Evaluating public education messages aimed at monitoring and responding to social interactive technology on smartphones among young drivers

Accident Analysis and Prevention 104 (2017) 24–35

Contents lists available at ScienceDirect

Accident Analysis and Prevention journal homepage: www.elsevier.com/locate/aap

Evaluating public education messages aimed at monitoring and responding to social interactive technology on smartphones among young drivers

MARK

⁎

Cassandra S. Gaulda, , Ioni Lewisa, Katherine M. Whiteb, Judy J. Fleiterc, Barry Watsona a b c

Queensland University of Technology, Centre for Accident Research and Road Safety – Queensland (CARRS-Q), Kelvin Grove Campus, Kelvin Grove 4059, Australia Queensland University of Technology, School of Psychology and Counselling, Kelvin Grove Campus, Kelvin Grove 4059, Australia Global Road Safety Partnership, International Fédération of Red Cross & Red Crescent Sociétés, Route de Pré-Bois 1, CH-1214 Vernier, Switzerland

A R T I C L E I N F O

A B S T R A C T

Keywords: Public education messages SatMDT Monitoring Responding Soical interactive technology Young drivers

Young drivers are more likely than any other age group to access social interactive technology (e.g., Facebook, Email) on a smartphone while driving. The current study formed part of a larger investigation and was guided by The Step Approach to Message Design and Testing (SatMDT) to evaluate the relative effectiveness of three different public education messages aimed at reducing smartphone use among young drivers. The messages were each adapted to the specific behaviours of monitoring/reading and responding to social interactive technology on smartphones. Participants (n = 288; 199F, 89M) were drivers aged 17–25 years who resided in the Australian state of Queensland. Message acceptance (i.e., intention and effectiveness) and message rejection were both assessed using a self-report survey. Multivariate analyses found that, overall, the messages targeting monitoring/ reading behaviour were considered more effective than those targeting responding behaviour. The message that challenged the underlying motivation that believing you are a good driver makes it easier to monitor/read social interactive technology while driving was considered particularly effective by young male drivers.

The percentage of Australian mobile phone users who own smartphones is expected to reach 91% by 2017 (Telstra, 2014). The increased functionality of smartphones (e.g., access to social networking sites and emails) has meant that they have a greater potential to distract a driver. Despite the illegal nature of hand-held mobile phone use for all Australian drivers, the extra capabilities of smartphones are mostly accessed in hand-held mode, leading to an increase in crash risk (RudinBrown et al., 2013). In addition, it is possible that drivers are increasingly concealing their use from outside view, making detection (and enforcement) difficult (Gauld et al., 2014; Rudin-Brown et al., 2013). This concealment, in addition to other factors such as tinted car windows, heightens the need for other countermeasures, such as public education messages, to raise awareness of the dangers of smartphone use while driving. 1. Young drivers Despite being over-represented in road crash statistics (Department of Infrastructure and Regional Development, 2014), young drivers aged 18–25 years are more likely than any other age group to use a smartphone while driving (AAMI, 2012). Simulator studies have shown that young drivers distracted by their phones are more likely to run

⁎

Corresponding author. E-mail address: [email protected] (C.S. Gauld).

http://dx.doi.org/10.1016/j.aap.2017.04.011 Received 24 November 2016; Received in revised form 3 April 2017; Accepted 16 April 2017 0001-4575/ © 2017 Elsevier Ltd. All rights reserved.

yellow-lights (Haque et al., 2013) and take a substantially longer time to detect events originating in the driver's peripheral vision, such as a pedestrian entering a crossing (Haque and Washington, 2013). This evidence indicates that young drivers have an increased risk of being involved in road trauma as a result of using their smartphones (Neyens and Boyle, 2008). 2. Social interactive technology The term ‘interactive technology’ broadly encompasses functions that respond to user actions which, in turn, may cause the user to respond further (Interactive Technology Learning Curriculum Online, 2012). ‘Social interactive technology’ refers to smartphone functions that allow the user to communicate with other people via, for example, social networking sites (e.g., Facebook, Twitter), emails, and also texting and calling. As most Australians own smartphones, their ability to communicate with others through a variety of functions beyond texting and calling has increased. This expansion in communication channels has been termed ‘media multiplexity’ and is typical of modern relationships (Baym, 2015 p. 156). It is, therefore, possible, that young drivers are also communicating with others through a variety of applications on their smartphones. Indeed, in addition to being twice

Accident Analysis and Prevention 104 (2017) 24–35

C.S. Gauld et al.

Fig. 1. The SatMDT (Lewis et al., 2009, 2016b).

Groeger, 2011). Given that fear is an aversive affect which people wish to remove/avoid feeling, they will, therefore, be motivated to change especially if they are equipped with effective strategies for reducing the threat (Witte, 1992). Eliciting fear, however, is not the only way to persuade. Depicting compliance with the desired behaviour, and the associated positive consequences (e.g., approval from others) may also be effective (Lewis et al., 2007b, 2008a, 2013b; Tay, 2011). This modelling of the desired behaviour (which is often associated with positive emotion) can also act to reinforce the behaviour of drivers who are already acting in the desired manner. Public education messages have been associated with several limitations. These limitations include a lack of theoretical guidance on what constitutes effective message content, failure to segment the audience and to gain a thorough understanding of the target population, and, of relevance to the current study, a lack of scientifically rigorous evaluations measuring different outcome measures (Hoekstra and Wegman, 2011; Lewis et al., 2009; Slater, 1999; Stead et al., 2005). Indeed, the evaluation of the effectiveness of public education messages is not yet standard practise (Elliott, 2011; Hoekstra and Wegman, 2011; Hutchinson and Wundersitz, 2011; Phillips and Torquato, 2009). Evaluations have the potential to provide specific information regarding which key factors in the message design were effective and which were not, thereby building an evidence base regarding how to make public education messages more effective. If evaluations are not conducted, resources may continue to be directed towards unevaluated methods and not towards developing newer, and potentially more effective, methods (Hoekstra and Wegman, 2011; Plant et al., 2011).

as likely to make a phone call and four times more likely to text than drivers over 50 years, young drivers are also more likely to read emails while driving (AAMI, 2012) and 14% have admitted to taking a ‘selfie’ and uploading it onto social media while driving (AAMI, 2015). 3. Monitoring/reading and responding behaviours Recent research has investigated discrete behaviours associated with mobile phone use, such as reading and responding, as these behaviours have different rates of prevalence, and have been associated with different underlying motivations and different risk perceptions (Atchley et al., 2011; Shi et al., 2016; Waddell and Wiener, 2014). For example, young drivers report responding to communications more often than initiating them while driving (Atchley et al., 2011; Waddell and Wiener, 2014), suggesting that an underlying motivation may be the experience of social pressure to respond (Nemme and White, 2010). Young drivers perceive that sending text messages and replying to text messages are more risky than reading text messages (Shi et al., 2016). However, while this perception may be encouraging young drivers to read communications more often than initiate or respond to them (Gauld et al., 2016a), recent research has shown that simply hearing a notification can significantly disrupt performance on an attentiondemanding task (Stothart et al., 2015). It is possible, therefore, that reading a communication may not be as safe as young drivers perceive it to be. While these previous studies were limited to calling and texting behaviours, it is possible that the differences in prevalence and risk perception may also apply to the other social interactive technologies (e.g., Facebook messaging, emailing). The current study addresses this gap in knowledge by investigating the specific behaviours of monitoring/reading and responding to the range of social interactive technology on smartphones among young drivers.

5. The step approach to message design and testing The SatMDT (Lewis et al., 2009, 2016b) is a relatively new and innovative framework specifically designed for guiding the development and evaluation of health message content including road safety (see Fig. 1). While it is acknowledged that many behaviour change models/theories do exist (see Tay (2011) for a comprehensive review of these models) as well as manuals to guide the development and evaluation of campaigns (e.g., Delhomme et al., 2009; WHO, 2016), the SatMDT is unique in that it draws together empirical and multitheoretical evidence to guide the development and evaluation of road safety message content. Slater (1999) highlighted the need for direction and guidance on message development that specifically draws together complementary aspects of various theories. In addition, the SatMDT attempts to address some of the prior limitations of road safety public education message research and practise. The underlying psychological theories of decision making and attitude–behaviour relations that guide the SatMDT are the Theory of Planned Behaviour (TPB; Ajzen, 1985, 1991), the Extended Parallel Process Model (EPPM; Witte, 1992), The Elaboration Likelihood Model (ELM; Petty and Cacioppo, 1986), and

4. Road safety public education messages Road safety public education messages aim to modify or encourage safer road user behaviours (Elliott, 1993; Lewis et al., 2009; Watson et al., 1996). This persuasive effect can occur either directly by attempting to motivate behaviour change or indirectly through supporting other initiatives such as enforcement, through agenda-setting, or by simply normalising safe road user behaviours (Elliott, 1993; Lewis et al., 2009; Watson et al., 1996). In Australia, historically, road safety advertising campaigns endeavour to change behaviour through the use of threat appeals that elicit fear. Typically, these appeals aim to motivate through the depiction of the possible outcomes of non-compliance with the safe driving behaviour (e.g., injury or death) that the message is promoting (Dillard et al., 1996; Witte, 1992). For example, these outcomes may be physical injury, death, or legal sanctions (Donovan et al., 1999; Elliott, 1993; 25

Accident Analysis and Prevention 104 (2017) 24–35

C.S. Gauld et al.

5.2. Emotional appeal type

Social Learning Theory (Bandura, 1969). Notably, the framework acknowledges the impact of both message related (e.g., emotion, response efficacy) and individual characteristics (e.g., gender, age) on message effectiveness, includes primary research on the target audience, and incorporates an evaluation of message effectiveness. The four main steps of the framework are: (1) identification of preexisting individual characteristics (e.g., gender, current level of engagement in behaviours, and knowledge of strategies to reduce crash risk); (2) development of message-related characteristics focusing on the underlying beliefs and strategies elicited in Step 1 as well as other factors (e.g., emotional appeal type, positive or negative modelling of behaviour); (3) assessment of individual emotional and cognitive responses; and (4) evaluation of the effectiveness of the message by assessing both message acceptance (i.e., intention to adopt the message's recommendations) and message rejection (i.e., an avoidance or denial reaction) (Lewis et al., 2009, 2016b). Given the limited number of studies focusing on evaluating road safety public education messages, and similar to previous studies (Lewis et al., 2016a), the current study was guided by the evaluation phase, Step 4 of the SatMDT. Key factors that are considered in the evaluation are outlined below. These factors comprise the outcome measures of message acceptance and message rejection, as well as factors that may influence message effectiveness: emotional appeal type, gender, and response efficacy.

Generally speaking, approaches which model positive, safe behaviours elicit a positive emotional response and messages that model negative behaviours elicit a negative emotional response (Lewis et al., 2016b, 2007b). It is unusual, however, for a message to elicit just one emotional response. Dillard et al. (1996) found that, of 31 AIDS public service announcements, 30 elicited two or more emotions and 16 elicited three emotions, suggesting that there are a variety of factors that can influence how an individual responds. In addition, different audiences (e.g., males versus females) may report a variety of emotional responses to the same message. While some studies have conducted emotion checks in response to road safety messages (e.g., Kaye et al., 2016; Tay, 2011) it is not yet common practise and the anticipated emotional response often relies on the researchers’ assumptions (Lewis et al., 2016b; Plant et al., 2011). As previously stated, threat appeals that elicit fear are the most common form of road safety public education message in Australia, but they may not be as effective as desired (Carey et al., 2013) and the utility of other forms of threat appeals have been investigated. Social threat messages, for example, may focus on social disapproval resulting from failure to comply with a road safety message and financial threats may focus on the monetary costs resulting from illegal on road behaviours (Donovan and Henley, 1997). These other types of threat appeal may be more likely to elicit negative emotions such as shame or loss of peer approval, rather than fear (Rossiter and Percy, 1997). Appeals eliciting positive emotion, such as humour and pride, can also persuade (Dillard et al., 1996; Lewis et al., 2007b). Messages that elicit a predominantly positive emotional response have been developed in other countries (e.g., New Zealand and Belgium). Further, current research suggests that young people may respond better to positive appeals as they perceive them as less condescending in nature (Lewis et al., 2007b). Regardless of the emotion appeal type, however, it is vital that the emotion/s elicited within the message context is deemed appropriate. If they are not, the audience may respond to the message in a maladaptive manner (Lewis et al., 2013a; Rossiter and Percy, 1997).

5.1. Message acceptance and message rejection In accordance with Step 4 of the SatMDT, the effectiveness of the public education messages was evaluated in terms of message acceptance and message rejection (Lewis et al., 2016b). Similar to Lewis et al. (2008c), the term message acceptance, in the current study, refers to the degree to which a message has a persuasive impact and, therefore, the extent to which the message is effective. Message acceptance is often assessed using indirect measures of behaviour, such as selfreported degree of attitudinal, behavioural intention, or behavioural change (Elliott, 1993; Witte, 1992) which derive from theories such as the TPB (Ajzen, 1991). Specifically, the TPB posits that attitude underpins behavioural intention which is a good predictor of actual behaviour. These indirect measures and, in particular, behavioural intention, have previously been used to assess the degree of message acceptance within the road safety context (Kaye et al., 2013; Lewis et al., 2008b, 2016a; Tay and Watson, 2002). Alternatively, direct measures, such as asking participants to report how effective they perceive a message to be, have also been used to assess message acceptance in the road safety context (e.g., Kaye et al., 2013). Perceived effectiveness has been shown to be a reliable indicator of actual effectiveness (Dillard et al., 2007). In order to understand the impact of a message more thoroughly, however, message rejection should also be assessed (Witte, 1992). Message rejection refers to self-reported maladaptive responses to a message that an individual may experience such as ignoring, avoiding, or minimising a message (Witte, 1992). Message rejection, however, is rarely assessed (Lewis et al., 2008c, 2010; Tay and Watson, 2002). While it is possible that in the past message rejection and message acceptance were assumed to be opposite ends of the same continuum (Tay and Watson, 2002), more recent research suggests that different factors predict the extent to which individuals accept and reject a message (Lewis et al., 2008c, 2010; Tay and Watson, 2002). For example, a higher degree of fear elicited by a message has been shown to increase message rejection but not have any impact on message acceptance (Tay and Watson, 2002). In accordance with Step 4 of the SatMDT, the current study assessed both message acceptance, operationalised by behavioural intention and perceived effectiveness, and message rejection as outcome measures of persuasiveness.

5.3. Gender differences Physical threat appeals that elicit fear appear to be less effective for young males than young females (Lennon et al., 2010; Lewis et al., 2007b; Tay and Ozanne, 2002). They have even been found to have the opposite effect whereby young males carry out the behaviour contrary to that which is advocated in the message (Goldenbeld et al., 2008; Lennon et al., 2010). Further, young people may respond better to positive appeals as they perceive them as less condescending in nature (Lewis et al., 2007b). In addition, young drivers, particularly males, often display a degree of self-enhancement bias (or optimism bias) whereby they overestimate their driving ability compared to other drivers (e.g., Sibley and Harre, 2009; White et al., 2011). Threat appeals eliciting fear appear to have little impact on this bias in young drivers whereas positive appeals have been successful in reducing the level of self-enhancement bias (Sibley and Harre, 2009). Positive appeals (e.g., messages eliciting humour or pride) may be particularly persuasive for young male drivers (Lewis et al., 2007a, 2008a). In light of these gender differences, the current study investigated the differences in message outcomes for young male and female drivers. 5.4. Response efficacy Response efficacy refers to the inclusion of strategies within a message that, when utilised, the audience believes will avert the threat (Witte, 1992). Response efficacy has been shown to be a vital factor for message effectiveness (Lewis et al., 2010, 2007b, 2013a,b; Tay and Watson, 2002; Witte, 1992). As for threat-based appeals eliciting fear, 26

Accident Analysis and Prevention 104 (2017) 24–35

C.S. Gauld et al.

when driving, 28.5% reported using it in the illegal hand-held mode, and 20.5% of participants reported not using their smartphone while driving. Only 1.4% (n = 4) of all participants reported being caught for smartphone use while driving and 0.7% (n = 2) reported being fined. Most participants (77.1%) had completed high school as their highest level of education with 16% having completed an undergraduate degree. As commonly occurs in Australian universities, all first year psychology students received partial course credit for their participation and other participants were eligible to enter the draw to win one of five $AUD50 store vouchers.

the persuasiveness of messages eliciting positive emotion is the greatest when response efficacy is high (Lewis et al., 2007b, 2010, 2013a,b). Despite the importance of response efficacy, very few advertisements contain any reference to specific strategies (Cismaru, 2014; Phillips and Torquato, 2009). Presenting strategies within messages that target real issues that may elicit a low level of perceived response efficacy can be viewed as unethical (Popova, 2012). Some studies, therefore, do not manipulate response efficacy and present it as consistently high (Witte and Morrison, 1995). As smartphone use while driving is a prevalent and risky behaviour, the messages evaluated in the current study included the same three strategies (i.e., pull over, put it on silent, put it right out of sight) and it was assumed that these strategies would elicit a high level of response efficacy.

7.2. Materials/measures 7.2.1. The messages Nine different messages had been previously developed and piloted with young drivers (n = 33) in an earlier phase of the study and in accordance with Steps 2 and 3 of the SatMDT (see Gauld et al., under review). In this prior study, three messages were identified (each targeting a different underlying belief) as candidates for the final evaluation study on the basis of having had the most positive persuasive effects/ratings. The messages were presented as written outlines suitable for development into audiovisual messages for an online platform (e.g., YouTube) with the message medium (i.e., the online format) previously determined in consultation with young drivers. Participants were informed that the written messages were in an early stage of development and may eventually be developed into video format. In accordance with the SatMDT, the messages challenged a previously identified critical belief (see Gauld et al., 2016a) and included key features that have been shown to enhance message effectiveness including modelling of behaviour, emotional appeal type, strategies to prevent young drivers accessing social interactive technology while driving (i.e., response efficacy), and relevant contextual features (e.g., driving in the suburbs).

6. Current study The current study followed on from studies which identified key underlying beliefs in accordance with Step 1 of the SatMDT (see Gauld et al., 2016a) and a study that developed and piloted public education messages that targeted these underlying beliefs in accordance with Steps 2 and 3 of the SatMDT (Gauld et al., under review). Subsequently, the current study was guided by the final message evaluation phase (Step 4) of the SatMDT (see Fig. 1). Specifically, it evaluated the relative effectiveness of the three previously developed public education messages (see Gauld et al., under review) which were each adapted to target both monitoring/reading and responding behaviours. The influence of the key constructs of emotional appeal type, gender, and response efficacy were also considered. In addition to exploring the relative effectiveness of each message as assessed by the outcome measures of message acceptance and message rejection, the following hypotheses were proposed: H1. Intention to monitor/read and respond to social interactive technology (as a measure of message acceptance) would be lower for the intervention groups (i.e., groups exposed to a message) for each behaviour than for the control group (i.e., group not exposed to a message); and

7.2.2. The survey There were 3 versions of the survey: a control group (that received no message), an intervention group (that received one of three messages) for the behaviour of monitoring/reading, and an intervention group (further divided into three more groups that each received one of three messages) for the behaviour of responding. These messages were ‘Good Driver’, ‘Animated Smartphone’, and ‘Voice Your Opinion’ (see Table 1). One version of each message targeted monitor/read and one version targeted respond. There were minor changes to the message wording to differentiate between the two behaviours. As stipulated by the SatMDT, each message modelled positive behaviour and challenged a different belief, as indicated. Consistent with earlier phases of the study (see Gauld et al., 2016a,b), the key definitions were presented at the beginning of each version of the survey. Specifically, social interactive technology was defined as “functions accessed on smartphones through which the user communicates with other people. Examples of social interactive technologies include, but are not limited to, social networking sites (e.g., Facebook, Twitter), text messages, emails, phone calls, and Instagram”. “Monitor/read means to check your smartphone for communications and/or to read them while driving” and “respond to means to reply to a communication that was started by someone else while driving”. It was also noted that “while driving includes being stopped at traffic lights or in traffic; anywhere other than being in a parked vehicle”. After collecting various demographics (e.g., age, gender, type of car, type of licence), one of the messages was presented as a written outline for each intervention group. Effectiveness was measured both indirectly (i.e., behavioural intention to monitor/read and respond) and directly (i.e., perceived effectiveness). Message rejection was also assessed as well as perceived response efficacy and emotional responses. A manipulation check and an emotion check were also included in the

H2. For males, messages that elicit positive emotion for them would be more persuasive than messages that elicit negative emotion. 7. Method 7.1. Participants Approximately half (48%) of all participants (n = 288, 199F, 89M) were first year psychology students (n = 139, 113F, 26M) who had selfselected via an online recruitment system at a large Australian university. The other participants (n = 149, 86F, 63M) were recruited via various methods including face to face recruitment on the university campus grounds, from email lists, social media posts, websites, and from snowballing of the researchers’ family and friends. All participants were aged between 17 and 25 years (M = 19 years, SD = 1.87), 17.4% had an open licence, 31.9% had a P1 licence, and 36.1% had a P2 licence (P1 and P2 are intermediate licences [P1 is the first provisional licence before moving onto P2] in Queensland issued to a newly licensed driver for duration of up to 3 years before they receive an open licence).1 All participants owned a smartphone with the average length of ownership of 4.8 years (SD = 1.6 years) and most participants (99%) reported using their smartphone for personal use. Just over half of the participants (51%) reported using their smartphone in hands-free mode 1 Please note: 246 participants answered the question regarding licence type as the question was not included initially and was added after commencement of the study.

27

Accident Analysis and Prevention 104 (2017) 24–35

C.S. Gauld et al.

Table 1 Summaries of public education message content and underlying belief challenged. Message name

Message summary

Underlying belief challenged

‘Good driver’

A young driver waves goodbye to their Dad as they leave for university. The Dad waves back and shouts out ‘drive safely’ to which the young driver replies ‘Yeah, yeah, dad, I am a good driver you know’. As the young driver leaves a notification ‘ding’ is heard on their smartphone. The driver reaches over to check it but hesitates as they remember that they told their Dad they were a good driver. A dog runs in front of the car and the car stops safely. Voiceover: Be the good driver you say you are. Tagline: Pull over. Put it on silent. Put it right out of sight. Good drivers don’t < check/ answer > their smartphone while behind the wheel. A young driver is in a slow-moving traffic jam and feeling frustrated A ‘ding’ is heard on their smartphone and so they reach over to the back seat to retrieve their smartphone and < check/answer > the notification. Their eyes are still focused on the road ahead. An animated smartphone is seen desperately crawling away from the driver's hand and hides under the driver's jacket which is also on the back seat. A pedestrian suddenly runs out in front of the car. The driver is surprised, but brings their hand back to the wheel and stops in plenty of time. A tired smartphone peeps out from under the jacket and appears relieved. Voiceover: Think it's OK to answer your smartphone in a traffic jam? It's not. There are unexpected dangers in slow- moving traffic. Tagline: Pull over. Put it on silent. Put it right out of sight. Stay alert, and away from your smartphone, in all traffic situations. A young driver is alone in their car in a suburban area on their way to a friend's place. A notification ‘ding’ is heard on their smartphone which the driver ignores. Another notification ‘ding’ is heard and again, the driver ignores it. They arrive at their destination and join 4 other friends. Two of the friends ask if the driver received their notifications < and why the driver didn’t answer their notifications > to which the driver replies ‘Oh, I was driving’. The two friends who had sent messages mildly ridicule the driver. The other three friends support the driver by describing various negative consequences of such action (e.g., crashing, demerit points). The driver checks their phone and is surprised at how trivial the message actually is (i.e., picking up pizza on the way over). Voiceover: Keep it real. If you don’t like your friends < checking/answering > their smartphone while driving, tell them so. Tagline: Pull over. Put it on silent. Put it right out of sight. Voice your opinion – let your friends know what you really think.

Believing you are a good driver makes it easier to < monitor/read or respond > to social interactive technology while driving

‘Animated Smartphone’

‘Voice your Opinion’

Slow-moving traffic makes it easier to < monitor/read or respond > to social interactive technology while driving.

Friends/peers approve of a young driver < monitoring/reading or responding > to social interactive technology while driving.

7.2.5. Response efficacy Response efficacy comprised three items (e.g., “The public education message was effective in providing a strategy (or strategies) to stop me monitoring/reading social interactive technology while driving”) which were measured on a seven-point Likert scale ([1] strongly disagree to [7] strongly agree). These items were adapted from previous studies (e.g., Lewis et al., 2010) and formed a reliable scale in the current study for both monitoring/reading (α = .83) and responding (α = .77).

intervention surveys. As an indirect measure, behavioural intention was also able to be assessed in the control survey as it did not contain any reference to a message. In addition, the control survey also measured demographics and prior behavioural engagement.

7.2.3. Message acceptance Two scales measured message acceptance. For behavioural intention, three items were framed in terms of the target behaviour, action, context, and time (Fishbein and Ajzen, 2009) and were assessed on a seven-point Likert scale ([1] strongly disagree to [7] strongly agree). Thus, the target behaviour was “ < monitoring/reading or responding to > social interactive technology on a smartphone while driving” (e.g., “I intend to monitor/read/respond to social interactive technology while driving in the next week”) and formed a reliable scale for both monitoring/reading (α = .91) and responding (α = .90). Similar to previous studies (Dillard et al., 1996; Kaye et al., 2013), message effectiveness comprised two items (e.g., “How persuasive do you think the message was?” [1] not at all persuasive to [7] very persuasive) which were strongly correlated for both monitoring/reading (r(122) = .89, p < .01) and responding (r(108) = .80, p < .01). Both measures of message acceptance were measured after the presentation of the public education message.

7.2.6. Emotion check In response to the question stem “We are interested in knowing exactly how the public education message made you feel. Please indicate on the scale provided, the extent that the message made you feel” participants were asked to rate 13 discrete emotions (e.g., competent, anxious, flattered) on a seven point Likert scale ([1] strongly disagree to [7] strongly agree). This scale had been adapted from previous studies (e.g., Lewis et al., 2016c). 7.2.7. Manipulation check To provide an indication of whether the manipulation of the underlying beliefs and behaviours had occurred as anticipated, participants were asked, ‘In a few words, can you tell us what you thought the main message was?’

7.2.4. Message rejection Five items comprised the message rejection scale which were assessed on a seven-point Likert scale ([1] extremely unlikely to [7] extremely likely). The items had been adapted from previous studies (Lewis et al., 2008c) (e.g., “If it was on TV, I’d change channels”; “I’d leave the room”) and formed a reliable scale for both monitoring/ reading (α = .79) and responding (α = .69).

7.3. Procedure When approval had been obtained from the University's Human Research Ethics Committee, recruitment materials which included a link to the online survey were circulated to potential participants. Information at the beginning of the survey described the project, including what participation involved, expected benefits and risks, 28

Accident Analysis and Prevention 104 (2017) 24–35

C.S. Gauld et al.

reading or responding that was being targeted. Responses included ‘Don’t be a hypocrite; good drivers don’t check their phones’ and ‘If your friend is answering their smartphone while driving, put their best interest at heart and ensure you tell them that it is not acceptable’. In addition, almost all of the responses made reference to messages discouraging the general use of smartphones while driving.

Table 2 Self-reported average daily frequency of monitoring/reading and responding to social interactive technology while driving (n = 288). Daily frequency

Monitoring/reading (n)

Responding (n)

0 1 2 3 4 5 6 7 8 10 12 15 20 100

75 67 59 32 9 20 6 0 2 12 1 1 3 1

169 57 31 7 1 13 1 2 0 6 0 0 1 0

Total

288

288

8.3. Data pre-check To determine whether there were any differences between those who reported engaging with social interactive technology, on average, at least once per day and those who reported not engaging at all, independent samples t-tests were conducted for the key outcome measure of behavioural intention. Prior to the analysis, the measure of behavioural engagement was dummy coded into participants who reported not engaging (‘1’) and participants who reported engaging at least once on a daily basis (‘2’) for both monitoring/reading and responding (see Table 4). For monitoring/reading, results of the t-test found a significant difference between those who did not report monitoring/reading (M = 1.75, SD = 1.41) and those who did (M = 3.39, SD = 1.45; t(178) = −6.95, p < .001, two-tailed). For responding, there was also a significant difference between those who did not report responding (M = 1.90, SD = 1.04) and those who did (M = 3.07, SD = 1.32; t(164) = −6.41, p < .001, two-tailed). These results suggested that there were differences between these two samples for each behaviour on the study's key outcome measure. Subsequent analyses were, therefore, conducted on each sample separately. Table 4 describes the number of participants allocated to each message for each of the two surveys based on reported behavioural engagement.

and confidentiality. Participants were randomly allocated via computer to one of the two behaviour groups (i.e., monitoring/reading or responding) or the control condition. If the participant was allocated to one of the behaviour groups, they were then randomly assigned to one of the three public education messages adapted for that behaviour. Participants in the control condition did not read a public education message. Completion and submission of the survey to the researchers was considered provision of consent to participate. 8. Results

8.4. Emotion check

8.1. Descriptive statistics

As in previous studies (Lewis et al., 2016c), Table 5 lists the top 3 emotions elicited for each message. This table shows that a variety of emotions was elicited for each message for males and females with and without reported behavioural engagement. For males with reported behavioural engagement, ‘Good Driver (respond)’ elicited positive emotions; specifically, the male participants reported feeling relaxed, competent, and happy. For females with no reported behavioural engagement, ‘Good Driver (monitor/read)’ elicited positive emotions; specifically, the reported feeling relieved, proud, and happy. For both males and females who reported no behavioural engagement, ‘Good Driver (monitor/read)’ elicited positive emotions; specifically, participants reported feeling relieved, proud, and happy. The same emotions were elicited for ‘Good Driver (respond)’ for males and females together who reported behavioural engagement.

Of the 288 participants, 167 reported an average of 0 times per day that they responded to communications while driving, while 75 reported 0 for monitoring/reading. For responding, of those who reported a daily average of greater than 0, the majority (n = 109) reported responding to between 1 and 5 communications per day. For monitoring/reading, of those who reported a daily average of greater than 0, the majority (n = 187) also reported monitoring/reading between 1 and 5 communications per day (see Table 2). The means and standard deviations for each of the outcome variables are presented in Table 3, revealing some general trends. All intervention groups scored lower on behavioural intention than the control groups. ‘Good Driver (monitor/read)’ and ‘Good Driver (respond)’ scored the highest on effectiveness, regardless of the participants’ reported behavioural engagement. For response efficacy, all intervention groups perceived the strategies provided in the message would be effective in reducing monitoring/reading and responding to social interactive technology while driving, thereby eliciting a reasonably high level of response efficacy.

8.5. Effects of message exposure and gender on message acceptance and rejection The data were analysed separately for monitoring/reading and responding. Within each behaviour group, analyses were conducted separately for the sample of participants who reported behavioural engagement and those who did not. In order to provide an indication of the relative effectiveness of each message compared to the other messages within each behavioural group, three two-way between groups ANCOVAs were carried out for the samples of participants who reported behavioural engagement. Gender and group (i.e., message or no message) were the independent variables and the two measures of message acceptance (i.e., behavioural intention and effectiveness) and message rejection were the dependent variables. The reported level of behavioural engagement was the covariate. For the samples who did not report behavioural engagement, there was no equivalent covariate as the level of behavioural engagement for each participant in this sample was zero; therefore, three two-way between groups ANOVAs were conducted with the same variables as listed

8.2. Manipulation C Manipulation check Participants’ responses to the question ‘In a few words, can you tell us what you thought the main message was?’ underwent a content analysis to provide an indication of whether the manipulation of the underlying beliefs and behaviours had occurred as anticipated. Results showed that the majority of participants correctly identified the relevant underlying belief that was being challenged. Responses included, ‘That if you claim to be a good driver, then you should respect road safety rules’ (for ‘Good Driver’); ‘It is dangerous to operate your phone while you are driving, even if the traffic is slow’ (for ‘Animated Smartphone’); and, ‘Not to cave to peer pressure and to support your friends for not using phones while driving’ (for ‘Voice Your Opinion’). Many participants identified the specific behaviour of monitoring/ 29

Accident Analysis and Prevention 104 (2017) 24–35

C.S. Gauld et al.

Table 3 Mean ratings and standard deviations for intention, effectiveness, message rejection and response efficacy for young male and female drivers for each message.

Intention

Reported behavioural engagement

Monitor: good driver

Monitor: animated smartphone

Monitor: voice your opinion

Control: monitor

Respond: good driver

Respond: animated smartphone

Respond: voice your opinion

Control: respond

Yes (Monitor: n = 130)

3.09 (1.37) M:3.17 (1.49) F:3.06 (1.35)

3.21 (1.29) M: 3.22 (1.60)

2.95 (1.11) M: 2.71 (1.04)

2.84 (1.11) M:3.83 (1.80)

3.01 (0.75) M:2.90 (0.76)

2.64 (1.09) M:2.58 (1.10)

F: 3.21 (1.23)

F: 3.01 (1.14)

F:2.48 (1.09)

F:2.42 (0.96)

F:2.67 (1.14)

1.79 (1.85) M: 1.00 (0.00) F: 2.10 (2.13)

1.30 (0.58) M: 1.56 (0.96)

1.50 (0.82) M: 1.92 (1.03)

1.97 (1.10) M: 2.43 (1.10)

1.71 (0.67) M: 2.00 (0.76)

1.67 (0.81) M: 2.00 (0.76)

F: 1.19 (0.38)

F: 1.08 (0.17)

4.09 (1.66) M: 5.24 (1.00) F: 3.66 (1.66) 2.07 (1.56) M: 1.73 (0.98) F: 2.21 (1.75)

F: 2.03 (1.24)

F: 1.52 (0.56)

F: 1.49 (0.68)

3.70 (1.94) M: 3.27 (2.42) F: 3.89 (1.80) 2.05 (1.20) M: 2.09 (1.16) F: 2.03 (1.24)

4.64 (1.30) M:5.00 (1.41) F:4.47 (1.26) 4.61 (1.84) M: 4.88 (0.63) F: 4.50 (2.17)

4.18 (1.63) M: 3.25 (1.47)

4.17 (1.33) M:3.29 (1.41)

4.73 (1.68) M:4.50 (1.78)

4.54 (1.28) M:4.14 (1.07)

4.21 (1.48) M: 3.38 (1.80)

F: 4.71 (1.50) 3.90 (1.81) M: 3.17 (2.25)

F:4.42 (1.22) 3.81 (1.77) M: 3.75 (1.89)

F:4.82 (1.72) 4.70 (1.32) M: 4.43 (1.30)

F:4.71 (1.34) 4.43 (1.37) M: 4.00 (1.55)

F: 4.55 (1.28) 4.35 (1.33) M: 4.36 (1.46)

F: 4.21 (1.68)

F: 3.88 (1.93)

3.49 (1.04) M:3.13 (0.92) F:3.66 (1.07) 3.80 (1.30) M: 3.45 (1.11) F: 3.94 (1.40)

3.79 (1.33) M: 4.42 (1.03)

3.66 (0.92) M:3.77 (0.65)

F: 3.43 (1.36) 4.18 (1.15) M: 4.47 (1.42)

F:3.62 (0.99) 4.03 (0.86) M: 4.20 (0.99)

F: 4.06 (1.12)

3.85 (0.81)

4.57 (1.08) M: 4.79 (0.73) F: 4.47 (1.22) 4.70 (1.44) M: 5.17 (1.11) F: 4.50 (1.57)

4.23(1.54) M: 3.88 (1.75)

4.53 (1.28) M: 3.67 (1.43)

F: 4.41 (1.44) 4.03 (1.87) M: 3.67 (1.89)

F: 4.77 (1.15) 4.25 (1.16) M: 4.58 (0.83)

F: 4.24 (1.98)

F: 3.92 (1.48)

(Respond: n = 70) No (Monitor: n = 50) (Respond: n = 96) Effectiveness

Yes (Monitor: n = 90) (Respond: n = 53) No (Monitor: n = 32) (Respond: n = 55)

Message rejection

Yes (Monitor: n = 90) (Respond: n = 53) No (Monitor: n = 32) (Respond: n = 55)

Response efficacy

Yes (Monitor: n = 90) (Respond: n = 53) No (Monitor: n = 32) (Respond: n = 55)

N/A

N/A

N/A

N/A

N/A

N/A

F: 4.85 (1.36)

F: 4.72 (1.25)

F: 4.35 (1.31)

4.00 (1.02) M:4.80 (1.39)

3.57 (1.08) M: 3.63 (1.16)

3.66 (1.05) M: 4.30 (1.01)

F:3.71 (0.73) 3.82 (0.94) M: 4.69 (0.92)

F: 3.54 (1.08) 3.51 (0.93) M: 3.67 (1.29)

F: 3.40 (0.99) 3.69 (0.98) M: 3.51 (1.31)

F: 3.35 (0.56)

F: 3.40 (0.67)

F: 3.78 (0.80)

4.78 (1.19) M: 4.92 (1.26)

4.28 (1.14) M: 4.05 (1.48)

4.67 (1.00) M: 3.92 (1.10)

F: 4.73 (1.22) 4.80 (0.96) M: 4.48 (0.96)

F: 4.37 (1.01) 4.62 (1.41) M: 3.78 (1.64)

F: 4.97 (0.82) 4.28 (1.33) M: 3.90 (1.65)

F: 4.97 (0.96)

F: 5.19 (0.94)

F: 4.49 (1.14)

N/A

N/A

N/A

N/A

N/A

N/A

Note: Mean scores are based on a 7-point Likert scales. Lower intention scores indicate less accepting views of monitoring/reading or responding to social interactive technology while driving. Higher effectiveness scores indicate a perceived higher level of message persuasiveness. Higher message rejection scores indicate a stronger intention to avoid and/or deny the message recommendations. Higher response efficacy scores indicate the strategies in the messages would be effective in reducing risk. These are the raw descriptive scores.

previously. Table 6 shows the results for all the analyses. The significant findings are then described in more detail.

Table 4 Number of participants assigned to each message and grouped by reported behavioural engagement. Behaviour

Message

Reported behavioural engagement

Number of participants

Controla

Nil

Monitor/read

Good driver

Monitor/read

Animated smartphone Voice your opinion Good driver

Yes – M/R No – M/R Yes – R No – R Yes No Yes No Yes No Yes No Yes No Yes No

40 (29F, 11M) 18 (13F, 5M) 17 (12F, 5M) 41 (30F, 11M) 25 (17F, 8M) 14 (10F, 4M) 33 (21F, 12M) 10 (7F, 3M) 32 (25F, 7M) 8 (4F, 4M) 15 (11F, 4M) 20 (13F, 7M) 24(17F, 7M) 15 (9F, 6M) 14 (10F, 4M) 20 (13F, 7M)

Monitor/read Respond Respond Respond

Animated smartphone Voice your opinion Total

8.5.1. Monitor/read for the sample reporting behavioural engagement For the outcome measure of intention, there was a significant main effect for message. Post hoc comparisons, using the Bonferroni adjustment, revealed that the mean for each message was significantly lower than the control group. This result was in the expected direction. Specifically, participants scored significantly lower on intention for ‘Good Driver (monitor/read)’ (M = 3.09, SD = 1.37, p < .01), ‘Animated Smartphone (monitor/read)’ (M = 3.21, SD = 1.29, p < .01), and ‘Voice Your Opinion (monitor/read)’ (M = 2.95, SD = 1.11, p < .01) than control group (M = 4.09, SD = 1.66) (see Fig. 2). There were no other significant findings for intention. For the outcome measure of effectiveness, there was a significant interaction effect between gender and message. Specifically, as Fig. 3 shows, males were more likely to rate ‘Good Driver’ (M = 5.00, SD = 1.41) as more effective than both ‘Voice Your Opinion (monitor/read)’ (M = 3.29, SD = 1.41) and ‘Animated Smartphone (monitor/read)’ (M = 3.25, SD = 1.47). For females, however, inspection of the mean scores showed no significant differences between the messages for the mean effectiveness ratings. The main effect for gender was also significant with mean scores suggesting that, overall, females (M = 4.53, SD = 1.32) were more likely to rate the messages more effective than males (M = 3.78, SD = 1.60). There were no other significant findings for the outcome measure of effectiveness. For the

288

Note: ‘M/F’ denotes gender and the number denotes how many. For examples, ‘19M’ means there were 19 males in the sample. a All participants in the control group reported on both monitoring/reading and responding. ‘M/R’ refers to monitor/read and ‘R’ refers to respond.

30

Accident Analysis and Prevention 104 (2017) 24–35

C.S. Gauld et al.

Table 5 Top 3 self-reported emotions for each message and as a function of behavioural engagement. Behaviour

Message

Reported behavioural engagement?

Males

Females

Total sample

Monitor/read

Good driver

Yes

Relieved, competent, anxious

Relieved, anxious, competent

Animated smartphone

No Yes No

Relieved, anxious, happy/proud/ fearful Relieved, anxious, fearful Amused/competent, fearful Relieved/agitated/fearful

Relieved, proud, happy Anxious, fearful, relieved Annoyed, anxious, agitated

Voice your opinion

Yes

Agitated/annoyed/sad

No Yes

Sad/proud/competent/annoyed/ fearful Relaxed/competent/happy

Annoyed, agitated, competent/ anxious Annoyed, sad/fearful/relieved

Relieved, proud, happy Fearful, anxious, relieved Agitated, annoyed, anxious/ fearful Annoyed, agitated, competent Annoyed, sad, fearful

No

Relieved, anxious, proud

Yes No Yes

Competent, annoyed/anxious Relieved, amused/anxious Agitated, annoyed, sad/relieved/ anxious Sad, annoyed, fearful

Respond

Good driver

Animated smartphone Voice your opinion

No

Relieved, proud, agitated/happy/ anxious Competent/anxious, fearful/ agitated Fearful, anxious, annoyed Anxious, fearful, relieved Annoyed, agitated/amused/proud/ competent Annoyed, agitated, relieved

Relieved, happy, proud Anxious, relieved, competent Fearful, anxious, annoyed Anxious, relieved, fearful Annoyed, agitated, proud Annoyed, agitated, sad

Note: In the cases where the mean scores for the emotions were equal, more than 3 emotions are presented. Table 6 Summary of results for outcome measures by message and gender as a function of behavioural engagement. Behaviour

Monitoring/reading

Outcome measure

Intention

Effectiveness

Results Analysis

Sample with behavioural engagement

Sample without behavioural engagement

Covariate

F(1, 121) = .86, p = .356, ηp2 = .01

n/a

Message × gender

F(3, 121) = 2.62, p = .054, ηp2 = .06

F(3, 42) = .86, p = .469, ηp2 = .06

Message

F(3, 121) = 7.76, p < .001, ηp2 = .16 *

F(3, 42) = .39, p = .764, ηp2 = .03

Gender

F(1, 121) = 2.06, p = .154, ηp2 = .02

F(1, 42) = .04, p = .839, ηp2 = .00

Covariate

F(1, 83) = .24, p = .624, ηp2 = .00

n/a

Message × gender Message Gender Message rejection

Covariate Message × gender Message Gender

Responding

Intention

F(2, 83) = 2.61, p = .080,

= .06

F(1, 83) = 4.14, p = .045,

ηp2

F(1, 83) = .49, p = .488,

ηp2

F(2, 83) = 2.75, p = .070, F(2, 83) = 1.24, p = .294,

ηp2

F(1, 83) = .43, p = .512,

F(2, 26) = .84, p = .444, ηp2 = .06 F(1, 26) = .14, p = .717, ηp2 = .01

*

n/a

= .01

ηp2

ηp2

= .02

F(2, 26) = .35, p = .711, ηp2 = .03

=

.06 ŋp2

= .06

F(2, 26) = .44, p = .650, ηp2 = .03

=

.03ŋp2

= .03

F(2, 26) = .55, p = .584, ηp2 = .04

= .01ŋp = .01 2

F(1, 26) = .04, p = .848, ηp2 = .00

Covariate

F(1, 61) = .02, p = .968, ηp2 = .00 ŋp2 = .00

n/a

F(3, 61) = 1.28, p = .288, ηp2 = .06 ŋp2 = .06

F(3, 88) = .45, p = .720, ηp2 = .02

Message

F(3, 61) = 1.13, p = .343, ηp2 = .05ŋp2 = .05

F(3, 88) = .61, p = .611, ηp2 = .02

Covariate Message × gender Message Gender

Message rejection

= .08

ηp2

*

Message × gender

Gender Effectiveness

F(2, 83) = 3.39, p = .039,

ηp2

F(1, 88) = 3.42, p = .068, ηp2 = .04

F(1, 61) = .53, p = .469,

ηp2

F(1, 46) = .04, p = .853,

ηp2

=

F(2, 46) = .26, p = .775,

ηp2

2

= .01ŋp = .01

F(2, 49) = .30, p = .741, ηp2 = .01

F(2, 46) = .62, p = .542,

ηp2

= .03ŋp = .03

F(2, 49) = .25, p = .779, ηp2 = .01

F(1, 46) = 2.16, p = .149,

= .01ŋp = .01 2

.00 ŋp2

= .00

2

ηp2

=

.05ŋp2

= .05

n/a

F(1, 49) = .96, p = .332, ηp2 = .02

Covariate

F(1, 46) = .07, p = .791,

Message × gender

F(2, 46) = 1.04, p = .360, ηp2 = .04 ŋp2 = .04

Message

F(2, 46) = 1.46, p = .243, ηp2 = .06 ŋp2 = .06

F(2, 49) = 1.39, p = .260, ηp2 = .05

Gender

F(1, 46) = 4.52, p = .039, ηp2 = .09 ŋp2 = .09*

F(1, 49) = 3.12, p = .084, ηp2 = .06

ηp2

=

.00 ŋp2

= .00

n/a F(2, 49) = 3.82, p = .029, ηp2 = .14 *

* Result is significant at the p < .05 level.

outcome measures. Inspection of the mean scores, however, suggested that, although not significant, intention to monitor/read social interactive technology were generally lower for all of the intervention groups relative to the control group.

outcome measure of message rejection, no significant main or interactional effects were found. 8.5.2. Monitor/read for sample reporting no behavioural engagement There were no significant findings for this sample on any of the 31

Accident Analysis and Prevention 104 (2017) 24–35

C.S. Gauld et al.

Fig. 2. Main effect for monitoring/reading on the outcome measure of intention (for the sample reporting behavioural engagement).

Fig. 5. Gender × message interaction for responding on the outcome measure of message rejection (for the sample reporting no behavioural engagement).

8.5.4. Respond for sample reporting no behavioural engagement There was a significant gender × message interaction found for the outcome measure of message rejection for this sample. As shown in Fig. 5, males were more likely to score ‘Good Driver (respond)’ (M = 4.69, SD = 0.92) higher on message rejection than both ‘Animated Smartphone (respond)’ (M = 3.67, SD = 1.29) and ‘Voice Your Opinion (respond)’ (M = 3.51, SD = 1.31). Females, however, scored Voice Your Opinion (respond)’ (M = 3.78, SD = 0.80) higher on message rejection than both ‘Animated Smartphone (respond)’ (M = 3.40, SD = 0.67) and ‘Good Driver (respond)’ (M = 3.35, SD = 0.56). There were no other significant main or interactional effects for this sample on any of the remaining outcome measures. Inspection of the mean scores, however, showed that, while not significant, intentions to respond to social interactive technology were generally lower for all of the intervention groups relative to the control group.

Fig. 3. Gender x message interaction for monitoring/reading on the outcome measure of effectiveness (for the sample reporting behavioural engagement).

8.5.3. Respond for sample reporting behavioural engagement There was a significant main effect for gender on message rejection. As shown in Fig. 4, regardless of the message, males (M = 4.12, SD = 1.21) were significantly more likely than females (M = 3.55, SD = 0.95) to report higher levels of message rejection. For this sample, there were no other significant main or interactional effects found for any of the other outcome measures (i.e., intentions, effectiveness). Inspection of the mean scores, however, showed that, while not significant, intention to respond to social interactive technology were generally lower for all of the intervention groups relative to the control group.

9. Discussion The main aim of this study was to evaluate the relative persuasive effects of three messages aimed at stopping young drivers monitoring/ reading or responding to social interactive technology on their smartphone while driving. Overall, the SatMDT (Lewis et al., 2009, 2016b) successfully guided the evaluation in the current study. Although the importance of message evaluation has been acknowledged in some literature, it is not yet standard practice, particularly in the use of scientifically based evaluations that measure different outcome measures as conducted in the current study (Elliott, 2011; Hoekstra and Wegman, 2011; Hutchinson and Wundersitz, 2011; Lewis et al., 2009; Phillips and Torquato, 2009; Stead et al., 2005). In addition to assessing meassage acceptance (i.e., behavioural intention and effectiveness), the current study also measured message rejection. Key features which have been shown to influence message effectiveness (i.e., emotional appeal type, gender, and response efficacy) were also assessed. Overall, the messages targeting monitoring/reading were found to be more persuasive than those targeting responding. The behaviour of monitoring/reading was reported to be more prevalent with 74% of participants reporting monitoring/reading on a daily basis compared to 41% of participants who reported responding. This higher prevalence is supported in previous studies (Atchley et al., 2011; Shi et al., 2016). These results support the suggestion that monitoring/reading and responding may be distinct behaviours and, therefore, may require different message content when attempting to persuade young drivers to stop engaging with social interactive technology. The results for each behaviour are discussed in the sections below.

Fig. 4. Main effect of gender for responding on the outcome measure of message rejection (for the sample reporting behavioural engagement).

32

Accident Analysis and Prevention 104 (2017) 24–35

C.S. Gauld et al.

may experience such as ignoring, avoiding, or minimising a message (Witte, 1992). For the sample with reported behavioural engagement, males were more likely, overall, to score all messages higher on message rejection than females. For the sample with no reported behavioural engagement, a significant interaction effect was found whereby males were more likely to reject ‘Good Driver (respond)’ and females were more likely to reject ‘Voice Your Opinion (respond)’. This finding is interesting in that it is the only significant finding relating to the samples who reported no behavioural engagement. Little is currently known about the underlying influences of message rejection (Lewis et al., 2008c) and the reasons for the findings in this current study are also unclear. It may be possible however, that, given these young male drivers were already complying with the desired behaviour, that they reported maladaptive responses, such as skipping the ad if it was on YouTube or leaving the room when the ad came on as they did not perceive the messages as relevant to them. These maladaptive responses represent common operationalisations of the message rejection construct. Together, these findings reinforce the importance of measuring both acceptance and rejection when evaluating public education messages (Witte, 1992). More research is needed to gain a greater understanding of the factors that influence message rejection which may ultimately help to create more effective messages.

9.1. Monitoring/reading For monitoring/reading, significant results were found for the sample who reported behavioural engagement. The first hypothesis, that each intervention group would score significantly lower on intention than the control group was supported for this behaviour. It is likely that these messages were highly relevant to this sample, thereby improving their persuasiveness (Lewis et al., 2012). As all participants in this sample reported monitoring/reading on a daily basis, this finding supports prior research suggesting that the lower the base level of engaging in the desired behaviour, the more persuasive impact a message is likely to have (Elliott, 1993). For the current study, the higher the base level of smartphone use while driving, the greater potential persuasive impact the messages had. There was also a significant gender x message interaction found for the outcome measure of effectiveness among the sample who reported behavioural engagement. Specifically, males were more likely to score ‘Good Driver (monitor/read)’ as the most effective message. ‘Good driver (monitor/read)’ challenged the underlying belief that believing you are a good driver would encourage monitoring/reading social interactive technology while driving. This message depicted a young driver's father encouraging them to drive safely and the young driver reassuring their father that they are a good driver (see Table 1). Parental influence on young driver behaviour has been documented in previous research (e.g., Gil et al., 2016; Prato et al., 2010; Scott-Parker et al., 2009, 2012). More recent research has highlighted that the samesex parent may have a greater influence than the opposite-sex parent (Gil et al., 2016; Scott-Parker et al., 2014). While Scott-Parker et al. (2014), for example, investigated the young driver's perception of the riskiness of their parents’ driving behaviour, it is possible that, in the current study, the encouragement of the father to drive safely had more influence on young male drivers than on young female drivers. For the emotions elicited by this message, the male participants reported feeling relieved, anxious, happy, proud, and fearful (see Table 5). Hypothesis two, that males would report that the messages eliciting a positive emotional response for them as more persuasive was, therefore, only partially supported given that negative emotions (i.e., anxious, fearful) were also reported by males for this message. Table 5 shows that the only message that males reported as eliciting all positive emotions as the top three strongest emotions elicited was ‘Good Driver (respond)’. These three positive emotions were reported for males with reported behavioural engagement and who reported feeling relaxed, competent, and happy. An inspection of the mean score for effectiveness for this message showed that this sample scored ‘Good Driver (respond)’ highest on effectiveness compared to the other messages that all elicited some negative emotions. Arguably, the lack of significant findings for the sample without behavioural engagement with their smartphone while driving could have been expected. Given that this sample reported already carrying out the desired behaviour, the base level of compliance could be considered already high and, therefore, the messages’ ability to persuade could be somewhat limited (Elliott, 1993). Although not significant, Table 3 revealed trends in outcome scores such that all messages for both samples (with and without reported behavioural engagement) were associated with mean intention scores which were lower than the control group (i.e., in the expected direction). Given that all of the messages modelled positive behaviour, it is possible that, while there were no significant findings, the messages acted to reinforce these participants’ behaviour and promote prevention (Lewis et al., 2007b).

9.3. Monitoring/reading and responding Perhaps the most interesting finding to emerge from the current study was that young male drivers who reported behavioural engagement with smartphone use while driving rated ‘Good Driver (monitor/ read)’ as the most effective message while young male drivers who reported no behavioural engagement rated ‘Good Driver (respond)’ with the highest level of message rejection. Given that the only difference between the message content relates to the subtle changes made to target the particular behaviour (e.g., ‘he/she reaches over to pick it up and check their notification’ for monitoring/reading was adjusted to ‘he/she reaches over to pick it up and answer the communication’ for responding), this result provides further support for the view that monitoring/reading and responding may be distinct behaviours with respect to smartphone use while driving. For both behaviours, there was a variety of emotional responses reported for each message. While all the messages modelled positive behaviour, a variety of emotions was elicited (see Table 5), supporting previous research that has shown that messages are likely to elicit more than two emotions (e.g., Dillard et al., 1996). Of note, and similar to Lewis et al. (2016c), the prevalence with which the emotion of fear was reported was greater than anticipated. It may be possible that, due to the past reliance on threat-appeals eliciting fear in road safety advertising campaigns, the participants in the current study were anticipating that some sort of threat would be involved and, therefore, experienced a fear-based reaction. Alternatively, fear may have been elicited in response to what might happen if they engage with their smartphone while driving, such as being involved in a crash or receiving a fine. For response efficacy, all messages scored reasonably high for both genders with and without reported behavioural engagement (see Table 3). This finding is encouraging given that a high level of response efficacy is a key determinant of message effectiveness (Lewis et al., 2007b, 2010, 2013a,b; Tay, 2002; Tay and Watson, 2002; Witte, 1992). As the strategies had been elicited in earlier phases of the research with the end users (i.e., young drivers) in accordance with the SatMDT (Lewis et al., 2009, 2016b), it was anticipated that they would elicit a high level of response efficacy.

9.2. Responding 9.4. Strengths and limitations For responding, significant results were found for the outcome measure of message rejection. Message rejection refers to the extent of self-reported maladaptive responses to a message that an individual

The current study applied the evaluation step of the SatMDT (Lewis et al., 2009, 2016b) which is a theoretical framework incorporating 33

Accident Analysis and Prevention 104 (2017) 24–35

C.S. Gauld et al.

ment of smartphone use while driving is associated with some unique and rather challenging aspects (e.g., as drivers try to conceal their use), highlighting a potentially important role to be played by public education messages to support enforcement efforts. Theoretically-based insights of message persuasiveness such as that offered by the current study are important in helping to identify key factors which may heighten the effectiveness of messages, not just targeting young drivers, but all drivers.

scientific methodology. The importance of evaluation, a task which is otherwise rarely conducted, cannot be underestimated (Elliott, 2011; Hoekstra and Wegman, 2011; Lewis et al., 2009; Phillips and Torquato, 2009; Stead et al., 2005; Wundersitz and Hutchinson, 2011). Measuring both message acceptance and message rejection as well as other key factors, such as response efficacy and emotional appeal type, can provide insights into the messages’ impact on these outcome measures. The inclusion of a control group to compare with the responses of the intervention groups on the outcome measure of intention enhanced the robustness of the conclusions. The current study also contained manipulation checks of both message content and emotion. These checks are not always conducted; rather, many researchers rely on assumptions regarding a message's content and the assumption that the desired effect will be achieved (Dillard et al., 1996; Eveland and McLeod, 1999; LaTour and Rotfeld, 1997; Lewis et al., 2009; Phillips and Torquato, 2009; Plant et al., 2011). If the emotions are not checked, the degree of message acceptance or rejection may be attributed to an assumed emotion rather than the actual emotion/s that is elicited (Dillard et al., 1996; Lewis et al., 2009). There were also limitations associated with the current study. The messages were presented as written outlines with a single exposure. It is possible that, when compared to a fully developed audio-visual advertisement, these one-off presentations of written outlines may not have the potential to elicit the strongest possible impact (see Elliott, 2011, for a discussion of effective levels of message exposure). This limitation has been acknowledged in previous applications of the evaluation step of the SatMDT (e.g., Lewis et al., 2016a). When the sample was split with regard to behavioural engagement, some of the cells became quite small which may have reduced the statistical power to detect effects. The self-report nature of the surveys may have been subject to bias, particularly as the behaviours being investigated were illegal (Beck and Ajzen, 1991). Although it is possible that participants may have responded in a socially desirable manner, the anonymous nature of the survey should have minimised reporting bias. Although every effort was made to recruit more young male drivers, the sample was approximately two-thirds female. Females are less likely to be involved in road crash (Queensland Government, 2016) and are less likely to be apprehended for mobile phone use while driving in the Australian state of Queensland (Department of Transport and Main Roads, 2016). In addition, more than half of the participants in this study were university students. University students may be more educated than the general population, thereby potentially limiting the generalisability of the findings. Future research should include a followup measure of intention and/or behaviour to ascertain whether the messages had a persuasive impact over time.

References AAMI, 2012. Young driver Index 2012. Retrieved from: www.aami.com.au/sites/default/ files/fm/news/Young%20Driver%20Index%22012_12_14.pdf. AAMI, 2015. AAMI Targets Social Stigma to Tackle Distracted Driving. Retrieved from: https://www.aami.com.au/media-centre/aami-targets-social-stigma-tackledistracted-driving.html. Ajzen, I., 1985. From intentions to actions: a theory of planned behavior. In: Kuhl, J., Beckmann, J. (Eds.), Action Control: From Cognition to Behaviour. Springer-Verlag, Berlin, pp. 11–39. Ajzen, I., 1991. The theory of planned behavior. Organ. Behav. Hum. Decis. Process. 50 (2), 179–211. http://dx.doi.org/10.1016/0749-5978(91)90020-T. Atchley, P., Atwood, S., Boulton, A., 2011. The choice to text and drive in younger drivers: behavior may shape attitude. Accid. Anal. Prev. 43 (1), 134–142. http://dx. doi.org/10.1016/j.aap.2010.08.003. Gauld, C., Lewis, I., White, K., Watson, B., 2016a. Young drivers’ engagement with social interactive technology on their smartphones: critical beliefs to target in public education messages. Accid. Anal. Prev. 96, 208–218. Gauld, C., Lewis, I., White, K., Watson, B., 2016b. Key beliefs influencing young drivers’ engagement with social interactive technology on their smartphones: a qualitative study. Traffic Inj. Prev. 17 (2), 128–133. http://dx.doi.org/10.1080/15389588.2015. 1047014. Gauld, C., Lewis, I., White, K., Fleiter, J., Watson, B., under review. Public education messages aimed at smartphone use among young drivers: a mixed methods exploration of their effectiveness. Traffic Inj. Prev. Bandura, A., 1969. Principles of Behavior Modification. Holt, Rinehart and Winston, New York. Baym, N., 2015. Personal Connections in the Digital Age, 2nd ed. Polity Press, Cambridge, UK. Beck, L., Ajzen, I., 1991. Predicting dishonest actions using the theory of planned behaviour. J. Res. Personal. 25, 285–301. http://dx.doi.org/10.1016/0092-6566(91) 90021-H. Carey, R.N., McDermott, D.T., Sarma, K.M., 2013. The impact of threat appeals on fear arousal and driver behavior: a meta-analysis of experimental research 1990–2011. PLOS ONE 8 (5). http://dx.doi.org/10.1371/journal.pone.0062821. Cismaru, M., 2014. Using the extended parallel process model to understand texting while driving and guide communication campaigns against it. Soc. Mark. Q. 20 (1), 66–82. http://dx.doi.org/10.1177/1524500413517893. Delhomme, P., De Dobbeleer, W., Forward, S., Simoes, A., 2009. Manual for Designing, Implementing, and Evaluating Road Safety Communication Campaigns. Belgian Road Safety Institute, Belgium. Department of Infrastructure and Regional Development, 2014. Road trauma Australia 2014 Statistical Summary. Retrieved from: http://bitre.gov.au/publications/ ongoing/files/Road_trauma_Australia_2014_statistical_summary_N_ISSN.pdf. Department of Transport and Main Roads, 2016. Mobile Phone Infringements . Dillard, J.P., Plotnick, C.A., Godbold, L.C., Freimuth, V.S., Edgar, T., 1996. The multiple affective outcomes of aids PSAs: fear appeals do more than scare people. Commun. Res. 23 (1), 44–72. http://dx.doi.org/10.1177/009365096023001002. Dillard, J.P., Shen, L., Vail, R.G., 2007. Does perceived message effectiveness cause persuasion or vice versa? 17 consistent answers. Hum. Commun. Res. 33 (4), 467–488. http://dx.doi.org/10.1111/j.1468-2958.2007.00308.x. Donovan, R., Henley, N., 1997. Negative outcomes, threats and threat appeals: widening the conceptual framework for the study of fear and other emotions in social marketing communications. Soc. Mark. Q. 4 (1), 56–67. http://dx.doi.org/10.1080/ 15245004.1997.9960987. Donovan, R., Jalleh, G., Henley, N., 1999. Executing effective road safety advertising: are big production budgets necessary? Accid. Anal. Prev. 31 (3), 243–252. http://dx.doi. org/10.1016/S0001-4575(98)00074-8. Elliott, B., 1993. Road Safety Mass Media Campaigns: A Meta Analysis (9780642512529;0642512523). Federal Office of Road Safety, Canberra. Elliott, B., 2011. Beyond reviews of road safety mass media campaigns: looking elsewhere for new insights. J. Australas. Coll. Road Saf. 22 (4), 11–18. Eveland, W.P., McLeod, D.M., 1999. The effect of social desirability on perceived media impact: implications for third-person perceptions. Int. J. Public Opin. Res. 11 (4), 315–333. Fishbein, M., Ajzen, I., 2009. Predicting and Changing Behavior: The Reasoned Action Approach. Psychology Press, New York. Gauld, C., Lewis, I., White, K.M., 2014. Concealing their communication: exploring psychosocial predictors of young drivers’ intentions and engagement in concealed texting. Accid. Anal. Prev. 62, 285–293. http://dx.doi.org/10.1016/j.aap.2013.10. 016. Gil, S., Taubman-Ben-Ari, O., Toledo, T., 2016. A multidimensional intergenerational model of young males’ driving styles. Accid. Anal. Prev. 97, 141–145. http://dx.doi.

10. Conclusion The current study applied the evaluation step of the SatMDT (Lewis et al., 2009, 2016b) to three public education messages targeting monitoring/reading and responding to social interactive technology among young drivers. The messages aimed at monitoring/reading social interactive technology by young drivers were deemed more effective and, in particular, ‘Good Driver (monitor/read)’ was found to be most effective for males. It may, therefore, be worthwhile targeting monitoring/reading in future messages and developing ‘Good Driver’ into an audiovisual message format. The findings also support previous theoretical and empirical evidence which suggested that message acceptance and message rejection are different constructs. More research, however, is required to clarify the underlying factors influencing message rejection in order to increase message effectiveness. The importance of conducting emotions checks and including strategies with a high level of response efficacy was also reinforced. Accessing social interactive technology, and especially the behaviour of monitoring/reading, is prevalent among young drivers; thus efforts are needed to reduce this high risk behaviour. Police enforce34

Accident Analysis and Prevention 104 (2017) 24–35

C.S. Gauld et al.

Neyens, D.M., Boyle, L.N., 2008. The influence of driver distraction on the severity of injuries sustained by teenage drivers and their passengers. Accid. Anal. Prev. 40 (1), 254–259. http://dx.doi.org/10.1016/j.aap.2007.06.005. Petty, R., Cacioppo, J., 1986. The elaboration likelihood model of persuasion. In: Berkowitz, L. (Ed.), Advances in Experimental Social Psychology, vol. 19. Academic, New York, pp. 123–205. Phillips, R., Torquato, R., 2009. An Overview of 45 Anti-Speeding Campaigns. Oslo, Norway . Plant, B., Reza, F., Irwin, J.D., 2011. A systematic review of how anti-speeding advertisements are evaluated. J. Australas. Coll. Road Saf. 22 (4), 18–33. Popova, L., 2012. The extended parallel process model: illuminating the gaps in research. Health Educ. Behav. 39 (4), 455–473. http://dx.doi.org/10.1177/ 1090198111418108. Prato, C.G., Toledo, T., Lotan, T., Taubman-Ben-Ari, O., 2010. Modeling the behavior of novice young drivers during the first year after licensure. Accid. Anal. Prev. 42 (2), 480–486. http://dx.doi.org/10.1016/j.aap.2009.09.011. Queensland Government, 2016. Driver Demographics. Retrieved from: https://data.qld. gov.au/dataset/crash-data-from-queensland-roads/resource/dd13a889-2a48-4b918c64-59f824ed3d2c. Rossiter, J.R., Percy, L., 1997. Advertising Communications & Promotion Management, vol. 2 McGraw-Hill Companies, New York. Rudin-Brown, C.M., Young, K., Lenne, M., 2013. Melbourne drivers’ observed use of mobile phones: could there be unintended consequences of partial bans? In: Regan, M., Victor, T. (Eds.), Human Factors in Road and Rail Transport: Driver Distraction and Inattention: Advances in Research and Countermeasures. Ashgate Publishing Ltd., Farnham, Great Britain, pp. 313–326. Scott-Parker, B., Watson, B., King, M.J., 2009. Exploring How Parents and Peers Influence the Behaviour of Young Drivers . Scott-Parker, B., Watson, B., King, M.J., Hyde, M.K., 2012. “They’re lunatics on the road”: exploring the normative influences of parents, friends, and police on young novices’ risky driving decisions. Saf. Sci. 50 (9), 1917–1928. http://dx.doi.org/10.1016/j.ssci. 2012.05.014. Scott-Parker, B., Watson, B., King, M.J., Hyde, M.K., 2014. Young novice drivers and the risky behaviours of parents and friends during the provisional (intermediate) licence phase: a brief report. Accid. Anal. Prev. 69, 51–55. http://dx.doi.org/10.1016/j.aap. 2013.11.016. Shi, J., Xiao, Y., Atchley, P., 2016. Analysis of factors affecting drivers’ choice to engage with a mobile phone while driving in Beijing. Transp. Res. F: Traffic Psychol. Behav. 37, 1–9. http://dx.doi.org/10.1016/j.trf.2015.12.003. Sibley, C., Harre, N., 2009. The impact of different styles of traffic of traffic safety advertisement on young drivers’ explicit and implicit self-enhancement bias. Transp. Res. F: Traffic Psychol. Behav. 12 (2), 159–167. Slater, M.D., 1999. Integrating application of media effects, persuasion, and behavior change theories to communication campaigns: a stages-of-change framework. Health Commun. 11 (4), 335–354. http://dx.doi.org/10.1207/S15327027HC1104_2. Stead, M., Tagg, S., MacKintosh, A.M., Eadie, D., 2005. Development and evaluation of a mass media theory of planned behaviour intervention to reduce speeding. Health Educ. Res. 20 (1), 36–50. http://dx.doi.org/10.1093/her/cyg093. Stothart, C., Mitchum, A., Yehnert, C., 2015. The attentional cost of receiving a cell phone notification. J. Exp. Psychol.: Hum. Percept. Perform. http://dx.doi.org/10.1037/ xhp0000100. Tay, R., 2002. Exploring the effects of a road safety advertising campaign on the perceptions and intentions of the target and nontarget audiences to drink and drive. Traffic Inj. Prev. 3 (3), 195–200. http://dx.doi.org/10.1080/15389580213651. Tay, R., 2011. Drivers’ perception of two seatbelt wearing advertisements with different emotional appeals and cultural settings. J. Australas. Coll. Road Saf. 22 (4), 82–89. Tay, R., Ozanne, L., 2002. Who are we scaring with high fear road safety advertising campaigns? Asia Pac. J. Transp. 4, 1–12. Tay, R., Watson, B., 2002. Changing drivers’ intentions and behaviours using fear-based driver fatigue advertisements. Health Mark. Q. 19 (4), 55–68. http://dx.doi.org/10. 1300/J026v19n04_05. Telstra, 2014. Cross-Platform Consumers: Media Multi-Tasking and Multi-Screening in Today's Mobile Landscape [Press Release]. Retrieved from: http://www.telstra.com. au/aboutus/download/document/telstra-cross-platform-consumers.pdf. Waddell, L.P., Wiener, K.K.K., 2014. What's driving illegal mobile phone use? Psychosocial influences on drivers’ intentions to use hand-held mobile phones. Transp. Res. F: Traffic Psychol. Behav. 22, 1–11. http://dx.doi.org/10.1016/j.trf. 2013.10.008. Watson, B., Fresta, J., Whan, H., McDonlad, J., Dray, R., Bauermann, C., et al., 1996. Enhancing Driver Management in Queensland. Retrieved from: Brisbane. White, M., Cunningham, L., Titchener, K., 2011. Young drivers’ optimism bias for accident risk and driving skill: accountability and insight experience manipulations. Accid. Anal. Prev. 43 (4), 1309–1315. Witte, K., 1992. Putting the fear back into fear appeals: the Extended Parallel Process Model. Commun. Monogr. 59 (4), 329–349. Witte, K., Morrison, K., 1995. Using scare tactics to promote safer sex among juvenile detention and high school youth. J. Appl. Commun. Res. 23 (2), 128–142. http://dx. doi.org/10.1080/00909889509365419. World Health Organisation, 2016. Road Safety Mass Media Campaigns: A Toolkit. Retrieved from: http://www.who.int/violence_injury_prevention/publications/road_ traffic/media-campaigns/en/. Wundersitz, L., Hutchinson, T., 2011. What can we learn from recent evaluations of road safety mass media campaigns? J. Australas. Coll. Road Saf. 22 (4), 40–50.

org/10.1016/j.aap.2016.09.004. Goldenbeld, C., Twisk, D., Houwing, S., 2008. Effects of persuasive communication and group discussions on acceptability of anti-speeding policies for male and female drivers. Transp. Res. F: Traffic Psychol. Behav. 11 (3), 207–220. Groeger, J., 2011. How many E's in road safety? In: Porter, B. (Ed.), The Handbook of Traffic Psychology. Elsevier Science, UK. Haque, M.M., Ohlhauser, A.D., Washington, S., Boyle, L.N., 2013. Examination of distracted driving and red light running: analysis of simulator data. In: Paper presented at the 92nd Annual Meeting of Transportation Research Board. Washington DC, USA. . Haque, M.M., Washington, S., 2013. Effects of mobile phone distraction on drivers’ reaction times. J. Australas. Coll. Road Saf. 24 (3), 20–29. Hoekstra, T., Wegman, F., 2011. Improving the effectiveness of road safety campaigns: current and new practices. IATSS Res. 34 (2), 80–86. http://dx.doi.org/10.1016/j. iatssr.2011.01.003. Hutchinson, T.P., Wundersitz, L.N., 2011. Road safety mass media campaigns: why are results inconclusive, and what can be done? Int. J. Inj. Control Saf. Promot. 18 (3), 235–241. http://dx.doi.org/10.1080/17457300.2010.540330. Interactive Technology Learning Curriculum Online, 2012. Interactive Technology. Retrieved from: http://www.wiu.edu/itlc/resources/itlc_glossary.html. Kaye, S., Lewis, I., Algie, J., White, M., 2016. Young drivers’ responses to anti-speeding advertisements: comparison of self-report and objective measures of persuasive processing and outcomes. Traffic Inj. Prev. 17 (4), 352–358. http://dx.doi.org/10. 1080/15389588.2015.1084419. Kaye, S., White, M., Lewis, I., 2013. Individual differences in drivers’ cognitive processing of road safety messages. Accid. Anal. Prev. 50, 272–281. http://dx.doi.org/10.1016/ j.aap.2012.04.018. LaTour, M.S., Rotfeld, H.J., 1997. There are threats and (maybe) fear-caused arousal: theory and confusions of appeals to fear and fear arousal itself. J. Advert. 26 (3), 45–59. Lennon, R., Rentfro, R., O’Leary, B., 2010. Social marketing and distracted driving behaviors among young adults: the effectiveness of fear appeals. Acad. Mark. Stud. J. 14 (2), 95–113. Lewis, I., Ho, B., Lennon, A., 2016a. Designing and evaluating a persuasive child restraint television commercial. Traffic Inj. Prev. 17 (3), 271–277. http://dx.doi.org/10.1080/ 15389588.2015.1072626. Lewis, I., Watson, B., White, K.M., 2016b. The Step approach to Message Design and Testing (SatMDT): a conceptual framework to aid road safety message development and evaluation. Accid. Anal. Prev. 97, 309–314. Lewis, I., White, K.M., Ho, B., Elliott, B., Watson, B., 2016c. Insights into targeting young male drivers with anti-speeding advertising: an application of the Step approach to Message Design and Testing (SatMDT). Accid. Anal. Prev. 103, 129–142. Lewis, I., Watson, B., Tay, R., 2007a. Examining the effectiveness of physical threats in road safety advertising: the role of the third-person effect, gender, and age. Transp. Res. F: Traffic Psychol. Behav. 10 (1), 48–60. http://dx.doi.org/10.1016/j.trf.2006. 05.001. Lewis, I., Watson, B., White, K.M., Tay, R., 2007b. Promoting public health messages: should we move beyond fear-evoking appeals in road safety? Qual. Health Res. 17 (1), 61–74. http://dx.doi.org/10.1177/1049732306296395. Lewis, I., Watson, B., White, K.M., 2008a. An examination of message-relevant affect in road safety messages: should road safety advertisements aim to make us feel good or bad? Transp. Res. F: Traffic Psychol. Behav. 11 (6), 403–417. http://dx.doi.org/10. 1016/j.trf.2008.03.003. Lewis, I., Watson, B., White, K.M., 2008b. Predicting future speeding behaviour: the appeal of positive emotional appeals for high risk road users. In: Paper Presented at the National Conference of the Australasian College of Road Safety and the Travelsafe Committee of the Queensland Parliament. Brisbane, QLD. . Lewis, I., Watson, B., White, K.M., 2008c. Predicting the acceptance and rejection of emotion-based anti-speeding messages: the role of attitudinal beliefs and personal involvement. In: Paper Presented at the Australasian Road Safety Research, Policing and Education Conference. Adelaide, SA. . Lewis, I., Watson, B., White, K.M., 2009. What do we really know about designing and evaluating road safety advertising? Current knowledge and future challenges. In: Paper presented at the Australasian Road Safety Research Policing and Education Conference Sydney. NSW. . Lewis, I., Watson, B., White, K.M., 2010. Response efficacy: the key to minimizing rejection and maximizing acceptance of emotion-based anti-speeding messages. Accid. Anal. Prev. 42 (2), 459–467. http://dx.doi.org/10.1016/j.aap.2009.09.008. Lewis, I., Watson, B., White, K.M., 2013a. Extending the explanatory utility of the EPPM beyond fear-based persuasion. Health Commun. 28 (1), 84–98. http://dx.doi.org/10. 1080/10410236.2013.743430. Lewis, I., Watson, B., White, K.M., Elliott, B., 2013b. The beliefs which influence young males to speed and strategies to slow them down: informing the content of antispeeding messages. Psychol. Mark. 30 (9), 826–841. http://dx.doi.org/10.1002/ mar.20648. Lewis, I., Watson, B., White, K.M., Elliott, B., Thompson, J., Cockfield, S., 2012. How males and females define speeding and how they’d feel getting caught for it: some implications for anti-speeding message development. In: Paper Presented at the Australasian Road Safety Research, Policing and Education Conference 2012. Wellington, New Zealand. . Nemme, H.E., White, K.M., 2010. Texting while driving: psychosocial influences on young people's texting intentions and behaviour. Accid. Anal. Prev. 42 (4), 1257–1265. http://dx.doi.org/10.1016/j.aap.2010.01.019.

35