INTRODUCTION

Background
The Motor Vehicle Occupant Safety Survey is conducted biennially for the National Highway Traffic Safety Administration (NHTSA). It is a national telephone survey composed of two questionnaires, each administered to several thousand randomly selected persons age 16 and older. The Version 1 Questionnaire emphasizes safety belt issues while Version 2 emphasizes child restraint issues. The questionnaires also contain smaller modules addressing such areas as air bags, emergency medical services, and crash injury experience. For the 2003 survey, each questionnaire was administered to approximately 6,000 individuals.

NHTSA conducted the first Motor Vehicle Occupant Safety Survey in 1994. Subsequent versions of the survey have included modest revisions to reflect changes in information needs. Thus the 2003 survey contained numerous items from the earlier surveys allowing the agency to monitor change over time in knowledge, attitudes, and (reported) behavior related to motor vehicle occupant safety. The 2003 survey also included new questions dealing with such areas as wireless phone features and use while driving, inspection stations for child restraints, and new LATCH and tether child car seat attachments.

The following report presents findings from the 2003 Motor Vehicle Occupant Safety Survey pertaining to child car seats. Specifically, it explores the following areas: 1) seating position of children age 12 and younger in motor vehicles; 2) transporters of young children; 3) car seat use by children under age 9; 4) car seat installation, inspection, and training; 5) reasons for non-use of car seats by children; 6) booster seat issues; and 7) enforcement of child restraint laws. An eighth section examines MVOSS trends between 1994 and 2003 on selected child restraint issues.

Methodology
The 2003 Motor Vehicle Occupant Safety Survey was conducted by Schulman, Ronca, & Bucuvalas, Inc. (SRBI), a national survey research organization. SRBI conducted a total of 12,377 telephone interviews among a national population sample. To reduce the burden on respondents, the survey employed two questionnaires. A total of 6,180 interviews were completed with Version 1 and 6,197 interviews were completed with Version 2. Although some questions appeared in both versions (e.g., demographics, crash injury experience, safety belt use), each questionnaire had its own set of distinct topics. Each sample was composed of approximately 6,000 persons age 16 and older, including oversamples of persons age 16-39. The procedures used in the survey yielded national estimates of the target population within specified limits of expected sampling variability, from which valid generalizations can be made to the general public.

The survey was conducted from January 8, 2003 to March 30, 2003. For a complete description of the methodology and sample disposition, including computation of weights, refer to the 2003 Motor Vehicle Occupant Safety Survey, Volume I. Methodology Report. This report includes English and Spanish language versions of the questionnaires.

The percentages presented in this report are weighted to reflect accurately the national population age 16 and older. Unweighted sample sizes (“N”s) are included so that readers know the exact number of respondents answering a given question, allowing them to estimate sampling precision (see Appendix A for related technical information).

Percentages for some items may not add to 100 percent due to rounding, or because the question allowed for more than one response. In addition, the number of cases involved in subgroup analyses may not sum to the grand total who responded to the primary questionnaire item being analyzed. Reasons for this include some form of nonresponse on the grouping variable (e.g., “Don’t Know” or Refused), or use of only selected subgroups in the analysis. Moreover, if one of the variables involved in the subgroup analysis appeared on both versions of the questionnaire but the other(s) appeared on only one questionnaire, then the subgroup analysis was restricted to data from only one version of the questionnaire.

The survey employed two questions to categorize cases for subgroup analyses involving race and ethnicity. The first asked respondents if they considered themselves to be Hispanic or Latino. Those who said “Yes” composed the Hispanic analytic subgroup in the study, those who said “No” composed a non-Hispanic comparison group. The second question was treated independently of the ethnicity question, i.e., it was asked of every respondent. The interviewers recited several different racial categories, and asked respondents which categories described them. Respondents could select more than one. For purposes of analysis, a respondent was assigned to a specific racial category if s/he selected only that category. The few respondents who selected multiple categories (fewer than 350 out of more than 12,000 cases) were analyzed as a separate multi-racial group that could include both Hispanics and non-Hispanics, and the Hispanic analytic subgroup included both African Americans/Blacks and Whites.

The abbreviations DK and Ref are frequently listed as response categories in the report. DK stands for “Don’t Know” and Ref stands for Refused. For most questions, the persons who answered “Don’t Know” vastly outnumbered those who refused to answer the question.