(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.