2011-7632. Uniform Criteria for State Observational Surveys of Seat Belt Use

AGENCY:

National Highway Traffic Safety Administration (NHTSA), Department of Transportation (DOT).

ACTION:

Final rule.

SUMMARY:

This Final Rule amends the regulation establishing uniform criteria for designing and conducting State seat belt use observational surveys and the procedures for obtaining NHTSA approval of survey designs, and provides a new form for reporting seat belt use rates to NHTSA. Since the adoption of the Uniform Criteria in 1998, NHTSA and the States have accumulated substantial experience in the design and implementation of seat belt use surveys. This experience has provided insight into factors that could affect survey accuracy and reliability. In addition, technological improvements in road inventories have made it possible to select observation sites that are more representative of the road segments in the State in a more cost effective manner. For these reasons, NHTSA is revising the Uniform Criteria so that future surveys will give States more accurate data to guide their occupant protection programs.

DATES:

This Final Rule becomes effective on May 2, 2011.

FOR FURTHER INFORMATION CONTACT:

For program issues: Mr. Jack Oates, Chief, Program Implementation, Regional Operations and Program Delivery, National Highway Traffic Safety Administration, 1200 New Jersey Avenue, SE., NTI-200, Washington, DC 20590. Telephone number: 202-366-2730; E-mail: [e-mail].

For statistical issues: Dr. Chou-Lin Chen, Chief, Mathematical Analysis Division, National Center for Statistics and Analysis, National Highway Traffic Safety Administration, 1200 New Jersey Avenue, SE., NVS-421, Washington, DC 20590. Telephone number: 202-366-1048; E-mail: [e-mail].

For legal issues: Ms. Jin Kim, Attorney-Advisor, Office of the Chief Counsel, National Highway Traffic Safety Administration, 1200 New Jersey Avenue, SE., NCC-113, Washington, DC 20590. Telephone number: 202-366-1834; E-mail: [e-mail].

SUPPLEMENTARY INFORMATION:

Table of Contents

I. Background

II. Notice of Proposed Rulemaking

III. Comments

A. In General

B. General Cost

C. Definitions

D. Selection of Observation Sites

E. Assignment of Observation Times

F. Observation Procedures

G. Quality Control

H. Computation of Estimates

I. Submission and Approval of Seat Belt Survey Design

J. Re-Selection of Observation Sites

K. Annual Reporting Requirements

IV. Statutory Basis for This Action

V. Regulatory Analyses and Notices

A. Executive Order 12866 and Regulatory Policies and Procedures

B. Regulatory Flexibility Act

C. Executive Order 13132 (Federalism)

D. Executive Order 12988 (Civil Justice Reform)

E. Paperwork Reduction Act

F. Unfunded Mandates Reform Act

G. National Environmental Policy Act

H. Executive Order 13175 (Consultation and Coordination With Indian Tribes)

I. Regulatory Identifier Number (RIN)

J. Privacy Act

K. Congressional Review of Agency Rulemaking

I. Background

Section 1403 of the Transportation Equity Act for the 21st Century (TEA-21) (Pub. L. 105-178) authorized a seat belt incentive grant program that awarded grant funds to States based on a State's seat belt use rate. On September 1, 1998, the National Highway Traffic Safety Administration (NHTSA) published as an interim final rule the criteria to ensure accurate and representative measurements of a State's seat belt use rate, known as the Uniform Criteria for State Observational Surveys of Seat Belt Use (the Uniform Criteria). See 63 FR 46389. On March 14, 2000, NHTSA published a Final Rule, adopting the Uniform Criteria with one clarifying change.^[1] See 65 FR 13679.

The Safe, Accountable, Flexible, Efficient Transportation Equity Act: A Legacy for Users (SAFETEA-LU) (Pub. L. 109-59), enacted on August 10, 2005, did not reauthorize the seat belt incentive grant program. However, SAFETEA-LU established new administrative requirements relating to a State's qualification for a highway safety grant under 23 U.S.C. 402. One such requirement is that the State must provide satisfactory assurances that it will conduct an annual Statewide seat belt use survey in accordance with the criteria for State seat belt use rate measurement established by the Secretary of Transportation.^[2] In August 2005, NHTSA notified the States and Territories that the Statewide surveys conducted in accordance with the Uniform Criteria for State Observational Surveys of Seat Belt Use, as published at 23 CFR part 1340, would satisfy the administrative requirements of Section 402. In addition, the implementing guidelines for the safety belt performance grant program under 23 U.S.C. 406 provide that seat belt use surveys conducted in accordance with the Uniform Criteria serve as the basis for an award under the seat belt performance provisions of that grant program.

Since the adoption of the Uniform Criteria in 1998, NHTSA and the States have accumulated substantial experience in the design and implementation of seat belt use surveys. This experience has provided insight into factors that could affect survey accuracy and reliability. In addition, technological improvements in road inventories have made it possible to select observation sites that are more representative of the road segments in the State in a more cost effective manner. For these reasons, NHTSA proposed to revise the Uniform Criteria so that future surveys would give States more accurate data to guide their occupant protection programs.

II. Notice of Proposed Rulemaking

On January 28, 2010, NHTSA published a notice of proposed rulemaking (NPRM) to amend the Uniform Criteria. See 75 FR 4509. NHTSA proposed several key changes to the 1998 Uniform Criteria. In particular, the agency proposed to revise the geographic coverage of the sampling frame from the population-based exclusion criterion to a fatality-based exclusion criterion and to identify the road types that are required to be included in a State's sampling frame. The proposal also changed the precision requirement from a five percent relative error to a 2.5 percentage point standard error. In addition, the agency proposed quality control procedures, such as quality control monitors, training and statistical review, to help ensure accuracy and consistency across all State surveys. Finally, the agency proposed submission of additional information from the survey results as part of a State's annual certification, including the data source of the sampling frame, exclusions applied to the sampling frame, procedures for collecting additional data to reduce the nonresponse rates, explanation of any imputation methods, procedures to adjust the sampling weight, and procedures to be followed if the standard error is exceeded.

III. Comments

By the close of the comment period on March 29, 2010, the agency received submissions from 27 commenters in response to the NPRM. Commenters included the following State agencies: California Office of Traffic Safety (CA OTS), Colorado Department of Transportation (CO DOT), Idaho Transportation Department (ID DOT), Illinois Department of Transportation (IL DOT), Iowa Governor's Traffic Safety Bureau—Department of Public Safety (IA TSB), Kansas Department of Transportation (KS DOT), Louisiana Highway Safety Commission—Department of Public Safety and Corrections (LA HSC), Maine Bureau of Highway Safety (ME DPSC), Missouri Highway Safety Division—Department of Transportation (MO DOT), Nevada Department of Public Safety (prepared by University of Nevada—Las Vegas) (NV DPS), New Hampshire Highway Safety Agency (NH HSA), New York Governor's Traffic Safety Committee—Department of Motor Vehicles (NY TSC), North Dakota Department of Transportation (ND DOT), Oregon Department of Transportation (OR DOT), Pennsylvania Department of Transportation (PA DOT), Texas Department of Transportation (TX DOT), Washington Traffic Safety Commission (WA TSC), West Virginia Governor's Highway Safety Program (WV HSP), Wisconsin Division of State Patrol, Bureau of Transportation Safety—Department of Transportation (WI State Patrol), Wyoming Highway Safety Program—Department of Transportation (WY HSP). Additional commenters included two associations—Governor's Highway Safety Association (GHSA) and International Association of Chiefs of Police (IACP); three professors and staff—Mississippi State University (MS State Univ.), New Jersey Institute of Technology (NJIT) and Old Dominion University (ODU); one consultant to a State—Peters and Associates Engineers Inc. (Peters & Assoc.); and one interested member of the public.

A. In General

Several commenters expressed general support for revising the criteria. These commenters stated that the changes to the protocol are appropriate and timely and will enhance the accuracy and consistency of seat belt use surveys. (E.g., GHSA at 1; WI State Patrol at 1; WA TSC at 1; TX DOT at 1; CA OTS at 1. See also NV DPS at 9; ODU at 1.) In addition to expressing general support for revising the criteria, these commenters also had more specific comments regarding different aspects of the proposal. The agency addresses these comments below under the appropriate heading.

Some commenters expressed general concern with revising the Uniform Criteria. One commenter suggested reducing the frequency of State observational surveys, and one commenter suggested expanding the National Occupant Protection Use Survey (NOPUS) to each State instead of requiring States to conduct independent surveys. (WI State Patrol at 1; CO DOT at 1.) We decline to adopt these commenters' suggestions as Section 402 requires each State to provide assurances that it will conduct annual Statewide seat belt use surveys in accordance with the Uniform Criteria to ensure that the measurements are accurate and representative.

One commenter believed that changing the survey criteria at the end of the authorization was not cost effective unless a seat belt incentive program formed a part of the future authorization. (LA HSC at 1.) In the NPRM, we stated that the purpose of revising the criteria was to improve the accuracy and reliability of surveys conducted by States. We believe it is necessary to do so now based on our experience reviewing State survey results. Regardless of whether a seat belt incentive program is part of a future authorization, improved data will enable States to guide their highway safety program evaluation and program management more effectively now.

The NH HSA stated that the most significant effect of the proposed change would be damage to trend information. (NH HSA 1.) The commenter further stated that policy analysis based on previous methodology would no longer be relevant as a tool to measure seat belt usage. Id. For some States, seat belt use rate estimates obtained from a survey meeting the new criteria may depart from the trend of survey outcomes in recent years. However, any departure from the trend will reflect the fact that the data will be more accurate and more reliable. Specifically, observation sites will be drawn from a more up-to-date and comprehensive road inventory. The seat belt survey will also be less biased toward urban areas due to the shift from a population-based exclusion to a fatality-based exclusion. (See discussion in Section III.D.1 below.) Finally, the survey will have greater precision due to the shift from a five percent relative error to a 2.5 percentage point standard error. (See discuss in Section III.D.5 below.) NHTSA believes that the need for more accurate and reliable data outweighs concerns about departure from trends reflected under the 1998 Uniform Criteria.

The CO DOT stated that the proposal did not address the large gap in data through lack of nighttime observations. (CO DOT at 1.) As we stated in the NPRM, although nighttime observations of seat belt use may provide States with useful data, the agency believes that several factors weigh against extending the sampling requirements. First, extending the sampling requirement to nighttime observations would reduce the value, for comparison purposes, of survey results from previous years' data. States and other interested parties use this information to determine the impact of various seat belt use programs and activities. In addition, seat belt use is difficult to reliably observe in the dark, even in the most well-lit areas. Nighttime observations are also less safe for data collectors than daytime observations because data collectors are less conspicuous and exposed to an increased presence of impaired drivers. For these reasons, the agency declines to change the rule in response to this comment.

The LA HSC suggested that traffic cameras positioned in predetermined locations will provide higher data quality at a fraction of the costs of manual collection and will reduce exposure of data collectors to highway hazards. (LA HSC at 1.) The Uniform Criteria do not prohibit States from using traffic cameras to conduct seat belt use observations. However, States must still comply with the other provisions of the Uniform Criteria, such as observation procedures (§ 1340.7) and quality control (§ 1340.8). With this clarification, no change is made to the rule.

The OR DOT suggested that States should be allowed to continue using their current survey methodology but adjust sampling to re-weight observation sites using the proposed fatality-based criterion. (OR DOT at 2.) Much of the survey methodology in the revised criteria is a clarification of the 1998 Uniform Criteria. The major change for most States will be in the sampling frame—changing from a population-based exclusion of counties to a fatality-based exclusion of counties and an updated road inventory. Because of this change, we expect that the large majority of States will have to re-select a probability sample of observation sites. However, some States already may be in close compliance with the revised criteria and may not need to make significant changes to their current survey design.

NHTSA received four comments requesting additional guidance. The IACP and WV HSP requested greater guidance and technical assistance on conducting surveys, including common data collection procedures and approaches for calculation of sampling error estimates. (IACP at 1; WV HSP at 2.) The OR DOT and NV DPS requested a sample of an acceptable survey design for States. (OR DOT at 2; NV DPS at 7.) The agency intends to provide technical assistance, such as providing county-by-county breakdowns of passenger motor vehicle occupant fatalities and an inventory of roads. In addition, NHTSA is developing a sample of an acceptable survey design to assist the States' redesign efforts. This sample design will provide general guidance for designing a seat belt use survey, including the calculation of survey standard error. However, the Final Rule still requires States to rely on their own statistician to design, conduct and analyze the data. As we discuss in Section III.G below, the purpose of requiring statistician involvement is to ensure that both the survey design and the annual reporting of seat belt use rates are carried out in a methodologically-sound manner. No change to the Final Rule is made in response to these comments.

Two commenters mentioned a “self-report survey” and a “public opinion survey” as further increasing costs to the States. (IACP at 2; CO DOT at 1.) We believe the commenters are referring to the annual public opinion survey that States voluntarily agreed to conduct and report on as part of their core highway safety program performance measures. We note that the public opinion survey is not mandatory and not related to the seat belt survey criteria. Therefore, we do not address these comments here.

The OR DOT stated that the revised criteria require new performance measures. (OR DOT at 2.) The revised criteria do not impose a performance measure—States are not required to meet a specific seat belt use rate. Rather States are required to conduct surveys that are consistent with the revised criteria so that seat belt use rate estimates are more accurate and reliable.

The agency received more specific comments regarding different aspects of the proposal, including requests for clarification and recommendations to change the proposal. The agency addresses these comments below under the appropriate heading.

B. General Cost

One interested member of the public stated that conducting such surveys is a waste of taxpayer's money. (Jean Public at 1.) As discussed in Section III.A. above, States are required by statute to provide assurances that it will conduct annual Statewide seat belt use surveys in accordance with the Uniform Criteria to ensure that the measurements are accurate and representative. See 23 U.S.C. 402.

A number of commenters expressed general concerns regarding increased costs as a result of the revised Uniform Criteria, especially related to the number of observation sites, and the requirements for quality control monitoring and additional observations as a result of nonresponse rates. (WA TSC at 4; OR DOT at 2; IACP at 2; ID DOT at 1; WY HSP at 1; LA HSC at 1; NJIT at 3; CO DOT at 1; GHSA at 1; ODU at 1-2; MO DOT at 1; CA OTS at 1.) Some of these commenters stated that the additional costs would divert resources away from other programs. (CO DOT at 1; LA HSC at 1; GHSA at 1; ID DOT at 1.) NHTSA understands that the new criteria may impose additional costs for some States, especially States that will need to conduct observations in more counties and at more observation sites. Based on our review of the changes required under the new criteria and States' current seat belt use surveys approved under the 1998 Uniform Criteria, we do not believe that this will significantly increase costs for most States.^[3] However, the changes to the Uniform Criteria will yield better data, and this will improve the States' identification of low seat belt use problem areas and permit more effective targeting of countermeasures to increase seat belt use. Accordingly, States will be able to target their life-saving programs more effectively, resulting in fewer fatalities. For these reasons, we believe that the improved quality and reliability of the survey outweighs the additional costs.

Several commenters suggested that NHTSA provide additional funding to assist States or that NHTSA redesign 56 surveys and analyze the data from those surveys. (ODU at 1-2; MO DOT at 2; WA TSC at 2, 4.) Although NHTSA intends to provide technical assistance to States, including a road inventory and a fatality distribution, NHTSA does not have the resources to provide States with additional funds, and NHTSA does not have the resources to redesign all 56 surveys and analyze the data. Currently, all States receive NHTSA grant funds, such as Section 402 program funds (23 U.S.C. 402), which may be used to design and conduct surveys and analyze data. States may also use other grant funds, such as Section 406 program funds (23 U.S.C. 406), to redesign and conduct surveys. Consequently, NHTSA declines to adopt these commenters' suggestion.

C. Definitions (§ 1340.3)

GHSA asked for an explanation of the source of roadway-related definitions. (GHSA at 1.) The agency relied on the U.S. Census Bureau's TIGER/Line files, Second Edition (2006) for the definition of most of the roadway-related terms. Specifically, the definition for access ramp, cul-de-sac, vehicular trail, service drive, and traffic circle comes from the TIGER/Line files. The TIGER/Line files are typically used in conjunction with geographic information system (GIS), or similar, software. Because the database of road segments that NHTSA will provide to States will be the GIS population of roads, we relied on the definitions used in the TIGER/Line files. The other two terms (nonpublic road and unnamed road) are not defined in the TIGER/Line files, but are defined to reflect a common understanding of these terms.

The TX DOT asked whether the definition of passenger motor vehicle included limousines and other for-hire vehicles, whether a pickup truck included a wrecker tow vehicle, a flatbed 3 or 4 ton truck or a utility service truck, and whether a van included any size or type a van. (TX DOT at 3.) In the NPRM, passenger motor vehicle was defined as “a passenger car, a pickup truck, van, minivan or sport utility vehicle.” Generally, most passenger cars, pickup trucks, vans, minivans and sport utility vehicles (SUVs) are motor vehicles with a gross vehicle weight rating (GVWR) of less than 10,000 pounds (lbs). In the proposal, NHTSA intended those vehicles with a GVWR of under 10,000 lbs to be included in the seat belt use survey. To clarify this point and in response to this comment, we amended the definition of “passenger motor vehicle” to read as follows: “A motor vehicle with a gross vehicle weight rating of less than 10,000 pounds, including a passenger car, pickup truck, van, minivan or sport utility vehicle.” Accordingly, those vehicles, including limousines, for-hire vehicles, wrecker tow vehicles, flatbed 3 or 4 ton trucks, utility service trucks and vans that are under 10,000 lbs GVWR must be included in the seat belt use survey.

D. Selection of Observation Sites (§ 1340.5)

1. Sampling Frame: Exclusion

The agency received many comments regarding the sampling frame requirements. It appears that most of these comments reflect a misunderstanding of the sampling frame requirements. Much of this misunderstanding appears to stem from the fatality-based exclusion at the county level, as specified in § 1340.5(a). (E.g., ID DOT at 1; Peters & Assoc. at 1-2; IA TSB at 1; MS State Univ. at 1; ND Dot at 1; NV DPS at 5; OR DOT at 1; PA DOT at 1; TX DOT at 1; WA TSC at 4-5; WV HSP at 1-2; WY HSP at 1; NH HSA at 1.) Many commenters mistakenly referred to a “fatality-based survey” or a “fatality-based criterion”. (E.g., Peters & Assoc. at 1-2; GHSA at 2; IA TSB at 1; NV DPS at 5.) For example, one commenter stated that the switch from a population-based to a fatality-based seat belt survey is flawed and will produce far less representative results of actual safety belt use for a number of reasons: (1) The location of fatalities is not necessarily representative of where people live, work and drive; (2) fatal crashes are not representative of where injury crashes occur and not representative of where property damage crashes occur; and (3) a fatality-based survey is not representative of exposure or population, which are two critical components of a measurement of seat belt use. (IA TSB at 1.) Some commenters asked how site selection can be based on both a fatality and county-by-county basis. (E.g., OR DOT at 1; Peters & Assoc. at 1.) We believe that these comments capture the general misunderstanding among commenters regarding the fatality-based exclusion.

As a general matter, in a survey that covers a large geographic area, such as a Statewide seat belt use survey, it may be costly to send data collectors to a random sample of observation sites across the large geographic area. To reduce travel costs and time of collection, the large geographic area is divided into subareas, or primary sampling units, and a sample of these primary sampling units is selected. Data are then collected within these primary sampling units. For Statewide seat belt use surveys, the large geographic area is the State itself, and the primary sampling units are counties (or county equivalents). Because States believed that the costs were too great to send data collectors to a randomly selected sample of observation sites across the State, in the 1998 Uniform Criteria, NHTSA allowed the State to select a probability-based sample of its counties and then to select observation sites within the selected counties in which to count seat belt use observations. The selection of counties is called the first stage of the sample. A number of States had many small counties with little population and road traffic. Thus, the 1998 Uniform Criteria allowed States to exclude the smallest counties with a cumulative 15 percent of the population from the first stage sampling frame, i.e., a population-based exclusion.

Over time, NHTSA became aware that some small counties had measurable road traffic because they contained major roads. The traffic on these roads resulted in a disproportionate share of the motor vehicle related fatalities compared to the population of the county. This circumstance was the basis for NHTSA's proposal to change the exclusion criteria in the revised criteria from a population-based exclusion to a fatality-based exclusion in the first stage sampling frame. With this change, counties that have few fatalities may be excluded from the seat belt survey. We note, however, States are not required to exclude counties from the sampling frame, and may include and sample randomly from all counties. (See KS DOT at 1.) Accordingly, under the new criteria, the State may identify any set of counties that collectively account for 15 percent of the State's passenger motor vehicle occupant fatalities, and that set of counties may be excluded from the first stage sampling frame.

The main purpose for allowing any exclusion, whether fatality-based or population-based, is to control operational costs. As explained above, the purpose of the exclusion is to help States reduce travel costs and time of collection by excluding areas where little data are likely to be collected. The fatality-based exclusion does not directly affect the selection of the actual observation sites within eligible counties. In other words, neither the number of fatalities in a county nor the specific locations where those fatalities occurred should serve as the basis for selecting the observation sites. (See LA HSC at 2.)

Some commenters stated that shifting from a sampling frame of counties accounting for at least 85 percent of the Statewide population to one that includes at least 85 percent of the passenger vehicle occupant fatalities seems to replace a potentially biased sampling frame with one that is almost certainly biased. (E.g., WA TSC at 1-2; GHSA at 1-2. See also IACP at 2; NV DPS at 2, 4; ODU at 2; NJIT at 1-2.) NHTSA believes that, by permitting the systematic exclusion of the State's most sparsely populated counties from the sampling frame, the 1998 Uniform Criteria created an urban bias. While we are not eliminating the urban bias, we believe we are reducing it in many States by replacing the population-based exclusion with the fatality-based exclusion. For some States, the change to a fatality-based exclusion may not have any impact on the sampling frame. For example, under a population-based exclusion, States systematically set aside sparsely populated counties from the sampling frame. Therefore, States end up with more urban areas in the first stage sampling frame. A fatality-based exclusion is less urban-biased with respect to seat belt use because States are not systematically eliminating all low population areas, but are required to include those low population counties which have enough fatalities to be included in the sampling frame. It is not uncommon in many States for a sparsely populated county to have high traffic volume, sometimes resulting in relatively frequent crashes and deaths. A population-based exclusion may eliminate that county while a fatality-based exclusion may keep that county in the sampling frame.

Other variables, such as crash or fatality rates, registered vehicles, vehicle miles traveled (VMT), could be used as the basis for excluding counties from the sampling frame. (See MS State Univ. at 1.) NHTSA decided to use passenger vehicle occupant fatalities in motor vehicle crashes as the measure because these other measures are similar to population and would result in a similar urban bias. For example, vehicle registrations likely are closely correlated with population so that a registration-based exclusion probably would produce the same urban bias as the current population-based exclusion. On the other hand, VMT or VMT-based crash or fatality rates could be used to exclude counties in a way that would avoid an urban bias. However, many States do not have accurate counts of VMT for all counties. For this reason, NHTSA believes that allowing fatality-based exclusion of counties is the preferred method of balancing survey efficiency and cost considerations while continuing to ensure a representative sample. Therefore, no change is made in response to these comments.

One commenter stated that it is likely that oversampling and overweighting the observations in rural counties with high fatality rates and low belt use rates will erroneously depress the State's seat belt use rate in an attempt to focus attention on problem areas. (WA TSC at 2.) The fatality-based exclusion is not an attempt to focus on problem areas. Rather, as discussed above, its intent is to help reduce the current urban bias and ensure that seat belt use rate estimates are more representative. We note that oversampling need not lead to overweighting if the weight is calculated properly. In our opinion, properly weighted observations will not introduce error into the Statewide seat belt use rate estimate.

Commenters expressed concern that the fatality-based exclusion would result in additional counties being included in the sampling frame, especially in more rural areas. (NJIT at 1; WA TSC at 2; WY HSP at 1; MO DOT at 1; ND DOT at 1; GHSA at 2.) For example, the MO DOT stated that the “fatality based sampling frame” will result in 62 counties and 1,426 observation sites being included in the sampling frame when the State's current sampling frame includes 20 counties and 460 observation sites. (MO DOT at 1. See also CA OTS at 1.) We note that when more counties are in a sampling frame, more counties are eligible for selection, but this does not necessarily mean that more counties will be selected. Although it is not clear how the MO DOT estimated the number of counties and observation sites, we believe that the estimate may be a result of a misunderstanding of the fatality-based exclusion as discussed above, and we do not expect that Missouri or any State would be required to more than triple the number of observation sites at which it collects data. Some States will need to increase the number of counties included in their county sampling frame. This may result in an increase in the number of counties selected in the first stage of their sample, which, in turn, may impose additional costs for these States. However, we do not believe it would approach the magnitude described by MO DOT. Based on our review of current seat belt survey designs, we do not believe that change to a fatality-based exclusion will significantly increase costs in most States. Because these new Uniform Criteria were designed to give States flexibility in designing seat belt use surveys, States should work with their survey statisticians to develop a design that best meets their needs and circumstances.

Some commenters stated that the shift to a fatality-based exclusion may impact low population States, where fewer motor vehicle fatalities occur, and suggested that five years of fatality data should be averaged. (ND DOT at 1; ME DPSC at 1; WY HSP at 1; NH HSA at 1.) We recognize that in small rural States, there can be sharp annual fluctuations in the numbers and distribution of fatalities. Accordingly, after careful consideration, we have decided to amend the rule to allow States the option to average three, four or five years of FARS data to determine passenger vehicle occupant fatalities. We believe that allowing States this flexibility will reduce the potentially disproportionate impact of any single multi-fatality crash in any given year. Accordingly, we have amended the rule in response to these comments.

Several States asked for clarification about the kinds of fatalities that are to be counted for the fatality-based exclusion. Two commenters asked for clarification about which occupants and which vehicles counted in the fatality-based exclusion. (OR DOT at 1; WV HSP at 1.) Two commenters suggested that the sampling frame should be limited to unrestrained fatalities. (NY TSC at 1-2; PA DOT at 1.) To clarify, the county-by-county fatality counts that will be used to identify counties that may be excluded from the sampling frame will be based on passenger motor vehicle occupant fatalities only (and not on commercial vehicle occupant, motorcycle, bicycle or pedestrian fatalities) that occur in each county on roads to which the public has the right of vehicular access. We recognize that counts of unrestrained fatalities could be used as the basis for determining which counties may be excluded from the sampling frame. However, especially in sparsely populated States with relatively low fatality totals, the counts of unrestrained fatalities even when using three, four or five year totals may produce unstable distributions. For consistency, NHTSA believes that all States should use counts of all passenger motor vehicle fatalities to determine which counties may be excluded from the sampling frame. As noted before, the purpose of the fatality exclusion is not to direct data collection to specific locations of fatalities but to reduce the bias toward urban areas. The agency declines to adopt these commenters' suggestions.

The LA HSC commented that if crashes are used in lieu of population as a factor in determining site location, States should be permitted to account for serious injury crash data. (LA HSC at 2.) Another commenter recommended that the rule be changed to measure “fatal crashes” instead of “crash fatalities.” (IACP at 1.) As stated above, neither the number of fatalities in a county nor the specific locations where those fatalities occurred is a factor in selecting the actual observation sites. The fatality counts are used solely to determine which counties a State may exclude from its sampling frame in order to control operational costs. NHTSA believes that fatality counts serve as an adequate basis for that determination. We do not believe that counts of serious injuries or of fatal crashes (in lieu of crash fatalities) would provide additional value in the identification of counties that may be excluded from the sampling frame. We decline to change the rule in response to these comments.

One commenter asked if day and night differences in fatalities will be considered in selecting observation sites. (NV DPS at 4.) As noted above, we believe that the commenter misunderstood the purpose of the fatality-based exclusion. The actual fatalities do not determine where the observation sites are located. Rather, the three-, four- or five-year fatality counts will be used to allow the State to exclude counties that make up 15 percent of the State's passenger vehicle occupant fatalities. Both day and night passenger vehicle occupant fatalities will be counted in determining the fatality counts. With this clarification, no change to the rule is made.

One State asked if States could have the option to use their own non-FARS fatality data because the most recent FARS data are often less current than State's own traffic fatality counts. (WI State Patrol at 1.) In the preamble to the NPRM, NHTSA stated that the agency would provide States with county-by-county passenger motor vehicle occupant fatality counts for the relevant time period. We believe that allowing States to use their own fatality data would have little or no impact on defining the State's sampling frame. After careful consideration, we have decided to allow States the option to use the NHTSA-provided fatality data or their own fatality data for the most recent three, four or five years, provided that the State fatality data reflect the FARS definitions and are approved by NHTSA. We have revised the Final Rule accordingly.

One commenter stated that under the existing rule, the population and average VMT based sampling criteria took into account an “infinite sample size,” but with the proposed fatality-based sampling, “the fatalities become the universe.” (NV DPS at 4.) The commenter further stated that the statistical concepts related to the “central limit theorem” and the “normal distribution” might lose their “essence,” and that we cannot derive statistical results. Id. The commenter asked if it was necessary to look into the “finite population theory for sampling.”

This commenter appears to assume, incorrectly, a change from population-based exclusion to fatality-based exclusion, in which the central limit theorem (which states that as the sample size becomes larger, the sample mean will have an approximately normal distribution) would cease to have applicability because the sample size would be finite. As discussed above, the proposal did not change to fatality-based exclusion. It changed only the way counties are excluded from the sampling frame. In other words, the proposal did not require observations to be conducted at locations of fatal crashes. In fact, locations of fatal crashes should not serve as the basis for selecting actual observation sites. Under both the 1998 Uniform Criteria and the proposal, both population and sample sizes are finite, not infinite. This does not mean that the central limit theorem cannot be applied to the seat belt use rate estimate. In a seat belt use survey, the sample mean is the estimated seat belt use rate. As the sample size in a State's seat belt use survey becomes large enough, the central limit theorem would still be valid, i.e., the estimated seat belt use rate will have an approximately normal distribution.

Several commenters stated that the change from a population-based exclusion to a fatality-based exclusion would make it difficult for States to compare the seat belt survey results to previous years' estimates. (NY TSC at 1; PA DOT at 1; GHSA at 2; IA TSB at 1; TX DOT at 2. See discussion in Section III.A above.) While the change in the sampling frame exclusion may make comparing estimates from previous years more challenging, we believe that the data still remains valuable. The new seat belt use rate estimates will be more representative of the actual seat belt use rate in the State than previous estimates. We believe that the need for more accurate and reliable data outweighs the challenges in comparing survey results. (As noted above, this is not a change from a population-based survey to fatality-based survey. Rather, the change only affects the exclusion from the sampling frame that is allowed for controlling operational costs.)

2. Sampling Frame: Database of Roads

Some commenters supported the requirement that the database of road inventories includes all roads with a few exceptions, and agreed that NHTSA should provide the database to the States. (WA TSC at 3; Peters & Assoc. at 2; NV DPS at 6.) Two commenters requested more specific information about the database of road inventories. (ND DOT at 1; GHSA at 2.) The commenters asked whether the NHTSA-provided database would be a GIS population of roads, what variables would be included in the database, and whether roads on tribal lands and national parks would be included. The road database that NHTSA intends to provide will include road type and location, but is unlikely to include vehicle miles traveled. Roads on tribal lands and national parks are included in the database. As discussed in more detail below, NHTSA has amended the rule to allow the exclusion of rural local roads in counties that are not included in a Metropolitan Statistical Area (MSA) and, as a result, many roads in tribal lands and national parks may be excluded. No change to the rule is made in response to these comments.

One commenter requested clarification on whether the State could use its own map, aerial photos or satellite images in site selection. (WI State Patrol at 2.) As stated in the rule, a State may use its own map, aerial photos or satellite images if it is approved by NHTSA, as provided in § 1340.5(a)(2).

One commenter asked whether the State must abandon the practice of making seat belt use observations only at stop lights or stop signs. (WI State Patrol at 2.) As specified in the proposal, the sampling frame may not be limited only to roads having a stop sign or stop light. Accordingly, States may not confine the data collection to stop lights and stop signs. Such a practice would exclude observation of a large universe of road segments, and would produce a biased seat belt use estimate. Consequently, the agency declines to adopt the suggestion in response to this comment.

Another commenter agreed with the proposed exclusions except for the exclusion of access ramps to interstate roadways. (Peters & Assoc. at 2.) GHSA requested an explanation of the road exclusions. (GHSA at 2.) As stated in the NPRM, the agency excluded these categories of roads for reasons of safety and practicality. In addition, some of these road categories are not likely to be in the database of road inventories, and they are low traffic volume roads that will not affect the overall estimate of the seat belt use rate. We note that a State may choose to exclude the roads identified in § 1340.5(a)(2)(iii), but is not required to do so. Therefore, a State may include access ramps in its database of road inventories. With this clarification, no change to the rule is made.

Several commenters expressed concern that the proposed road coverage requirement would require States, especially those with more rural areas, to sample on road segments with very little traffic volume. (ID DOT at 1; IL DOT at 1; NH HSA at 1; IACP at 1; ME DPSC at 1; GHSA at 1-2.) NHTSA researched this issue and found that while local roads (as defined by the Federal Highway Administration) have more than two-thirds of road miles in rural areas, these roads experience only 13 percent of the VMT in rural areas. Sending data collectors to observation sites on some of these local roads might result in few, or even no, observations and would increase costs to the State. However, NHTSA determined that excluding all local roads in rural areas would have a measurable impact on the accuracy of the seat belt use rate estimate.

FHWA has classification standards for roads (“Functional Classification Guidelines”) that all States use. In FHWA's Functional Classification Guidelines, “local roads” are a category of roads which are not collectors or arterials, but permit travel between end points of a trip. FHWA further classifies local, collector and arterial roads as being in an urban or rural area. Based on our review of available data, allowing States to exclude all “rural local roads” from the sampling frame would significantly impact the seat belt use rate estimate because a substantial number of rural local roads are in areas that have high traffic volumes and potentially high fatality rates. For this reason, NHTSA determined that allowing a more limited exclusion of rural local roads would be more appropriate.

Pursuant to 44 U.S.C. 3504(e)(3), 31 U.S.C. 1104(d), and Executive Order No. 10253, the Office of Management and Budget (OMB) defines Metropolitan Statistical Areas (MSAs) and other statistical areas. (E.g., OMB Bulletin No. 10-02.) MSAs are statistical areas that include multiple counties and county-equivalents with at least one urbanized area that has a population of at least 50,000. See 75 FR 37246, 37252 (June 28, 2010). Rural local roads in counties or county-equivalents that are not part of an MSA are those roads that are likely to have very low traffic volume. Consequently, excluding rural local roads in non-MSA areas from the sampling frame would have a limited impact on the seat belt use rate estimate, but would significantly reduce data collection costs. Therefore, NHTSA has amended the Final Rule to allow an additional exclusion of rural local roads that are not included in an MSA.

3. Sampling Selection Requirements

A number of commenters requested clarification on the selection of observation sites, as provided in § 1340.5(b). (E.g., TX at 2; Peters & Assoc. at 2; ND DOT at 1.) One commenter asked about the selection of roads and assignment of weights in order to avoid oversampling one road type. (GHSA at 2. See also NV DPS at 6.) We believe that this commenter is seeking clarification of two issues—(1) how should the road inventory be stratified by functional classifications for purposes of this selection; and (2) how should the observations at each observation site be weighted to ensure an accurate estimation of seat belt use rate. (See also WY HSP at 3, 11; MO DOT at 2; NY TSC at 2.) One commenter stated that if all roads in the State have an equal probability of being selected, streets and other roads with relatively low traffic volumes are likely to be overrepresented in the sample. (NY TSC at 2. See also NJIT at 1-2.)

States are not required to stratify their sampling frame, whether by road type or by another variable. The only requirement is that the selection of observation sites be probability-based, i.e., each observation site should have a known probability of selection. A State may draw a simple random sample without stratification from the entire road sampling frame, i.e., from the listing of all road segments within a county selected in the first stage of sampling. A State also may stratify the sampling frame in a variety of ways, using such variables as road classification or VMT, and then draw a probability based sample of observation sites from each stratum. Although NHTSA believes that stratifying road inventories by functional classification and selecting road segments with probability proportional to VMT within these strata is good practice, there are other valid approaches to forming a probability sample of observation sites.

Regardless of the method used for selecting observation sites, once observations are made, the data must be weighted in a manner consistent with the survey design. One commenter stated that because the State does not have VMT for all local roads, it would be forced to exclude these roads from the sampling frame. (LA HSC at 1.) States are not permitted to exclude systematically all local roads from the sampling frame. If the State does not have VMT for some or all eligible road segments in the selected counties, the State must use some other method for designing and selecting its sample. The survey statistician should be able to help States select an appropriate method for weighting observation sites. With this clarification, no change to the rule is made.

Several commenters requested additional information on the number of observation sites States need to include in their survey. The WY HSP asked whether States will be given any guidelines as to the recommended number of observation sites to be selected based on the number of counties. (WY HSP at 3.) The MO DOT requested that provisions be added for States to randomly select a representative group of counties from those representing 85 percent of the State's fatalities, similar to the table provided in the appendix to the current criteria. (MO DOT at 1.) The commenter further requested that the rule specify the required number of road segments to be sampled in each county. (MO DOT at 2.) The OR DOT requested a formula for determining the total number of sampling sites that would be needed. (OR DOT at 1.)

NHTSA does not intend to provide a table or formula specifying the number of observation sites per county based on the number of road segments available within that county. One table or formula will not optimally serve to determine the number of observation sites for all survey designs. We revised the Uniform Criteria to give States the flexibility to design a survey in a manner that best meets the specific environment in each State. States should rely on their survey statistician to develop a survey design that will meet the Uniform Criteria and meet the State's needs. The survey statistician can help the State determine the necessary sample sizes for selection of counties at the first stage and selection of observation sites (road segments) within the selected counties. NHTSA is developing a sample of an acceptable survey design to assist the States' redesign efforts, and will post the sample survey design on NHTSA's Web site. Accordingly, we decline to adopt these commenters' suggestion.

4. Substitution and Resecheduling of Observation Sites

Generally, commenters stated that the agency's proposal in § 1340.5(c) regarding protocols when observation sites are not available was reasonable. (WA TSC at 3; Peters & Assoc. at 3.) One commenter suggested that pre-selecting alternate observation sites before the start of data collection would be more practical. (Peters & Assoc. at 3.) The pre-selection of alternate observation sites is not precluded under the Uniform Criteria. The alternate observation sites must be in the same county and the same road classification as the observation site the State is replacing. Another commenter asked if it would be acceptable to combine both options—returning to the observation site on the same day of the week and at the same time of the day and selecting an alternate observation site—as part of the State's protocol when an observation site is temporarily unavailable. (TX DOT at 2.) States may include one or both options as part of their protocol in their survey design. With this clarification, no change to the rule is made.

The commenter stated that requiring the State to return to an observation site at the same time and day of the week if an observation site is missed, especially if the observation site is hundreds of miles away, would increase commuting time and other costs. (IACP at 2. See also ODU at 3.) As stated in the NPRM, if conditions preclude data collection at an observation site at the scheduled time and day, States may elect to return to the observation site at the same time and day or collect data from preselected alternate observation sites to replace the observation site that is unavailable. See § 1340.5(c). If adverse weather precludes data collection, it is likely to affect data collection at multiple observation sites, especially if observation sites are cluster sampled. With adverse weather conditions, States generally should be able to anticipate whether the conditions would affect data collection before data collectors travel to the observation sites, and therefore, should be able to plan accordingly to mitigate commuting time and costs. States are encouraged to consider these issues and develop a protocol that best fits the conditions in its State. The agency makes no change to the rule in response to this comment.

5. Precision

The agency received a number of comments regarding the standard error in §§ 1340.5(d) and 1340.9(g). Some commenters agreed with the change. (WY HSP at 2-3; WA TSC at 3; IL DOT at 1.) Others stated that the change in the precision requirement would require a larger sample size or longer observation times, and this would result in higher survey costs. (Peters & Assoc. at 1, 3, 5; TX DOT at 2; GHSA at 2; NV DPS at 7; WV HSP at 2; ND DOT at 1; NH HSA at 1.) One of these commenters further stated that reducing the standard error by half would require a four-fold increase in sampling size, and asked whether it would be possible to get four times the current sample size with the same observation time. (NV DPS at 7.) This commenter may have misunderstood the change in the precision requirement—the 1998 Uniform Criteria specify that the relative error not exceed 5 percent while the proposed rule specifies that the standard error not exceed 2.5 percentage points. The standard error is an absolute measurement whereas the relative error measures the standard error as a fraction of the actual seat belt use estimate. Unlike the relative error, the standard error will not change regardless of a State's seat belt use rate estimate. A standard error holds all States to the same standard.

Our review of States' current surveys indicates that most States will be able to meet the required standard error of 2.5 percentage points with sample sizes comparable to their current surveys. Therefore, we believe that most States will not have to add a significant number of observation sites, and costs will not be significant. We also believe that any additional costs are outweighed by the improved quality and reliability of the survey data.

One commenter also stated that the precision calculation is made after the data are collected, and that an inadequate precision would require returning to some number or all of the observation sites to collect additional data, which could significantly increase the costs. (Peters & Assoc. at 3.) The commenter is correct that the final precision estimate cannot be made until after the data are collected. While a State would have to collect additional data if the precision is above 2.5 percentage points, the State would not have to return to all the observation sites to collect additional data. We note that at the survey design stage, the State may use information from its current survey or a similar survey from another State to estimate the number of observation sites that may be needed to provide a reasonable assurance that the precision requirement will be met. Consistent with standard practice in survey collection, States should select a sufficient number of observation sites to account for issues that may affect the standard error. We believe that a survey statistician should be able to help the State determine the proper sample size for the required precision. The agency declines to change the rule in response to these comments.

A commenter stated that the change in precision requirement will be difficult and expensive to achieve on rural roads. (NJIT at 2.) According to this commenter, fewer interstate and freeway roadways will be selected for performing data collection, which will lead to more roads of lower classifications and lower volumes being selected for observation. Id. The commenter further stated that this will result in longer time periods for data collection and in some cases require more than one visit to a location for data collection. Id. The precision requirement applies to the entire survey—not to individual observation sites. As stated above, for most States, we do not believe that the change will substantially affect the number of observation sites that will be required. Consequently, the periods of data collection should not substantially change for most States. We note that NHTSA is revising the proposal to allow States to exclude rural local roads in counties that are not included in an MSA from the sampling frame, which will help mitigate the concerns of this commenter. Accordingly, no further change to the rule is made in response to this comment.

One commenter stated that consistency in training and observation would be a much better place to invest resources than in reducing sampling error. (NH HSA at 1.) NHTSA assumes the commenter is asserting that improved training would reduce survey sampling error. There are two kinds of errors—measurement errors and sampling errors. Consistency in training to improve observation skills can reduce measurement errors, i.e., the accuracy of the observations. However, it will not reduce sampling error, which is a result of a subset of the target population is being observed. Consequently, the agency declines to adopt this commenter's suggestion.

E. Assignment of Observation Times (§ 1340.6)

Two commenters expressed support for the proposed observation times. (WA TSC at 3; Peters & Assoc. at 3.) One commenter stated that requiring data collection until 6 p.m. could pose difficulties during late fall and winter, when it gets dark by 5 p.m. (IACP at 1.) In the NPRM, the agency proposed that data collection must take place during daylight hours, not all hours, between 7 a.m. and 6 p.m. Accordingly, States would not be required to collect data at an observation site during non-daylight hours even if they fall between the hours 7 a.m. and 6 p.m. No change to the rule is made in response to these comments.

Two commenters stated that requiring random assignment of the day of the week for observations would increase costs, especially if data collectors were sent to opposite corners of the State in a 24-hour period. (IACP at 2; ODU at 3.) While the NPRM proposed random assignment of the day of the week for observations, it also proposed allowing States to cluster sample or group observation sites in close geographic proximity in order to reduce the costs of random assignment of observation sites. (See also OR DOT at 1.) This would reduce the need for data collectors to be sent to opposite corners of the State in a 24-hour period.

These commenters also suggested treating weekdays as equal and weekends as equal. (IACP at 2; ODU at 3.) NHTSA does not have data to support the assumption that there is no seat belt use rate difference between the days of the week and between Saturday and Sunday. We believe that allowing States to cluster sample observation sites in close proximity should sufficiently reduce the operational costs of random assignment of the day of the week. Therefore, NHTSA declines to adopt this recommendation.

The NV DPS asked how many observation sites can be grouped together into geographical clusters. (NV DPS at 7.) NHTSA did not propose to limit the number of observation sites that can form a cluster. As long as the State can allocate an appropriate time period for data collection at each observation site, those observation sites may be cluster sampled. No change to the rule is made in response to this comment.

F. Observation Procedures (§ 1340.7)

1. Data collection dates. The TX DOT commented that the proposal did not specify that the surveys should be conducted uniformly after Click It or Ticket or Memorial Day Monday. (TX DOT at 2.) This commenter recommended a wider window of survey time, such as May 1 through June 30. Id. In the NPRM, the agency did not propose a change from the requirement in 1998 Uniform Criteria that all observations take place during the calendar year, i.e., January 1 to December 31. Although most States choose to conduct seat belt use surveys after Click It or Ticket or Memorial Day Monday, States may conduct surveys anytime during the calendar year. No change to the rule is made in response to this comment.

2. Roads with two-way traffic. Peters & Assoc. commented that on low volume two-way streets, limiting an observer to observing one direction of traffic flow would result in unnecessary additional survey cost. (Peters & Assoc. at 4.) NHTSA believes that consistent methods for data collection should be applied at all observation sites, regardless of the volume of traffic. Moreover, requiring the data collector to observe traffic in one direction will help reduce measurement error. Accordingly, the agency declines to change the rule in response to this comment.

3. Vehicle coverage. The agency received a number of comments regarding the proposal's vehicle coverage requirements. The commenters were generally supportive of the proposal, but a few questioned the inclusion of certain vehicles. Several commenters appeared to misunderstand the requirements, and some commenters requested clarification.

One commenter stated that the proposal did not address State seat belt law exemptions for certain vehicles and recommended that all vehicles operated on a public roadway should be included in the survey regardless of a State's exemptions. (LA HSC at 2.) In contrast, two commenters questioned the inclusion of vehicles that are exempt under the State's seat belt law because it would not provide an accurate measure of compliance. (NY TSC at 2; GHSA at 2.) In the NPRM, the agency stated that all passenger motor vehicles must be included in the survey, including vehicles that are exempt by the State's seat belt law. The purpose of the seat belt use rate survey is not to determine compliance with the State's seat belt law, but rather to estimate the actual seat belt use of drivers and right front passengers. Consequently, the agency makes no change to the rule in response to these comments.

One commenter supported the inclusion of “passenger vehicles for commercial use” in the seat belt use survey. (WY HSP at 2.) Three commenters requested clarification of the term “commercial passenger motor vehicle.” (OR DOT at 1; TX DOT at 3; GHSA at 2.) Earlier in this notice, we clarified that passenger motor vehicles include a passenger car, pickup truck, van, minivan or sports utility vehicle with a GWVR of less than 10,000 lbs. Data from motor vehicles over 10,000 lbs GWVR need not be included in this survey. As stated in the preamble of the NPRM, data must be collected for passenger motor vehicles being used for commercial purposes. In other words, data from such passenger motor vehicles as taxis, flower delivery vans, and pizza delivery cars that are under 10,000 lbs GVWR must be included in the seat belt use survey. To clarify this point, in § 1340.7(c), the agency has amended the phrase “commercial passenger motor vehicles” to read “passenger motor vehicles used for commercial purposes.”

Three commenters stated that it would be difficult to collect or report data separately for various types of vehicles, such as passenger vehicles, commercial vehicles and out-of-state vehicles. (OR DOT at 1; WA TSC at 2; NV DPS at 8.) Although some States may want to collect such data separately in order to better serve their problem identification and program evaluation needs, the NPRM did not propose, and the Final Rule does not require, States to collect or report the data separately for these various vehicles.

4. Occupant coverage. Two commenters supported the agency's proposal to include right front passengers in booster seats in the seat belt use survey and exclude right front passengers in child safety seats. (Peters & Assoc. at 4; WY HSP at 2.) Three commenters disagreed with the proposal, with one recommending that children in child safety seats and booster seats be included in the seat belt use survey, another recommending that children in child safety seats and booster seats be excluded from the seat belt use survey, and a third recommending that both be excluded or both be included. (NY TSC at 2-3; WA TSC at 3; GHSA at 2-3.) As stated in the NPRM, we believe that data on passengers in child safety seats should be excluded because it is not possible to observe whether a child safety seat is properly installed or the child is properly restrained in the child safety seat. Unlike child safety seats, however, booster seats require the use of the readily-observable shoulder belt to secure the passenger. To clarify any misunderstanding, data on children in child safety seats should not be collected and reported in the seat belt survey conducted. (See GHSA at 3.) We do not believe that this will have much impact on the seat belt rate estimate since we expect the number of children in child safety seats in the front right passenger side of passenger motor vehicles to be limited. With this clarification, the agency makes no change to the rule.

Two commenters supported the proposal's requirement to record belted and unbelted occupants. (GHSA at 3; WY HSP at 2.) Two commenters disagreed with recording the belt status as unknown if it could not reasonably be determined whether the driver or the right front passenger is belted. (WI State Patrol at 2; Peters & Assoc. at 4.) These commenters suggested recording unknowns as unbelted. Id. We believe that the practice of recording unknown belt use as unbelted would underestimate the actual seat belt use rate estimate. For this reason, the agency declines to change the rule in response to these comments.

NJIT stated that there are many cases when shoulder belt use is “unknown,” and that recording unknowns will gather unusable data, making it difficult for data collectors to know when they have collected enough observations. (NJIT at 3.) We believe that the commenter is concerned that observations must continue at all observation sites until the percent of unknown seat belt use observations is below 10 percent. To clarify, the nonresponse rate is determined based on the entire survey sample, not individual observation sites. In addition, we do not agree that recording unknowns results in collecting unusable data. For quality control purposes, it is good practice to count and report the number of unknown belt use observations that occur. If the percent of unknown observations is high, this may indicate a need to improve observer training. The agency makes no change to the rule in response to this comment.

The agency did not receive specific comments regarding survey variables in § 1340.7(e). However, the agency has decided to change the term “survey variable” to “survey data” because this term describes the information more accurately.

The agency received only supportive comments regarding the data collection environment in § 1340.7(f). Accordingly, the agency made no changes to this provision.

G. Quality Control (§ 1340.8)

Although some commenters agreed that random unannounced visits would ensure more accurate data, several commenters stated that requiring quality control monitors at five percent of the observation sites would increase the survey costs. (WA TSC 3; Peters & Assoc. at 4; NV DPS at 8; GHSA at 3.) The LA HSC stated that unannounced visits by quality control monitors would require increased supervision of personnel. (LA HSC at 1.) Requiring quality control monitors to conduct random unannounced visits will increase survey costs to some States, especially to those States that do not currently conduct such visits. However, some level of supervision of data collectors is necessary for quality control. We believe that quality control monitors at five percent of the observation sites strikes the proper balance to minimize the costs while still ensuring that good data are being collected. The agency makes no change to the rule in response to these comments.

One commenter stated that sending quality control monitors would be difficult, especially in States with limited resources, and suggested that standards for inter-accuracy ratio testing be required for all States instead of quality control monitors. (WY HSP at 2.) We assume that the commenter is referring to “inter-rater reliability.” Inter-rater reliability is a measure of rating or coding consistency among different raters or coders, i.e., data collectors. While inter-rater reliability would be helpful for training observers, independent quality control monitors provide reasonable assurance that observation protocols are being properly implemented. As stated above, while this may increase some survey costs, we believe that this requirement would help ensure that good data are being collected. Consequently, the agency declines to adopt this recommendation.

The WY HSP asked whether “random unannounced visits” means randomly selected sites within each county or throughout the State, whether quality control monitors must visit each observer, and whether there are any criteria for these quality control monitors. (WY HSP at 5.) In the NPRM, the agency did not specify how to conduct random unannounced visits. Instead of requiring States to conduct these visits in a specific manner, the only requirements are that States conduct these visits to five percent of the observation sites and that the same individual cannot both collect data and monitor the collection of data at the same time. We believe that States should have flexibility in how to conduct these visits, taking into account each State's survey design and specific conditions. For example, a State may elect to conduct an unannounced visit in each county, for each survey crew or using some other factor. We note that random unannounced visits serve not only to check if data are being collected in accordance with the survey protocol, but also as a deterrent to bad data collection. The ME DPSC requested clarification regarding whether the quality control must be conducted by a vendor and not by the observer. (ME DPSC at 1.) As noted above, the same individual cannot both collect data and monitor data collection at the same time. Other than that restriction, the only requirements for a quality control monitor are that the individual be trained in the observation protocols and the substitution and rescheduling of observation sites. No change to the rule is made in response to these comments.

One commenter supported the training requirements for survey observers. (GHSA at 3.) Another commenter suggested that once observers have extensive training, the trained observers should be allowed to have refresher training via telephone conference call, Webinar, etc. every alternate year in order to keep the costs down. (WY HSP at 2.) We agree with the commenter that, once trained, observers may have refresher training via telephone conference calls, Webinars, or other methods that minimize costs. However, we believe that annual training is important to ensure accurate data collection, especially since data collectors are often not regularly engaged in data collection throughout the year. Accordingly, we decline to adopt the commenter's suggestion.

Three commenters questioned the requirement for a survey statistician and sought clarification of the statistician's qualifications. (ODU at 3; WY HSP at 3; IACP at 2-3.) The proposed requirement for a statistician is not to provide an additional level of checks and balances, as one commenter suggested, but to ensure that both the survey design and the annual reporting of seat belt use rates are carried out in a methodologically-sound manner. (See WY HSP at 6.) Over the years, we have encountered numerous instances where data were analyzed incorrectly, resulting in inaccurate seat belt use rate estimates. We believe that part of the reason for these errors was the lack of statistician involvement in the data analysis. For this reason, the rule requires that adequate statistical expertise be applied to the analysis of the survey data. As stated in the NPRM, the survey statistician should have knowledge of designing probability-based multi-stage samples, statistical estimators from such designs, and variance estimation of such estimators. To clarify, the statistician does not have to have a specific degree in statistics, but rather have sufficient training and experience in designing research surveys and analyzing the data, as described in the Uniform Criteria. However, the agency is removing “seat belt” from the term “seat belt survey statistician” in this section and throughout the rule to clarify that the statistician must be a person trained in statistical methodology.

Several commenters stated that requiring a statistician would add significant costs. (ODU at 3; IACP at 2-3; WA TSC at 3; Peters & Assoc. at 6; GHSA at 3.) As noted above, we have found errors in some State seat belt use rate estimates, and we believe that requiring a statistician to review and confirm the estimate annually will help reduce the errors we have seen. Although some States may incur additional costs, especially those that currently do not employ a statistician, the requirement is necessary in light of the errors noted above. States may use Federal grant funds, such as Section 402 and Section 406, to defray the costs of designing and conducting seat belt use surveys. Conducting an annual Statewide seat belt use survey in accordance with the uniform criteria “to ensure that measurements are accurate and representative” is an administrative requirement under Section 402. While NHTSA is ready to provide technical assistance, we believe that States should rely on their own statistician to ensure that the State seat belt use rate estimate is overseen and properly implemented in accordance with the approved survey design. Consequently, the rule remains unchanged with regard to this requirement.

H. Computation of Estimates (§ 1340.9)

One commenter generally supported all of the requirements in the proposal regarding computation of estimates. (WA TSC at 3.)

The agency received no comment regarding § 1340.9(b) (data editing). Therefore, NHTSA is making no substantive change to the requirement in this provision. (The agency is making a minor amendment to remove the phrase “or statistically edited,” because it is redundant.)

NHTSA received three comments regarding the imputation of unknown values of variables. Peters & Assoc. questioned the need for data imputation and GHSA stated that no imputation should be allowed. (Peters & Assoc. at 4; GHSA at 3.) The proposal does not require imputation, but rather allows States to use imputation if it is pre-approved by NHTSA. In general, NHTSA does not believe that imputation of unknown values will be necessary. However, in order to provide flexibility to States, NHTSA is allowing imputation if it is necessary to improve the estimates and the methodology is approved by NHTSA prior to data analysis. As noted by the IACP, there are a number of imputation methods. (IACP at 1.) NHTSA is not specifying which methods are acceptable because the acceptability of an imputation method depends on the survey design. NHTSA will ensure the proper use of imputation in surveys by requiring approval before imputation methods are used. No change to the rule is made in response to these comments.

One commenter asked for clarification of data-weighting if the survey is fatality-based as proposed in the NPRM. (Peters & Assoc. at 4.) As discussed in detail in Section III.D.1 above, we believe this type of concern arises from a misunderstanding regarding the fatality-based exclusion from the sampling frame. Fatality counts will be used to determine which counties may be excluded from the sampling frame. States will weight the data based on the selection probability of the sample observation site, which is determined by the way that the samples are selected. For example, observation sites may be selected by a simple random sample or by a probability proportional to sample size. Various measures such as VMT or traffic flow counts could be used for the measurement of size.

However, to clarify the sampling weight requirements, the agency is changing the reference to “inverse of the selection probability of the observation site at which the data were obtained” to “sampling weights as required by the sample design and any subsequent adjustments.” This change is necessary to clarify that the weights used in the estimation process reflect both the original sampling process and any subsequent weighting necessary, e.g., selection of direction of travel or of travel lanes for observation, non-response adjustments, among others. The Final Rule now reads as follows: “The estimation shall weight observed data by the sampling weights as required by the sample design and any subsequent adjustments.”

NHTSA received two comments regarding the requirement to include a procedure to adjust the sampling weights for observation sites with no usable data. One commenter stated that using an alternate observation site would be the most practical method. (Peters & Assoc. at 5.) In the proposal, the agency identified several methods to adjust for observation sites with no usable data, including using alternate observation sites. However, as stated in the preamble to the NPRM, allowing the States flexibility for selecting the method based on their survey design will enable them to determine which method or combination of methods best meets their needs.

Another commenter stated that States will incur additional costs for the suggested protocols for handling observation sites where data are not collected. (WY HSP at 3.) Although States have flexibility to select among several methods to adjust for observation sites with no usable data, some States may need to return to observation sites at a later time or visit an alternate observation site for data collection. Generally, observation sites at which no data are collected are very low traffic volume observation sites. As noted in Section III.D.2 above, rural local roads in counties that are not included in an MSA are roads with very low traffic volume. Because the agency is allowing States to exclude these roads, we believe that the incidence of returning to observations sites will be limited. For this reason, the agency is making no change in response to this comment.

The agency received numerous comments regarding the nonresponse rates. To summarize, the NPRM proposed that the nonresponse rates for the entire survey must not exceed 10 percent (for the total number of recorded unknown values of passenger presence to the number of passenger motor vehicles observed and for the ratio of the total number of recorded unknown values of belt use to the drivers and right front passengers observed). The NPRM further proposed that the State must include a procedure to collect additional observations if the nonresponse rates exceed 10 percent.

One commenter thought that a 10 percent nonresponse rate was too high, especially if imputation methods are used. (ME DPSC at 1.) States should strive to hold their unknown values well below 10 percent, and we believe that most States would be able to meet that requirement. As discussed above, we do not require imputation, and NHTSA will review any imputation proposals to ensure that imputation methods do not impair the accuracy of the data. We believe that allowing States a certain percentage of unknowns is necessary to ensure that any increased costs are not substantial. The agency declines to change to rule in response to this comment.

NHTSA received numerous comments expressing concern about increased survey costs related to the need to oversample observation sites or collect additional observations when the nonresponse rate exceeds 10 percent. (E.g., ND DOT at 1; GHSA at 3; ODU at 2.) Two commenters stated that this requirement would result in a longer overall survey timeline or slow the survey results. (Peters & Assoc. at 5; WY HSP at 3.) Although some States may incur increased survey costs and additional data collection time, with a properly designed survey and observation protocols, NHTSA does not anticipate that many States will need to return to observation sites to conduct additional observations. As a general principle, a properly designed survey should include a sufficient number of observation sites, anticipating that a few or some observation sites may produce no usable data. We believe that the increased costs that some States may incur are necessary to ensure a more accurate, representative seat belt use rate estimate. However, to reduce the reporting burden on States, NHTSA is deleting the proposed requirement that the nonresponse rate of passenger presence must not exceed 10 percent. While States must still collect data on passenger presence to help monitor quality control of data collection, States would not be required to report the nonresponse rates for passenger presence to NHTSA. In the Final Rule, the agency makes amendments in § 1340.9 and the Appendix to reflect this change.

Two commenters had concerns about nonresponse rates exceeding 10 percent in free-flowing traffic or on interstates with high speed travel. (IACP at 2; IL DOT at 1.) There are several methods to reduce the nonresponse rates in high speed travel or in high traffic areas. For example, observations may follow a protocol that does not attempt to observe all vehicles passing by, such as observing only a single lane, observing and recording every other vehicle or every third vehicle, among others. Vehicles that are not attempted to be observed would not be counted in the survey, i.e., they would not be counted as unknown. Only vehicles that are attempted to be observed must be recorded for the seat belt use status of the driver and right front seat passenger, if present.

Currently, most States use one of two methods for observations on interstate highways or other high speed roadways. One method is to conduct the observations at the base of the first exit ramp within or beyond the selected road segment. Another method, which may be less cost effective, is to collect observations while travelling in a vehicle at such locations—one member of the observation team would drive along the road segment at a speed below the posted limit while another member of the observation team would collect the data by observing the belt use of drivers and passengers in overtaking vehicles. Either of these methods or a combination of methods should help to reduce the incidence of unknown belt use observations, and therefore, help keep the nonresponse rate below 10 percent. With this explanation, the agency has made no change to the rule in response to these comments.

One commenter suggested that requiring States to record unknown variables may require in-field calculations. (Peters & Assocs. at 5. See also ODU at 2.) The requirement that unknown values must not exceed 10 percent applies to the entire survey, not to data collected at individual observation sites. There is no need to conduct in-field calculations to verify that the observations at a given site are below 10 percent. Specifically, the nonresponse rate is computed by dividing the total number of drivers and right front seat passengers with unknown belt use status by the total number of drivers and right front seat passengers observed. At the end of the survey, if the State still exceeds the nonresponse rate requirement, the State must collect additional data. No change to the rule is made in response to this comment.

NHTSA received a number of comments regarding the variance estimation. For purposes of discussion, these comments are addressed under the precision requirement of the selection of observation sites in Section III.D.5 above.

I. Submission and Approval of Seat Belt Survey Design and Post-Approval Alterations to Survey Design (§§ 1340.10 and 1340.11)

One commenter supported the requirements for submission and approval of seat belt survey designs. (GHSA at 3.) Another commenter stated that the new survey design requirements will require input from a survey statistician. (Peters & Assoc. at 5.) The agency anticipated that States would need a survey statistician to help design and conduct surveys and analyze the collected data. As discussed in detail in Section III.G. above, we have encountered instances of problematic survey results over the years, and we believe that these instances were the result of insufficient statistical expertise in the design and/or analysis phase of reporting the seat belt use rate estimate. For this reason, the agency proposed that survey results be reviewed and approved by a survey statistician. We believe that this will result in improved and more accurate survey results. The Final Rule retains the requirement for a survey statistician.

In § 1340.10(a), NHTSA has corrected cross references, and in § 1340.10(a)(1)(v), NHTSA changed the language “Define an observation site” to “Specify the method used to select the road segments for observation sites as provided by § 1340.5(b)”. This change clarifies what was being requested of the State and reflects the sequence generally followed in designing and selecting samples.

Several commenters stated that States needed more time to develop a new survey design and recommended delaying the implementation of the revised criteria. (See IACP at 1, WV HSP at 2, ND DOT at 1, GHSA at 4, and WY DOT at 3.) NHTSA agrees and has decided that the revised criteria will apply to seat belt use surveys conducted during calendar year 2012 and thereafter, not during calendar year 2011. In response to these comments, the agency has made changes to § 1340.2 (Applicability) and has revised the deadline for submission of proposed survey designs in § 1340.10(b) to January 3, 2012.

One commenter asked whether a State must submit its survey design every year, even if there are no substantive changes from the previous survey design approved by NHTSA. (WI DOT at 2.) The commenter agreed that States should resubmit the survey design when they propose to re-select observation sites or make other substantive changes to the survey design or data capture/processing protocol. Id. The agency did not intend States to submit survey designs every year. For calendar 2012 seat belt use surveys, the first year under the new requirements, States are required to submit proposed survey designs by January 3, 2012 so that NHTSA will have sufficient time to review the survey design before the surveys are conducted. Once a State's survey design has been approved by NHTSA, the State is not required to resubmit the survey design unless the State proposes alterations to a NHTSA-approved survey design. (See § 1340.11.) This is consistent with the annual reporting requirements, under which States certify that the survey was conducted using a survey design that was approved by NHTSA and that the survey design has not changed since NHTSA approval. (See § 1340.13(b); Appendix.) If a State chooses to redesign its seat belt survey, it should follow the procedures identified in § 1340.11. To clarify this point, the agency has added language in § 1340.10(b).

J. Re-Selection of Observation Sites (§ 1340.12)

One commenter disagreed with the requirement to update the survey design every five years, and two commenters stated that requiring States to re-select observation sites from updated sampling frame data would cause States to incur additional costs. (WY HSP at 1; Peters & Assoc. at 6; WA TSC at 3.) States are not required to redesign their surveys every five years. Rather, in the NPRM, the agency proposed requiring States to re-select observation sites every five years. Under the current seat belt use surveys, many States may be using an inventory of road segments that is years out-of-date. The inventory of road segments changes over time as new roads are constructed and existing roads are closed or changed. An up-to-date inventory of road segments is necessary to ensure that the seat belt use estimate is accurate and representative of Statewide seat belt use. We believe that the additional costs for re-selecting observation sites every five years will not be significant because States are required only to re-select observation sites from updated sampling frame data, not to redesign their surveys. In order to minimize the costs, NHTSA intends to provide States with the updated three, four or five year fatality distribution and inventory of road segments. The agency makes no change to the rule in response to these comments.

One of the commenters also expressed concerns that the re-selection of observation sites would require a new survey design, sampling of sites, pilot testing and travel to new sites to assess visibility and safety. (WA TSC 3-4.) The agency does not believe that the re-selection of observation sites requires a new survey design. While we believe that it is good practice to travel to new observation sites to assess visibility and safety, NHTSA is not specifying how States should determine where observations should be conducted and what other implementation measures should be adopted. These are decisions that the State is best positioned to make. We do not believe that these costs would be significant. The reasonable additional costs associated with re-selecting observation sites are necessary to ensure that the survey continues to be representative of the current inventory of road segments. No change to the rule is made in response to this comment.

One of the commenters stated that the requirement to reexamine geographic distribution of fatalities from the most recent three years could result in an inclusion or exclusion of different counties every five years, causing site selection changes and efficiencies associated with sample clustering to change every five years. (Peters & Assoc. at 2.) Although the sampling frame of counties could change every five years, we do not expect significant changes based on our review of historical FARS data. As discussed in Section III.D.1 above, NHTSA is amending the Final Rule to allow States the option of using a fatality distribution of three, four or five years instead of three years, which will help further reduce changes in the sampling frame. We note that the actual observation sites would change, because of re-selection requirements, regardless of whether the sampling frame of counties changes. States will still be able to cluster sample with the new observation sites. The agency makes no change to the rule in response to this comment.

A commenter questioned whether NHTSA would have the capacity to review all State and territorial seat belt survey designs every five years. (GHSA at 4.) The commenter suggested that after the initial review, NHTSA should stagger subsequent State reviews so that one-fifth of all State and Territory survey protocols are reviewed every year. Id. NHTSA will review and approve survey designs from all States and Territories for surveys conducted beginning calendar year 2012. The NPRM did not propose and the Final Rule does not require States to redesign surveys for NHTSA approval every five years. Rather, States are required to re-select observation sites using updated sampling frame data. It is the updated sampling frame data that requires NHTSA approval. NHTSA will deploy the necessary resources to review updated sampling frame data from all States and Territories.

K. Annual Reporting Requirements

NHTSA received two positive comments in support of the annual reporting requirements.^[4] (TX DOT at 1; GHSA at 3.) However, one commenter seemed to suggest that the current “research report” describing the survey methodology and results produced by States was sufficient. (WA TSC at 4.) Under the current reporting requirements, NHTSA does not have sufficient information to evaluate the computation of the seat belt use rate estimate. Based on our experience, some of the reported results were not consistent with the computation formula in the NHTSA-approved survey design. For this reason, NHTSA is requiring additional information in the annual reports submitted by States. NHTSA believes that the additional information is necessary in order to carry out the agency's responsibilities for grant management and oversight. No change to the rule is made in response to this comment.

One commenter suggested requiring a certification from the Governor's Representative for Highway Safety instead of the certification by the statistician. (GHSA at 3.) After careful consideration, the agency has decided to amend the rule to require only a certification from the Governor's Representative and to remove the requirement for certification by the statistician. However, a qualified survey statistician is still required to review and approve the survey results. See § 1340.8(c). Accordingly, NHTSA is amending the certification by the Governor's Representative to certify that a qualified statistician has reviewed the reported seat belt use rate estimate and information reported in Part B, and has determined that they meet the Uniform Criteria for State Observational Surveys of Seat Belt Use, 23 CFR Part 1340. The agency has made changes to § 1340.13 and corresponding changes to the certification in the Appendix in response to this comment.

IV. Statutory Basis for This Action

The Final Rule amends the uniform criteria for the measurement of State seat belt use rates for surveys that States are required to conduct annually under a grant program in accordance with 23 U.S.C. 402(b)(1)(E)(iii).

V. Regulatory Analyses and Notices

A. Executive Order 12866 and DOT Regulatory Policies and Procedures

Executive Order 12866, “Regulatory Planning and Review” (58 FR 51735, October 4, 1993), as amended by Executive Order 13563, “Improving Regulation and Regulatory Review” (76 FR 3821, January 21, 2011), provides for making determinations whether a regulatory action is “significant” and therefore subject to OMB review and to the requirements of the Executive Order. The Order defines a “significant regulatory action” as one that is likely to result in a rule that may:

(1) Have an annual effect on the economy of $100 million or more or adversely affect in a material way the economy, a sector of the economy, productivity, competition, jobs, the environment, public health or safety, or State, local, or Tribal governments or communities;

(2) Create a serious inconsistency or otherwise interfere with an action taken or planned by another agency;

(3) Materially alter the budgetary impact of entitlements, grants, user fees, or loan programs or the rights and obligations of recipients thereof; or

(4) Raise novel legal or policy issues arising out of legal mandates, the President's priorities, or the principles set forth in the Executive Order.

This Final Rule was not reviewed by the Office of Management and Budget under Executive Order 12866. The Final Rule is not considered to be significant within the meaning of E.O. 12866 or the Department of Transportation's Regulatory Policies and Procedures (44 FR 11034 (Feb. 26, 1979)).

This Final Rule does not affect amounts over the significance threshold of $100 million each year. This Final Rule sets forth the criteria for designing and conducting State seat belt use observational surveys, procedures for obtaining NHTSA approval of survey designs, and a new form for reporting seat belt use rates to NHTSA. The costs to design and conduct observation surveys under the criteria are well below the annual threshold of $100 million. This Final Rule does not adversely affect in a material way the economy, a sector of the economy, productivity, competition, jobs, the environment, public health or safety, or State, local, or Tribal governments or communities. This Final Rule does not create an inconsistency or interfere with any actions taken or planned by other agencies. This Final Rule does not materially alter the budgetary impact of entitlements, grants, user fees, or loan programs or the rights and obligations of recipients thereof. Finally, this Final Rule does not raise novel legal or policy issues arising out of legal mandates, the President's priorities, or the principles set forth in the Executive Order.

Currently, States are required to provide satisfactory assurances that they will conduct an annual Statewide seat belt use survey in accordance with the uniform criteria as part of the administrative requirements for a highway safety grant under 23 U.S.C. 402(b)(1)(E)(iii). This Final Rule does not change the statutory requirement to provide assurances that the State will conduct an annual Statewide seat belt use survey, but does change the way States collect and report survey data and the allowable error rate. Specifically, this Final Rule requires States to draw observation sites from an updated sampling frame. This Final Rule also improves quality control of the data collected by requiring States to train observers before data collection, to have quality control monitors conduct unannounced visits, and to have a statistician review the data collected. Finally, this Final Rule requires States to submit additional information in their annual certifications.

The agency has determined that this Final Rule is not significant. If a State does not provide assurances that it will conduct an annual Statewide seat belt use survey in accordance with the uniform criteria in a given year, a percentage of Section 402 grant funds could be withheld. However, States rely on statistically valid observational surveys of seat belt use to plan and evaluate their highway safety programs and have committed, through their highway safety offices, to conduct annual Statewide seat belt use surveys as part of the core performance measurement process. The agency believes that no State will decline to provide the required assurances. Because the impacts of this Final Rule are minimal, the agency is not required to prepare a full regulatory evaluation.

B. Regulatory Flexibility Act

Pursuant to the Regulatory Flexibility Act (5 U.S.C. 601 et seq., as amended by the Small Business Regulatory Enforcement Fairness Act (SBREFA) of 1996), whenever an agency publishes a notice of rulemaking for any proposed or Final Rule, it must prepare and make available for public comment a regulatory flexibility analysis that describes the effect of the rule on small entities (i.e., small businesses, small organizations, and small governmental jurisdictions). The Small Business Administration's regulations at 13 CFR part 121 define a small business, in part, as a business entity “which operates primarily within the United States.” (13 CFR 121.105(a)). No regulatory flexibility analysis is required if the head of an agency certifies that the rulemaking action would not have a significant economic impact on a substantial number of small entities. SBREFA amended the Regulatory Flexibility Act to require Federal agencies to provide a statement of the factual basis for certifying that an action would not have a significant economic impact on a substantial number of small entities.

NHTSA has considered the effects of this Final Rule under the Regulatory Flexibility Act. This Final Rule applies to States and they are not considered to be small businesses under the Regulatory Flexibility Act. States may employ contractors to collect survey data (which may be small businesses), but this Final Rule merely changes the procedures of collecting survey data and will not have a significant impact on the costs or profits of small businesses. Therefore, I certify that this Final Rule would not have a significant economic impact on a substantial number of small entities.

C. Executive Order 13132 (Federalism)

Executive Order 13132, “Federalism” (64 FR 43255, August 10, 1999), requires NHTSA to develop an accountable process to ensure “meaningful and timely input by State and local officials in the development of regulatory policies that have federalism implications.” “Policies that have federalism implications” are defined in the Executive Order to include regulations that have “substantial direct effects on the States, on the relationship between the national government and the States, or on the distribution of power and responsibilities among the various levels of government.” Under Executive Order 13132, the agency may not issue a regulation with Federalism implications that imposes substantial direct compliance costs and that is not required by statute unless the Federal government provides the funds necessary to pay the direct compliance costs incurred by State and local governments or the agency consults with State and local governments in the process of developing the proposed regulation. The agency also may not issue a regulation with Federalism implications that preempts a State law without consulting with State and local officials.

The agency has analyzed this Final Rule in accordance with the principles and criteria set forth in Executive Order 13132 and has determined that this Final Rule does not have sufficient Federalism implications to warrant consultation with State and local officials or the preparation of a Federalism summary impact statement. This Final Rule does not impose substantial direct compliance costs. While the costs to the States may vary depending on such factors as the State's current survey design and the size of the State, the agency estimates that the average cost to a State would be at most in the tens of thousands of dollars. We note that Federal funds from a number of NHTSA grant programs may be used to defray these costs. This Final Rule also does not preempt any State law or regulation or affect the ability of States to discharge traditional State government functions.

While the agency has determined that this Final Rule does not have sufficient Federalism implications to warrant formal consultation with State and local officials, the agency is aware that the revised criteria will impact States. For a number of years, the agency has had ongoing discussions with State officials about the seat belt use survey criteria. Several of these State officials expressed concerns about the accuracy and consistency of the survey results. Before the NPRM was published in January 2010, the agency discussed the possibility of revising the seat belt use survey criteria with officials from State Highway Safety Offices at Governors Highway Safety Association (GHSA) meetings. The agency sought their views on the need to change the criteria and potential areas of revision. Generally, these State officials were supportive of revising the seat belt use survey criteria to make the survey results more accurate and consistent.

In addition, when the NPRM was published, the agency reached out to the States and encouraged States to review the NPRM and provide comments. NHTSA received extensive comments from many States and interested parties, such as associations and universities and contractors, who assist States in conducting and analyzing the results of seat belt use surveys. As discussed in the preamble of the Final Rule, NHTSA revised the criteria to reduce further the impact on States in response to the States' comments.

D. Executive Order 12988 (Civil Justice Reform)

Pursuant to Executive Order 12988, “Civil Justice Reform” (61 FR 4729, February 7, 1996), the agency has considered whether this rulemaking would have any retroactive effect. This rulemaking action would not have any retroactive effect. This action meets applicable standards in sections 3(a) and 3(b)(2) of Executive Order 12988, Civil Justice Reform, to minimize litigation, eliminate ambiguity, and reduce burden.

E. Paperwork Reduction Act

Under the Paperwork Reduction Act of 1995, a person is not required to respond to a collection of information by a Federal agency unless the collection displays a valid Office of Management and Budget (OMB) control number. There are reporting requirements contained in the Final Rule that are considered to be information collection requirements under the Paperwork Reduction Act, as that term is defined by OMB in 5 CFR Part 1320. The estimated total annual burden is 19,040 hours. The total estimated number of respondents is 56 (50 States, the District of Columbia, Puerto Rico and 4 territories).

Pursuant to the Act, the agency solicited public comments on the proposed collection of information, with a 60-day comment period, in the notice of proposed rulemaking published on January 28, 2010 (75 FR 4509). The agency will publish a separate Federal Register Notice when we submit the information collection request to OMB for approval.

F. Unfunded Mandates Reform Act

Section 202 of the Unfunded Mandates Reform Act of 1995 (UMRA) requires Federal agencies to prepare a written assessment of the costs, benefits, and other effects of proposed or Final Rules that include a Federal mandate likely to result in the expenditure by State, local, or tribal governments, in the aggregate, or by the private sector, of more than $100 million annually (adjusted for inflation with a base year of 1995 (about $118 million in 2004 dollars)). This Final Rule does not include a Federal mandate resulting in annual State expenditures that would exceed the $100 million threshold.

G. National Environmental Policy Act

NHTSA has reviewed this rulemaking action for the purposes of the National Environmental Policy Act. The agency has determined that this Final Rule does not have a significant impact on the quality of the human environment.

H. Executive Order 13175 (Consultation and Coordination With Indian Tribes)

The agency has analyzed this Final Rule under Executive Order 13175, and has determined that the Final Rule does not have a substantial direct effect on one or more Indian tribes, does not impose substantial direct compliance costs on Indian tribal governments, and does not preempt tribal law. Therefore, a tribal summary impact statement is not required.

I. Regulatory Identifier Number (RIN)

The Department of Transportation assigns a regulation identifier number (RIN) to each regulatory action listed in the Unified Agenda of Federal Regulations. The Regulatory Information Service Center publishes the Unified Agenda in April and October of each year. You may use the RIN contained in the heading at the beginning of this document to find this action in the Unified Agenda.

J. Privacy Act

We will post all comments we receive, without change, to http://www.regulations.gov,, including any personal information you provide. Using the search function of our docket Web site, anyone can find and read the comments received into any of our dockets, including the name of the individual sending the comment (or signing the comment for an association, business, labor union, etc.). You may review DOT's complete Privacy Act Statement in the Federal Register published on April 11, 2000 (65 FR 19477-78), or you may visit http://DocketsInfo.dot.gov.

K. Congressional Review of Agency Rulemaking

The agency has not submitted the Final Rule to the Congress and the Government Accountability Office under the Congressional Review of Agency Rulemaking Act, 5 U.S.C. 801 et seq. This Final Rule is not a “major rule” within the meaning of the Act.

List of Subjects in 23 CFR Part 1340

• Grant programs—Transportation
• Highway safety
• Intergovernmental relations
• Reporting and recordkeeping requirements

For the reasons discussed in the preamble, the National Highway Traffic Safety Administration revises 23 CFR part 1340 to read as follows:

PART 1340—UNIFORM CRITERIA FOR STATE OBSERVATIONAL SURVEYS OF SEAT BELT USE

Subpart A—General

1340.1

Purpose.

1340.2

Applicability.

1340.3

Definitions.

Subpart B—Survey Design Requirements

1340.4

In general.

1340.5

Selection of observation sites.

1340.6

Assignment of observation times.

1340.7

Observation procedures.

1340.8

Quality control.

1340.9

Computation of estimates.

Subpart C—Administrative Requirements

1340.10

Submission and approval of seat belt survey design.

1340.11

Post-approval alterations to survey design.

1340.12

Re-selection of observation sites.

1340.13

Annual reporting requirements.

Appendix A to Part 1340—State Seat Belt Use Survey Reporting Form

Authority: 23 U.S.C. 402; delegation of authority at 49 CFR 1.50.

Subpart A—General

§ 1340.1

Purpose.

This part establishes uniform criteria for State surveys of seat belt use conducted under 23 U.S.C. 402, procedures for NHTSA approval of survey designs, and administrative requirements relating to State seat belt surveys.

§ 1340.2

Applicability.

This part applies to State surveys of seat belt use, beginning in calendar year 2012 and continuing annually thereafter.

§ 1340.3

Definitions.

As used in this part—

Access ramp means the segment of a road that forms a cloverleaf or limited access interchange.

Cul-de-sac means the closed end of a road that forms a loop or turn-around.

Non-public road means a road on which members of the general public are not allowed to drive motor vehicles.

Nonresponse rate means, for any survey variable, the percentage of unknown values recorded for that variable.

Observation site means the physical location where survey data are collected.

Passenger motor vehicle means a motor vehicle with a gross vehicle weight rating of less than 10,000 pounds, including a passenger car, pickup truck, van, minivan or sport utility vehicle.

Service drive means the segment of a road that provides access to businesses and rest areas.

Traffic circle means the segment of a road or intersection of roads forming a roundabout.

Unnamed road means a road, public or private, that has no name or number designation and is often a farm or logging road.

Vehicular trail means a road designed or intended primarily for use by motor vehicles with four-wheel drive.

Subpart B—Survey Design Requirements

§ 1340.4

In general.

This subpart sets forth the minimum design requirements to be incorporated in surveys conducted under this part.

§ 1340.5

Selection of observation sites.

(a) Sampling frame requirements—

(1) County coverage. The sampling frame from which observation sites are selected shall include counties or county-equivalents (including tribal territories), as defined by the U.S. Census Bureau, that account for at least 85 percent of the State's passenger vehicle occupant fatalities, provided that the average of the last three, four or five years, at the State's option, of available Fatality Analysis Reporting System (FARS) data or State fatality data approved by NHTSA shall be used to determine the State's passenger vehicle occupant fatalities.

(2) Road coverage.

(i) States shall select observation sites from a database of road inventories approved by NHTSA or provided by NHTSA.

(ii) Except as provided in paragraph (a)(2)(iii) of this section, all roads in the State shall be eligible for sampling. The sampling frame may not be limited only to roads having a stop sign, stop light or State-maintained roads.

(iii) The sampling frame need not include: rural local roads, as classified by the Federal Highway Administration's Functional Classification Guidelines, in counties that are not within a Metropolitan Statistical Area (MSA), as published by the Office of Management and Budget; non-public roads; unnamed roads; unpaved roads; vehicular trails; access ramps; cul-de-sacs; traffic circles; or service drives.

(b) Sampling selection requirements. The set of road segments selected for observation sites shall be chosen based on probability sampling, except that—

(1) The specific observation site locations on the sampled road segments may be deterministically selected;

(2) An alternate observation site may be used to replace an observation site selected based on probability sampling if it is located in the same county or county-equivalent, and has the same roadway classification (e.g., local road segment, collector road segment) when using the protocol of substitution and rescheduling of observation sites pursuant to paragraph (c) of this section.

(c) Requirements for substitution and rescheduling of observation sites. The survey design shall include at a minimum the following protocols:

(1) Protocol when observation site is temporarily unavailable for data collection.

(i) Observers shall return to the observation site at another time provided that it is on the same day of the week and at same time of the day or select an alternate observation site, as described in paragraph (b)(2) of this section, provided the data are collected on the same day and at approximately the same time as the originally-scheduled observation site.

(ii) The original observation site must be used for future data collections.

(2) Protocol when observation site is permanently unavailable for data collection.

(i) Except as provided in paragraph (c)(2)(ii), another observation site shall be selected in accordance with paragraph (b) of this section.

(ii) If it is not feasible to select another observation site based on probability sampling for the current data collection, an alternate observation site, as described in paragraph (b)(2) of this section, may be selected, provided the data is collected on the same day and at approximately the same time as the originally-scheduled observation site.

(iii) For future data collections, another observation site must be selected based on probability sampling in accordance with paragraph (b) of this section.

(d) Precision requirement. The estimated seat belt use rate must have a standard error of no more than 2.5 percentage points.

§ 1340.6

Assignment of observation times.

(a) Daylight hours. All daylight hours between 7 a.m. and 6 p.m. for all days of the week shall be eligible for inclusion in the sample.

(b) Random assignment. Except as provided in paragraph (c) of this section, the day-of the week and time-of-the-day shall be randomly assigned to observation sites.

(c) Grouping of observation sites in close geographic proximity. Observations sites in close geographic proximity may be grouped to reduce data collection burdens if:

(1) The first assignment of an observation site within the group is randomly selected; and

(2) The assignment of other observations sites within the group is made in a manner that promotes administrative efficiency and timely completion of the survey.

§ 1340.7

Observation procedures.

(a) Data collection dates. All survey data shall be collected through direct observation completely within the calendar year for which the Statewide seat belt use rate will be reported. Except as provided in § 1340.5(c), the survey shall be conducted in accordance to the schedule determined in § 1340.6.

(b) Roadway and direction(s) of observation—

(1) Intersections. If an observation site is located at an intersection of road segments, the data shall be collected from the sampled road segment, not the intersecting road segment(s).

(2) Roads with two-way traffic. If an observation site is located on a road with traffic traveling in two directions, one or both directions of traffic may be observed, provided that—

(i) If only one direction of traffic is observed, that direction shall be chosen randomly;

(ii) If both directions of traffic are observed at the same time, States shall assign at least one person to observe each direction of traffic.

(c) Vehicle coverage. Data shall be collected by direct observation from all passenger motor vehicles, including but not limited to passenger motor vehicles used for commercial purposes, passenger motor vehicles exempt from the State's seat belt use law and passenger motor vehicles bearing out-of-State license plates.

(d) Occupant coverage. Data shall be collected by direct observation of all drivers and right front passengers, including right front passengers in booster seats, but excluding right front passengers in child safety seats. Observers shall record a person as—

(1) Belted if the shoulder belt is in front of the person's shoulder;

(2) Unbelted if the shoulder belt is not in front of the person's shoulder;

(3) Unknown if it cannot reasonably be determined whether the driver or right front passenger is belted.

(e) Survey data. At a minimum, the seat belt use data to be collected by direct observation shall include—

(1) Seat belt status of driver;

(2) Presence of right front passenger; and

(3) Seat belt status of right front passenger, if present.

(f) Data collection environment. When collecting seat belt survey data—

(1) Observers shall not wear law enforcement uniforms;

(2) Police vehicles and persons in law enforcement uniforms shall not be positioned at observation sites;

(3) Communications by signage or any other means that a seat belt survey is being or will be conducted shall not be present in the vicinity of the observation site.

§ 1340.8

Quality control.

(a) Quality control monitors. Monitors shall conduct random, unannounced visits to no less than five percent of the observation sites for the purpose of quality control. The same individual shall not serve as both the observer and quality control monitor at the same observation site at the same time.

(b) Training. Observers and quality control monitors involved in seat belt use surveys shall have received training in data collection procedures within the past twelve months. Observers and quality control monitors shall be trained in the observation procedures of § 1340.7 and in the substitution and rescheduling requirements of § 1340.5(c).

(c) Statistical review. Survey results shall be reviewed and approved by a survey statistician, i.e., a person with knowledge of the design of probability-based multi-stage samples, statistical estimators from such designs, and variance estimation of such estimators.

§ 1340.9

Computation of estimates.

(a) Data used. Except as otherwise provided in this section, all data collected pursuant to § 1340.7(e) shall be used, without exclusion, in the computation of the Statewide seat belt use rate, standard error, and nonresponse rate.

(b) Data editing. Known values of data contributing to the Statewide seat belt use rate shall not be altered in any manner.

(c) Imputation. Unknown values of variables shall not be imputed unless NHTSA has approved the State's imputation procedure prior to data analysis.

(d) Sampling weights. The estimation formula shall weight observed data by the sampling weights as required by the sample design and any subsequent adjustments.

(e) Sampling weight adjustments for observation sites with no usable data. States shall include a procedure to adjust the sampling weights for observation sites with no usable data, including observation sites where no data were collected and observation sites where data were discovered to be falsified.

(f) Nonresponse rate.

(1) Subject to paragraph (f)(2) of this section, the nonresponse rate for the entire survey shall not exceed 10 percent for the ratio of the total number of recorded unknown values of belt use to the total number of drivers and passengers observed.

(2) The State shall include a procedure for collecting additional observations in the same calendar year of the survey to reduce the nonresponse rate to no more than 10 percent if the nonresponse rate in paragraph (f)(1) of this section exceeds 10 percent.

(g) Variance estimation.

(1) Subject to paragraph (g)(2) of this section, the estimated standard error, using the variance estimation method in the survey design, shall not exceed 2.5 percentage points.

(2) If the standard error exceeds this threshold, additional observations shall be conducted in the same calendar year of the survey until the standard error does not exceed 2.5 percentage points.

Subpart C—Administrative Requirements

§ 1340.10

Submission and approval of seat belt survey design.

(a) Contents: The following information shall be included in the State's seat belt survey design submitted for NHTSA approval:

(1) Sample design—The State shall-

(i) Define all sampling units, with their measures of size, as provided in § 1340.5(a);

(ii) Specify the data source of the sampling frame of road segments (observation sites), as provided in § 1340.5(a)(2)(i);

(iii) Specify any exclusions that have been applied to the sampling frame, as provided in § 1340.5(a)(2)(iii);

(iv) Define what stratification was used at each stage of sampling and what methods were used for allocation of the sample units to the strata;

(v) Specify the method used to select the road segments for observation sites as provided by § 1340.5(b).

(vi) List all observation sites and their probabilities of selection;

(vii) Explain how the sample sizes were determined, as provided in § 1340.5(d);

(viii) Describe how observation sites were assigned to observation time periods, as provided in § 1340.6; and

(ix) Identify the name and describe the qualifications of the State survey statistician meeting the requirements in § 1340.8(c).

(2) Data collection—The State shall—

(i) Define an observation period;

(ii) Specify the procedures to be implemented to reschedule or substitute observation sites when data collection is not possible on the date and time assigned, as provided in § 1340.5(c);

(iii) Specify the procedures for collecting additional data to reduce the nonresponse rate, as provided in § 1340.9(f)(2);

(iv) Describe the data recording procedures; and

(v) Specify the number of observers and quality control monitors.

(3) Estimation—The State shall—

(i) Describe how seat belt use rate estimates will be calculated;

(ii) Describe how variances will be estimated, as provided in § 1340.9(g);

(iii) Specify imputation methods, if any, that will be used, as provided in § 1340.9(c);

(iv) Specify the procedures to adjust sampling weight for observation sites with no usable data, as provided in § 1340.9(e); and

(v) Specify the procedures to be followed if the standard error exceeds 2.5 percentage points, as required in § 1340.5(g).

(b) Survey design submission deadline. For calendar year 2012, States shall submit proposed survey designs to NHTSA for approval no later than January 3, 2012. Thereafter, States should submit survey designs for NHTSA approval as specified in § 1340.11.

§ 1340.11

Post-approval alterations to survey design.

After NHTSA approval of a survey design, States shall submit for NHTSA approval any proposed alteration to their survey design, including, but not limited to, sample design, seat belt use rate estimation method, variance estimation method and data collection protocols, at least three months before data collection begins.

§ 1340.12

Re-selection of observation sites.

(a) Re-selection of observation sites. States shall re-select observation sites using updated sampling frame data, as described in § 1340.5(a), no less than once every five years.

(b) Re-selection submission deadline. States shall submit updated sampling frame data meeting the requirements of § 1340.5(a) for NHTSA approval no later than March 1 of the re-selection year.

§ 1340.13

Annual reporting requirements.

(a) Survey data. States shall report the following information no later than March 1 of each year for the preceding calendar year's seat belt use survey, using the reporting form in Appendix A to this part:

(1) Spreadsheet in electronic format containing the raw data for each observation site and the observation site weight;

(2) Statewide seat belt use rate estimate and standard error;

(3) Nonresponse rate for the variable “belt use,” as provided in § 1340.9(f);

(4) Dates of the reported data collection;

(5) Observation sites, identified by type of observation site (i.e., observation site selected in the original survey design, alternate observation site selected subsequent to the original survey design), and by characteristics of the observation site visit (i.e., at least one vehicle observed, no vehicles observed); and

(6) Name of the State survey statistician meeting the qualification requirements, as provided in § 1340.8(c).

(b) Certifications by Governor's Highway Safety Representative. The Governor's Highway Safety Representative (GR) or if delegated in writing, the Coordinator of the State Highway Safety Office, shall sign the reporting form certifying that—

(1) ________has been designated by the Governor as the GR, and if applicable, the GR has delegated the authority to sign the certification in writing to ________, the Coordinator of the State Highway Safety Office;

(2) The reported Statewide seat belt use rate is based on a survey design that was approved by NHTSA, in writing, as conforming to the Uniform Criteria for State Observational Surveys of Seat Belt Use, 23 CFR Part 1340;

(3) The survey design has remained unchanged since the survey was approved by NHTSA; and

(4) ________, a qualified survey statistician, reviewed the seat belt use rate reported in Part A (of the certification) and information reported in Part B and has determined that they meet the Uniform Criteria for State Observational Surveys of Seat Belt Use, 23 CFR part 1340.

(d) Audits. NHTSA may audit State survey results and data collection. The State shall retain the following records for five years and make them available to NHTSA in electronic format within four weeks of request:

(1) Computation programs used in the sample selection;

(2) Computation programs used to estimate the Statewide seat belt use rate and standard errors for the surveys conducted since the last NHTSA approval of the sample design; and

(3) Sampling frame(s) for design(s) used since the last NHTSA approval of the sample design.

APPENDIX A TO PART 1340—STATE SEAT BELT USE SURVEY REPORTING FORM

PART A: To be completed by the Governor's Highway Safety Representative (GR) or if applicable, the Coordinator of the State Highway Safety Office.

State:

Calendar Year of Survey:

Statewide Seat Belt Use Rate:

I hereby certify that:

• ________has been designated by the Governor as the State's Highway Safety Representative (GR), and if applicable, the GR has delegated the authority to sign the certification in writing to _________, the Coordinator of the State Highway Safety Office.
• The reported Statewide seat belt use rate is based on a survey design that was approved by NHTSA, in writing, as conforming to the Uniform Criteria for State Observational Surveys of Seat Belt Use, 23 CFR Part 1340.
• The survey design has remained unchanged since the survey was approved by NHTSA.
• ________, a qualified survey statistician, has reviewed the seat belt use rate reported above and information reported in Part B and has determined that they meet the Uniform Criteria for State Observational Surveys of Seat Belt Use, 23 CFR Part 1340.

Signature

Date

Printed name of signing official

Part B—Data Collected at Observation Sites

Site ID	Site type 1	Date observed	Sample weight	Number of drivers	Number of front Passengers	Number of occupants 2 belted	Number of occupants unbelted	Number of occupants with unknown belt use
Total

Standard Error of Statewide Belt Use Rate ^[3] ___

Nonresponse Rate, as provided in § 1340.9(f)

Nonresponse rate for the survey variable seat belt use: ____

Issued on: March 28, 2011.

David L. Strickland,

Administrator, National Highway Traffic Safety Administration.

Footnotes

1.  In 2000, NHTSA clarified that States are permitted to “cluster sample,” i.e., group observation sites according to geographic areas to minimize travel time and distance required to conduct the observations.

2.  49 CFR 1.50 (delegation of authority to Administrator of National Highway Traffic Safety Administration).

3.  The WA TSC stated that the revised criteria will substantially increase its costs, up to $100,000 the first year. (WA TSC at 1, 4.) NHTSA does not believe that many, if any, States will incur costs of $100,000 to redesign the survey. Based on our estimate, we expect that many States will spend tens of thousands of dollars to redesign the survey. The CA OTS stated that it expected increased costs due to a 50 percent increase in the number of survey sites. (CA OTS at 1. See also MO DOT at 1.) It is not clear how CA OTS determined that it would need a 50 percent increase in the number of observation sites under the new criteria. Because we have revised the survey criteria in order to provide States with design flexibility, we believe that CA OTS would not need a 50 percent increase in the number of observation sites and suggest California and other States consult its statistician for survey design alternatives. (See discussion in Section III.D.1 below.)

4.  NV DPS stated that the raw data from all States should be available online to the public. (NV DPS at 1.) We believe that the idea of making raw data available online deserves further consideration, but that it is a policy decision that is ancillary to the rulemaking. Therefore, we do not make changes to the rule in response to this comment.

1.  

 Identify if the observation site is an original observation site or an alternate observation site.

2.  

 Occupants refer to both drivers and passengers.

3.  The standard error may not exceed 2.5 percent.