Five teens did not complete the entire 10-week study; one teen vehicle had mechanical issues and
discontinued participation mid-way through the study, one system had consistent hardware failures
such that removal of the system was required, and three teens requested early removal of the system
that resulted in gaps in certain sections of the treatment and transfer data collection periods. The
hardware failures appeared to be combination of a display issue in addition to instances of system
tampering where power cables or cables were removed from the system. The data for these
individuals is noted or removed within the analysis section where appropriate. Data sets were not
separated based on gender and were collapsed to provide additional power for statistical tests.
Prior to statistical analyses the entire data set was reviewed to identify anomalies (e.g., outliers due to
missing data or hardware/software data collection failures) and, in those instances, the data were
removed from the data set.
Data analyses were conducted on miles driven by teens as identified by the absence of an RFID
marker in the data. An RFID card was detected for 15.7 percent of the total mileage (17.8% of the
GPS mileage) and varied between participants (see e.g., Appendix E). If an RFID card was present
but teen mode was selected on the TDIS, the data were still excluded from the analysis because the
presence of RFID card and the absence of adult mode selection did not provide enough information
to discriminate if a teen was driving with an adult, an adult was driving in teen mode, or a teen was
driving without an adult but the RFID card was still present. An assumption was made that data
without any RFID card contained valid teen driving data and were used for the analysis.
Valid GPS information allowed the STC to identify speed limit zones by querying the onboard
NAVTEQ map database. When the system database did not provide a speed limit for a particular
section of roadway or if there was a speed range (e.g., no absolute speed limit defined) then the
speed subsystem did not provide feedback to the teen driver. For teen drivers the speed limit was
known for 66 percent (25,216 miles) of all recorded miles driven by teens (38,125 miles). The
number of miles travelled by teens where the speed limit was not known was higher in Minnesota
(6,996 miles) compared to Maryland (5,913 miles). Instances of GPS signal loss or GPS signal
without map data occurred that impacted the speed feedback subsystem. If speed limit data were not
available, the related variable information was not calculated for these sections of data and
subsequently was not used in the overall analyses. The mileage driven by teens during the day or
night is reported in Appendix F: Vehicle Miles Traveled Night and Day by Teens.
Only significant results for each dependent variable within each subsystem are presented in the
following sections.
31
Table 4-3. Overview of outcomes of analyses of variance.
Variable Occupancy Factor
St