Continuous, Long Term Roadside Surveys to Evaluate Community Safety Programs

Robert B. Voas, Peter Roeper and Paul Gruenewald

Prevention Research Center, 2150 Shattuck Avenue, Suite 900, Berkeley, CA 94704, USA

ABSTRACT

Voluntary Breath Test Surveys of motorists between 9 pm and 2 am on alternate weekend nights are being employed to measure changes in driver behavior related to alcohol safety programs in three U.S. communities; Oceanside and Salinas, California and Florence, South Carolina. These cities are implementing programs directed at strengthening their enforcement of drunk driving laws; at promoting responsible serving practices; at reducing service to youths under age 21; and at limiting the density of liquor outlets through local zoning regulations. Initial data collected through the first two years of the program suggest that some of the programs which were initiated after a baseline year of data collection, are having an effect on intermediate measures such as perceived risk of arrest. Origin and destination studies permit the determination of where drinking occurred and allow these data to be related to participation of the outlets in responsible beverage service programs. This survey also allows the tracking of changes in the knowledge of significant elements of drunk driving legislation such as the legal BAC limit. Changes in knowledge and perceived risk can be related to on-going surveys of newspaper coverage of drunk driving issues.

INTRODUCTION

The Random Digit Dialing (RDD) telephone survey is a widely used technique for collecting information from a random sample of households in industrialized countries in which the vast majority of households have at least one telephone line. This technique has been used in the United States to contact a representative sample of licensed motorists who can be interviewed at some length about their drinking and their driving behavior. While refusal rates can be reasonably high as much as 40%-50% of those contacted, these surveys generally collect a rich set of information including a number of demographic variables which permit the weighing responses to conform to census or license data or the use of demographic and other key variables as covariates in the comparison of different groups of drivers. The technique is reasonably expensive, running between $30-$35 for a completed 15-20 minute interview.

An alternative, less expensive ($20 per interview) technique for collecting data on drivers is the use of roadside, voluntary breath test surveys. This procedure employs a police officer to direct a random sample of drivers off the road into a safe interview area. There motorists are approached by the survey conductor and requested to participate in the short interview. Acquiescence to such interviews typically runs at least 90%. In some areas this may be as high as 95%. An advantage of such face-to-face interviews is that the drivers can be asked to provide a breath test for BAC analysis and approximately 90% of those approached will agree to provide a breath sample. Another important feature of the roadside survey is that it is a behavioral measure since those contacted and interviewed are actual drivers using the roads at the time and places of the interview. Telephone surveys must depend upon both the memory and the frankness the individual interviewed. The objective of this preliminary study was to compare the populations contacted by these two survey methods.

METHODS

The Prevention Research Center has implemented both types of surveys as part of the evaluation of alcohol trauma prevention programs in three communities; Oceanside and Salinas in California and in Florence, South Carolina. In each community, roadside surveys are conducted on Friday and Saturday nights of alternate weekends between the hours of 9 p.m. and 2 a.m. Typically, one hundred motorists are interviewed each night for a total of approximately 400 in the course of a month at each site. In each of these communities, there is also a continuous random digit dialing survey of adults 18 and older which results in about 100 interviews per month.

The data used in the present study are shown in Table 1. Just under 4,000 telephone interviews have been conducted in each of the three intervention communities in the 33 months from April 92 to January 95. Roadside surveys began in June 91 in Oceanside and nearly 13,000 interviews have been collected through November 94. The roadside interviews started about a year later in Salinas, California and Florence, South Carolina. Because surveys are not conducted on rainy days, a larger number of interviews have been collected in the California cities of Oceanside and Salinas than in Florence where rain is frequently a problem. The primary purpose of both surveys was to detect change in the number of drinking drivers as a measure of program outcome, therefor the analysis was limited to drivers who were "drinkers" (i.e.reported that they had had at least one drink in the past four weeks). The proportion of all interviewees who reported drinking in the last twenty eight days is shown in Table 1. Oceanside, a beach resort adjacent to the Camp Pendleton Base housing 40,000 marines has a higher proportion of "drinkers" that the other communities. Florence, located in the traditionally more religious southern states had more infrequent drinkers and abstainers on the telephone survey, but the proportion of drinkers in the roadside survey was similar to the agricultural community of Salinas.

Table 1
Overview of Telephone and Roadside Surveys Covered in Current Study

In order to compare the drivers who drink contacted in the telephone survey with drinkers interviewed at the roadside, a set of questions were inserted into the telephone survey identifying respondents who would be likely to be driving between 9 a.m. and 2 a.m. on Friday and Saturday nights. The procedure was to ask the telephone respondent whether he or she had driven between 9 p.m. and 2 a.m. on the previous Friday and Saturday night. Those that indicated that they had driven on those nights were asked whether they were driving before or after midnight. This set of questions was used to identify those among the telephone respondents who were most likely to be from the same population sampled on Friday and Saturday nights at the roadside survey.

GENDER OF NIGHTTIME DRIVERS WHO DRINK

Table 2 provides the proportion of male respondents reporting that they drove on weekends at each of the three intervention sites. As would be expected, there are fewer males reporting that they did not drive on the weekend in the telephone survey. Those who drove before the midnight on the weekend divided fairly evenly between males and females, while males predominated as the drivers in the time period after midnight. The gender variation between those who did not drive on the weekend and those who drove after midnight on the weekend is greatest for the Southern city of Florence and less pronounced in the two California cities. Relative to the roadside survey, the telephone survey underestimated the proportion of males driving both before and after midnight on Friday and Saturday nights. The roadside survey confirms the tendency for males to comprise a larger proportion of the drivers after midnight on weekends which is not surprising given traditional dating patterns in the United States.

Table 2
Gender of Weekend Nighttime Drivers Who Drank in the Last Four Weeks

AGE OF NIGHTTIME DRIVERS WHO DRINK

Table 3 compares the age of telephone and roadside respondents in Oceanside. The results shown are typical of the other two communities. As might be expected, youthful drivers in the 18-25 age group were the largest group reporting weekend driving after 12 midnight, while they composed one of the smallest groups among those who stated they did not drive on the preceding weekend. The elderly comprised a larger portion of the telephone respondents who reported they did not drive on the previous weekend and the smallest portion of those who drove after midnight. In comparing the telephone with the roadside survey, the two methods provide reasonably similar estimates of involvement for the 26-35 age group, but the RDD technique appears to significantly under-estimate the portion drivers in the 18-25 age group and over-estimate the proportion of drivers above age 36 in the nighttime population. This pattern of relationship between the telephone and the roadside survey was repeated in all three communities.

Table 3
Nighttime Drivers Who Drank in the Last Four Weeks in Salinas: Percent in Each of Six Age Groups

ETHNICITY

As might be expected, the ethnic composition of the both the telephone and roadside survey populations varied significantly from community to community. One example of the relationship of the telephone to the roadside survey is shown in Table 4 which covers surveys in Salinas. Salinas is a city which according to census figures has a population which is 45% Hispanic, 45% Caucasian and 2% African American with the remaining 8% divided among other demographic groups. As can be seen in Table 4, there is an indication that the telephone survey roughly parallels the census data with the exception that proportionately fewer Hispanic households were contacted than would be expected from the census population. This may be due to a lower proportion of Hispanic households with telephones and/or a larger number of individuals within the household in Hispanic families. Hispanics appear to be better represented in the roadside surveys, where, before midnight, they out numbered the Caucasian drivers.

Table 4
Ethnicity of Nighttime Drivers Who Drank in Last Four Weeks in Salinas: Percent of All Drivers

DISCUSSION

The data summarized in this preliminary study demonstrate that there is a significant difference in gender, age and ethnicity between drinkers who report that they drove on weekend evenings in a telephone survey and drinkers actually interviewed on weekends at roadside surveys. Two features of roadside surveys probably contribute to these differences. First, the probability of being chosen for interview in a roadside survey provides a natural measure of the amount of driving exposure because those individuals who drive more are more likely to be stopped and selected for the roadside survey. The telephone survey on the other hand provided no measure of vehicle milage. It is probable that males and young persons not only drive more frequently but also greater distances on weekends. Secondly, the roadside surveys can not be mounted at random geographical locations because they must be established at points on the roadway where vehicles can be safely stopped and were there is sufficient traffic to make them reasonably efficient. This biases the locations to urban areas where there may be more evening recreational driving to and from drinking establishments than if survey sites included low traffic neighborhoods or high speed freeways. This could also produce a bias toward male and youthful drivers.

These differences between telephone and roadside surveys can serve to highlight issues that might be missed by the use of one technique alone. From Table 1, it appears that the extent of the potential drinking driving problem might be significantly underestimated in Florence if reliance were placed on the telephone survey alone. From Table 4, it appears that the role of Hispanics in the drinking driving problem might be under estimated for the telephone survey. The proportion of Hispanics in the roadside survey appears to be closer to that expected from census data that is the household survey. These issues will be explored further in additional analyses of the data collected in both surveys form the three communities.