Observed Day and Night Belt Use
Method
The sample sites used in the night (9 p.m. until 3:59 a.m.) belt observations were the same sites approved for use in Connecticut’s “S. 157 compliant” full statewide belt use survey. For both day and night surveys there were 100 sites in seven of Connecticut’s eight counties. These counties contain 97 percent of the State’s population (See Geary and Chaudhary, 2004).
The daytime procedures followed the Uniform Criteria for State Observational Surveys of Safety Belt Use. The nighttime procedures were developed to mimic the guidelines set forth for daytime observations.
Day and night observations were conducted twice; once each pre- and post-campaign. Prior to Connecticut’s May 2004 Click It or Ticket mobilization campaign a “mini” statewide observation of 17 sites, all of which were also part of the full statewide survey, was conducted. This mini survey was designed to give a “snapshot” of Connecticut belt use and was not intended to provide detailed analysis beyond this purpose.
Immediately following the mobilization, a second round of observations was conducted using all 100 sites included in Connecticut’s statewide safety belt survey.
Relatively sophisticated night vision equipment was used for nighttime observation when roadway lighting was insufficient to make the observations. This near-military-grade equipment allows for vision in both light and dark, whereas earlier versions of night vision equipment would have been essentially blinded by the headlights of the observed vehicles. Specifically, an XR5 “Image Intensifier” tube, manufactured by Delft Electronic Products, was mounted in Unitec GS7 night vision goggles. According to the manufacturer, the XR5 tube contains an “auto-gating” feature that results in no “blooming” and the smallest “halo” available. That is, the tube does not shut down in bright light and headlights make only a relatively small ring of light.
To supplement this equipment, handheld infrared spotlights, visible only with the use of the night-vision goggles and not to the naked human eye, further illuminated the roadway, making vehicle occupants visible for belt observations even in total darkness.
The goggle/spotlight combination requires practice to use effectively. As such, all observers for the night survey were experienced with daytime observations. Each performed night observations at several sites prior to the start of data collection.
Each observation site was observed during the day for 45 minutes. Observers coded belt use for cars, pickup trucks, sport utility vehicles, and vans for the driver and the outboard front-seat passenger (children in a front-seat child restraint are excluded from the survey). Each lane of traffic in one direction was observed for equal amounts of time. Where traffic was moving too quickly on high-volume roadways, a reference point some distance away on the road was chosen by which the next qualifying vehicle must pass before being recorded on the data sheet.
Nighttime procedures were consistent with daytime procedures, with some changes. Nighttime observations were conducted with the use of the night vision goggles and handheld infrared spotlights. Because of the equipment, observations were done by a two-person team, with one person observing traffic and the other recording the results as stated by the observer. For night observations, each lane of traffic in one direction was observed for equal amounts of time unless vehicles were passing at an extremely slow rate, in which case all lanes were observed simultaneously. Also, due to limitations in the equipment to observe vehicles at high speeds, observers conducted some observations on exit ramps of limited access highways as opposed to the main travel lanes used for daytime observations. Observations were to be made on the highway itself, typically at the nearest lighted overpass or rest area, whenever possible.
Weighting
In order to render the observations at the 100 sites representative of the entire State, the data are weighted based on Average Annual Daily Traffic (AADT). This weighting results in sites that account for a higher proportion of statewide traffic holding more “weight” than those which account for less traffic. The weighting for the daytime observations are based on 24-hour traffic counts. Given that (in Connecticut, for example) about 90 percent of the travel occurs during daytime hours, this method is not problematic. Using 24-hour traffic counts to determine traffic volume for night, however, is problematic. In Connecticut, traffic patterns shift at night. Relatively more traffic at night is on interstate highways than on other types of roads.
Given this shift in traffic patterns, traffic volume for the hours 9 p.m. to 3:59 a.m. was calculated by using the traffic counts for the night hours only. Hourly counts were provided by the Connecticut DOT Bureau of Policy and Planning. Thus, unlike day observations, night observations were weighted based on the traffic counts occurring only during the times at which night observations were conducted. The weights are designed to produce an estimated belt use for the entire “night” and not for any specific hour of the night. Two sites resulted in no night observations (i.e., no vehicles passed the observers within the 45-minute observation period) and therefore did not contribute to the night belt use rate.
Results
For the full statewide (100-site) survey, for both drivers and passengers, daytime safety belt use in Connecticut produced a weighted safety belt usage rate of 83.0 percent (N = 28,269). Nighttime weighted belt use was 76.6 percent (N = 9,075). Thus, belt use was 6.4 percentage points lower at night than during the day. Chi square analyses were run on unweighted data only as weighted data are not amenable for statistical analyses because of the transformation of the underlying N. The chi square (on the unweighted data; day: 80.6 percent; night: 73.6 percent) demonstrates a significant difference between day and night belt use (χ2 = 202.06, p < .001).
As described above, highway sites during the day were all measured on the highways themselves. Some of the highway sites (N = 11) during the night were observed at off ramps because poor lighting and high vehicle speeds made on-highway observations impossible. As a result, another calculation was conducted removing both day and night data from sites where relocation occurred to ensure that the difference in belt use between day and night was not due to this methodological difference. The unweighted difference between day (80.2 percent) and night (73.9 percent) belt use was 6.3 percentage points, a statistically nonsignificant difference.
Chi square analyses on vehicle type, gender, driver, and passenger belt use showed that for all categories belt use at night was significantly lower (p < .05) than day belt use. The greatest absolute difference in belt use by time of day was for SUV occupants, where belt use was almost 9 percentage points lower at night. According to the analyses, pickup truck occupant belt use, while lowest of the vehicle types, showed the smallest difference between day and night belt use (see Table 1).
There was an interrelationship between time of day and population density. The difference in day and night belt use was much greater on urban roadways than rural roadways. On rural roadways the difference between day (85.8 percent) and night (83.5 percent) was less than 3 percentage points (χ2 = 3.04, p > .05). On urban roadways the difference between day (79.6 percent) and night (72.6 percent) was 7 percentage points (χ2 = 177.75, p < .01). That is, the finding that safety belt use is lower during the night than during the day is most characteristic of urban locations in Connecticut (see Figure 2).
Table 1. Percent Observed Belt Use by Time of Day (100 Post Sites in CT)
Night |
Day |
Dif |
χ2 p value |
||
|---|---|---|---|---|---|
| Vehicle Type | |||||
| car | % |
74.5 |
82.1 |
7.6 |
<.001 |
N |
6,516 |
17,315 |
|||
| pickup | % |
56.6 |
62.3 |
5.7 |
<.02 |
N |
512 |
2,521 |
|||
| SUV | % |
76 |
84.8 |
8.8 |
<.001 |
N |
1,431 |
5,293 |
|||
| van | % |
72.4 |
79.5 |
7.1 |
<.001 |
N |
615 |
3,133 |
|||
| Gender | |||||
| male | % |
68.7 |
76 |
7.3 |
<.001 |
N |
5,657 |
16,298 |
|||
| female | % |
81.7 |
86.8 |
5.1 |
<.001 |
N |
3,397 |
11,934 |
|||
| Occupant Type | |||||
| driver | % |
72.8 |
80.7 |
7.9 |
< .001 |
N |
7,002 |
23,500 |
|||
| passenger | % |
76.4 |
79.9 |
3.5 |
=.001 |
N |
2,073 |
4,762 |
|||
Further examination using FARS for Connecticut crashes demonstrates a similar pattern of belt use among fatally injured front seat occupants of passenger vehicles. Specifically, there was no difference (χ2 = 0.00, p > .05) between day and night belt use on rural roads (day: 51.7 percent; night: 51.9 percent), but there was a relatively large difference (χ2 = 4.036, p < .05) between belt use during the day (55.0%) and night (40.0%) on urban roads. A backward stepwise logistic regression on the observed and FARS data combined produced a significant interaction of time of day by population density (p < .05).
Figure 2. CT FARS (2002 – 2003) versus Observed Belt Use by Population Density (2004)
[d]
Statewide belt use observations were conducted immediately following an enforcement mobilization aimed at safety belt use. This mobilization could have had a differential affect on day versus night belt use.
A comparison of pre-campaign and post-campaign belt use was conducted. The pre-campaign data came from the 17 “mini” sites while the post-campaign data contained the full 100 sites that include the 17 mini sites. A backward binary logistic regression produced a significant (p = .01) interaction between times of day and pre/post campaign (see Figure 3). The difference, pre-campaign, between day (78.7 percent; N = 2,914) and night (66.5 percent; N = 872) was greater than the difference post campaign between day (80.6%) and night (73.6%). According to chi square analyses, the differences between daytime and nighttime observations were significant for both pre-campaign (χ2 = 54.02, p < .001) and post-campaign (χ2 = 199.37, p < .001). Clearly there was a smaller—yet still significant—difference between day and night belt use following the mobilization than before. Thus, the enforcement and media campaign appeared to have an impact on nighttime safety belt use.
Figure 3. Percent Observed Belted Pre- and Post-Campaign by Time of Day
[d]