Supporting Statement B

Status: Original and derived artifacts are available for this document.

Download: docx | pdf | html



Supporting Statement for United States Energy and Employment Report Data Collection
     Part B: Collections of Information Employing Statistical Methods
OMB No. 1910-5179
 

Table of Contents
Part B: Collections of Information Employing Statistical Methods	i
B.1. Respondent Universe	1
B.2. Statistical Methods	2
B.3. Maximizing Response Rates	3
B.4. Test Procedures and Form Consultations	6
B.5. Statistical Consultations	6


B.1. Respondent Universe
Describe (including a numerical estimate) the potential respondent universe and any sampling or other respondent selection methods to be used. 
The USEER survey universe is built from establishments in the 50 states, District of Columbia, and U.S. territories (i.e., Puerto Rico and U.S. Virgin Islands) that are within a defined set of North American Industry Classification System (NAICS) codes and have at least one permanent employee (includes owner operated or sole proprietor establishments).1 The major NAICS sectors represented in USEER include, but are not limited to:
21 – Mining, Quarrying, and Oil & Gas Extraction (oil, gas, coal, mining) 
22 – Utilities (electric power generation, transmission, distribution, and storage) 
23 – Construction (energy efficiency, transmission, generation facility construction) 
31–33 – Manufacturing (equipment, vehicles, batteries, components, fuels processing) 
42 – Wholesale Trade
48–49 – Transportation and Warehousing (including pipeline-related activities) 
54 – Professional, Scientific, and Technical Services (engineering, energy consulting, R&D) 
56 – Administrative and Support Services

A Qualifying Energy Establishment is a business location in the U.S. that is directly involved with researching, developing, producing, manufacturing, distributing, selling, implementing, installing, or repairing components, goods, or services related to:
    • TDS;
    • EPG;
    • Fuels, including extraction, processing, production, and distribution;
    • EE, including heating, cooling, and building envelope; and 
    • Transportation, including MV&CP. 
This definition also includes supporting services such as consulting, finance, tax, and legal services related to energy. To qualify for the base USEER survey, firms must explicitly state their involvement in energy as outlined above. 
     
The 2027 USEER retains the core USEER sectors—Fuels; Electric Power Generation; Electric Power Transmission, Distribution, and Storage; Energy Efficiency, including heating, cooling, and building envelope work; and Transportation, including Motor Vehicles and Component Parts—and adds or clarifies coverage for process heat, carbon capture, carbon utilization and carbon storage.     Unlike the 2026 instrument, the 2027 instrument does not use a separate facilities-construction eligibility or routing path. Construction establishments qualify through the same energy-activity screening process as other establishments when their work directly supports a qualifying energy activity.                         
A Qualifying Energy Worker is an employee of a Qualifying Energy Establishment who spends some portion of his or her time supporting the qualifying energy portion of the business. 
It is important to note that the USEER sampling frame expressly excludes any employment in retail trade NAICS codes except for fuel dealers (NAICS 454310). This approach excludes motor vehicle dealerships, gas stations, appliance and hardware stores and other retail establishments. Any sales language in the survey instrument applies only to in-frame wholesale/non-retail establishments. Establishments can be of any ownership, and include federal, state, and local government entities.
The 2027 USEER will use a stratified sampling plan that is representative by industry code, establishment size, and geography to determine the proportion of establishments that work with specific energy-related technologies, as well as the proportion of workers in such establishments that use the same. The sample will include approximately 45,406 selected establishments, with a target of approximately 10,000-12,000 completed surveys (an expected qualifying/full-survey completion rate of 22-26%).2 The survey data are then analyzed and applied to existing public data published by the Bureau of Labor Statistics, effectively constraining the potential universe of energy establishments and employment.

Statutory methodology continuity. Consistent with 42 U.S.C. § 18841(b)(2), the 2027 USEER retains the core design of the USEER methodology approved by OMB in 2016 under OMB Control Number 1910-5179. The design remains establishment based and uses geography, NAICS industry, and establishment size as core frame, stratification, or sample-allocation variables. QCEW establishment and employment totals continue to inform sample allocation and statistical estimation, while the operational sampling frame combines a known universe of establishments with evidence of energy-related activity and a broader national establishment frame used to identify additional qualifying establishments. The 2027 implementation updates data sources and statistical procedures where appropriate while retaining the historical state and NAICS domains necessary to support comparison with the 2016 and 2017 USEER surveys. 
B.2. Statistical Methods 
Describe the procedures for the collection of information including: 
Summary of changes from the previously approved instrument. The 2027 USEER revises the currently approved collection under OMB Control Number 1910-5179. Major changes include revised energy-activity and establishment screeners; clearer establishment-level employment reporting; revised workforce demographic, hiring, recruitment, apprenticeship, and workforce-development questions; new environmental-compliance, establishment-activity, and artificial-intelligence measures; and replacement of the prior occupation wage module with an occupation staffing module. The separate facilities-construction eligibility and routing path used in 2026 has been removed. As described in Supporting Statement A, DOE estimates 45,406 annual responses and 8,742 annual burden hours. A qualifying energy-establishment response is estimated at 45 minutes for burden-calculation purposes. Supporting Statement A provides the overall burden estimate; this Part B describes the statistical, quality-control, response-rate, and testing procedures relevant to the revised instrument.

Continuity with the 2016 OMB-approved methodology and statutory comparability. Section 18841(b)(2) requires the USEER survey and analysis to use the methodology approved by OMB in 2016 under OMB Control Number 1910-5179, employ representative stratified sampling, and be designed to elicit a comparable number of responses from businesses in each State and from businesses with the NAICS codes represented in the 2016 and 2017 USEERs. The 2027 design preserves the principal features of the historical USEER design, including establishment-level data collection; use of geography, NAICS industry, and establishment size in sample design and allocation; use of QCEW establishment and employment totals as statistical controls; and use of both a known universe of establishments with evidence of energy-related activity and a broader establishment universe used to identify additional qualifying establishments. 

The 2027 design strengthens this framework through probability-based sample selection, explicit analysis weights, nonresponse adjustment, calibration, and variance estimation that accounts for the stratified sample design. To address the statutory comparability requirement, DOE will establish historical benchmarks using the best available 2016 and 2017 USEER sample and response-disposition records. The benchmarks will identify the number and distribution of completed responses by state and by the historical NAICS codes represented in those surveys. Expected 2027 incidence (qualification) rates will be used to translate these completion benchmarks into sample-allocation and fieldwork targets. 

The design will seek to preserve the state and NAICS composition of the 2016 and 2017 surveys while accounting for changes in the establishment population, frame coverage, response propensity, and NAICS classification. Any 2027 NAICS domains added to address current energy activities will supplement, rather than substitute for, the historical NAICS domains used to meet the statutory comparability objective. 
Sampling Framework. As stated above, the target population for the USEER includes establishments in the 50 states, District of Columbia, and U.S. territories (i.e., Puerto Rico and Virgin Islands)  that are within a defined set of NAICS codes and have at least one permanent employee. BLS’ Quarterly Census of Employment and Wages (QCEW) data provide establishment and employee totals across in-scope industries defined by NAICS; this external dataset establishes a baseline employment universe on which sample allocation decisions may be based. 
Because the QCEW establishment sampling frame is confidential, the USEER sampling frame of establishments will be formed from two sources: (1) a listing of establishments known to have employees performing energy activities in the industries of interest—this listing is herein referred to as the “known universe” and may incorporate industry association databases, approved utility contractor lists, responding establishments from prior iterations of USEER, and other public and private sources —and (2) a commercially available list of establishments with national coverage from a vendor, such as Data Axle, and referred to as the “unknown universe.” The industry coverage of the unknown universe will be defined based on a set of approximately 200 NAICS codes (6-digit) of interest provided by DOE.      
To identify and mitigate duplicate coverage across the two frame sources, the known universe will be matched against the commercially available list by the frame vendor. The matching will also provide a consistent designation for industry code and estimated employment size across the two frame sources.           To preserve the project timeline for the 2027 USEER, the remaining establishments in the known universe may be excluded from the sample rather than traced for updated information. It is assumed that, if the unmatched establishments truly exist as part of the study’s target population, they will be reasonably represented on the commercial list.       
Conceptually, the full sampling frame will consist of the unique set of reverse-matched known universe establishments and those from the commercial list within the NAICS codes of interest. At a minimum, the frame will include the following variables to be used in sampling and weighting:
    • State / Physical address
    • Contact information (e.g., mailing address, phone, email as available)
    • Estimated number of employees
    • NAICS code (6-digit)
     
Stratification. Consistent with the historical USEER methodology, the 2027 sample will use geography, NAICS industry or documented energy-sector groupings of NAICS industries, and establishment size as core design variables. Energy-sector groupings may be used as higher-level operational strata, but each establishment’s six-digit NAICS code will be retained on the sampling frame and used, as appropriate, for sample allocation, weighting, fieldwork monitoring, and evaluation of comparability with the 2016 and 2017 NAICS response benchmarks. Establishments will be further stratified by employment size where appropriate, and large establishments or other defined groups may be selected with certainty. For example, establishments may be classified into three categories (e.g., small, medium, large). Large establishments (e.g., 250+ employees) may be selected with certainty (probability = 1) because they account for an outsized share of total energy employment relative to their count. 

Prior-cycle information on qualifying incidence, response rates, and variance may be used to improve sampling efficiency, including through differential allocation or oversampling of detailed NAICS codes. These adjustments will be implemented in a manner that maintains the representative stratified design and the state and historical-NAICS comparability objectives described above. 

The extent to which industry sectors (i.e., groups of detailed NAICS codes within the five key energy sector-based strata) on the commercial list vary in terms of their incidence rates of qualifying establishments is unknown. Oversampling detailed industry codes (6-digit NAICS) with higher incidence rates is beneficial for fieldwork efficiency and statistical precision. If incidence rate estimates by 6-digit NAICS code are available from prior USEER cycles, this information may be used to inform additional substrata to oversample those industries. The final stratification scheme would be developed in collaboration with DOE according to their analytic priorities. 
In practice, some strata may be collapsed due to a small number of frame establishments in the geographies and size categories of interest. This is especially true for less populous states and territories. Ideally, stratification and collapsing decisions will be informed by precision requirements for key estimates and tabulations rather than fixed minimum sample sizes per stratum. If variance estimates from prior iterations of the USEER are available, they may be used to inform the level of stratification granularity for which reliable estimates are likely to be supported.
Sample Size & Allocation. The 2027 USEER sample will consist of approximately 45,406 selected establishments with a target of approximately 10,000-12,000 completed qualifying/full-length surveys. This will help ensure that two key statistical targets are achieved:
    • accurate and reliable results with a statistical margin of error of +/-4% or less on a national level and at a 95% confidence interval. 
    • sufficient sample size such that the distribution of completed responses is comparable to the 2016 and 2017 USEER surveys by state and by the historical NAICS domains represented in those surveys, as required by 42 U.S.C. § 18841(b)(2)   
     
For efficiency in identifying establishments with qualifying employees, the known universe portion of the frame will be selected with certainty. Based on prior iterations of the USEER, the known universe is expected to include about 12,000 establishments. Among the unknown universe (excluding known universe establishments in common), the remainder of the selected sample size (i.e., 45,406 minus the size of the known universe) must be allocated to the approximately 1,113 strata described above (assuming 53 states/territories x 5 energy sectors x 3 size classes). This results in an average of approximately 42 sampled establishments, and 9 responding establishments, per stratum. 
If readily available, incidence rates (i.e., proportion of qualifying firms in each previously used stratum) and variance estimates from prior cycles will be used to inform sample allocation for the 2027 USEER, Otherwise, initial allocation across strata for the 2027 USEER will be done proportionally to the square root of the employee counts represented in each stratum (i.e., total employment across both unknown and known universes after deduplication), treating the known universe as a certainty component, and assigning the remaining sample to the unknown universe in proportion to its residual need for representation. 

The square-root allocation gives larger strata more sample while preserving greater representation in smaller domains than would generally result from a strictly proportional-to-employment allocation. This approach is intended to balance precision for national employment estimates with DOE’s need to produce reliable estimates for states, territories, energy sectors, and other priority domains. 

QCEW employment totals will be used when available and sufficiently aligned with the sampling strata; otherwise, the best available frame employment measures will be used. Ideally, these employee counts will come from the QCEW, as employment information from the QCEW is believed to be a more reliable source than the employment estimates from the commercial list. Other sources may be integrated to inform allocation depending on the level of granularity available from QCEW. Minimum sample sizes for each stratum for which precise estimation is required may be specified. In this case, the sample will still be allocated proportionally, subject to the constraint that each stratum has at least the minimum sample size. At least two responding units per stratum are      required to estimate variance for that stratum.           
Occupation staffing module. The 2027 USEER replaces the prior occupational wage module with an occupation staffing module. The module collects the distribution of the sampled establishment’s energy-related employees across occupations and does not ask respondents to report occupation-level wages, salaries, benefits, or payroll amounts. Respondents report the complete staffing profile for the energy-related employee total previously reported for the establishment.
Replacing the 2026 wage module with an occupation staffing module reduces the need for respondents to report detailed, potentially sensitive wage and benefit information that may not be readily available. It also provides a more complete picture of the occupations supporting energy-related work and allows DOE to combine establishment staffing data with reliable federal occupational wage data for more consistent analysis.
The web instrument will use the survey-year O*NET-Standard Occupational Classification (SOC) occupation list as a controlled lookup rather than displaying the full detailed occupation list as a static grid. Respondents using the detailed occupation path will search, browse, review descriptions, compare candidate occupations, and select only occupations present at the sampled establishment. The staffing grid will display only the occupations or major groups selected by the respondent.
The module uses an adaptive burden-reduction design. Establishments reporting 50 or fewer energy-related employees begin with detailed occupation reporting but may switch to broader O*NET-SOC major groups if detailed information is not reasonably available. Establishments reporting more than 50 energy-related employees may provide a detailed occupation profile or report using broader O*NET-SOC major groups. Respondents report the complete staffing profile at one level of detail rather than mixing detailed occupations and major groups within the same establishment profile. Detailed occupation responses can be aggregated to O*NET-SOC major groups, permitting a major-group profile to be constructed for all respondents while preserving detailed information where available.
The module permits counts or approximate percentages; uses instructions based on primary job duties; provides plain-language and full occupation descriptions; supports comparison of ambiguous occupations; includes total checks and correction opportunities; permits “Other or uncertain” follow-up text; records reporting granularity and classification-difficulty flags; and asks a short burden question. External wage benchmarks may be attached to the reported occupational level through a separately documented statistical method. Any optional wage calibration study would be treated as a separate collection component or separately documented change, as appropriate.
Data Editing and Handling of Item Nonresponse. Data editing will retain comparable edits used in prior USEER cycles where questions remain comparable, but the 2027 instrument also flags inconsistencies and allows correction or continuation when answers are best estimates while the occupation staffing module applies stricter total checks for counts and percentages. The instrument will allow a combination of soft edits and hard edits. This means respondents can provide best estimates when exact records are not readily available and so item nonresponse and breakoff are minimized.
For employee-count, demographic, occupational, and similar allocation items, the instrument will display or read consistency checks when component counts or percentages do not match the applicable total. Respondents will be given an opportunity to revise the component entries, correct the earlier total, or continue if the responses are best estimates. Raw responses will be preserved. Where analytic scaling is applied after an unresolved nonzero mismatch, scaling factors and flags will be retained in the analytic file.
For the occupation staffing module, the data file will preserve the reporting level selected by the respondent, the selected detailed O*NET-SOC occupations or O*NET-SOC major groups, raw counts or percentages, any scaled analytic values, the amount reported in “Other or uncertain,” open-ended classification text, total-edit flags, and burden/classification feedback. Detailed O*NET-SOC responses may be aggregated to O*NET-SOC major groups. Major-group responses will be retained at the major-group level unless any disaggregation is performed through a separately documented statistical method.
For race and ethnicity, DOE will collect the seven SPD 15 minimum categories using employee self-identification information available in payroll, human resources, or other business records. The instrument will not require the seven category counts to sum to the employee total because employees may be identified with more than one category. Record-status rows for generic multiracial/multiethnic records and unavailable or unmappable records will be retained separately from race/ethnicity categories. DOE will not impute an employee’s race or ethnicity from name, appearance, language, occupation, geography, or other survey responses.
Methods to address item nonresponse will be determined after assessing the level and pattern of nonresponse and the potential for nonresponse bias. Imputation may be considered for a limited number of key outcome variables when appropriate. Streamlined methods such as weighted sequential hot deck, mean imputation, or carry-forward imputation for recent known-universe respondents may be used where methodologically justified. Some survey respondents may be treated as unit nonrespondents because of high item nonresponse or other data-quality issues.
     Race and ethnicity data and requested OIRA authorization for minimum-category collection. The 2027 race and ethnicity item is designed for an establishment survey in which a knowledgeable business respondent reports counts from payroll, human resources, or other business records. DOE will request OIRA authorization through the PRA process to collect the seven SPD 15 minimum categories without detailed race and ethnicity subcategory follow-up. DOE’s rationale is that the respondent is reporting by proxy from establishment records; many establishments may not maintain detailed subgroup information for the energy-related employee subset; collecting all detailed categories would require many additional employee-count fields; detailed establishment-level cells would often be very small, increasing confidentiality risk and limiting publishable utility; and the additional burden would likely produce substantial missing, estimated, or inconsistent data. The instrument preserves the seven minimum categories, permits “alone or in combination” reporting where records support it, records whether information is available for all, some, or none of the employees, and separately records data availability limitations.
Weighting. Three weight components will be developed to account for use of a complex sample design, mitigate bias due to observed survey nonresponse, and reduce coverage differences between the sample and target population.
    • Base weight. For each establishment, the base weight is the reciprocal of the selection probability within its sampling stratum, which was calculated using the sample sizes and the employee counts by stratum. For establishments selected with certainty, such as known universe establishments and large establishments, the base weight is 1.  
    • Nonresponse adjustment factor. A raking-type adjustment, such as the one implemented by the WTADJUST procedure in SUDAAN or the survey package in R, will be used to adjust the weights for unit nonresponse and reduce bias due to differential response rates among subpopulations. Frame variables that are related to the response propensity, the key outcome variables, or both will be used as auxiliary variables. 
    • Calibration adjustment factor. The counts obtained from the QCEW (and other sources as needed) will be used in weight calibration to ensure that the estimates agree with other US government reports to the extent possible. This adjustment can also correct for coverage error in the frame.
     
The final analysis weight is the product of the three weight components. Weight trimming may be used to cap extreme analytic weights, which are often driven by large adjustments or rare selection probabilities; trimming reduces variance inflation, trading a small increase in potential bias for gains in estimate stability and precision.
Estimation. Estimates will be calculated using the final analysis weights. The variance of each estimate will itself be estimated using Taylor Series or replication methods via a statistical software package that accounts for the complex sample design (e.g., disproportionate sampling, stratification). Stratified random sampling will be implemented; the allocation of the overall sample to strata will help to keep the variance estimates for key analytic domains within an acceptable level of precision based on analytic priorities set by DOE. 
Certain estimates may be flagged or suppressed due to low precision (i.e., high variance). Small area estimation may be explored, as possible within project timeline and available funds, to improve reliability and precision of subnational estimates; small area estimation is a model-based estimation approach that borrows strength across related geographic areas and incorporates auxiliary information, producing more stable estimates for domains with limited sample sizes.

DOE will produce national, state, and county-level employment estimates consistent with the geographic reporting requirements of 42 U.S.C. § 18841. County-level estimates will be calculated using the approved estimation procedures described above and will be evaluated for statistical reliability and disclosure risk before release. Estimates that do not meet applicable precision, quality, or confidentiality standards may be flagged, suppressed, aggregated, or otherwise protected. Where methodologically appropriate and sufficiently validated, small-area estimation may be used to improve the reliability of county-level estimates with limited sample sizes.
     Occupation staffing estimates will be calculated using the final analysis weights. Detailed occupation responses will be used at the detailed O*NET-SOC level when sample size, precision, and disclosure review permit. Detailed O*NET-SOC responses may also be aggregated to O*NET-SOC major groups. Establishments reporting only O*NET-SOC major groups will contribute to major-group estimates and will be flagged by reporting granularity. DOE will not treat major-group responses as if the establishment provided detailed O*NET-SOC data unless a disaggregation method is separately documented and evaluated. Estimates may be flagged or suppressed because of high variance, high item nonresponse, high “Other or uncertain” share, unresolved total mismatches, or confidentiality risk. If DOE attaches external wage benchmarks to occupational staffing profiles, that process will be documented as a statistical estimation procedure and will be distinguished from directly collected occupation-level wage data.
B.3. Maximizing Response Rates
Describe methods to maximize response rates and to deal with issues of non-response. 
Methods to maximize response will focus on facilitating and encouraging participation. To facilitate participation, sample members will      first be asked to complete the survey online. But they will also be given the opportunity to participate in a phone interview if they prefer. As respondents complete the survey (in either mode) they will be prompted to provide missing data or resolve discrepancies between select items. For example, if the sum of workers reported by demographic characteristic exceeds the total number of workers reported earlier in the questionnaire, the respondent will be asked to reconsider/revise their answers to resolve the discrepancy.3 This approach helps to minimize item nonresponse and dealing with such instances during the interview mitigates the need for later follow-up with the respondent (i.e., minimizes burden).     
The 2027 instrument incorporates additional burden-reduction and response-quality features because several sections are new or substantially revised. These include establishment-level routing, skip patterns that display only applicable energy activities and technologies, explicit permission to provide best estimates when exact records are not readily available, and web/phone instructions that allow respondents to consult records or another knowledgeable employee for specialized sections.                
For the occupation staffing module, burden reduction is central to the design. Respondents will not be shown a full static list of all detailed O*NET-SOC occupations. Instead, the web survey will use search, autocomplete, browse-by-major-group navigation, plain-language descriptions, comparison tools for ambiguous search results, and a roster that displays only selected occupations or major groups. Respondents may report counts or approximate percentages, and establishments with more than 50 energy-related employees may report at the O*NET-SOC major-group level. Any respondent who cannot reasonably provide detailed occupational information may switch to the major-group path.
Based on information available before programming and testing, the occupation staffing section is expected to require roughly 1 to 2 minutes per occupation or occupational group reported by the establishment. The number of categories will vary substantially by establishment size and workforce complexity. The survey will collect respondent-reported time for the occupation section and web paradata, where available and approved, to refine the burden estimate and identify improvements for future cycles.
In addition, the data collection schedule and outreach methods will be designed to encourage survey participation. Survey invitations will be mailed to sample members in consecutive releases every 3 to 4 weeks. This will allow for efficient case management and responsiveness to sample members. About 1 week after the invitation is mailed, a reminder will be sent by mail and email (if an email address is available). After the reminder is sent, nonresponding cases will be assigned for phone prompting, with priority given to cases in the known universe. Cases will be monitored so that when a stratum has achieved its quota, cases in that stratum will be pulled from the prompting caseload so that resources can be applied to any remaining underperforming strata. 
Finally, statistical methods will be utilized so that the collection ultimately yields reliable data that can be generalized to the study universe. As noted in Item 2 above, imputation will be considered to address item nonresponse. In addition, DOE will assess whether respondents and nonrespondents differ on available frame characteristics and adjust weights when appropriate. A nonresponse bias analysis will be conducted. Frame data on establishment type, size, and geography will be used in the nonresponse bias analysis. For each establishment characteristic, the distribution of respondents to nonrespondents will be compared. A Cohen’s Effect Size statistic will be calculated for each characteristic. If any characteristic has an effect size that falls into the “medium” or “high” category, as defined by Cohen, then there is a potential for bias in the estimates.4 Frame characteristics associated with response and key survey outcomes will be considered in the nonresponse-adjustment model.
B.4. Test Procedures and Form Consultations
Describe any tests of procedures or methods to be undertaken. 
The 2027 USEER contains substantial revisions relative to the previously approved instrument.      
During programming the survey subcontractor may revise screen text, help text, examples, search aliases, occupation descriptions, instructions, or layout to improve clarity and reduce burden, provided the revisions do not change the substantive intent of the question, respondent universe, data elements collected, or approved burden in a material way. If testing or programming identifies a change that is more than clarifying or otherwise material to the approved collection, DOE will consult OIRA and submit an appropriate non-substantive change request or revision, as applicable.
The survey subcontractor will also use respondent-reported burden, web timing data, paradata where approved, item nonresponse, total-edit rates, and open-ended “Other or uncertain” text after the 2027 field period to evaluate the revised modules and inform future USEER cycles.     
 No large-scale pretest is planned for the 2027 collection. Depending on available resources, DOE may conduct targeted cognitive, usability, or programming tests with fewer than 10 in-scope establishments or respondents before fielding. Testing would focus on revised screeners, establishment-reporting instructions, workforce demographic questions, business-activity and environmental-compliance questions, AI questions, and the occupation staffing module. DOE will also evaluate the revised instrument after data collection using respondent-reported completion time, approved web timing and paradata, item nonresponse, edit rates, and other data-quality indicators. Material changes identified before fielding will be submitted to OIRA as appropriate.
B.5. Statistical Consultations
     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. 
Margaret Taylor, Research Scientist, Lawrence Berkeley National Laboratory 
[email protected] | 510-847-1879 
Megan Waggy, Research Survey Scientist, RTI International
[email protected] | 919-541-8744