Family Communication Patterns and Teen Driving Intervention Effectiveness

Abstract

Objectives:

Teen drivers are at increased crash risk, largely due to lack of experience. Parents play a key role in influencing teen behaviors and attitudes around driving safety. Parent-involved interventions may improve teen driving safety but tend to be resource intensive and have limited scalability. In this study, we examined how family communication patterns (FCPs) impact teen risky driving and the effectiveness of a parent-focused teen driving intervention.

Methods:

Our data came from a large randomized controlled teen driving intervention trial. We randomized parent-teen dyads into one of 3 groups: parent communication intervention plus in-vehicle event recorder feedback; in-vehicle event recorder feedback only; or control. The primary outcome variable was teen risky driving (self-reports and triggered events); the primary exposure variables were FCPs and intervention group. We used generalized linear models to calculate effect estimates.

Results:

Teens’ baseline risky driving did not vary by family communication pattern. The impact of the parent-focused intervention was stronger in families with a laissez-faire FCP. The laissez-faire FCP focuses little on child conformity and downplays communication.

Conclusions:

These results provide a framework for targeting high-resource teen driving interventions (event recorder feedback and parent-communication training) to families with laissez-faire communication patterns to attain the greatest risk reductions.

Motor vehicle crashes represent the leading cause of death among adolescents and young adults (ages 15-24) in the United States.¹ Each day, 17 young people die and more than 1500 are injured in motor vehicle crashes.¹ Although the increased risk for motor vehicle crash persists through young adulthood, the first 6 months of driving is a particularly risky time for novice drivers due to limited driving experience and underdeveloped decision-making skills.^2,3

Parents play a crucial role in shaping the attitudes, skills, and behaviors of young drivers. Several researchers have demonstrated that youth driving behaviors often mimic the driving behaviors of their parents.^4-6 Young drivers may observe and emulate the driving behaviors of their parents, but shared personality, attitudes, and knowledge likely contribute to the similarities between young drivers and their parents.⁷ Parents may shape the driving behaviors of their teens further through parenting practices. They may impose or reinforce limits on risky driving conditions, such as limiting the number of passengers allowed in the vehicle or restricting late night driving.⁸ Teens who perceive their parents’ style as authoritative or authoritarian are more likely to use seatbelts and less likely to speed compared with passive parents.⁹

Teens’ attitudes and behaviors also may be shaped through their family’s communication style.¹⁰ According to the Family Communications Patterns Theory, patterns of communication within the family may be categorized on 2 dimensions: socio-orientation and concept-orientation (Figure 1).^11,12 Socio-orientation refers to the level of harmony within the communication between family members. Families at the higher end of the socio-orientation dimension are likely to avoid conflict and emphasize obedience. Concept-orientation refers to the openness within communication between family members. Families at the higher end of the concept-orientation dimension often have frequent and lengthy interactions.

Figure 1. Family Communication Patterns²³.

Socio-orientation: level of harmony within the communication between family members

Concept-orientation: openness within communication between family members

Based on the 2 dimensions of communication, families may be categorized into one of 4 family communication patterns (FCP): pluralistic, protective, consensual, or laissez-faire (Figure 1).¹³ Pluralistic families are high in concept-orientation and low in socio-orientation. They value open communication and encourage consideration of multiple perspectives when developing opinions. Protective families are low in concept-orientation and high in socio-orientation. They encourage children to adopt the opinions of the parents through strong emphasis on obedience and conflict avoidance. Consensual families are high in both socio- and concept-orientation. They often discuss issues, but children are ultimately expected to adopt the opinions of their families. Laissez-faire families are low in both socio- and concept-orientation. They tend to downplay communication, place limited emphasis on child conformity, and act without consideration of other family members. For example, parents of teens in laissez-faire families may not place much emphasis on rule setting related to driving and even if they do set limits, the teen may decide not to conform without any real consequence for their non-conformity.

The 4 FCP categories align with the 4 well-known Baumrind¹⁴ parenting styles: permissive, authoritarian, authoritative, and neglectful. Pluralistic families often have permissive parents; protective families often have authoritarian parents; consensual families often have authoritative parents; and laissez-faire families often have neglectful parents. However, Baumrind’s parenting styles and FCPs have important differences. FCP theory operates under the premise that all family members’ contributions are important to consider, not just parental behavior.¹⁵ FCP theory also underpins the importance that a family’s shared social reality or understanding of each other is vital to optimal functioning.^15,16

Previous research examining how FCPs influence teens’ attitudes toward driving indicates that parents that use a consensual communication approach tend to have more frequent conversations about driving safety.¹⁷ As a result of these frequent conversations, teens tend to have more positive attitudes towards driving safety. Although family communication patterns are associated with more frequent conversations about driving safety in previous studies, it is not clear if these conversations result in safer driving among teens.¹⁷ Moreover, no studies have examined the relationships of family communication patterns and driving behaviors among teens using both subjective (eg, self-report) and objective measures (eg, triggered events from an event recorder). It is important to address these gaps to inform intervention development and delivery.

Many past teen driving interventions have focused on improving parent-teen communication around driving through either passive dissemination of information (content delivered with little or no engagement with family) or direct engagement with families. One review found that interventions with direct engagement with families (eg, Steering Teens Safe, Teen Driving Plan, and STEER) to be more promising in improving teen driving compared with passive dissemination of materials.¹⁸ To date, no reviewed studies of the impact of passive dissemination on teen crash rates have found significant changes to those rates. Because these direct engagement programs target parent communication and behaviors, it is likely that parent-teen communication styles moderate the effectiveness of the programs. However, to our knowledge, no studies have examined the influence of communication style on the effectiveness of these programs to reduce teen risky driving behaviors.

In this study, we aimed to: (1) examine the associations between family communication patterns and teens’ risky driving identified through self-report and monitoring equipment installed in their cars; and (2) determine whether family communication style moderated the effects of the parent-communication focused intervention (ie, Steering Teens Safe)¹⁹ on teens’ risky driving. We hypothesized that teen risky driving and the impact of a teen driving intervention would vary by family communication pattern.

METHODS

Conceptual Model

The research questions and hypotheses for this study were based on our a priori conceptual model (Figure 2), which indicates teen attitudes and behaviors are influenced by teen (eg, sex), parent (eg, education), and family characteristics (eg, family communication patterns). The introduction of a teen driving intervention also can influence teen attitudes and behaviors and all of these factors may serve as predictors of a teen’s risky driving behavior.

Figure 2. Conceptual Model of Teen Risky Driving.

Study Design

Study data used for analyses were collected from a larger, randomized controlled intervention trial whose aim was to improve teen driving by training parents to use motivational interviewing when discussing driving with their teen.¹⁹ The motivational interviewing-based communication intervention was called Steering Teens Safe and was coupled with a kinematically triggered in-vehicle event recorder that served as the basis for providing weekly feedback to teens and their parents about driving errors.^20,21

Study participation began with a 4-week baseline period, as close as possible to the date the teen received his or her intermediate license (first opportunity to drive without adult supervision, but with time-of-day restrictions). During this period, teens drove as they normally would. After baseline, driving events were recorded for an additional 16-week consecutive follow-up period for a total study enrollment of 20 weeks. For purpose of analysis, the 16-week follow-up period was later divided into 4, 4-week follow-up segments. Event recorders were installed in teens’ cars upon enrollment for all study groups.

Study Participants

Parent and teen dyads were passively recruited (eg, school mailings, handouts at school activities) via 13 high schools and one healthcare employer in a Midwest state between August 2011 and December 2014. Parents and teens were eligible if the teen: (1) obtained their intermediate license (allows unsupervised driving, with restrictions) during the study period, (2) had access to a vehicle and were the primary driver of that vehicle, (3) had vision that met legal driving standards, (4) were fluent in English (teen and at least one parent), (5) provided proof of vehicle insurance, and (6) did not have any siblings already enrolled in the study (one dyad per family could be enrolled).

Interested persons contacted the study team and the team screened them for eligibility. We obtained informed consent from all eligible parent-teen dyads that were enrolled in the study.

Randomization and Intervention Groups

Following consent, dyads were randomly assigned to one of 3 groups: (1) Steering Teens Safe parent intervention and in-vehicle event recorder feedback (ERF+STS), (2) in-vehicle event recorder feedback only (ERF), or (3) control (C). Randomization was conducted in blocks of 3, representing each of the study groups. This approach was used to account for seasonal weather effects and school session.

Parents in the ERF+STS group received the evidence-based Steering Teens Safe communication training,^19-21 which teaches and encourages parents to communicate frequently with their teens using motivational interviewing techniques, during baseline and prior to receiving any feedback from the in-vehicle system.

Upon system activation after baseline, parents received feedback regarding their teen’s driving in the form of driving reports. These ‘Report Cards’ were mailed to the parent along with a DVD including video clips of events triggered by the in-vehicle event recorder (see below for a description of the event recorder). Weekly reports also included the following components: number of overall events recorded that week, a list of triggered behaviors (eg, hard braking), a summary of safety-related behaviors observed from the triggered event video clips (eg seat belt use, failure to follow traffic signals), possible goals for the next week, summary graphs of the teen’s cumulative events (updated weekly), number of events compared to other teens in the study, and seat belt use by the teen compared to other teens in the study. Those assigned to the feedback only (ERF) condition began receiving driving reports following baseline, but were not trained in motivational interviewing. Finally, those in the control condition received neither training in motivational interviewing nor driving reports.

Variables and Measures

Triggered event recorders.

The event recorder system (DriveCam^® by Lytx) captured audio and video data (one camera facing into the vehicle cabin and one facing out toward the road) when braking, acceleration, or steering exceeded a 0.5g threshold, which triggered recording. Using a buffer, the recorder saved a total of 12 seconds of video – 8 seconds before the trigger and 4 seconds after.

Family communication patterns.

Evaluation of family communication pattern employed the Family Communication Pattern Scale.²² The FCP Scale consists of 10 items that evaluate communication on 2 dimensions, socio-orientation (5-items) and concept-orientation (5-items), with reliability for 3-to 6-item versions ranging from 0.61 to 0.84 and 0.54 to 0.82 respectively.²³ The reliability (Cronbach alphas) for the subscales in the current study was 0.74 for socio-orientation and 0.73 for concept-orientation. The scale appears to have good face validity, but criterion-related validity of FCP and teen risky driving have not been established. Both teens and their parents indicated how often they used certain types of statements on a 4-point Likert scale that ranged from “never” to “often.” For example: “I have driven for years – don’t argue with me” and “Say that every member of your family should have some say in family decisions.” Teen and parent scores were combined and median scores were used to divide families into one of the 4 communication patterns according to Fitzpatrick, Ritchie:²⁴ (1) consensual (high on socio-orientation and high on concept-orientation), (2) pluralistic (low on socio-orientation, but high on concept-orientation), (3) laissez-faire (low on both socio-orientation and concept-orientation), or (4) protective (high on socio-orientation, but low on conceptual-orientation). Combined parent and teen scores were used to allow for representation of both perspectives.

Risky driving inventory – self-report measure.

Our subjective outcome measure was teen-reported risky driving, which was measured using an adapted 25-item version of the Risky Driving Inventory.²⁵ The original version of this inventory had acceptable reliability (alphas ranging from 0.62 to 0.77) and the reliability for scale in the current study was good (alpha = 0.78). The scale appears to have good face validity, but other measures of validity are not available. Teens were asked how many times they performed a list of driving behaviors in the past week. The reports used in this study were taken at the end of the 4-week baseline period. Scores were computed by adding together the total number of times of each of the risky driving behaviors reported.

The risky driving behaviors included:

• Exceed the speed limit in residential or school zones

• Change lanes without signaling

• Make an illegal U-turn

• Drive through a stop sign without stopping completely

• Switch lanes to weave through slower traffic

• Drive 10 to 19 miles per hour over the speed limit

• Drive through an intersection when the light was red or just turning red

• Tailgate or follow someone too closely

• Change lanes without enough room between cars

• Cut in front of a car to turn

• Drive 20 or more miles per hour over the speed limit

• Pass 2 or 3 vehicles at a time on a road with 2-way traffic

• Pass a car in a no-passing zone

• Drive after using illegal drugs

• Drive after drinking alcohol

• Drive without a seat belt

• Talk on the phone while you were driving

• Turn up the music so loud you could not hear another car honk

• Drive in a way to show off to other people

• Race another car for a short distance

• Drive when you were upset or emotionally disturbed

• Lose traction while on a gravel road

• Take a turn on a roadway so quickly that you felt the car tilt

• Drive through an uncontrolled intersection without slowing or stopping

• Drive without paying attention to unusual circumstances (eg, work zone, emergency vehicles)

Triggered risky driving events – objective measure.

Our objective outcome measure was the total number of valid events captured by in-vehicle event recorders. Valid, triggered events included any driver actions that resulted in a g-force greater than 0.5 and were subjectively confirmed as valid by raters during video coding. Triggered events were coded and categorized into the following categories: hitting a curb while parking, shifting error, near-crash evasive action (hard braking, close proximity to another vehicle, squealing tires, startled expression), and crashes (eg, making contact with another object, leaving the roadway, lane departures resulting in crash).

Demographics.

Demographics used in model fitting procedures and final analysis included teen sex (male, female), parental education (college graduate, other), and parental marital status (married, divorced, other). Parent and teen data were linked by dyad ID.

Data Analysis

To test the research question: “How do family communication patterns impact risky driving (independent of an intervention, at baseline)?” both baseline self-reported risky driving and baseline objective triggered events were used as dependent measures. Because the Risky Driving Inventory (RDI), one of the outcomes of interest, was highly right-skewed and is greater than or equal to zero by definition, generalized linear models (GLMs) were used for the analysis. A GLM based on the gamma distribution was found to fit the data best. Teen sex and parent age, education level, marital status, and employment status were considered as potential confounders. Teen sex and parental marital status were found to be important confounders. AIC was used to select the GLM distribution and confounders in the final model. The aforementioned gamma GLM was used for analyzing the actual events as well. However, in modeling the events, the log of the miles driven was used as an offset.

To test the research question: “Does the impact of teen driver interventions (ERF, ERF +STS) vary by FCP?” generalized linear models were fit using generalized estimating equations (GEE). Because the outcome of interest (triggered events) is a count, models based on Poisson and negative binomial distributions were considered. The negative binomial distribution was found to fit the data best as determined by QIC. An exchangeable working correlation matrix was used to account for the within-subject correlation of events data collected on each teen longitudinally. Candidate models, offset by the log of the number of miles driven, were specified a priori and QIC was used to determine a final model that best balanced goodness-of-fit (ie, conformity of the model to the data) and parsimony (ie, simplicity or number of terms in the model) among the models under investigation. A reduction in QIC of 2-units was considered meaningful, using the recommendations by Burnham and Anderson.²⁶ We considered using the 4-level categorical variable of FCP and the 2 separate indicators of concept-orientation and socio-orientation. We found FCP to be a better fit than concept- and socio-orientation.

Although teens were randomized to one of 3 treatment groups (ERF, ERF +STS, and control), the control group was found to have an appreciably higher rate of valid trigger events at baseline. Thus, we sought to control for a few baseline characteristics in our modeling strategy. Particularly, we investigated controlling for the teen’s sex, parental education, and the baseline rate of valid trigger events. We considered adjusting for the baseline rate of valid trigger events as categorical or continuous during model selection. Treating the rate as continuous had the best fit. Rate ratio estimates comparing intervention groups within a level of FCP were computed using a least squares means approach that adjusted for other effects in the model by using observed means and proportions for continuous and categorical predictors, respectively. We conducted all analyses using SAS (version 9.4) and R (version 3.3.1).

RESULTS

Characteristics of Study Participants

Of the 161 dyads enrolled in the study, 147 were eligible for analysis. Eleven dyads were removed because they did not have complete follow-up data and 3 were missing the family communication pattern survey items from which the FCP categories are derived.

The study population had slightly more females (51.3%) than males and most were in 10^th grade (73.3%) and Caucasian (89.3%). Over 80% of the included teens reported first driving at age 14, which is the earliest age graduated driver licensing can begin in Iowa.²⁷ Most (82%) teens reported they drive to school. Average miles driven per teen each month of the study ranged from 500 to 600 miles and increased slightly over time across all of the study groups, but varied widely from teen to teen. Participating parents were primarily the teen’s mother (74%) and 95% were Caucasian. The parent population was also highly educated (66% college graduates) and married (88%).

Self-reported Risky Driving

Responses to the 25 items of the self-reported risky driving inventory were largely right-skewed (large positive value for skewness) and some items had a few observations with large values (large positive value for kurtosis). The most frequently cited risky behaviors (number of times in the past week) included exceeding the speed limit in residential or school zones (mean = 1.94, SD = 3.28, skewness = 3.00, kurtosis = 3.00), talking on the phone while driving (mean = 1.25, SD = 1.95, skewness = 2.28, kurtosis = 6.00), and turning up the music so loud they could not hear another car honk (mean = 1.10, SD = 2.99, skewness = 4.46, and kurtosis = 22.58). The least cited behaviors were drive after drinking alcohol (mean = 0.01, SD = 0.08, skewness = 11.92, kurtosis = 140.52), drive after using illegal drugs (mean = 0.01, SD = 0.13, skewness = 13.60, kurtosis = 140.52), and pass in a no passing zone (mean = 0.03, SD = 0.21, skewness = 7.44, kurtosis = 59.00).

FCPs as Rated by Teens and Their Parents

The distribution of families into the 2 FCP dimensions was nearly evenly split, with 45% of families categorized as high on concept-orientation, and 48% high on socio-orientation (these are not mutually exclusive). After dividing persons into the 4 combinations of concept-orientation (high/low) by socio-orientation (high/low), results showed that laissez-faire (33.3%) was the most prevalent family communication pattern, followed by consensual (26.5%), pluralistic (24.5%), and then protective patterns (15.6%; Table 1). The distribution of FCPs within each study group was similar to the distribution of FCP across the whole study population, with the exception of the ERF+STS study arm, which had slightly more pluralistic families (30.0%) and fewer protective families (10.0%), compared to the ERF only and control groups.

Table 1.

Distribution of Family Communication Patterns across Intervention Groups

	Intervention Group
Family Communication Pattern	Control	ERF+STS	ERF	Total
Consensual	13 (28.3%)	13 (26.0%)	13 (25.5%)	39 (26.5%)
Laissez-Faire	15 (32.6%)	17 (34.0%)	17 (33.3%)	49 (33.3%)
Pluralistic	10 (21.7%)	15 (30.0%)	11 (21.6%)	36 (24.5%)
Protective	8 (17.4%)	5 (10.0%)	10 (19.6%)	23 (15.6%)
Total	46	50	51	147

Note.

ERF = event recorder feedback; STS = Steering Teens Safe

Risky Driving (both Self-reports and Triggered Events) by FCPs

Model results did not find family communication patterns to be predictive of teens’ self-reported risky driving or triggered events (Tables 2 and 3), after adjusting for teen sex and parent education.

Table 2.

Self-reported Risky Driving Rate Ratios by Teen and Parent Characteristics and Family Communication Pattern

Variable	Estimated Mean Ratio	95% CI	p-value
Teen Sex: Female	1.18	(0.70, 2.01)	.54
Parent Education: College graduate	0.83	(0.46, 1.5)	.54
FCP: Protective	1.39	(0.60, 3.22)	.44
FCP: Pluralistic	0.96	(0.49, 1.88)	.91
FCP: Consensual	0.76	(0.38, 1.53)	.44

Note.

FCP = Family Communication Pattern

Referent categories: Male, Some college or less, Laissez-Faire

Table 3.

Triggered Event (Offset by Log (Miles) Driven) Rate Ratios Teen and Parent Characteristics and Family Communication Pattern

Variable	Estimated Rate Ratio	95% CI	p-value
Teen Sex: Female	0.62	(0.33, 1.15)	.13
Parent Education: College graduate	0.87	(0.43, 1.75)	.70
FCP: Protective	1.17	(0.43, 3.20)	.76
FCP: Pluralistic	0.75	(0.35, 1.61)	.46
FCP: Consensual	1.08	(0.47, 2.47)	.86

Note.

FCP = Family Communication Pattern

Referent categories: Male, Some college or less, Laissez-Faire

FCPs and Triggered Event Rates by Intervention Group

The final model of FCPs and event rates by intervention group included covariates for parental education, teen sex, and the interaction between continuous baseline event rate and intervention group. Predictors included FCP and intervention group. Results from the generalized linear models revealed that teens in the combined ERF+STS intervention group had lower event rates than those in the control group, regardless of FCP. Adjusted for teen sex, parental education, baseline rate, and study segment, teens with a laissez-faire family communication pattern experienced the largest benefit from the addition of STS to the intervention. Teens with a laissez-faire family communication pattern who received the ERF+STS intervention experienced, on average, a 29% reduction (p < .01) in their event rate compared to teens who received the ERF intervention alone. As Figure 3 shows, this comparison was not statistically significant in the other 3 family communication pattern groups. The ERF alone intervention group also had lower event rates compared to the control group, among the subgroup of persons with laissez-faire and protective FCPs.

Figure 3. Triggered Event Rate Ratios by Family Communication Patterns and Intervention Group Adjusted for Teen Sex, Parental Education, Baseline Event Rate, and Study Segment.

Note.

ERF = event recorder feedback; STS = Steering Teens Safe; C = Control

DISCUSSION

In this study, we aimed to understand the influence of family communication patterns (FCPs), both as a predictor and as a moderator of the parent-focused intervention, on teens’ risky driving (self-reports and triggered events). Previous research indicates that parent-child communication processes are important contributors to teen health-risk behaviors^28,29 and that parents can play an important role in a teen’s driving safety.^30-32 However, parents often are either not well trained to communicate about and reinforce safe driving behaviors of their teens, or underestimate driving risks and do not exert the full extent of their control toward managing those risks.³³

During adolescence, the parent-child relationship may become more egalitarian, as the adolescent becomes more autonomous.^34,35 Despite these changes, family communication patterns are considered predictable and stable over time;^36,37 they also have been linked to stable and enduring personality traits,³⁸ and receptivity to intervention may differ by those patterns and traits.

Our results indicate that teen driving behavior did not vary by FCP at baseline, prior to an intervention. However, the impact of the teen driving intervention was moderated by FCP. Teens in the intervention groups (event recorder, event recorder plus Steering Teens Safe) had lower triggered event rates, compared to control group teens. This suggests that the combined intervention (ERF + STS) had some beneficial aspects across all teens.

When we examined the specific components of the intervention (ERF, STS) we found that laissez-faire families experienced the most benefit when STS (ie, parents’ use of motivational interviewing) was added to ERF, as the STS+ERF group had lower event rates compared to the ERF alone group. Families with laissez-faire communication patterns tend to have limited communication and do not have a strong focus on conformity of their children.^24,36,37 Our results suggest that laissez-faire parents benefit from the communication training included in Steering Teens Safe, which is reflected in improvements in their teens’ driving, possibly because they had the most room for improvement related to communication. These families likely experienced an improvement in communication because motivational interviewing trains the “interviewer” to ask open-ended questions that encourage more detailed responses from the “interviewee.”

For families with a protective FCP, the ERF part of the intervention had the most impact. It is possible that electronic recorder feedback may better align with the traits of protective families and less with STS, given their focus on the child being obedient to their parent’s authority and low priority for open communication.^10,24 Protective families are likely to have more authoritarian parents, which are associated with low warmth/support and high rules and monitoring, which also aligns more with an ERF-based intervention component versus the STS communication-based component.¹⁴

For consensual and pluralistic families, the combined STS+ERF intervention was more effective compared to controls; however, ERF alone versus control and ERF+STS versus ERF alone were not statistically significant and the explanation of these findings is unclear. Pluralistic families tend to have less consistency in parental messaging and place a low value on child conformity, which may make teens less likely to adopt parental views.²⁴ The combined ERF+STS might have provided both the parental communication tools and consistency in messaging needed to see an impact among those teens.

Parents in consensual families often have a permissive parenting style, which is characterized by high warmth and support, but few restrictions.¹⁴ Consensual families also value open communication, but differ from pluralistic families, in that they also expect high conformity from their children.²⁴ Consensual families align with the authoritative parenting style, which is high in warmth and support, but also high in monitoring and restrictions.¹⁴ Teens in consensual families may be more like more likely to adopt parental views because they are presented in a consistent manner and are expected to conform. Previous research shows that the consensual FCP and authoritative parenting styles are the most ideal in terms of increased communication about safe driving¹⁷ and lower reports of risky driving behaviors among teens.³⁹ It is possible that this group may have had the least room for improvement, so the combined ERF+STS provided the most comprehensive tools for which to see an incremental impact in this group.

Our results highlight the value of understanding a family’s communication pattern, characterized by levels of conversation and conformity⁴⁰ to understand the impact of teen driving interventions. Measuring a family’s communication pattern prior to intervention also may be useful for tailoring an intervention to address areas for improvement, namely increasing communications about safe driving and/or increasing parental monitoring, restrictions, and limit-setting. In addition, understanding which families may benefit most from the intervention may assist agencies in deciding how to distribute this type of program when resources are limited. In protective families, for example, teens in either intervention had lower rates of triggered events compared to the control group. However, there were no differences in protective families between the 2 intervention groups. Conversely, in laissez-faire families, teens in the event recorder plus Steering Teens Safe intervention had lower triggered event rates than teens in the event recorder only intervention, although teens in either intervention had lower rates than teens in the control group. Taken together, these results suggest that protective families can receive the lower resource intervention (event recorder only) without significantly reduced benefit whereas laissez-faire families benefit most from the higher resource intervention (event recorder plus Steering Teens Safe).

Limitations

The sample size for this study was not powered to answer hypotheses based on subsets of the randomized groups; therefore, the results should be interpreted with caution, as they are likely under-powered. The generalizability of study results may be limited, given the study sample was drawn from one state and the majority of participants were white, well-educated, and high-income. The FCPs for this study were measured for driving safety only, which does not provide a full picture of the family dynamics and parental influence, though it may have relevance to teen driving safety. Future studies are needed to capture a more diverse sample and also to measure FCPs for a wider array of topic areas.

Our results may be used to tailor and target teen driving interventions based on the parent-teen relationship and existing family communication styles. These findings address the issue of scalability of effective, yet resource-intensive, teen driving interventions by identifying families and teens that gained the greatest safety benefits from the intervention. Families with a laissez-faire communication pattern may see more benefit from teen driving interventions with communication-components, compared to families with other communication patterns.