedure; the degree of accuracy needed for the purpose in the proposed justification; any unusual problems requiring specialized sampling procedures; and any use of periodic (less frequent than annual) data collection cycles to reduce burden.

PROCEDURES FOR THE COLLECTION OF INFORMATION

Estimation

For categorical variables, we can be 95 percent certain that survey responses are within 5 percentage points of the estimate in each cell. Any cell may be compared with another cell to detect significant differences. If the differences are of medium size (half a standard deviation or 5 points in the above referenced scaled) the sample sizes are large enough to detect effects at the .01 level with power of .95, and most are sufficiently large to detect small effects (.2 standard deviations or 2 points on the above referenced scales) at the .05 level with power of .7. Small effects will not be detected in cells that use the finite population correction factor.

The previous section presents the stratification and allocations of this sample. The sample sizes are designed to obtain point estimates for proportions with a 95 percent confidence interval of 5 percent under the worse case scenario.

Specialized Sample Development Procedures

Survey of Disabled Veterans and Survey of Survivors of Disabled Veterans

A sample of five times the desired sample size will be obtained as an initial sample. The sample will be divided into 50 or fewer replicate groups, each with a size of 10 percent of the allocated sample size. The groups will be selected so each group has approximately the same proportion of combat disabled veterans. Initially ten groups will be released. Each member of a released group will be called and the protocol will be followed until a final disposition is reached. From these early groups, an expected response rate will be calculated and additional groups will be released so as to obtain the allocated number of respondents. This approach will produce the precise sample size.

VA will provide addresses from administrative records for all veterans and survivors in the sample, based on their receipt of disability benefits. The address information on the population files is the address of record for benefit payment purposes. Because VA electronically transfers benefits to banks in most cases, this address might be as old as the address when the individual first began receiving disability benefits. Given that addresses will be obsolete for many respondents, it will be necessary to perform extensive address verification before the survey. Even after that, we anticipate that it will not be possible to obtain current addresses for substantial proportions of the sample.

VA will assist in improving address currency by matching the disability benefits samples with other VA benefit files that have more recent addresses, such as VHA files for veterans receiving VHA health benefits, and VBA home loan files for veterans and spouses who have home loans. The contractor (ORC Macro) will establish the current address of sampled veterans and spouses obtained from VA through multiple methods, beginning with U.S. Postal Service data. The contractor will access the National Change of Address (NCOA) database maintained by the U.S. Postal Service to obtain correct addresses, ZIP Codes, and telephone numbers. For those not found through these sources, the contractor will consult telephone directory services, credit bureau searching services, and other market research databases to obtain current addresses and phone numbers. Even with these efforts, we anticipate that it will not be possible to obtain current addresses for substantial proportions of some of the cells.

To minimize nonresponse bias, we will draw replicate samples at the time of the original sampling. This will guarantee that once we proceed to contact a respondent, we will continue with a sufficient number of calls until we contact him/her or determine that s/he cannot be reached. This will avoid bias associated with selecting easily reachable respondents.

We will begin with an initial sample that equals the response target for each cell. We expect that most of the nonresponse will come from unreachable veterans. We will determine the expected response rate on the basis of response to the initial sample. As we approach exhaustion of a cell, we will estimate the percent nonresponse and be able to examine its distribution. We will then estimate the size of the next group to be released. A third group may become necessary to achieve the targeted number of respondents.

Survey of National VSOs

CNAC will conduct the survey of the National VSOs who are co-located at VA facilities. The Commission will obtain a list of current e-mail addresses for this population from the central headquarters offices of National Veterans Service Organizations that employ the targeted VSOs and supply this information to the contractor (CNAC). These offices will also be requested by the Commission to assist in legitimizing the survey and ensuring participation of the respondents.

Data Collection Procedures

Survey of Disabled Veterans and Survey of Survivors of Disabled Veterans

Veterans and survivors in the released replicate groups will receive an introductory, informed consent letter that will outline the project, provide the contact information, and assure the confidentiality of the responses before the initiation of the CATI surveys of these populations. A sample letter for service disabled veterans and survivors are included in Appendix A of this supporting statement. Respondent calls will begin 1 week after mailing the letter.

The surveys will be administered via CATI and are expected to last an average of 30 minutes. If we reach a respondent who is physically or mentally unable to complete the survey by phone, we will ask to speak to a family member who can answer the questions for him/her.

Survey of National VSOs

The survey will be implemented electronically via the internet using standard survey software. The data collection process will use a modified Dillman method for initiating contact with the targeted respondents via a pre-survey e-mail introduction letter from the Commission. The letter will introduce the project and objectives of the survey as well as CNAC as the survey contractor. CNAC will follow up with a second e-mail letter that provides directions for how to participate in the survey. Pre-survey activities will also include coordination with national offices of VSOs so they are fully aware of the survey and can encourage participation. The survey will be accessed by targeted respondents by using a unique link to the survey website provided to them in the e-mail letter from CNAC. The survey will be fielded for 2 months, with periodic reminders sent via e-mail to encourage non-responding individuals to respond. The total time for completing the survey is estimated at 40 minutes.

3. Describe methods used to maximize the response rate and to deal with issues of non-response. The accuracy and reliability of information collected must be shown to be adequate for intended uses. For collections based on sampling, a special justification must be provided for any collection that will not yield “reliable” data that can be generalized to the universe studied.

Maximizing Response Rates

Our approach to maximizing response rates begins with extensive efforts to locate respondents, followed with actions to legitimize the survey and make it convenient for respondents.

The use of information technology facilitates maximizing the response rate for this evaluation. The greatest threat to response bias in this study is failure to locate respondents. Steps that the contractor will follow to locate correct telephone numbers as mentioned above include updated addresses from other VA benefit files that have more recent addresses, such as VHA files for veterans receiving VHA health benefits, and VBA home loan files for veterans and spouses who have home loans and the U.S. Postal Service, phone directories, market research databases, and credit bureaus.

A modified Dillman approach will be used to conduct the survey, including an introductory letter from the Chairman of the Veterans Disability Benefits Commission that legitimizes the study and encourages responses. ORC Macro will include postage-paid return postcards for phone number confirmations and updates. ORC Macro will offer respondents the option of completing the survey by calling ORC Macro toll free at the respondent’s convenience or scheduling a preferred time to complete the survey. A toll-free number will also be provided to answer any questions that respondents might have about the survey.

The CATI system that will be used for the survey implementation has the capacity to conference in family members, for instance, when the respondent is a nursing home patient, to maximize response rates. CATI interviewers and supervisors are well trained and have broad experience, making them sensitive to the needs of respondents. The experienced interviewers can guide respondents gently toward completion of the survey. Senior interviewers are assigned to nonresponse conversion teams, and interviewers experiencing a difficult respondent may forward the case to this specialized team. The system will schedule interview times at appointments requested by respondents who call the toll-free number or indicate a preferred time on the return postcard.

The automated version of the instrument always allows refusal and “don’t know” as an option. To maximize item response, these options are typically not read, although they are accepted and recorded.

The CATI survey implementation will also will follow an OMB approved telephone protocol for the surveys of veterans and survivors:

1. Eight call back attempts: Interviewers will make eight attempts to reach an eligible household and interview the respondent or surrogate. Each call attempt will allow a minimum of five rings.

2. Different times of day: To minimize bias, attempts will be distributed among three day-parts: weekday days (Monday through Friday from 9:00 a.m. to 5:00 p.m. respondent time), weekday evenings (Monday through Friday from 5:00 p.m. to 9:00 p.m. respondent time), and weekends (Saturdays from 10:00 a.m. to 9:00 p.m. respondent time and Sundays from 10:00 a.m. to 9:00 p.m. respondent time). We assume that one-third of the calling will be done during weekdays.

3. One attempt per day: The contractor will attempt a contact once a day during the week, and (at most) twice on weekends. The only exceptions to this will be when the respondent requests a specific appointment or with “busy” numbers.

4. Complete history of attempts: The disposition of each call attempt will be recorded on all sample records. The CATI system automatically stores the disposition of each attempt, providing a complete call history for each telephone number. This call history is displayed on the interviewer’s screen during each new attempt on a record in the sample.

After the callbacks are exhausted, replicates will be released to reach the desired number of respondents.

The Commission will request endorsement by veterans’ service organizations for the survey of National VSOs. The National VSOs are drawn from these organizations and are very committed to serving veterans. We believe that they will view responding to this survey as part of their commitment.

Using a modified Dillman approach, the Commission will notify respondents of the survey with an introductory letter attached to an e-mail. The contractor, CNAC, will follow with an e-mail providing instructions for participating in the survey and a unique link to access it. Respondents will be able to complete the survey at their convenience, which will also enhance response rates. The contractor will send weekly e-mail reminders to nonrespondents over a 2-month period, encouraging their response.

Nonresponse Analysis

Given that the study may need to use a substantial number of replacements, a detailed nonresponse analysis will be conducted, one that will examine the relationship between sampling status and frame variables. Nonresponse bias exists primarily when the characteristics that determine response are associated with the variables being estimated.

Fortunately the frame includes a large number of variables that could be associated with responding. This permits extensive nonresponse analysis to identify any potential bias. The simplest analysis will be the comparison of frame variables estimated from the sample with those obtained from the population. This will indicate whether there is potential for bias. If there is, we will develop an equation predicting response from the frame variables. We will then correlate the predictive equation with the actual responses to key questions.

Should the presence of nonresponse bias be identified, the predictive equation obtained above will be used to develop preponderance scores (probabilities that a person would be a respondent). These scores can then be used to adjust the results of our estimates. Estimates can be obtained with and without adjustments, and a decision on whether to use them will be made by CNAC in consultation with the Commission.

A limited non-response bias analysis will also be performed for the National VSO survey. We will have only limited information about the targeted population (e.g., how many VSOs were contacted from each service organization) and will compare respondents with the total targeted population on whatever attribute information we have in order to identify potential non-response bias.

4. Describe any tests of procedures or methods to be undertaken. Testing is encouraged as an effective means of refining collections to minimize burden and improve utility. Tests must be approved if they call for answers to identical questions of 10 or more individuals.

A pretest of the veteran and survivor survey instruments was conducted to refine the data collection procedures and minimize respondent burden. Each pretest included nine respondents. The pretest findings indicated that respondents were put off by their lack of familiarity with the Commission and the extensive disclosure statements in the survey introduction, that they thought the instrument failed to capture the information stated in the objectives, and that some questions were irrelevant to the circumstances of elderly respondents. We made the following revisions to improve the survey procedures as a result of the pretest:

• We revised the letter from the Commission to make it more convincing

• We revised the survey introduction to reduce repetition with the initial contact letter, while still meeting nondisclosure requirements established by the Institutional Review Board

• We clarified wording of some survey items that did not come from extant surveys

• We altered branching criteria so that questions that are irrelevant to elderly respondents are skipped

• We added items on respondents’ general reaction to service-connected disabilities added to increase the face validity of the instruments.

Because the instruments are composed largely of items from extant surveys (the SF-12/36, NHIS Disability Supplement, CPS, and the General Social Survey) with established psychometric properties, we did not perform additional qualitative research.

The survey instrument for the National VSOs was pretested on August 2 and August 4. The pretest was conducted as a written paper-and-pencil survey with a total of nine respondents at two VA Regional Offices. Respondents were debriefed following their completion of the survey to identify any issues they may have with it. The instrument was then revised and refined.

5. Provide the name and telephone number of individuals consulted on statistical aspects of the design and the name of the agency unit, contractor(s), grantee(s), or other person(s) who will actually collect and/or analyze the information for the agency.

CNAC Staff—Design and Analysis

• Dr. Laurie May, Director, Health Care Operations and Policy Research Center, (703) 824-2966

• Dr. Joyce McMahon, Project Director, (703) 824-2334

• Dr. Eric Christensen, Research analyst, (703) 824-2212

• Dr. Dean Gerstein, Consultant, (202) 483-1131

• Dr. Dan Harris, Research Analyst, (703) 824-2283

• Dr. Bradley Gray, Research Analyst, (703) 824-2696

• Elizabeth Schaeffer, Associate Research Analyst, (703) 824-2992

ORC Macro Staff—Design and Data Collection

• JoAnn Kuchak, VDBC Survey Manager, (301) 572 0269

• Dr. Pedro Saavedra, Senior Sampling Statistician, (301) 572-0273

• Dr. Erika Gordon, Senior Analyst, (301) 572-0881

19. Terms of Clearance

Terms

VA shall provide OMB with a report detailing the findings and results of the study

including the results of any non-response bias analysis conducted by the Department.

Response

VA will provide OMB with the findings and results of the study when they become final. The non-response bias analysis is described on p. 27-28 of this justification.

13