A Strategic Approach to Transforming Traffic Safety Culture to Reduce Deaths and Injuries (2018)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

5 Chapter 1: An Introduction to Terminology and Theory 1.1 Traffic Safety and Public Health The World Health Organization (WHO) has identified traffic safety to be an international public health issue (Peden, et al., 2004). Traffic crashes are predicted to become the third leading cause of disability and lost years of life worldwide in the next 10 years (World Health Organization, 2014).1 However, in the United States (US), traffic crashes have already reached this ranking in the past decade (NHTSA, 2006; NHTSA, 2003; NHTSA, 2008a; NHTSA 2009; NHTSA, 2011). For example, as recently as 2007 the US Department of Transportation concluded: Because of the young lives consumed, motor vehicle traffic crashes ranks third overall in terms of the years of life lost (i.e., the number of remaining years that people were expected to live had they not died), behind only cancer and heart diseases (NHTSA, 2011, p. 1). The economic and societal impact of traffic crashes in the US has been estimated to be more than $871 billion each year (NHTSA, 2014a). Given its moral and economic significance, the reduction of fatal (and serious) traffic crashes should be a national public health priority. Indeed, this priority is the basis of the Toward Zero Death (TZD) national initiative to reduce traffic fatalities to zero that has been adopted by many state traffic safety agencies (TowardsZeroDeaths, 2012). Traffic safety culture is recognized to be a key element of this national strategy: “Creating a positive traffic safety culture is integral to helping our nation move toward a vision of a highway system with no fatalities (TowardsZeroDeaths, 2012, p. 61).” A positive traffic safety culture is expected to not only reduce the role of 1 Public health often measures the impact of disease or injury in terms of Disability-Adjusted Years Lost (DALY), which assumes a standard life span at each age, for example, assuming a 20 year old male has a standard expected life span of 65 years. In this case, this 20 year old would lose 45 years of quality life if they were killed or seriously injured (disabled) in a car crash while aged 20 years (World Health Organization, 2014) . The total loss of DAYLs in a population is the sum of loss for all cases in the population.

road u stake 1.2 T T requi an an crash differ Haza Figur crash T in the road roadw condi T intera envir of its choic sers in traff holder actio he Role o he developm res an under alysis of ha . For examp ent aspects rds. e 2. The ro risk (modi he physical roadway en geometry (e ay, trees cl tions (e.g., r he structure ctions cont onment. Eac group mem es and instil ic fatalities ns to improv f the Road ent of a stra standing of t zards withi le, Figure 2 of the tran le of factor fied from E environmen vironment t .g., restricte ose to shou educed visib of the phys ribute to t h social gro bers. Cultu ls rewards a and serious i e traffic saf User tegic appro he factors th n the transp represents sportation s in the phy iksund, 200 t can presen hat can incre d sight dista lders on lim ility or slip ical environ he formatio up is charac re influenc nd punishm 6 njuries, but ety. ach to reduc at cause cra ortation env two categor environmen sical and so 9). t naturally ase crash ris nces), road ited access pery road co ment also i n of socia terized by a es the valu ents for beh also increas e traffic fata shes. Such u ironment th ies of risk t: System cial enviro occurring “ k. Such haz side obstacl facilities), nditions). nfluences hu l groups, w culture that es and beli aviors that e e road user a lities and se nderstandin at increase hazard that Hazards an nments tha system haza ards may inc es (e.g., anim and unfavor man intera hich defin influences efs related t ither confor cceptance o rious injurie g comes from the risk of emerge from d Behaviora t increase rds” inheren lude reduce als crossin able weathe ctions. Thes e our socia the behavior o behaviora m or violate f s a l t d g r e l s l

7 the cultural expectations of the group (Fishbein & Ajzen, 2010). Traffic safety culture refers to that subset of culture that influences behavioral choices that influence traffic safety. “Behavioral hazards” refer to behaviors committed by road users (e.g., speeding, impaired driving, and not wearing a seat belt) that increase crash risk. Behavioral hazards contribute to a majority of traffic crashes (Evans, 1991) – and as a result – also account for a significant percentage of traffic fatalities:  Speeding is related to 30% of US traffic fatalities (NHTSA 2014b).  Driving while impaired from alcohol is related to 30% of US traffic fatalities (NHTSA 2013).  Non-use of passenger restraints account for 56% of US traffic fatalities (NHTSA 2014c). It is reasonable to assume that the injury or fatality resulting from a traffic crash was unintentional, but in many cases, the behavioral hazard that caused the crash was deliberate. “Deliberate behaviors” are those the actor (1) was willing (desired) to commit when the opportunity arose, or (2) had already planned to commit (intended) in anticipation of that opportunity. For example, traffic violations such as speeding represent the deliberate decision to commit a behavior that contradicts accepted standards for safe driving (often represented by laws) (Stradling et al., 2000). As a result, communities in which many drivers commit traffic violations also have a higher crash risk (Stradling et al., 2000). Such evidence suggests that removing behavioral hazards (such as traffic violations) should be an effective way to significantly improve traffic safety. However, to change these behaviors, it is necessary to understand the influence of the social environment and traffic safety culture on the deliberate decision to engage such behaviors (see Figure 2). 1.3 The Role of the Social Environment Health is not merely a product of individual biology, psychology, and behavioral factors; it is the sum of collective social conditions created when people interact with the [social] environment. Thus, preventing unintentional injury requires attention to the entire social system (Allegrante et al., 2010, p. 24).

A comp repre 2008 depen exam and a Figur T natur – our of cu withi Reid, famil man m cultur s shown in rised of a h senting a set ; Hogg & R ding on ou ple, one indi resident of e 3. Levels he need to e. As shown sense of se ltural influen n a group is 2006). For y culture wh ay instead e while fulf Figure 3, t ierarchy of of common eid, 2006). W r similarity vidual may a specific co of social gr identify with in Figure 4 lf and belon ce across th relevant to example, a m ile fulfillin assume the illing his du he social en social laye attributes th e belong t to the comm be a parent ( mmunity (c ouping with and belon , our identifi ging within ese groups d the context an may ass g the goal o role of “em ties as a del 8 vironment rs. These l at differenti o multiple g on attribut family layer ommunity l in the soci g to social g cation with t the social e epends on o (goal) of th ume the rol f driving his ployee” and ivery-truck d can be view ayers define ate them fro roups withi es associate ), a delivery ayer). al environm roups is fu hese groups nvironment ur percepti e behavior e of “father” children to be more in river. ed as a “so social cat m other grou n our social d with those driver (wor ent. ndamental t forms our s . The source on that our i (Hornsey, 2 and be influ school. Lat fluenced by cial system egories, eac ps (Hornsey environmen groups. Fo kplace layer o our huma ocial identit and strengt dentified rol 008; Hogg & enced by h er, that sam his compan ” h , t r ), n y h e is e y

Figur T cultur huma to beh they (Matt physi envir coope cultur behav L cultu beha to pro e 4. Examp hese social e: “the colle n group from ave in ways identify (Fo aini & Thye cal environ onments. In rate and su es of those ioral decisi ogically the ral influen viors and re vide a stand le of social groups with ctive progra another” ( that encour gg, 2009). r, 1996). Cu ment to su this way, c rvive in the groups that ons and our n, traffic s ces across ject behavi ard operatio identity de in our socia mming of t Kluckhohn, age accepta Culture fu lture also p stain a co ulture prov ir environm define our collective w afety could the social oral hazard nal definitio 9 rived from l environme he mind wh 1951, p. 86) nce and avo nctions as rovides a sh nsistent set ides the fou ents (Brand social ident ellbeing. be significa environme s. To achie n of "traffic membershi nt are distin ich distingu . Humans a id rejection a “system o ared unders of behavi ndation for imonte et a ity have imp ntly impro nt were al ve this align safety cult p in social c guished by ishes the me re intrinsica by the group f social re tanding of t oral respon members o l., 2006). T ortant influ ved and su igned to p ment, it is f ure." ategories. their uniqu mbers of on lly motivate s with whic inforcement he social an ses to thos f a group t herefore, th ences on ou stained if a romote saf irst necessar e e d h ” d e o e r ll e y

1.4 D A about share artifa in thi such predi 2011 In traffi them behin some logic cultur partic Figur envir efining Tr rguably, the what cultu d among a g cts (Cooper, s world that an inclusive ct other tha ). stead, if the c safety cultu in its own d d behaviors thing inform , Figure 5 de e, which i ular group i e 5. Definit onment tha affic Safet greatest cha re includes roup of peo 2000; Luria such a defi definition h n itself: “it goal is to c re must be efinition: “ are the top ed but sepa picts delibe s defined a n the social ion of cultu t influence y Culture llenge of de . Historicall ple, as wel & Faraeli, 2 nition does n as limited u covers alm hange behav able to expla if behaviors ic of inves rate from cu rate (willful s the thou environmen re of share s deliberate 10 fining traffi y, culture h l as with the 008). It is d ot already tility becau ost everythi ioral hazard in and predi are the targ tigation, the lture” (Mye and intenti ghts shared t. d thoughts behaviors. c safety cult as been eq ir common ifficult to im include (My se nothing ng and ther s affecting ct these beh et of chang n behavior rs et al., 20 onal) behav among pe among grou ure is being uated with behaviors a agine wha ers et al., 20 is left for it eby nothing crash risk, t aviors rathe e, and the c s must be u 14, p. 27). B iors to be th ople identi ps in socia too inclusiv the though nd generate t else remain 14). And so to explain o ” (Alvesson he concept o r than includ ultural force nderstood a ased on th e outcome o fying with l e ts d s , r , f e s s is f a

11 Conceptualized in this way, “culture is in the mind of the people” (Geertz, 1973, p. 86). In general, thoughts are our mental representations and interpretations of the physical and social environment (Figure 2). More specifically, thoughts include values (ideals to which group members collectively aspire) and beliefs (understanding of the physical and social environment). Values and beliefs are the foundation of other types of thoughts including our attitudes about behaviors, perceptions of normal behavior, and perceived control over our behavioral choices in the social and physical environment. Collectively, these thoughts influence our willingness and intention to behave in these environments. It is important to note that behavioral hazards are not the only behaviors relevant to traffic safety culture. For example, not wearing a seat belt is a behavioral hazard. However, a family (or workplace) establishing a rule about always wearing a seat belt is an important behavior or action that must be included as well. This is critical to the strategic approach proposed in this project to transform traffic safety culture (see Chapter 2): culture not only influences behavioral hazards but also stakeholder actions. In this context, we can now define traffic safety culture as the values and beliefs shared among groups of road users and stakeholders that influence their decisions to behave or act in ways that affect traffic safety. This definition includes the contribution of road user behaviors to fatal and serious injury traffic crashes as well as the contributions from all levels of the social environment, such as organizations whose actions can impact traffic safety. Indeed, the role of traffic safety culture in determining stakeholder actions is a key concept for the strategic approach proposed in this report (see 1.12 Developing a Strategic Approach to Transform Traffic Safety Culture). However, with this definition, it is the shared values and beliefs of road users and stakeholders that define traffic safety culture, not the resulting behaviors and actions themselves. The recognition that multiple groups and stakeholders across the social environment influence road user behavior is the foundation of the strategic approach discussed in this report to transform traffic safety culture. Traditionally, behavioral safety efforts have sought to answer a question like “How do we get people to wear seat belts?” Viewed through the lens of the social environment, the question is transformed to become “What actions do we want to grow across the entire social environment that will

result repor influe safety comp seek 1.5 M M is to i behav behav this p Reaso (Gerr mode Glanz envir Figur their in more pe t to transfo nces across culture w onents of tr to change. odeling T any values mprove traff ioral hazar ioral model urpose: Va n Action A ard et al., 2 ls into a sing et al., 20 onment and e 6. Integr predicted i ople wearin rm traffic the social en ithin this st affic safety raffic Safe and beliefs a ic safety, th ds (and st s that have a lue-Belief-N pproach (Fi 008). Comb le integrate 08). Figure traffic safet ated model nfluence on g seat belts safety cultu vironment t rategic app culture and ty Culture re shared am en it is critic akeholder lready been orm Theory shbein & Aj ining the su d model resu 6 represen y culture to used in this deliberate 12 ?” Thus, the re involves oward the s roach, it is their predic ong road u al to determi actions). Fo established (Stern, 20 zen, 2010), bstantive f lts in a mor ts an integ engaging in project to behaviors. strategic a a process ame traffic s necessary tive relation sers and stak ne which va rtunately, from prior r 00; Oreg & and the Prot eatures of th e comprehen rated mode behavioral h specify com pproach pro that aligns afety goal. to have a m ship to the eholders. If lues and bel a number esearch can Katz-Gerr otype Willin ese differe sive unders l that relat azards (see ponents of posed in th the cultura To use traffi odel of th behaviors w the intentio iefs influenc of validate be applied t o, 2006), th gness Mode nt behaviora tanding (e.g es the socia Figure 2). culture an is l c e e n e d o e l l ., l d

13 This model acknowledges the influence from the social environment with which the individual identifies (social identity). By representing the thoughts shared among a group of people with the same social identity (Figure 4), these models can also predict the effect of culture on collective group behaviors. When used to represent group culture, the predictive function of these models can provide guidance about which values and beliefs are most relevant to achieving specific traffic safety goals. In this model, behavioral willingness and intention predict engagement in behavior. Willingness and intention are directly influenced by:  The attitude (positive or negative) toward the behavior as evident by emotional reaction to the behavior and perceived utility of the behavior;  The perception of what is typical (Perceived Descriptive Norm) or expected (Perceived Injunctive Norm) as well as the desirability of the image associated with the type of person typically associated with that behavior (Prototype Image); and,  The perception of an individual's capacity and control to perform the behavior (Perceived Control). In turn, each of these emerges from beliefs about the behavior:  Attitudes are determined by behavioral beliefs about the consequences of a behavior (e.g., wearing a seat belt reduces injury in a crash) and how that consequence is evaluated (e.g., not being injured is important).  Perceived descriptive norms are determined by our beliefs about what others typically do or, in other words, what is perceived as “normal.” Perceived injunctive norms are determined by our beliefs about what others typically expect or approve. Beliefs about the attributes of people who typically commit a particular behavior determine the image of the prototypical person representing that behavior (Prototype Image).  Perceived control is determined by control beliefs about various factors or conditions that may impact an individual’s sense of being in control of the behavior. The formation of a belief system depends on cultural values. Values are ideals to which people aspire (Joffe, 2003). Values provide the criteria by which the desirability of different choices and possible outcomes are evaluated (Lee et al., 2007). By defining what

14 is considered important, values provide the impetus to develop beliefs about the physical and social environments relevant to achieving goals consistent with these values (Spates, 1983). In summary, the predictive model shown in Figure 6 provides a basis for understanding which thoughts influence behavioral hazards affecting traffic safety. By determining which thoughts influence behavior, strategies can be developed with the specific intention of changing these thoughts. By transforming the culture of a group, we also change the social identity of individuals in that group. Because our social identity is always present to influence our behavioral decisions, strategies based on the transformation of traffic safety culture can sustain behavior change. This is in contrast to traditional traffic safety methods that may be applicable to only certain times and locations within the roadway system that may produce only temporary results. To apply the traffic safety culture as a paradigm to sustain behavior change, we need measures of the types of shared thoughts that influence the behaviors we seek to change. Next, methods for measuring the components traffic safety culture are discussed. 1.6 Measuring Traffic Safety Culture As we have noted, traffic safety is impacted by many behavioral hazards (Figure 2). For example, not wearing a seat belt, speeding, and driving while impaired are three risky behaviors exhibited by drivers that significantly impact traffic safety (NHTSA, 2008a). Traffic safety is also affected by the actions of various stakeholders across the social environment. For example, safety legislation, enforcement issues, workplace traffic safety policies, and poor communication between parents and young drivers are also behaviors that can significantly impact traffic safety. Therefore, measuring traffic safety culture involves measuring the values and beliefs related to a variety of behaviors among relevant groups in the social environment. Because of the large number of behaviors and multiple groups within the social environment affecting traffic safety, it is not always sufficient to use a single instrument to measure traffic safety culture.

15 In regards to designing measurement instruments, Table 1 provides a summary of the components of traffic safety culture proposed in Figure 6. The measurement of these components is unique for each specific behavior (i.e., behavioral beliefs about speeding are different than behavioral beliefs about texting while driving). Guidance for designing specific questions to measure each of these components is provided in Appendix B. It is important to note that all of the components of traffic safety culture (as defined in this document) are thoughts. While the resultant behaviors of these thoughts can be observed, traffic safety culture is in people’s minds and cannot be directly determined by observation. For example, while an individual can be observed speeding, the mere observation does not reveal why the driver chooses to speed. Therefore, the actual components of traffic safety culture can only be gathered by asking people what they are thinking. Table 1. Summary of Traffic Safety Culture Components Intention Definition The deliberate decision to commit a behavior in an anticipated situation. Role in the Model Intention predicts behavior (Fishbein & Ajzen, 2009). Because some behaviors may be constrained by factors other than personal choice (“I don’t use a cell phone when I drive because I don’t have one”), adding intention to the model increases the predictive understanding of the other factors in the model. If intention is excluded, other factors in the model may not predict behavior because the behavior is constrained by unrelated circumstances. Willingness Definition The predisposition to commit a behavior if an unexpected situation arises. Role in the Model Willingness predicts behavior (Gerrard et al., 2008). Willingness addresses behaviors that may be more spontaneous – such as suddenly deciding to speed on a certain stretch of road. Combined, willingness and intention provide a better overall prediction of behavior than just intention alone. Attitudes Definition Subjective evaluation of an object or behavior in terms of emotional reaction (e.g., “Speeding is exciting”) and perceived utility (e.g., “Seat belts are useless). Role in the Model Attitudes predict intention (Fishbein & Ajzen, 2009) and willingness (Gerrard et al., 2008). Measuring attitudes reveals how people feel about a certain behavior and thus informs interventions. Behavioral Beliefs Definition Expectations about the physical and social consequences of a behavior (e.g., “If I speed, I will likely get an expensive fine,” “If I drink and drive, my friends will exclude me”).

16 Role in the Model Behavioral beliefs inform and predict attitudes (Fishbein & Ajzen, 2009). Since attitudes are an important predictor of willingness and intention, it is important to understand the beliefs that precede attitudes. Also, understanding behavioral beliefs informs interventions to change beliefs and attitudes. Prototype Image Definition The stereotype of people perceived to typically engage in the behavior (e.g., “People who speed are cool.”). Role in the Model Prototype image predicts willingness and intention (Gerrard et al., 2008). People are motivated and inhibited to engage in certain behaviors based on their perceptions of the typical person who engages (and doesn’t engage) in the behavior. Understanding motivating (and inhibiting) prototypical images is important for designing interventions. Perceived Norms Definition The behavior believed to be common and expected in a given context (e.g., wearing a seat belt when driving with parents). Role in the Model Perceived norms predict intention (Fishbein & Ajzen, 2009) and willingness (Gerrard et al., 2008). Often perceived norms may be inaccurate whereby risky behaviors are perceived as normal and protective behaviors as atypical. Normative Beliefs Definition Beliefs about (1) what behaviors are most common in a group (e.g., “All my friends speed”); (2) what important people in that group expect (e.g., “My parents expect me to wear a seat belt”); and (3) what are the shared characteristics of people perceived to typically engage (or abstain) in that behavior. Role in the Model Normative beliefs inform and predict perceived norms (Fishbein & Ajzen, 2009). Normative beliefs are defined relative to a specific referent group (“coworkers at my facility”) or individual (“my spouse”). When combined across many different referent groups, normative beliefs form perceived norms. Measuring normative beliefs reveals which referent groups dominate perceived norms and thus informs intervention design. Perceived Control Definition Perception of our ability to determine our own behaviors (e.g., “I can choose my own speedin traffic”). Role in the Model Perceived control informs and predicts intention (Fishbein & Ajzen, 2009) and willingness (Gerrard et al., 2008). For some behaviors (typically, those that are more complex), individuals may have supportive attitudes and perceived norms but are unable to engage in the behavior because they perceive they are not in control. Control Beliefs Definition Beliefs about my ability to engage or not engage in the behavior based on factors either internal or external to oneself (e.g., “Crashes are determined by fate,” “I am comfortable not speeding even if everyone around me is”). Role in the Model Control beliefs inform and predict perceived control. Control beliefs provide more detailed information about what increases and decreases an overall sense of control (e.g., perceived control). Control beliefs examine the impact of a certain factor or scenario on intention to engage in the behavior. Values Definition Ideals to which we aspire (Joffe, 2003) that define the goals for our behavioral choices and direct the formation of our belief systems (e.g., “I must protect my family,” “I desire a life without stress”).

17 Role in the Model Values influence the formation of our belief systems (e.g., behavioral, normative, and control beliefs) (Spates, 1983). Values such as “personal freedom” versus “collective safety” may result in very different attitudes about traffic safety issues. A variety of methods can be used to measure these components including: (a) representative surveys, (b) intercept surveys, (c) purchased internet survey panels, (d) focus groups, (e) key informant interviews, and (f) existing survey datasets. a. Surveys that have respondents who are randomly selected to measure the components of traffic safety culture (as well as appropriate behaviors) among a portion of a population. The portion, or sample, of the population of interest is selected so there is high probability that the responses of the sample represent the entire population of interest. To make this assumption, individuals should be selected randomly from the entire population such that everyone in the population has an equal likelihood of participating in the survey. Appendix C contains a table of the number of samples required for different confidence levels. b. Surveys with respondents who are selected by “convenience” can use the same surveys as those who are randomly selected. However, these convenience samples are not likely to be representative of the broader population like they would be if they were randomly selected. For example, data obtained from customers at a liquor store may not be representative of people who do not drink alcohol. c. Individuals can also be recruited to participate in online surveys using the internet. Various companies sell “panels” or access to individuals to complete an internet-based survey. These individuals are typically paid to complete surveys and may be recruited to participate in a variety of ways. d. Focus groups are typically composed of a small number of individuals who engage in a facilitated dialogue with a trained focus group facilitator to develop a deeper understanding of an issue. Focus groups allow for follow-up questions and much broader and deeper exploration. e. Key informant interviews are typically conducted one-on-one with a trained

18 interviewer and an individual who is considered a significant stakeholder. The interview typically has standard questions but may also allow for follow-up questions to develop a better understanding. f. In addition to gathering new data, there are many existing surveys already being collected by various agencies (see Appendix C for a list of existing surveys and questions asked). These include survey results collected among both youth (e.g., middle and high school students) as well as adults. Depending upon the purpose of the survey and the methods used, results may be available at the national, state, county, or school levels. Each method has advantages and disadvantages based on the following criteria (Table 2). While there are many characteristics of data gathering methods, these four (e.g., representative, accurate, sensitive, and predictive) were selected because they are highly relevant to measuring traffic safety culture (as defined above), they are important for gathering information that supports strategy development, and they are important if the data gathered are going to be used for evaluation purposes. Other characteristics such as time to prepare the method feasibility using the method with current staff skills are important to consider as well. 1. Representative – the degree to which the data collected represents the entire population of interest. 2. Accurate – the degree to which the data collected represents the actual beliefs and behaviors of those answering the questions. 3. Sensitive – the degree to which small changes in culture between respondents (and over time) can be measured. 4. Predictive – the degree to which the data collected are based on a theoretical model of behavior and can be used to reveal empirical relationships between traffic safety culture and behaviors. 5. Cost – the amount of resources (e.g., time, money etc.) to plan and execute the method.

19 Table 2. Advantages and Disadvantages of Various Methods of Measuring Traffic Safety Culture Method Representative Accurate Sensitive Predictive (a) Representative survey high high high high high (b) Intercept survey varied varied high high moderate (c) Purchased internet survey panel unknown unknown high high varied2 (d) Focus group moderate moderate varied low low (e) Key informant interviews low moderate varied low low (f) Existing data sources varied varied varied low varied 3 a. Surveys using random samples can provide a high likelihood that the responses collected represent the overall population; can provide a high degree of accuracy (assuming the survey questions were designed carefully to measure what they are supposed to be measuring and that the methods used to perform the survey fostered a high sense of anonymity); provide a high level of sensitivity (assuming that the questions and answer choices were designed appropriately); and readily support empirical analysis to develop predictive models (assuming the questions were designed based on a theoretical model). b. Intercept surveys (which use a convenience sample) can be completed quickly and often at less cost, and can be especially helpful during early stages of the research when generating hypotheses. However, the convenience sample does not represent the entire population like a random sample would, so results are not generalizable. Furthermore, the level of accuracy may depend on how the survey is implemented. 2 Cost will depend on the type of sample requested and the cost-model used by the survey service. 3 The cost will depend on the possible charges involved in purchasing the existing data base. Cost

20 If individuals are selected and asked to complete the survey, the sense of anonymity may decrease and aspects of social desirability (the desire of the respondent to “please” the individual asking for the information) may increase. With reduced anonymity and heightened social desirability, respondents may be less likely to report accurate information (especially about illegal or undesirable behaviors). c. A purchased internet survey panel may use the same survey instrument as in the two previous cases. However, it is often unclear how the various companies recruit their participants, and these individuals may be “professional survey takers” – that is, they routinely complete surveys online. This may impact the accuracy of their responses. Clearly, individuals without internet access are not represented. d. A focus group provides an opportunity for a much more in-depth conversation with a small number of people that reveals more qualitative (as opposed to quantitative) information. In addition, new ideas that emerge can be explored providing a richer and broader understanding. Depending upon how the individuals are recruited to participate, they may or may not be representative of the population. However, because focus groups are typically small in size, it may be hard to generalize the results. Issues of group dynamics, lack of anonymity, and social desirability can reduce accuracy. As the data gathered are typically qualitative in nature, empirical predictive techniques are not able to be used. e. Key informant interviews provide an opportunity for a much more in-depth conversation with a critical stakeholder. The information gathered is often very valuable to understand that individual’s position. Understanding how critical stakeholders understand issues provides invaluable information to guide implementation in the future. However, the results may not be representative of other leaders or individuals in the population, and the informant’s perceptions of the beliefs of others may not be accurate. Lack of anonymity and social desirability can reduce accuracy. As the data gathered are typically qualitative in nature, empirical predictive techniques are not able to be used. f. Existing surveys provide certain data without requiring any additional time or resources to collect. Additionally, many existing surveys are conducted on a

21 regular basis, thus providing information over time that can be valuable in assessing impact and understanding trends. However, most of these surveys measure only self- reported behavior (Appendix D), which is not part of the proposed definition of traffic safety culture. Few surveys include measures of shared thoughts that could be defined as traffic safety culture. Indeed, no surveys include a complete set of measures based on a comprehensive model that predicts the relationship between traffic safety culture and road user behavior (e.g., Figure 6). Therefore, these surveys have limited application to measuring traffic safety culture for the purpose of developing transformation strategies. Furthermore, because these surveys are designed by independent entities, it may be difficult to customize them for local purposes. Decisions regarding which methods to use should be based on weighing these various advantages and disadvantages. For example, while representative surveys allow for the development of a behavioral model to inform interventions, key informant interviews are much less expensive and provide some degree of enhanced understanding of how certain individuals perceive traffic safety behavior. Thus, a combination of methods overcomes the disadvantages of any single approach and provides a better understanding. For example, including both quantitative methods from surveys and qualitative methods from interviews can provide both breadth and depth. Bringing multiple sources of information also bolsters the confidence in the data. Results that align across methods support higher validity, and results that contradict can inform important questions. References for best practices for each of the methods of measurement discussed above are provided in Appendix E. This includes an overview of a recommended process for developing customized surveys. Given that specialized skills to develop surveys and to interpret their results may not be present in all transportation agencies, these surveys may need to be developed by external experts. To assist transportation agencies in contracting with external experts, guidance for developing a request for proposal (RFP) to include customized surveys is included in Appendix F.

22 1.6.1 Measuring Behavior As discussed previously, our definition of traffic safety culture is based on shared thoughts that predict behavior – but behavior is not itself a component of traffic safety culture. Instead, we collect self-reported or actual measures of behavior as the outcome we are trying to change by transforming culture. Notably, we do not encourage the use of behaviors as a measure of culture itself for several reasons. First, while behaviors may result from aspects of culture (thoughts), we cannot directly determine which thoughts were involved by observing behaviors alone. Second, caution should be exercised when analyzing crash data and arrest data. Crash and arrest statistics are very important and enhance overall understanding; however, depending on the behavior, it may be difficult to determine the prevalence of a risky behavior only based on crash or arrest data. Arrest data are often indicators of enforcement practices and priorities. Many traffic related risky behaviors do not result in arrest, and thus, arrest data are often a poor indicator of risky behavior. Furthermore, neither crash data nor arrest data provide any information about the beliefs of those engaging in risky behaviors and therefore provide limited guidance on strategy selection to change beliefs. 1.7 Analyzing Traffic Safety Culture Data There are four common types of analyses that can be applied to traffic safety culture data in support of the development of traffic safety culture-based (TSCB) strategies: 1. Descriptive – provides a basic understanding of the current level and variability for different aspects of culture based on a review of simple analyses such as averages and frequency reports. 2. Prediction – provides a deeper understanding of which cultural components directly influence deliberate behaviors using statistical methods to test the predictions of the underlying model relating culture to behavior (see Figure 6). In this model, attitudes, norms, prototype image, and perceived control are all predicted to directly influence willingness and intention to behave (or act) which then predicts behavior. Those aspects of culture identified by this analysis to be the strongest predictor(s) then inform the selection of strategies to transform culture and change behavior.

23 3. Assessment – provides an understanding of which values and beliefs (behavioral beliefs, normative beliefs, control beliefs) included in the underlying model (see Figure 6) give rise to those aspects of culture that the previous analysis showed predicted behavior. Those values and beliefs that are shown to be most relevant then determine the content of the strategy. The purpose of the selected strategy is to leverage the values or change the beliefs that this analyses determined were most influential. 4. Evaluation – provides an understanding about the effectiveness of a strategy. In the following sections, each form of analysis is discussed in terms of its primary function and data requirements. In addition, an example of applying each analysis phase is provided based on hypothetical data. A simplified version of the behavioral model introduced in Figure 6 is included to help the reader understand component(s) of the model that are being analyzed in the example. 1.7.1 Descriptive The descriptive phase of analysis has two primary functions. First, it provides a simple characterization of the current level of culture. This can be used as a baseline with which to compare future culture after applying a traffic safety culture-based (TSCB) strategy (see 1.7.4 Evaluation). Recalling that culture is defined in terms of thoughts “shared” within a group, the second function of this analysis phase is to explore the extent to which particular thoughts are shared within the group. This can be used to identify which measured components represent the core definition of the group culture. However, because this phase of analysis does not seek to examine the relationship between measures of traffic safety culture and (reported) road user behaviors, it is not possible to logically assert which strategies may be effective in changing road user behavior. Next, we list the most important data requirements (see Table 2) for this form of analysis. Data Requirements:  representative of the population of interest  accurate  include measures of traffic safety culture

Exam boxed respo show peopl +1, + positi G the av about samp In sugge and th A yet an respo Exce speed Very Exce speed Very ple: item show ndents wou n in this exa e. Respons 2, or +3 ac ve (safer) at iven the dis erage attitu speeding w le, so this at contrast, th st a very po erefore, a s s a final hyp other group nse, but also eding the sp limits of 5 desirable | 0 eding the sp limits of 5 desirable | 1 n below to ld check the mple are th es would the ross the res titude towar tribution of de score for as neutral. titude would e next hypo sitive (safe) trong indica othetical ex . This exam describing eed limit b 5 mph or m | 0 | 0 | 0 | eed limit b 5 mph or m 0 | 10 | 10 | measure at box that b e number o n be scored ponse optio d speeding. responses to this sample However, not be a str thetical exam attitude tow tion of that ample, the ple shows th the range of y more tha ore would 0 | 0 | 70 | V y more than ore would 10 | 10 | 10 24 titude towa est represen f people giv for analysis ns. In this this item in would be 0 there appea ong indicato ple of samp ard speeding group cultur box below s e importan responses. n 10 mph w be: ery undesir 10 mph w be: | 10 | Very The si current l measured the avera For exam state resi rd speeding ted their ow ing each re by assignin case, a high the followi , indicating rs to be lit r of this pa le data from (average = e. hows the sa ce of not on Here, we ha hen driving able hen driving undesirable mplest way evel of cu component ge of the re ple, imagine dents that among roa n attitude. sponse in a g values of er score im ng hypothet that the bas tle consensu rticular grou a different +3) that is w mple respon ly describin ve the same on roads w on roads w to express th lture for is to expres sponse dat a survey o included th d users. Th The number sample of 7 -3, -2, -1, 0, plies a mor ical example eline attitud s within th p culture. group woul idely shared se data from g the averag overall ith posted ith posted e a s a. f e e s 0 e , e e d , e

25 average as the first example (neutral attitude), but clearly two distinct patterns of responding that may suggest two distinct sub-cultures within the same group. It is important to recognize that it is not possible to identify a strategy based on this descriptive phase alone because it does not analyze the (predictive) relationship among the components of traffic safety culture and behavior (see Figure 6). For example, knowing only that a group shares a risky attitude toward speeding does not necessarily mean that component predicts speeding. And so it would be an assumption rather than an empirical conclusion that an attitude-based strategy would reduce speeding in that group. Similarly for reasons we have already discussed (see 1.6.1 Measuring Behavior), only having data that speeding was prevalent in a particular group (e.g., self-report or observational data) would not necessarily imply that this group has a risky culture regarding speeding. 1.7.2 Prediction The primary function of the prediction phase of analysis is to empirically confirm the pathways adopted model that predicts the relationship between traffic safety culture and behavioral hazards. Specifically, the prediction phase of analysis uses advanced statistical methods (e.g., multiple regression, decision tree analyses, latent factor modeling) to specify which are the strongest pathways evident from the collected data that represents the traffic safety culture of the measured group. That is, which cultural pathways in the model are the strongest determinants of behavior? Data Requirements:  representative  accurate  predictive, that is, measures of traffic safety culture based on a theoretical behavioral model  data responses from individual respondents (i.e., data cannot be aggregated)  both traffic safety culture measures as well as behavioral data (in the same data record for each respondent) Exceeding the speed limit by more than 10 mph when driving on roads with posted speed limits of 55 mph or more would be: Very desirable | 35 | 0 | 0 | 0 | 0 | 0 | 35 | Very undesirable

26 Predictive modeling is limited by the data available. For example, data sets with only data about behaviors, but not values or beliefs (or vice-versa) cannot be analyzed to determine the predicted relationship between the behaviors and the underlying values and beliefs of the culture. Similarly, results from focus groups and key informant interviews do not provide quantitative data suitable for statistical predictive analysis. However, such methods may provide valuable information, a deeper understanding, and be used to inform the development of surveys. Example: Figure 7 presents the hypothetical results of a multiple regression analysis of a state representative survey to identify the relative predictive strength of those components of culture that the model predicts are direct influences on road user intention and speeding behavior. The values (ranging from 0 to 1) represent the relative predictive value of each component (with * indicating which component is statistically significant). The larger the value, the more strongly that component predicts behavioral intention. Perceived Norms .50 Attitudes .80* Perceived Control .20 Behavioral Intention

27 Figure 7. Example of predictive analysis to identify components of traffic safety culture that predict intention to speed. In this example, attitudes are the strongest predictor of speeding intention. This suggests the need to change these attitudes so that the intention to speed (and actually speeding behavior) would be reduced. However, to change attitudes, we need to assess which behavioral beliefs determine attitudes (see Figure 6). This understanding comes from the next phase of analysis (assessment). It is important to recognize that the conclusions of a predictive analysis can only apply to the particular group from which the data was collected. For another group with a different traffic safety culture, attitudes may have no predictive effect – and instead – perceived control may be the strongest predictor. This illustrates the need to measure and analyze traffic safety culture for each culturally unique group rather than generalize conclusions across disparate groups within the social environment. Indeed, this discussion also exemplifies why there rarely exists any "off the shelf" strategies that can be applied generically across different social groups. Instead, because each culture may have specific cultural components uniquely predictive of behaviors within that group, each group may require a different strategy to transform its culture to change behavior. 1.7.3 Assessment The primary function of the assessment phase of analysis is to examine which values and beliefs are most relevant to those components of culture (e.g., attitudes) the model and Perceived Norms .50 Attitudes .80* Perceived Control .20 Behavioral Intention

predi most the st D      E the in speed woul most of a value U drivin attitu attitu ctive analys relevant bel rategies sele ata Requir represent accurate predictiv behavior data resp both traff record fo xample: itial predic ing were de d conduct a relevant to 7-point scal represents a sing this sa g too fast m de that spee des about sp is indicates iefs evident cted in the p ements: ative e, that is, me al model onses from i ic safety cu r each respo tive analysi termined to separate pr the formatio e of level o stronger be me hypothe akes passe ding is und eeding wou directly inf from this ass redictive an asures of tr ndividual re lture measur ndent) s (see Figu be the stro edictive ana n of this att f agreement lief. tical data se ngers uncom esirable. Th ld seek to in 28 luence willi essment the alysis phase affic safety spondents ( es as well a re 7). In th ngest predi lysis to det itude. Here, with each t, Figure 8 fortable is erefore, a crease the b ngness and n become th . culture base i.e., data can s behavioral Gener of assessm used in th First, analysis c which componen e previous ctor of spee ermine whi belief stren belief state shows that t the stronge TSCB strat elief that sp intention to e focus for c d on a theor not be aggr data (in the ally, there a ent method is phase. a separat an be used beliefs pr ts of cultur example, at ding intenti ch behavior gth is meas ment. A hig he behavior st predictor egy to chan eeding is un behave. Th hange withi etical egated) same data re two type s that can b e predictiv to determin edict thos e revealed in titudes abou on. Thus, w al beliefs ar ured in term her respons al belief tha of the (safe ge road use comfortable e n s e e e e t e e s e t ) r

29 for their passengers. This example demonstrates the logic that it is not possible to change an attitude without addressing the underlying belief. Figure 8. Example of assessment analysis to determine which behavioral beliefs predict attitudes about speeding. Second, response to an item measuring behavior (e.g., self-reported frequency of behavior in the past 30 days) can be used to define two categories of road users: (A) Those who do engage the behavior and (B) Those who do not engage the behavior. Statistical methods can then be used to compare these groups in terms of their average level of agreement for each measure of belief (e.g., T-tests, ANOVA, etc.). Applying this assessment method, Figure 9 shows that (A) speeders agree significantly less than (B) non-speeders that speeding is uncomfortable for their passengers. This assessment again supports a TSCB strategy to increase the belief that speeding is uncomfortable for passengers, especially for road users who speed. “Speeding increases my chances of having a crash.” .40 “If I speed, it is likely I will get a ticket.” .50* “Driving too fast can make my passengers feel uncomfortable.” .65* Attitude (speeding is undesirable)

30 Figure 9. Example of assessment analysis to compare belief strength between (A) speeders and (B) non-speeders. 1.7.4 Evaluation The primary function of the evaluation phase of analysis is to quantify the effectiveness of strategies. As shown in Figure 10, an evaluation should address all the various stages underlying the strategy design and implementation: 1. Was the target group exposed to the strategy? A strategy will not be effective unless the target group has sufficient exposure to the active component of the strategy. 2. Was the component of culture that the strategy intended to change affected? In order for any strategy to claim responsibility for any observed change in behavior, it must demonstrate that the component of culture predicted to affect that behavior has itself changed. 3. Was there evidence of change in the target behavior? The purpose of any strategy is to change behavior. Therefore, this key objective must be evident by recording changes in behavior. 4. Was there a reliable reduction in crash risk? Ultimately, strategies are developed Belief: Speeding Belief: Speeding Belief: Speeding increases likelihood of increases risk of increases passenger ticket.* crash. discomfort.* 2 1 0 2 3 3 4 4 4 5 5 6 6 7 (A): Speeders (B) Non-Speeders Le ve l o f A gr ee m en t ( lo w - hi gh )

31 and implemented to support traffic safety goals. And so, the changes in road user behavior should culminate in an observed reduction in fatal and serious injury crashes. Figure 10. Illustration of elements of strategy design and implementation that should be included in evaluation phase of analysis. In addition to the final evaluation of the implemented strategy, forms of evaluation can be done throughout the design and implementation process. For example, the initial concept for the strategy can be reviewed by independent experts or cross-referenced with similar or related strategies documented in the literature to confirm the feasibility of the concept. As the strategy is designed, key components should be pilot tested to obtain initial evaluation data to validate and revise the design. And the implementation plan can also be independently reviewed to assure that it is viable. Data Requirements:  representative  accurate  sensitive, that is, able to reveal small changes  predictive, that is, measures of traffic safety culture based on a theoretical behavioral model  data responses from individual respondents (i.e., data cannot be aggregated)  both traffic safety culture measures as well as behavioral data (in the same data record for each respondent) Exposure to Strategy Cultural Transformation Change in behavior Reduction in crash risk

32 Example: Several authors have noted that road safety campaigns (one type of strategy used to change traffic safety culture) are rarely subjected to a formal and complete evaluation (Robertson & Pashley, 2015; Hoekstra & Wegman, 2011). This lack of accessible evaluation data severely restricts the advancement and adoption of effective campaigns because there is (1) no guidance on how to improve campaigns, (2) no evidence to discontinue ineffective campaigns, and (3) no impetuous to advance safety campaign techniques. Thus, rather than provide a hypothetical example of a complete evaluation, we instead wish to reference that the European CAST project has developed standard tools for evaluating roadway safety campaigns (Vaa et al., 2009) and reporting their effectiveness (Boulanger, 2009). Both these tools are supported by a comprehensive guidance manual for designing, implementing, and evaluating roadway safety campaigns (Delhomme et al., 2009). Transportation agencies are advised to review this tool and manual as part of the design, implementation, and evaluation steps of the strategic approach. 1.8 Traffic Safety Culture-Based Strategies Arguably, the basic purpose of a TSCB strategy is to create a culture within society whereby traffic safety is “highly valued and rigorously pursued” (AAAFTS, 2012). However, it is important to note that changing culture alone is not the defining criterion for a TSCB strategy. Society and its beliefs are naturally dynamic and adaptive to any significant changes in the environment. A change in beliefs over time is necessary but not sufficient to define a TSCB strategy. For example, many aspects of European beliefs and culture changed because of the dramatic effect the Black Plague had on the social environment (Whipps, 2008), but no one would argue that this plague was a culture-based strategy. Instead, a traffic safety culture-based (TSCB) strategy is intentionally designed to leverage values and change beliefs within a specific social group to change behaviors (and actions) that affect traffic safety (Figure 3). That is, the transformation of culture was the immediate goal of the strategy, with transformation itself being the specified mechanism of changing behavior rather than a long-term residual effect of the strategy.

33 The adoption of an appropriate theoretical model relating culture to behavior is not only an indication of such intention; it is also an important determinant of strategy effectiveness. For example, Robertson and Pashley (2015) recently reviewed the effectiveness of a variety of safety campaigns and their underlying theoretical models. According to this review: “The decision to adopt a theoretical approach to guide campaign development is important and can increase the likelihood of effectiveness compared to merely developing a campaign in an ad hoc or intuitive fashion that has no clear link to the actual behavior of concern. Moreover, a theoretical approach will serve to guide decision making related to each step of the campaign and help to ensure a coherent strategy is developed.” (Robertson & Pashley, 2015, p. iii). In practice, evidence of an adopted model (intention) may be documented in the formative research supporting the strategy development, a logic model describing how the strategy is supposed to result in change, or in the evaluation of the strategy effectiveness. As previously discussed, we propose the behavioral model shown in Figure 6 to be a suitable model for the intentional design of TSCB strategies. A logic model is a visual description of the “logic” behind how a strategy or initiative is supposed to lead to behavior change. For example, suppose surveys of youth indicate they do not understand the risks associated with cell phone use while driving. A curriculum is developed to teach students about divided attention and how even short moments of inattention can lead to crashes. A logic model for this strategy can then be described as (Figure 11): • If a policy is established that all students in driver education are required to take the new distracted driving segment, then all students should be exposed to the curriculum. • If all students are exposed to the curriculum and the curriculum has been designed and tested to change beliefs about distracted driving, then student’s beliefs (e.g., perception of harm) about distracted driving should change. • If perception of harm from distracted driving increases among students, then fewer students should engage in distracted driving. While this example is simple, it highlights how a logic model can clarify the underlying intention of a strategy and how that strategy leads to behavior change.

34 Figure 11. Example of logic model for student driver distraction curriculum to change beliefs about distracted driving. In these regards, TSCB strategies differ from other traditional strategies that may only address extrinsic factors in the physical environment. For example, research has shown that the effects of speeding enforcement often dissipate once the enforcement ends (Vaa, 1997; Christie et al., 2003; ERSO, 2007). This effect suggests that the fundamental beliefs influencing the decision to speed were not changed by the enforcement activity. In contrast, TSCB strategies are expected to produce sustained effects because they influence the cultural values and beliefs that form our social identity – and so – become ingrained in our decision making rather than a reaction to external factors specific to a location or time within the physical environment. 1.9 Examples of Traffic Safety Culture-Based Strategies Here, we provide examples of TSCB strategies based on the adopted behavioral model (Figure 6). Each example represents a specific pathway in the model relating the components of culture to behavior. A simplified version of the behavioral model introduced in Figure 6 is included to illustrate the active pathway for each example. Where possible, we also provide evidence of the effectiveness of these strategies (see 1.7.4 Evaluation). 1.9.1 Value-Based Strategy Example It is generally accepted that values cannot be changed other than socialization at an early age (Fishbein & Ajzen, 2009). TSCB strategies can leverage existing (positive) cultural values in three Change in distracted driving behaviors. Change in beliefs about harm of distracted driving. Exposure to curriculum content about attention and driving.

ways behav exam seatb 12 an Safer Figur Partn S that s camp behav mobi T can r strate For e based a con . First, thes ior by dem ple, the “E elt wearing d the ad cam Roads Partn e 12: Awar ership, 201 econd, cultu upport safe aigns that e iors that co lity (Grenob hird, becaus emind us of gy can be u xample, the on where th text for futu e strategies onstrating h mbrace Lif to the univer paign video ership, 201 d winning 5) ral socializa r road user spouse the nsider the s le, 2014): ( e our values our values sed to prepa importance ey live as sh re message can increa ow that be e” campaig sally accept : https://ww 5; Making t Embrace L tion proces behavior ca value of alt afety of oth https://www determine w as a way to re people f of children own in Figu s that woul 35 se the will havior satis n from Su ed value of w.youtube. he Link, 201 ife campaig ses within f n support T ruism could ers rather t .youtube.co hat is relev make our b or subseque can be emp re 13, whic d promote p ingness and fies an exi ssex Englan protecting f com/watch? 0). n (Sussex S amilies and SCB strateg be expect han one’s o m/watch?v= ant for us to eliefs more nt strategies hasized as h can be an rotective b intention sting cultur d deliberat amily as sho v=h-8PBx7 afer Roads schools to ies. For exa ed to promo wn immedi uaWA2Gbc believe, TS salient. In s to change a way to co effective wa ehaviors am to commit al value. Fo ely connec wn in Figur isoM (Susse instill value mple, medi te road use ate needs fo nJU). CB strategie o doing, th those belief nnect peopl y to establis ong parents a r ts e x s a r r s is s. e h :

https =nTz Figur Mont unde 1.9.2 “I be my c some belief align a med as by behav revea comm ://www.yout Bm8BZya8 e 13. Exam ana culture rage drinki Normativ lieve most o oworkers ex cases, the a that most p the belief sy ia campaign standers to p ioral hazar led signific unity mem ube.com/wa . ple of “Mo in prepara ng. e Belief-ba f my friend pect me to n ctual norm m eople are e stem to refl was design revent impa d of drinkin ant misperc bers over-es tch?list=PL ntana Care tion for a f sed Strate s speed” is a ever text w ay already ngaging in ect the actu ed to addres ired driving g and driv eptions abo timated the 36 efHXZ_EU s about Kid uture camp gy Examp normative hile driving represent a unsafe beha al safe norm s normative (an exampl ing). This s ut the actu prevalence 1nMyXev3s s” to emph aign on pr le (Percei TSCB perceived normative (important environme particular b descriptive ” is a norm safe behavio vior. In suc . In the exam beliefs amo e of a stakeh trategy was al norms i of driving a vi8_Ynu0gb asize defini otecting chi ved Norms strategies ca norms by beliefs a ) people i nt commit ehaviors. F belief. “I be ative injunc r, but there h cases, it i ple shown ng adults ab older action based on a n the com fter drinkin EBQ6m&v ng value of ldren from ) n change ou changing th bout whic n the socia and expec or example, lieve most o tive belief. I is a mistake s desirable t in Figure 15 out engagin to reduce th survey tha munity (e.g g and unde r e h l t f n n o , g e t ., r-

estim comp perce contr Figur inter 1.9.3 type Figur strate of se viewe seatb their comm ated those w leted after ived norma ol communi e 15. Exam vention. Normativ of person p e 16, the M gy intended atbelts with d positivel elt is also ex own social unity that t ho would tr the media tive beliefs ties) (Otto & ple of strat e Belief-ba erceived to ontana Depa to increase local Amer y as local pected to be status by e hey highly r y and preve interventio among inte McMahill, egy to chan sed Strate typically co rtment of T seat belt use ican Indian heroes in th viewed mor ngaging in espect. 37 nt someone n showed rvention co 2015). ge perceive gy Examp mmit the b ransportatio among Am rodeo cham eir native e positively the same b from drinki statistically mmunities d norms ab le (Protot TSCB changing increase th the behav normative characteri ehavior. Fo n (MDT) h erican India pions. Beca communitie . Communit ehaviors as ng and driv significant (and no ch out bystan ype Image) strategies prototype im e cultural d ior by lev beliefs ab stics associ r example, as created a ns by assoc use these ch s, the act o y members those mem ing). Survey changes i anges amon der based o age seek t esirability o eraging th out positiv ated with th as shown i traffic safet iating the us ampions ar f wearing can then rais bers of the s n g n o f e e e n y e e a e ir

Figur incre 1.9.4 For e encou Jack, contr exper that i condi e 16. “Safe ase seat bel Control B xample, a sa rage riders 2015). As s ol riders h imental eva t reduced in tion (Habya on all road t use in Mo elief-base fety campai to complain hown in Fig ad to spea luation (i.e., surance clai rimana & Ja s” campaig ntana on tr d Strategy gn in Kenya to drivers w ure 17, thes k up and randomized ms in the bu ck, 2015). 38 n using a p ibal lands ( Example used sticke ho were be e messages encourage control wit s fleet by 2 ositive prot MDT, 2013 (Perceived TSC perceive the und contextu affecting autonom rs posted w having dang focused on safe driver h placebo) o 5% to 33% otype imag ). Control) B strategie d control se erlying b al and per the perceiv y to commi ithin commu erously (Ha increasing behaviors. f this campa compared t e to s to chang ek to chang eliefs abou sonal factor ed ability an t behaviors. nity buses t byarimana & the perceive A rigorou ign indicate o the contro e e t s d o d s d l

Figur to int 1.10 T camp Road attitu (Delh transf Howe transf F of fe negat e 17. Exam ervene with TSCB Stra he precedin aigns. Medi way safety des and beha omme et al orm aspect ver, not all orming traf or example, ar-based me ive emotion ple of safet unsafe dri tegies and g example a-based cam campaigns h viors relate ., 2009). In s of traffic roadway saf fic safety cu some media ssages. Suc s (pain) rath y campaign ving on pub Roadway s of TSCB paigns can ave been d d to traffic s this regard safety cul ety campaig lture. campaigns h messaging er than cha 39 to increase lic buses. Safety Ca strategies be a cost-ef efined as in afety using o , roadway s ture and ch ns are desig seek to moti is predica nging the cu perceived mpaigns are media fective way tentional eff rganized m afety campa ange speci ned to chan vate behavi ted on basic ltural value control for -based, roa to reach lar orts to chan ethods of co igns can be fic road us ge road use or change th human na s and belief passengers dway safet ge audience ge road use mmunicatio designed t er behavio r behavior b rough the us ture to avoi s underlyin y s. r n o r. y e d g

40 the behavior. And so, this messaging strategy would not qualify as an example of a TSCB strategy, which themselves are based on transforming the social environment that is the basis of our social identity. In any case, evidence of the effectiveness of fear-based messages is “mixed and inconsistent” (Lewis et al., 2007). Recent reviews of the design and evaluation of roadway safety campaigns and mass media messaging for health behaviors suggest that the following factors are important for campaign effectiveness (Robertson & Pashley, 2015; Hoekstra & Wegman, 2011; Wakefield et al., 2010; Vaa et al., 2009; Williams & Haworth, 2007). These same factors should also support the success of TSCB strategies that use communication methods:  The campaign is based on a solid theoretical foundation (model) of the factors that determine the target behavior.  The campaign is based on an analysis of local data regarding the type of crash, relating behavior, and at risk group of road users.  The campaign is based on an understanding of the reasons and motivations for engaging in the behavior.  The campaign uses messaging and imagery that resonates with the “personal experiences” of the road users exhibiting the target behavior.  The campaign includes a message of efficacy by demonstrating behavioral alternatives to the target behavior.  The campaign utilizes a range of communication channels and tools aligned with the common types of communication access used by the targeted road user audience, including local and personalized messaging.  The campaign has sufficient resources to ensure exposure of the target audience to the message, including intermittent refreshing of the message content to maintain awareness of the issue conveyed by the message.  The campaign delivers the messages near the time and location of the opportunity to either engage or resist the target behavior.  The campaign combines both rationale and emotional content into the message.  The messages are based on extensive research and pilot testing using the target audience. This ensures that the campaigns are iteratively modified for maximum relevance and effectiveness within the targeted audiences.

  A Europ concl chang Tabl (Vaa In noted (Rob evalu becau disco techn devel repor comp safety meth If the cam depicted a audience The camp and enfor recent me ean Campa uded that as ing road us e 3. Summa et al., 2009 the long te that road sa ertson & P ation data s se there is ntinue ineff iques. In rec oped standa ting their ef rehensive g campaigns odologies us paign show s “ordinary to identify w aign is com cement. ta-analysis igns and Aw a result of er behavior ry of Meta- ) rm, evaluat fety campai ashley, 201 everely rest (1) no gui ective camp ognition of rd tools for fectiveness ( uidance ma (Delhomme ed to evalua s drivers c people rathe ith the repr bined with of different areness-rai including th as summariz analysis of ion is critic gns are rare 5; Hoekstra ricts the adv dance on h aigns, and this need to evaluating r Delhomme nual for de et al., 2009 te marketin 41 ommitting t r than villain esented driv other traffic roadway s sing Strateg ese success ed in Table Roadway S al to effecti ly subjected & Wegm ancement a ow to impr (3) no imp evaluate TS oadway safe et al., 2009) signing, imp ). Additiona g campaigns he target be s.” This ma ers. safety meth afety camp ies in Traffi factors, cam 3 (Vaa et al afety Camp ve campaig to a formal an, 2011). nd adoption ove campai etuous to a CB strategi ty campaign . Both these lementing, l methods m . havior, thes kes it easier ods includi aigns condu c Safety (C paigns are ., 2009). aign Effec ns. Several and comple This lack of effectiv gns, (2) no dvance safe es, the CAS s (Boulang tools are su and evalua ay also be e drivers ar for the targe ng educatio cted by th AST) projec successful i tiveness authors hav te evaluatio of accessibl e campaign evidence t ty campaig T project ha er, 2009) an pported by ting roadwa adapted from e t n e t n e n e s o n s d a y

42 1.11 Laws and Enforcement as TSCB Strategies Traffic laws and enforcement can be used as a TSCB strategy. The traditional perspective is that merely passing a law does not directly change beliefs (Geisinger, 2002). Instead, it is the way that passage of a law and its enforcement is communicated within society that can directly influence culture. Specifically, according to the Theory of Expressive Law, road user beliefs can be directly changed if (1) the law is communicated to reflect that most people do not approve of the behavior (i.e., there is a strong injunctive norm or, in other words, most people agree people should not engage in the prohibited behavior); and (2) the enforcement of that law is communicated to be effective and reliable in sanctioning non-compliance (Geisinger, 2002). In addition, communication strategies can change people’s expectations about being caught for non-compliance (i.e., a behavioral belief) (e.g., https://www.youtube.com/watch?v=SuATE6EUegY). Communication can also change the behavioral beliefs about the consequences of enforcement, including the social consequences of being apprehended that can create social motivation to comply with the law (e.g., http://www.nhtsa.gov/drivesober/get-involved/videos/ignition-interlock/). Thus, certain forms of high visibility enforcement campaigns that publicize both the normative aspect of the law and the effectiveness of the enforcement can directly change beliefs and represent a form of TSCB strategy. This example also demonstrates the advantage of combining multiple strategies including laws, enforcement, and safety campaigns to achieve traffic safety goals. For example, Australian road safety agencies often combine high visibility enforcement with messaging to communicate the increased risk of being apprehended (https://www.youtube.com/watch?v=0m5jD3iv_T8) and a safety campaign that seeks to transform behavioral beliefs about the negative consequences of the risky behavior addressed by the enforced law (https://www.youtube.com/watch?v=ud8p5t2sXOo). 1.12 Developing a Strategic Approach to Transform Traffic Safety Culture The examples of TSCB strategies discussed previously can transform the traffic safety culture of a specific group of road users within the social environment. However, the various layers of the social environment also influence one another. That is, the behavioral

43 hazards of any road user group are also impacted by the actions of other stakeholder groups within the same social environment. For example, consider strategic efforts to increase bike helmet use among young bicyclists. Healthcare providers in a community expand their role in improving traffic safety by explaining to parents the importance of bike helmet use. Law enforcement routinely reminds youth to always wear a bike helmet. And schools can encourage bike helmet use as well as providing places to store bike helmets in the classroom. By aligning the traffic safety culture and actions of stakeholders across the social environment with the goals of a TSCB strategy to change the behavior of a specific road user group, not only will that strategy be more effective, its safety outcomes will also be more sustainable. Therefore, a strategic approach to transform traffic safety culture should leverage the values and change the beliefs of all of the relevant traffic safety stakeholders across the social environment (Nation et al., 2003). In this way, traffic safety culture is transformed across the entire social environment, not just within one road user group. As a result of this transformation, traffic safety is valued and pursued at every level of society (AAAFTS, 2012). To illustrate this strategic approach, Figure 18 shows stakeholder groups at the (A) national, (B) state, (C) community, and (D) school levels whose actions can impact a behavioral hazard being aligned with the goals of a TSCB strategy to reduce that behavioral hazard by transforming the traffic safety culture within the families of a specific group of road users (see the ‘★’ in Figure 18).

Figur cultu strat A the U Figur seeks incre      B use w adult ackno major popu e 18. Illust re by integr egies for ro s an examp tah Departm e 19, this p to address s ase seat belt increase b increase increase w increase of seat be align sup workplac y growing t ill be reinfo s in these wledge tha ity of the po lation to use ration of pr ating strat ad user beh le of applyin ent of Publ rogram of in pecific traff use include ystander en family rules orkplace p law enforcem lt laws; and port among es, and valu hese actions rced by ma communitie t they shoul pulation by a seat belt. oposed stra egies to alig avior. g this strate ic Safety to tegrated TS ic safety go s strategies t gagement to about alway olicies and c ent use of , leaders to en e of a prima among stak ny member s are weari d. These str giving them 44 tegic appro n stakehold gic approac address sea CB strategi als. In this c o: get others s wearing a ompliance seat belts an courage sea ry seat belt eholders ac s of the com ng their se ategies buil specific ac ach to tran er actions w h, let us rev tbelt use in es spans the ase, a progr to wear a se seat belt; about alway d consistent t belt use, p law. ross the soc munities. T at belts. Ev d on the ex tions to “rem sform traff ith goals o iew an exam rural areas. social env am of TSCB at belt; s wearing a and visible olicies amo ial environm he data rev en those w isting cultur ind” the m ic safety f TSCB ple used b As shown i ironment an strategies t seat belt; enforcemen ng ent, seat be eal that mo ho are no e among th inority of th y n d o t lt st t, e e

Figur with U meas this s suppo stake So as cultur the c cultur e 19. Summ goals to inc p to this po ure and ana ame model rt the trans holder actio shown in F e on road u ulture of id e and behav ary of stra rease seat b int in the rep lyze traffic and metho formation o n, we need to igure 20, the ser behavior entified stak iors of the t tegic appro elt use. ort, we hav safety cultur ds can also f traffic saf transform same mod (see Figure eholders w argeted road 45 ach used in e presented e as it relat be applied ety culture. the traffic sa el used to re 6) can be us hose action user group Utah to al a model of es to road u to increase To increase fety culture present the ed to design s can posit . ign stakeho behavior an ser behavio stakeholder or instigat of the stake influence of strategies t ively influe lder actions d methods t rs. Howeve actions tha e the desire holder group traffic safet hat transform nce both th o r, t d . y e

Figur chan beha T using (TZD as the Death consi e 20. Illust ge actions o viors. he strategic TSCB strat ) National S unifying th s, 2012). A stent with th ration of tr f stakehold approach wi egies is cons trategy. Wi eme to “coo s shown in e safe-syste affic safety er A, B, C a th the goal o istent with t thin this na rdinate effo Figure 21 m approach 46 culture (see nd D to inf f reducing b he framewo tional strate rts among tr , the integr that involv Figure 6) a luence road ehavioral ha rk included i gy, traffic s affic safety ation of sta es the integ s the fram user cultu zards in roa n the Towar afety culture partners” (T keholder ac ration of mu ework to re and d user group d Zero Deat is portraye owards Zer tions is als ltiple traffi s h d o o c

safety zero f Figur Gove R a coa transf Foste indiv user inevit comb G traffi appro guida basis strategies, atality visio e 21. Safe-s rnment of S esearch on c lition involv orm culture r-Fishman e idual stakeh group beha ably require ined efforts iven that the c safety cultu aches that w nce on a pro of the propo which is the n including ystem stra outh Austr ommunity a ing a variet (Orwin et t al., 2007; d olders can im viors, the in s the forma . stakeholde re is contex ill apply i cess to ach sed strategi basis of sev Sweden, the tegic appro alia, 2012) nd organiza y of stakeho al., 2012; O eGruy & Et plement in tegration o tion of stak rs and action t specific, it n all circum ieve the inte c approach t 47 eral nationa Netherland ach to traff . tional transf lders is the rwin et al., z, 2010; Sch dividual TS f strategies eholder coa s relevant to is not possib stances. Ins gration acro o transform l strategic a s, and Austr ic safety (ad ormation pr most effecti 2014; Eddy armer, 2009 CB strategie across mul litions to al the goals f le to prescr tead, Chapt ss the socia traffic safet pproaches to alia (ACRS apted from ocesses reco ve leadershi et al., 2012 ; Kotter, 19 s that target tiple stakeh ign and coo or transform ibe the conte er 2 provid l environme y culture. achieve th , 2010). mmends tha p structure t ; Park, 2014 96). Wherea specific roa older group rdinate thes ing road use nt of specifi es high-leve nt that is th e t o ; s d s e r c l e