Recency Effects and Hazard Monitoring on attributions of other drivers

Transportation Research Part F 39 (2016) 43–53

Contents lists available at ScienceDirect

Transportation Research Part F journal homepage: www.elsevier.com/locate/trf

The impact of Primacy/Recency Effects and Hazard Monitoring on attributions of other drivers Dwight A. Hennessy a,⇑, Robert D. Jakubowski b, Brittany Leo a a b

SUNY Buffalo State, Department of Psychology, Buffalo, NY, USA Denver Public Schools, Denver, CO, USA

a r t i c l e

i n f o

Article history: Received 8 July 2015 Received in revised form 5 February 2016 Accepted 7 March 2016 Available online 31 March 2016 Keywords: Primacy Serial position effect Hazard Monitoring Attributions Traffic

a b s t r a c t The present study examined the impact of Primacy/Recency Effects and Hazard Monitoring on driver attributions. Participants viewed a simulated near collision from the perspective of a trailing motorist. The amount of error free driving prior to the near collision varied between two groups, where the near incident occurred either early or later in their viewing experience. They were then given the opportunity to provide judgments of the offending driver based on how safe, dangerous, risky, and skilled the driver was in general, and to evaluate their overall performance. Results showed a Primacy Effect dominance in that judgments of the driver were most negative in the early group, but this was moderated by high Hazard Monitoring for ratings of ‘‘dangerous” and ‘‘safe”. This suggests that judgments of other drivers are likely to be quick and based on early information, but are impacted by personal factors such as a tendency to monitor for hazards. Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction The study of attributions has a long and rich history in the social sciences, yet has received relatively little attention in traffic research. This is an important omission given that attributions are a crucial component to the thoughts and feelings that are constructed about others in terms of who they are as a driver and the underlying causes of their actions, and can subsequently alter behavior towards those individuals. As in any social setting, drivers routinely make evaluations of themselves, traffic events, and other drivers, and negative judgments can lead to dangerous and conflictual behavior in the traffic environment. However, the traffic context is unique in many respects compared to other environments, largely due to the transitory nature of driving interactions (high speeds, combined with visual isolation and anonymity among drivers) and the elevated potential for personal danger. These factors, combined with the limited amount of information available when attributions are formed, may impact the speed, focus, impetus, process, and ultimately outcome when judging other drivers. The few studies that have examined driver attributions have typically focused on the causal attribution approach (see Malle, 2011) which attempts to identify underlying cause and distinguish responsibility for negative driving behaviors/ outcomes, particularly aggressive driving and collisions (e.g. Davies & Patel, 2005; Groeger & Grande, 1996; Kouabenan, 1998). Most have provided support for Weiner’s (1986) attribution theory in which judgments of the responsibility of driving actions are concentrated on their locus (internal/external influences), stability (consistency/variability of behavior over time), and controllability (actions due to skill or to luck/fate). Typically drivers are perceived to be responsible for their

⇑ Corresponding author. E-mail address: [email protected] (D.A. Hennessy). http://dx.doi.org/10.1016/j.trf.2016.03.001 1369-8478/Ó 2016 Elsevier Ltd. All rights reserved.

44

D.A. Hennessy et al. / Transportation Research Part F 39 (2016) 43–53

negative activities and outcomes more frequently when said behavior is judged as internal, stable, and controllable (see Lennon, Watson, Arlidge, & Fraine, 2011; Lustman, Wiesenthal, & Flett, 2010). Further, these attributions have been found to impact subsequent negative thoughts, feelings, and actions towards other drivers. For example, Wickens, Wiesenthal, Flora, and Flett (2011) found greater anger directed towards drivers viewed as responsible for offensive behavior. Similarly, aggression, retaliation and punishment towards other drivers are often reported as more appropriate and acceptable when the perpetrator is viewed as personally responsible for the instigating actions (Baldwin & Kleinke, 1994; Feather & Deverson, 2000), when those actions are viewed as typical of that driver (Britt & Garrity, 2006) and when they are perceived to be intentional (Lennon & Watson, 2011). In contrast, the current study was focused more on the inferential approach to attributions (see Malle, 2011), which has not been widely studied in the traffic context, where qualities or traits are allocated to other drivers based on their driving behavior/outcomes. In this perspective, observers attempt to determine the reason for a state action by assigning trait qualities to an actor that would define its underlying motive. The emphasis in the present study was to potentially evaluate factors that might increase the assignment of negative driving characteristics (e.g. risky, dangerous) to other drivers after witnessing their commission of an undesirable driving event (e.g. a lane violation and near collision). Specifically, these factors included cognitive bias due to timing of available information (Primacy/Recency Effects) and the tendency towards Hazard Monitoring. Cognitive biases have long been known to alter the nature of attributions, often at the expense of accuracy. Roseborough, Wiesenthal, Flett, and Cribbie (2011) argued that Just World Beliefs (the biased view that an individual’s actions will justly lead to fair and deserving consequences) impact judgments of other drivers (i.e. ‘‘bad” things happen to ‘‘bad” drivers) and subsequent anger and aggression. Walton and Bathurst (1998) found evidence of a downward comparison bias (a form of social comparison in which others are perceived more negatively to inflate relative self evaluations) where participants judged other drivers as less safe than themselves in typical speed. The Defensive Attributions Theory (a tendency to assign greater blame to others for negative events/outcomes in order to protect impressions/evaluations of the self) has been used to explain the increased propensity to blame victims of collisions when they are more severe (Walster, 1966) and for drivers to blame others more often for their own severe collisions (Stewart, 2005). Finally, several studies have found support for the Fundamental Attribution Error in the traffic environment (assigning dispositional reasons to others and situational causes to the self more frequently for the same actions/outcomes), such as when rating skill in handling various driving scenarios (Lennon et al., 2011), for general risky driving tendencies (Harré, Brandt, & Houkamau, 2004), in committing traffic violations (Baxter, Macrae, Manstead, Stradling, & Parker, 1990), and in assigning fault for simulated near collisions (Hennessy, Jakubowski, & Benedetti, 2005). Another cognitive bias that might potentially impact the attribution process in the driving environment, but has yet to receive appropriate research attention, is the Serial Position Effect which holds that the first (Primacy Effect) and last (Recency Effect) bits of information are recalled more effectively than those in the middle of a larger continuous string of information (Robinson & Brown, 1926). While the dominance of either primacy or recency undoubtedly depend on a number of factors, it has been proposed that unique memory processes account for each independently. The Primacy Effect is likely due to greater rehearsal of information experienced early in a string (Baddeley, 1986) which would take advantage of deeper processing and potential links to previous experiences or events in episodic memory. Such information would then become more distinct and stand out during processing of subsequent information (Dolenc, Bon, & Repovš, 2013). In contrast, the Recency Effect often occurs due to working memory processes (Capitani, della Sala, Logie, & Spinnler, 1992) where new information competes with and displaces older primary information, making it easier to retrieve the most current details. It should be noted that a great deal of the research evidence for the serial position effect relies on recall of strings of words, syllables, or numbers, which are important in understanding errors in language and mathematical recall. However, Primacy and Recency Effects also occur in applied and social contexts, such as in making judgments and assigning characteristics to others. Copeland, Radvansky, and Goodwin (2009) argued that the natural process of forming first impressions of others may very well be grounded in the Primacy Effect. According to Asch (1946), impressions of the personal qualities of others is often formed quickly based on limited observable traits. However, given that individuals regularly display multiple qualities simultaneously, perceivers assign meaning and value to certain traits over others, where some come to stand out as more central. The centrality of these traits is determined by the context and personal meaning or relevance (see Nauts, Langner, Huijsmans, Vonk, & Wigboldus, 2014). Primacy in this view is not necessarily temporal but rather more functional, where succeeding traits are interpreted in relation to the central trait. According to Copeland et al. (2009) the first bits of information in such situations may actually aid in the comprehension of later information. So in the traffic environment, when early contact with another driver is negative (such as seeing a lane violation and near collision), this should come to represent a central ‘‘negative” trait and other information would be understood and interpreted in relation to that personal quality (i.e. subsequent negative driving should be seen as less positive). In contrast, the reverse should be true when early contact is positive. Thus it was expected in the present study that Primacy Effect would show a dominance over Recency Effect in attributions of other drivers, where primary information would come to dominate the qualities/characteristics assigned to another driver. There is also precedence to suggest that personal factors can alter the outcomes of attributions. In fact, Weiner (1972) argued that individual differences can lead to ‘‘disparities in perceptions of causality”. For example, with respect to the driving context, judgments of intentionality of offending driving behavior have been linked to narcissism (Lustman et al., 2010) and justice sensitivity (Roseborough, 2014) among observers. As a perceptual process, judgments of others are undoubtedly

D.A. Hennessy et al. / Transportation Research Part F 39 (2016) 43–53

45

impacted by a variety of cognitive, subjective, and individual facets, and while these might include variations in motivation, needs, or experience, the emphasis in the current study will be on Hazard Monitoring which is considered a component of trait driver stress. Stress is a common experience while driving which has been linked to a number of negative outcomes, including decreased attention, mood, and vehicle control, plus increased aggression, errors and collisions (Hennessy & Wiesenthal, 1999; Matthews, Dorn, & Glendon, 1991; Matthews et al., 1998), in addition to negative evaluations of others in subsequent environments (Hennessy, 2008; Hennessy & Chilicki, 2011). Based on the Transactional Model of Stress (see Matthews, 2002), Hennessy (2011) argued that those high in trait driver stress are more prone to experience elevated levels of state stress across various traffic situations, which would alter perceptions and interpretations of their experiences of other drivers. Hazard Monitoring, which represents a hyper vigilance to personal threat or danger while driving, has been considered one potential dimension in the development of trait driver stress (Matthews, Desmond, Joyner, Carcary, & Gilliland, 1997). The resulting stress from Hazard Monitoring should enhance the dominance of the Primacy Effect in judging other drivers, due to its impact on memory. According to Suhr, Demireva, and Heffner (2008), a key component of primacy is the capacity to rehearse items encountered early on in a ‘‘string” while newer items are being presented. This would allow initial items to be more readily moved to longer term storage. However, stress hormones, such as epinephrine and cortisol, have been found to alter the memory consolidation process, increase long term memory strength of early information (Cahill & Alkire, 2003) and longer term recall of primary items (Buchanan & Lovallo, 2001). Falk and Bindra (1954) argued that primacy may also occur as a result of increased vigilance to early stimuli under stressful condition, which is a hallmark of Hazard Monitoring. According to Lewandowsky and Murdock (1989) different levels of attention could account for the serial position effect where encoding of the first bits of information demand greatest attention, with diminishing degrees available for subsequent stimuli, hence leading to greater focus on, and memory of, primary information. Given that attention is also narrowed under stressful conditions (Cohen, 1980), information about another driver experienced early in an encounter should demand greater focus and represent the bulk of cognitive processing in forming judgments. This is particularly relevant for Hazard Monitors given that such drivers are under heightened caution regarding sources of perspective danger (thus dominating an already limited attention resource), and making them potentially more prone to experience negative affect from perceived threats from other drivers, which may increase the Primacy Effect further (see Demaree, Shenal, Everhart, & Robinson, 2004). Therefore, in the present study, it was expected that high Hazard Monitoring would alter the Primacy Effect in impacting judgments of other drivers. Specifically, those who viewed a negative driving event early in an encounter would be more likely to subsequently assign negative qualities to the offending driver, but predominantly when concurrently high in Hazard Monitoring. 1.1. Hypotheses Hypothesis 1. The present study will demonstrate a Primacy Effect (as opposed to a Recency Effect) in attributions of a target driver. Hypothesis 2. Judgments of a target driver will be more negative (i.e. poorer performance, more dangerous and risky, as well as less safe and skilled) when their negative actions are viewed early in an encounter (Primacy) and the participant is also high in Hazard Monitoring.

2. Method 2.1. Participants Participants (n = 206) were recruited from a student population, who were required to possess a valid driver’s license and commute regularly to campus. All participation was voluntary, with extra credit given at the discretion of course instructors. Overall, ages ranged from 18 to 53 years (M = 22.57 years, SD = 6.24), driving experience ranged from 1 to 36 years (M = 6.23 years, SD = 6.57), and average daily driving time ranged from 3 to 150 min (M = 27.36 min, SD = 28.28). For women (n = 114), ages ranged from 18 to 49 years (M = 21.85 years, SD = 5.51), driving experience ranged from 1 to 25 years (M = 4.79 years, SD = 4.96), and average daily driving time ranged from 3 to 120 min (M = 28.63 min, SD = 28.88). For men (n = 96), ages ranged from 18 to 53 years (M = 23.40 years, SD = 6.95), driving experience ranged from 2 to 36 years (M = 8.08 years, SD = 7.84), and average daily driving time ranged from 5 to 150 min (M = 25.80 min, SD = 27.64). 2.2. Materials 2.2.1. Videos A commercial driving simulation program (Need for Speed Porsche Unleashed) was used to record a near collision between an ‘‘offending” vehicle and a ‘‘victim” vehicle which was traveling in the opposite direction, where the offending vehicle swerved partially across the center line (both left side tires crossed the solid yellow center line) into the path of the oncoming victim vehicle (without collision but creating a situation where the victim vehicle had to deviate to their right

46

D.A. Hennessy et al. / Transportation Research Part F 39 (2016) 43–53

to avoid the offending vehicle). This software allowed the recorded video to be subsequently presented to participants from the visual perspective behind the offending vehicle (i.e. as if seen through the front windshield of a vehicle following the offending vehicle) and was displayed on a 27 in. HD widescreen monitor. Two versions of the same incident were created by varying the period of time that elapsed prior to and following the near collision. In the ‘‘early” timing condition, the offending vehicle traveled without incident for 30 s prior to the near collision event, then again without incident for another 4 min following the near collision. In the ‘‘late” timing condition the offending vehicle traveled for 4 min without incident prior to the near collision, and then continued for another 30 s without incident afterwards. Control participants watched the same video sequence but without witnessing the near collision. 2.2.2. Questionnaires In order to measure attributions of the offending vehicle driver, participants were asked to rate how dangerous, risky, safe, and skilled they thought the offending driver was in general. Each item was rated using a 0–5 scale (0 = ‘‘not at all”, 5 = ‘‘extremely”). They were also asked to rate the overall performance of the offending vehicle driver using a 0–5 scale (0 = ‘‘poor”, 5 = ‘‘excellent”). In order to help avoid possible expectancy effects, these questions were included among a series of distracter items which focused on other aspects of the video that were unrelated to the near collision (e.g. rating the reality of the events and recall of other events such as memory of a building or specific signs). Hazard Monitoring was measured using an eight item subscale of the Driver Stress Inventory (DSI) which is a self-report tool designed to measures various dimensions of trait driver stress (for a fuller description see Matthews, 2002). Items consist of events which involve some level of detection or monitoring of driving related hazards, and participants are asked to indicate (using a 0–5 scale) how often they engage in such activities when they typically drive. Consistent with the original scoring scheme, a weighted average was calculated where a higher score indicated greater tendency towards Hazard Monitoring. Open ended demographic questions were also included to identify their age, driving experience, and average daily driving time. The Hazard Monitoring subscale showed good internal consistency in the present study (Cronbach Alpha = 0.84). 2.3. Procedure All participation in the present study occurred in an individual setting in order to reduce social desirability. The researcher initially started their respective video while the participant watched in isolation. Assignment to conditions was random with a balance for gender, where a separate randomized list for the three conditions was generated for each gender in order to ensure equal numbers of men and women across each condition (38 women and 32 men per condition). Following completion of the video, the researcher returned and provided the battery containing the demographic/driving information questions, Hazard Monitoring scale, and driver attribution questions which were also completed in isolation. 3. Results Data were analyzed using SPSS 22. Descriptive statistics for the five attribution items showed consistent range from 0 to 5, and fairly analogous means near the mid-point of the six point scale: Dangerous (M = 2.25, SD = 0.86), Risky (M = 2.96, SD = 1.24), Safe (M = 3.43, SD = 1.27), Skilled (M = 2.94, SD = 1.35), Performance (M = 2.70, SD = 1.27). Prior to the main analyses, the authors verified that Hazard Monitoring levels (overall M = 1.70, SD = .922) were not significantly different across age, gender, and Timing Group, and that driving characteristics (age, driving experience, average daily driving time) were not significantly different across Timing and Hazard Monitoring Groups. A series of 2  2  2  3 partial factorial ANOVAs (comprising all main effect and two way interactions) were calculated using attributions of the offending driver (dangerous, risky, skilled, safe, performance) as separate DV, and including gender (male/female), age (median split younger/older), Hazard Monitoring (median split low/high), and timing (early/late/control) as the IV. Due to the use of multiple comparisons, a Bonferroni correction (alpha = .01) was used to control Type I error. Table 1 shows that attributions of how ‘‘dangerous” the offending vehicle driver is in general was predicted by the main effects of, and the two way interaction between, Timing and Hazard Monitoring. A Tukey post hoc analysis of interaction means using a Bonferroni correction (see Table 2) revealed that ‘‘dangerous” ratings were significantly higher among High Hazard Monitors compared to Low Hazard Monitors in both Early Timing and Late Timing groups (but not Control), however this difference was greater among the Early Timing group than the Late Timing group, where the greatest ratings of danger were provided by Early Timing/High Hazard Monitoring participants. Among the High Hazard Monitoring participants, dangerous ratings were greatest in the Early Timing group followed by the Late Timing group and then the Control group, whereas among Low Hazard Monitors, Early Timing participants did show greater dangerous ratings than Control but not Late Timing participants, and ratings among the Late Timing group were no different than Control. This demonstrates that early information had an impact on attributions compared to late information, but only among those high in Hazard Monitoring. As seen in Table 3, judgments of how ‘‘risky” the offending vehicle driver is in general were impacted by the main effects of Timing and Hazard Monitoring, but not their interaction. A Tukey post hoc analysis identified that Early Timing participants provided elevated ratings of riskiness (M = 3.70/SD = .89) compared to those in the Late Timing group

47

D.A. Hennessy et al. / Transportation Research Part F 39 (2016) 43–53 Table 1 ANOVA model for attributions of ‘‘How Dangerous Is Offending Vehicle Driver”. Source

SS

df

MS

F

g2

p

Timing Hazard Monitoring Timing  Hazard Monitoring Error Total

39.02 24.90 6.34 93.54 1228.00

2 1 2 204 210

19.51 24.90 3.17 .459

42.55 54.33 6.91

.294 .210 .064

.000 .000 .003

Table 2 Tukey post hoc of ‘‘Dangerous” attributions means across levels of Timing  Hazard Monitoring.

*

Timing

Hazard Monitoring

Control Early Late

High–low High–low High–low

1.61–1.84 1.94–2.70 2.25 – 3.33

Mean difference

Standard error

0.23 0.76* 1.08*

.163 .162 .164

Hazard Monitoring

Timing

Mean difference

Standard error

Low

Early–control Late–control Early–late

2.25–1.61 1.94–1.61 2.25–1.94

0.64* 0.33 0.31

.162 .166 .156

High

Early–control Late–control Early–late

3.33–1.84 2.70–1.84 3.33–2.70

1.49* 0.86* 0.63*

.164 .159 .170

p < .05.

Table 3 ANOVA model for attributions of ‘‘How Risky Is Offending Vehicle Driver”. Source

SS

df

MS

F

g2

p

Timing Hazard Monitoring Timing  Hazard Monitoring Error Total

130.00 11.29 1.93 186.07 2166.00

2 1 2 204 210

65.00 11.29 .968 .912

71.26 12.38 1.06

.411 .057 .010

.000 .001 .348

(M = 3.28/SD = .96), which were significantly greater than the Control group (M = 1.90/SD = 1.07). The High Hazard Monitoring group also reported the offending vehicle driver as more risky (M = 3.11/SD = 1.22) than the Low Hazard Monitoring group (M = 2.81/SD = 1.25). Attributions of how ‘‘safe” the offending vehicle driver is in general were predicted by the main effects of, and the two way interaction between, Timing and Hazard Monitoring (see Table 4). A Tukey post hoc analysis of interaction means using a Bonferroni adjustment (see Table 5) showed that judgments of offending vehicle driver safety were significantly lower for the Early Timing group than the Late Timing group, which in turn were lower than the Control group for both Low Hazard Monitoring and High Hazard Monitoring participants. However, this effect was more pronounced among High Hazard Monitors. Lowest offending vehicle driver safety ratings were offered by those who viewed the near collision early, but particularly when concurrently high in Hazard Monitoring. Ratings of offending vehicle driver ‘‘skill” were only predicted by the main effect of Timing (see Table 6). A Tukey analysis showed that the Early Timing group rated the offending vehicle driver as least skilled (M = 2.21/SD = 1.23); significantly lower than Late Timing participants (M = 2.68/SD = 1.24), which was subsequently significantly lower than the Control group (M = 3.94/SD = .91). While Hazard Monitoring did show a trend towards impacting attributions of skill, it did not reach significance based on the Bonferroni correction.

Table 4 ANOVA model for attributions of ‘‘How Safe Is Offending Vehicle Driver”. Source

SS

df

MS

F

g2

p

Timing Hazard Monitoring Timing  Hazard Monitoring Error Total

160.49 6.69 8.33 170.25 2822.00

2 1 2 204 210

80.24 6.69 4.16 .851

96.15 8.02 4.99

.485 .038 .047

.000 .005 .008

48

D.A. Hennessy et al. / Transportation Research Part F 39 (2016) 43–53

Table 5 Tukey post hoc of ‘‘Safe” attributions means across levels of Timing  Hazard Monitoring.

*

Timing

Hazard Monitoring

Control Early Late

High–low High–low High–low

Mean difference 4.51–4.41 1.86–2.75 3.32–3.61

Standard error

0.10 0.89* 0.29

.220 .221 .218

Hazard Monitoring

Timing

Low

Early–control Late–control Early–late

2.75–4.41 3.61–4.41 2.75–3.61

Mean difference 1.66* 0.80* 0.86*

.219 .224 .210

Standard error

High

Early–control Late–control Early–late

1.86–4.51 3.32–4.51 1.86–3.32

2.65* 1.19* 1.46*

.222 .214 .229

p < .05.

Table 6 ANOVA model for attributions of ‘‘How Skilled is Offending Vehicle Driver”. Source

SS

df

MS

F

g2

p

Timing Hazard Monitoring Timing  Hazard Monitoring Error Total

114.81 6.58 .583 263.45 2207.00

2 1 2 204 210

57.40 6.55 .292 1.29

44.45 5.10 .226

.304 .024 .002

.000 .025 .798

Table 7 ANOVA model for attributions of offending vehicle driver performance. Source

SS

df

MS

F

g2

p

Timing Hazard Monitoring Timing  Hazard Monitoring Error Total

82.53 .353 3.09 251.59 1876.00

2 1 2 204 210

41.26 .353 1.54 1.23

33.46 .286 1.25

.247 .001 .012

.000 .593 .288

Finally, judgments of the offending vehicle driver ‘‘performance” was impacted only by the main effect of Timing (see Table 7). Based on a Tukey analysis, those in the Control group rated higher performance (M = 3.58/SD = .73) than Early Timing (M = 2.11/SD = 1.29) and Late Timing participants (M = 2.41/SD = 1.22), but Early Timing and Late Timing did not significantly differ from one another. 4. Discussion 4.1. Primacy vs recency effects With respect to Hypothesis 1, the present outcomes generally provided support for a Primacy Effect in driver attributions. Specifically, primacy information had a greater impact than recency information on judgments of the target driver in the simulated near collision. Those placed in the Early Timing group were exposed to the near collision early in their encounter, which appeared to have a persistent negative impact on their longer term interpretations of the offending vehicle driver, even during a much longer period of error free driving afterwards, ultimately leading to poorer attributions of the offending vehicle driver relative to those of the Late Timing Group for four of the five attributes (more risky and dangerous, less safe and skilled). In a similar respect, those in the Late Timing group viewed problem free driving early in their exposure which appeared to temper negative evaluations of the offending vehicle driver on those same four factors compared to the Early Timing group. This suggests that drivers may be more prone to make quick judgments of other drivers early in their exposure, and based on relatively little information, rather than weigh the aggregate evidence of their total experience. Eyal, Hoover, Fujita, and Nussbaum (2011) argued that schemas invoked during early exposure can alter the attention to, expectations for, and priority of processing of subsequently experienced traits in others; thus the judgments of later actions would be filtered through these initially activated schemas. This is important given the prevalence of the stereotype within the general driving population that negative driving actions and outcomes stem from internal or personal factors, such as lack of skill, accident proneness, and risk taking, rather than situational factors (see McKenna, 1983; Rothe, 1994; Rumar,

D.A. Hennessy et al. / Transportation Research Part F 39 (2016) 43–53

49

1988). Hence, particularly for those in the Early Timing group, activation of a schema which posits that ‘‘dangerous drivers do dangerous things” after viewing the initial near collision relatively quickly, would dominate impression formation efforts even during the subsequent prolonged exposure to error/risk free driving. In fitting with the availability heuristic, which is typically activated outside of conscious awareness, the use of this ‘‘bad driver” schema would ease cognitive load by speeding processing of the offending vehicle driver’s subsequent actions, but compromise accuracy that comes with more intentional and effortful processing of state information (Nisbett & Ross, 1980). Therefore, quick judgments may not be truly representative of the offending vehicle driver, reflecting overly negative evaluations despite a preponderance of positive driving/performance evidence. This tendency may be exaggerated by Belief Perseverance (Anderson, Lepper, & Ross, 1980) which represents a fairly natural tendency for individuals to adhere to a set of ideas or beliefs they have constructed (e.g. those created when making initial judgments of the offending vehicle driver), even in the face of contradictory evidence (e.g. subsequent actions of the offending vehicle driver). The persistence of these initial judgments has been found to increase under conditions where causal reasoning has been used in their construction (Anderson & Kellam, 1992), such as in the case of forming attributions of why the offending vehicle driver would have been involved in the near collision early on for the Early Timing group. In fact, in some instances, individuals may even assimilate or alter newly received information in order to make it consistent with their initial evaluations (Richter & Kruglanski, 1998), such as could happen during exposure to subsequent positive driving among the Early Timing group, thus maintaining consistency and persistence in their initial beliefs. Another explanation for the dominance of the Primacy Effect in the present study could be the influence of diminishing attention whereby little consideration is given to subsequent traits when the initially revealed traits are viewed as definitive of that individual in that situation (Anderson, 1981). In this way participants in the Early Timing group would not have processed or considered the positive qualities as being more representative of the offending vehicle driver (even though comprising much more of their overall encounter) in lieu of the initial negative qualities, and vice versa for those in the Late Timing group. Another explanation could be the meaning change hypothesis which holds that the interpretation of subsequent traits is changed or altered to match initially presented traits. Thus when negative traits are displayed (such as ‘‘dangerous driver”) before more positive ones (such as ‘‘safe driver”), the subsequent positive traits are viewed as less positive (Hamilton & Zanna, 1974). The reverse could be argued for those in the Late Timing group as well. As a result, after viewing the near collision early on, the subsequent positive driving would be interpreted as less positive for the Early Timing group, while for the Late Timing group, a longer initial positive experience could make the subsequent near collision appear less negative, leading to less ‘‘negative” judgments of the offending vehicle driver. It should be noted, however, that no serial position effect was found for ratings of the offending driver’s ‘‘performance” in that neither a Primacy nor Recency Effect was found. Control participants did rate performance higher than both Early and Late Timing groups, but that would be expected given that they did not view the near collision. One reason why ratings among Early and Late Timing groups did not differ from one another could be that ‘‘performance” may have been perceived as more of a situational or state action, while ‘‘dangerous”, ‘‘risky”, ‘‘safe”, ‘‘skilled” were interpreted as more enduring dispositional or trait components. This is important because the inferential approach emphasizes that attributions are based on the assignment of traits as a way of explaining the actions of others. In this respect, if performance was not perceived as a trait, then participant ratings may not have been part of the attribution process or impacted in the same way as the other four attributes. Jones, Rock, Shaver, Goethals, and Ward (1968) have noted that situational and dispositional aspects of judgments are part of unique processes and do lead to different outcomes. As a result, performance could be rated lower by Early and Late Timing participants compared to the Control (who did not view the near collision) without being part of the inferential attribution process; hence, timing of the event (or the serial position effect overall) would not impact attributions in this instance. 4.2. Hazard monitoring The present study only partially supported Hypothesis 2, where Hazard Monitoring exaggerated the negative evaluations of drivers in the Early Timing group for only two of the five attributes (more ‘‘dangerous” and less ‘‘safe”). However, despite the absence of an interaction with Timing, High Hazard Monitoring was also found to independently lead to more negative ratings of ‘‘riskiness”. Thus the present study did demonstrate that Hazard Monitoring does have some impact on the driver attribution process. Given that Hazard Monitoring involves an elevated vigilance to danger while driving (Matthews, 2002), this heightened monitoring may lead to increased efforts to detect and remember signs of potential danger in other drivers. Thus, a near collision by another driver such as that experienced by the offending vehicle driver may increase the tendency for high Hazard Monitors to focus on and remember the incident because it represents, in their view, a potential future threat to themselves. With respect to judgments of how ‘‘dangerous” the offending vehicle driver is in general, High Hazard Monitoring increased ratings of dangerousness in both Early Timing and Late Timing groups, although the effect was more pronounced among Early Timing participants; with this group providing the highest ratings of dangerousness overall. In a similar respect, ratings of how ‘‘safe” the offending vehicle driver is in general, the Early Timing group provided lower ratings overall compared to Late Timing and Control groups. However, High Hazard Monitoring exaggerated the negative attributions of the Early Timing group, leading to the lowest ratings of safety overall. Further, while ratings did not differ from Control for Late Timing/Low Hazard Monitoring participants, those in the Late Timing/High Hazard Monitoring group did provide

50

D.A. Hennessy et al. / Transportation Research Part F 39 (2016) 43–53

significantly lower safety judgments compared to the Control (although still higher than Early Timing/High Hazard Monitoring participants). Thus, to some degree in the Early Timing group in particular, elevated Hazard Monitoring appeared to have magnified the already existing tendency to focus on, and report negative ratings of (more dangerous and less safe), the offending vehicle driver. More specifically, for those who viewed the near collision early and were concurrently high in Hazard Monitoring, the tendency to focus on potential danger and unsafe behavior from others may have inflated negative attributions even further because attention to the offending vehicle would have been activated relatively immediately in their exposure, which may have been more vivid and resolute due to the fact that they had just encountered a new potential ‘‘threat”. In addition, given their increased perception of negative experiences with other drivers in past encounters, their schemas of dangerous drivers may have been activated more quickly and intensely compared to those lower in Hazard Monitoring. As a result, they may have focused more on the details of the event, had a greater desire or need to make causal attributions of the offending vehicle driver’s actions due to self preservation and personal safety, solidifying the perseverance of their negative attribution to a greater degree. Even when confronted with contrary evidence that the offending vehicle driver was not currently driving in an unsafe manner, they would be more inclined to maintain their negative attribution because the offending vehicle driver had already established themselves as a potential risk, and thus the safest assumption for an individual overly concerned with monitoring for danger would be ‘‘once a threat, always a threat”. It is important to note that High Hazard Monitoring also had some impact on attributions of danger and safety among those in the Late Timing group in comparison to the Control group (albeit to a lesser extent than Early Timing/High Hazard Monitoring participants). One explanation is that the Recency Effect is due to working memory processes, given that a more current emphasis and heightened attention to ‘‘danger” among those high in Hazard Monitoring could potentially supplant previous or primary ‘‘harmless driver” information. This memory prominence could then lead to increased negative evaluations of ‘‘more dangerous” and ‘‘less safe”. However, for those low in Hazard Monitoring, who would not be excessively focused on threat from others, viewing positive driving for the majority of exposure to the offending vehicle driver may have been sufficient to activate a positive schema of the newly encountered driver or, as suggested by the meaning change hypothesis (Hamilton & Zanna, 1974), to alter the interpretation of the subsequent negative event (which was far outweighed in proportion of driving by positive/safe actions) to be ‘‘less negative”. In contrast, the need to guard against potential sources of danger among Late Timing/High Hazard Monitoring participants appears to have been sufficient to alter this process; to discount the bulk of the positive encounters to the extent that the offending vehicle driver crossed a threshold of representing a potential personal ‘‘threat”. Another explanation for the impact of Hazard Monitoring among Late Timing participants could be based on the Safety Margin Model (Summala, 2005), which posits that when drivers approach or exceed some learned boundary of safety (e.g. traffic moves more quickly or other vehicles are closer than desired for the conditions), they experience uncomfortable feelings, which then might alter subsequent affect, cognition, and actions. Drivers are impacted and act on these external driving stimuli only after such a threshold is reached. Implied in this is an individual difference aspect where the safety margin is based on personal experience and resulting driving personality. In this respect, a tendency to hazard monitor may have lowered the threshold of risk, activated negative feelings more readily, and increased negative judgments of the offending vehicle driver among many in the Late Timing/High Hazard Monitoring group. It should be noted as well, that Hazard Monitoring did not impact attributions for ratings of ‘‘performance” or ‘‘skill”, which does not support Hypothesis 2. With respect to ‘‘performance”, to some degree a similar argument could be made as with the Primacy/Recency Effects regarding the fact that performance may represent a state outcome rather than a trait characteristic which might alter the attribution process. However, given the heightened vigilance to personal threat among Hazard Monitors, some aspects of state performance should be of concern to this group. It is possible that since the ratings were offered several minutes after the actions were complete and no longer of personal relevance (and hence no need to ‘‘monitor”), neither dispositional nor situational cues were pressing or distinctive at the time of the ratings. In addition, the fact that they were not in any real danger during the incident due to the simulated presentation, may have minimized the relevance of state outcomes as a real threat to be monitored. In contrast, for ratings of ‘‘skill”, under a more liberal evaluation (alpha = .05) rather than the more conservative Bonferroni approach used here (alpha = .01), a significant main effect for Hazard Monitoring would have been identified. None the less, this may simply indicate that not all potential driving qualities are equally important to the search patterns of Hazard Monitors. Perhaps concepts such as danger, safety, and riskiness are more dominant in the schema of Hazard Monitors because, semantically, they stimulate more thoughts and emotions of potential threat and may be more reflective of dubious intentions among menacing drivers. Skill, on the other hand, represents a learned outcome pattern and represents a range of abilities at certain tasks (Haibach, Reid, & Collier, 2011). In this respect, a driver can be lower in skill and not necessarily place others in peril, whereas a dangerous, unsafe, or risky driver may be perceived as posing a more immediate personal threat and thus capture more attention and cognition from a high Hazard Monitor, ultimately leading to more negative evaluations. 4.3. Alternate explanations and future directions It should be noted that the methodology used in the present study did not encourage step by step processing in that participants did not need to make continual judgments of other drivers in real time over the course of multiple events, as might be warranted in actual driving conditions. According to the anchor-adjustment model (Hogarth & Einhorn, 1992) after an

D.A. Hennessy et al. / Transportation Research Part F 39 (2016) 43–53

51

initial belief is formed, adjustments are made based on new information. During step by step processing the Recency Effect is more dominant because the last piece of information will be given more weight in the judgment process. However, Hogarth and Einhorn (1992) did note that Primacy Effects could occur during step by step processing when the string of information was lengthy, which perhaps indicates greater demand on memory and processing capacity as would be more typical of driving scenes compared to strings of numbers or letters. A better understanding of the procedures used by participants over the course of the judgment made during driving events is needed in future research where multiple points of rating are gathered over a longer driving encounter. Another potential issue may be that the Late Timing group did not have as much time to process the near collision as did the Early Timing group, and with more time, may have come to a similar level of negative judgment. However previous research has suggested that attributions are typically constructed quickly, particularly during short term contact (Fiske, 2004; Gilbert & Osborne, 1989) which is consistent with Hennessy et al. (2005) who found that judgments of other drivers were present quickly following exposure to negative driving events. Further, the fact that those in the Late Timing/High Hazard Monitoring group provided significantly more negative attributions compared to the Control participants for some attributes suggests that they had sufficient time to create judgments during the allotted time frame. Alternatively it could be argued that there was too much time between the occurrence of the negative event and the process used to measure attributions, resulting in a threat to ecological validity. According to Gilbert and Malone (1995), after making quick attributions about others, there is typically a second stage of evaluation in which initial judgments of others are adjusted by contemplating possible situational forces. In this respect, negative attributions may have been understated in the current study as participants may have had time to adjust their initial judgments in a manner inconsistent with the attribution process during real driving situations, in which evaluations occur and responses are necessarily formulated rapidly and immediately following such incidents. Future research may benefit from attempting to measure the attribution process as it occurs, changes, and unfolds over time by using real time methods, such as think aloud protocols, and state measures, including tape recordings or videotaping. Another consideration is the fact that the present study did not include scenes of positive driving actions (e.g. slowing to accommodate merging traffic, showing gratitude by waving, or allowing another driver to proceed through an intersection). As previously stated, perceptions and processing of the near collision would likely have been impacted by events in the error free driving condition, and vice versa. Given that the error free driving was largely benign and uneventful, participant attention may have waned at that point, even temporarily, which might have altered processing of the contrasting subsequent near collision and impacted attributions of the driver significantly. However, it is not clear from the present study to what extent positive driving actions, as opposed to simply ‘‘error free” driving, might impact attributions; especially when experienced early. If the meaning change hypothesis is correct then perhaps positive driving actions could drastically modify evaluations and judgments of subsequent adverse actions and minimize negative attributions of those drivers. This could have important implications for understanding and improving inter-driver interactions and is thus worthy of further research investigation. It is also possible that the order of the questions presented in the battery, which were completed after viewing the video, may have created a priming effect. In other words, being asked general questions about personal patterns of Hazard Monitoring prior to more specific questions about the target driver in the video could potentially have stimulated an expectancy effect or primed more negative memories of the event which could exaggerate negative judgments. However, a reversed presentation would have simply present a similar issue with specific attribution questions potentially altering Hazard Monitoring scores, and if presented prior to the video, the Hazard Monitoring scale would likely have altered how participants viewed the events in the video. Finally, it must be noted that the sample was, on average, relatively young. This is relevant given that numerous negative driving processes have been linked to younger drivers. In particular, the experience and attribution processes of older drivers may be distinct from those of younger drivers (Holland, 1993) and novice drivers tend to monitor hazards differently than older drivers, such as detecting risk less quickly and assuming less personal danger across driving situation (Deery, 1999). As a result, the perception of the current events and the attributions of the offending vehicle driver may be unique among an older sample of drivers, therefore future research on driver attributions may benefit from a broader age representation and a focus on age related differences. 4.4. Conclusions The present study added to the growing research on driver attributions by demonstrating the presence of a Primacy Effect. Drivers appeared to make snap judgments in projecting dispositional qualities onto a target driver based on limited information encountered early in their experience rather than engage in active processing of their total experience with that diver. Further, the personal tendency to Hazard Monitor was found to amplify the Primacy Effect for some driver characteristics, highlighting the need to involve individual difference factors in interpreting the driver attribution process. These outcomes are relevant in the actual traffic context given that the amount of information available when attributions are formed can be extremely limited, creating circumstances conducive to judgment error and bias. Further, negative attributions of other drivers are often linked to subsequent dangerous attitudes and actions, such as downgraded perceptions of other drivers’ skill (Walton & Bathurst, 1998), elevated ratings of other drivers’ riskiness (Hennessy & Jakubowski, 2007), increased vengeance and aggression (Britt & Garrity, 2006; Wiesenthal, Hennessy, & Gibson, 2000),

52

D.A. Hennessy et al. / Transportation Research Part F 39 (2016) 43–53

decreased belief in personal collisions likelihood (Svenson, Fischhoff, & MacGregor, 1985), and reduced belief in personal risk of injury from traffic collisions (Dalziel & Job, 1997); all of which increase the potential for widespread danger and conflict in the traffic environment. Consistent with the Fundamental Attribution Error, Hennessy and Jakubowski (2007) argued that drivers are prone to attribute actions of other drivers (particularly negative activities) to internal qualities but provide more situational attributes for their own driving behavior. As such, viewing negative driving, in particular, from another driver early in an encounter could lead to the judgment that the action is representative of a stable and internal quality of ‘‘bad” or ‘‘unskilled” or ‘‘dangerous” driver. Such attributions could then alter how they interpret (e.g. the action was intentional, personal, threatening) and respond to (e.g. anger, aggression, vengeance) subsequent actions by that driver, thus leading to driving safety concerns. In the end, attributions are a complex experience with potential real life consequences, thus further research is still needed to better understand the development, process and progression of driving attributions, and potential implications for the thoughts, feelings, and actions of drivers. References Anderson, N. (1981). Foundations of information integration theory.Boston, MA: Academic Press. Anderson, C. A., & Kellam, K. L. (1992). Belief perseverance, biased assimilation, and covariation detection: The effects of hypothetical social theories and new data. Personality and Social Psychology Bulletin, 18, 555–565. Anderson, C. A., Lepper, M. R., & Ross, L. (1980). Perseverance of social theories: The role of explanation in the persistence of discredited information. Journal of Personality and Social Psychology, 39, 1037–1049. Asch, S. E. (1946). Forming impressions of personality. Journal of Abnormal and Social Psychology, 41, 258–290. Baddeley, A. D. (1986). Working memory.Oxford, England: Oxford University Press. Baldwin, M. R., & Kleinke, C. L. (1994). Effects of severity of accident, intent, and ‘‘alcoholism is a disease” excuse on judgments of a drink driver. Journal of Applied Social Psychology, 24, 2097–2109. Baxter, J. S., Macrae, C. N., Manstead, A. S., Stradling, S. G., & Parker, D. (1990). Attributional biases and driver behaviour. Social Behaviour, 5, 185–192. Britt, T. W., & Garrity, M. J. (2006). Attributions and personality as predictors of the road rage response. British Journal of Social Psychology, 45, 127–147. Buchanan, T. W., & Lovallo, W. R. (2001). Enhanced memory for emotional material following stress-level cortisol treatment in humans. Psychoneuroendocrinology, 26, 307–317. Cahill, L., & Alkire, M. T. (2003). Epinephrine enhancement of human memory consolidation: Interaction with arousal at encoding. Neurobiology of Learning and Memory, 79, 194–198. Capitani, E., della Sala, S., Logie, R. H., & Spinnler, H. (1992). Recency, primacy, and memory: Reappraising and standardising the serial position curve. Cortex: A Journal Devoted to the Study of the Nervous System and Behavior, 28, 315–342. Cohen, S. (1980). Aftereffects of stress on human performance and social behavior: A review of research and theory. Psychological Bulletin, 88, 82–108. Copeland, D. E., Radvansky, G. A., & Goodwin, K. A. (2009). A novel study: Forgetting curves and the reminiscence bump. Memory, 17, 323–336. Dalziel, J. R., & Job, R. F. S. (1997). Motor vehicle accidents, fatigue and optimism bias in taxi drivers. Accident Analysis & Prevention, 29, 489–494. Davies, G., & Patel, D. (2005). The influence of car and driver stereotypes on attributions of vehicle speed, position on the road and culpability in a road accident scenario. Legal and Criminological Psychology, 10, 45–62. Deery, H. A. (1999). Hazard and risk perception among young novice drivers. Journal of Safety Research, 30, 225–236. Demaree, H. A., Shenal, B. V., Everhart, D. E., & Robinson, J. L. (2004). Primacy and recency effects found using affective word lists. Cognitive and Behavioral Neurology, 17, 102–108. Dolenc, B., Bon, J., & Repovš, G. (2013). Serial position and distance effects in visual working memory. Studia Psychologica, 55, 67–82. Eyal, T., Hoover, G. M., Fujita, K., & Nussbaum, S. (2011). The effects of distance-dependent construals on schema-driven impression formation. Journal of Experimental Social Psychology, 47, 278–281. Falk, J. L., & Bindra, D. (1954). Judgments of time as a function of serial position and stress. Journal of Experimental Psychology, 47, 279–282. Feather, N. T., & Deverson, N. H. (2000). Reactions to a motor-vehicle accident in relation to mitigating circumstances and the gender and moral worth of the driver. Journal of Applied Social Psychology, 30, 77–95. Fiske, S. T. (2004). Social beings: A core motives approach to social psychology.Hoboken, NJ: Wiley. Gilbert, D. T., & Malone, P. S. (1995). The correspondence bias. Psychological Bulletin, 117, 21–38. Gilbert, D. T., & Osborne, R. E. (1989). Thinking backwards: Some curable and incurable consequences of cognitive busyness. Journal of Personality and Social Psychology, 57, 940–949. Groeger, J. A., & Grande, G. E. (1996). Self-preserving assessments of skill. British Journal of Psychology, 87, 61–79. Haibach, P. S., Reid, G., & Collier, D. H. (2011). Motor learning and development.Champaign, IL: Human Kinetics. Hamilton, D. L., & Zanna, M. P. (1974). Context effects in impression formation: Changes in connotative meaning. Journal of Personality and Social Psychology, 29, 649–654. Harré, N., Brandt, T., & Houkamau, C. (2004). An examination of the Actor-Observer effect in young drivers’ attributions for their own and their friends’ risky driving. Journal of Applied Social Psychology, 34, 806–824. Hennessy, D. A. (2008). The impact of commuter stress on workplace aggression. Journal of Applied Social Psychology, 38, 2315–2335. Hennessy, D. A., & Chilicki, J. (2011). Name discrimination of potential job applicants: The spillover of commuter stress and modern racist attitudes. In D. A. Hennessy (Ed.), Traffic psychology: An international perspective (pp. 175–194). New York: Nova Science Publishers. Hennessy, D. A., & Jakubowski, R. (2007). The impact of visual perspective and anger on the actor-observer bias among automobile drivers. Traffic Injury Prevention, 8, 115–122. Hennessy, D. A., Jakubowski, R., & Benedetti, A. (2005). The influence of the actor–observer bias on attributions of other drivers. In D. A. Hennessy & D. L. Wiesenthal (Eds.), Contemporary issues in road user behavior and traffic safety research (pp. 13–20). New York: Nova Science Publishers. Hennessy, D. A. (2011). Social, personality, and affective constructs in driving. In B. Porter (Ed.), Handbook of traffic psychology (pp. 149–164). New York: Elsevier. Hennessy, D. A., & Wiesenthal, D. L. (1999). Traffic congestion, driver stress, and driver aggression. Aggressive Behavior, 25, 409–423. Hogarth, R. M., & Einhorn, H. J. (1992). Order effects in belief updating: The belief-adjustment model. Cognitive Psychology, 24, 1–55. Holland, C. A. (1993). Self bias in older drivers’ judgments of accident likelihood. Accident Analysis & Prevention, 25, 431–441. Jones, E. E., Rock, L., Shaver, K. G., Goethals, G. R., & Ward, L. M. (1968). Pattern performance and ability attributions: An unexpected primacy effect. Journal of Personality and Social Psychology, 10, 317–340. Kouabenan, D. R. (1998). Beliefs and the perception of risks and accidents. Risk Analysis, 18, 243–252. Lennon, A., & Watson, B. (2011). ‘‘Teaching them a lesson?” A qualitative exploration of underlying motivations for driver aggression. Accident Analysis & Prevention, 43, 2200–2208. Lennon, A., Watson, B., Arlidge, C., & Fraine, G. (2011). ‘You’re a bad driver but I just made a mistake’: Attribution differences between the ‘victims’ and ‘perpetrators’ of scenario-based aggressive driving incidents. Transportation Research Part F: Traffic Psychology and Behaviour, 14, 209–221. Lewandowsky, S., & Murdock, B. B. (1989). Memory for serial order. Psychological Review, 96, 25–57.

D.A. Hennessy et al. / Transportation Research Part F 39 (2016) 43–53

53

Lustman, M., Wiesenthal, D. L., & Flett, G. L. (2010). Narcissism and aggressive driving: Is an inflated view of the self a road hazard? Journal of Applied Social Psychology, 40, 1423–1449. Malle, B. F. (2011). Attribution theories: How people make sense of behavior. In D. Chadee (Ed.), Theories in social psychology (pp. 72–95). New York: WileyBlackwell. Matthews, G. (2002). Toward a transactional ergonomics for driver stress and fatigue. Theoretical Issues in Ergonomics Science, 3, 195–211. Matthews, G., Desmond, P. A, Joyner, L. A, Carcary, B., & Gilliland, K. (1997). A comprehensive questionnaire measure of driver stress and affect. In J. A. Rothengatter & E. J. Carbonell Vay’a (Eds.), Traffic and transport psychology: Theory and applications (pp. 317–324). Amsterdam: Pergamon. Matthews, G., Dorn, L., & Glendon, A. I. (1991). Personality-correlates of driver stress. Personality and Individual Differences, 12, 535–549. Matthews, G., Dorn, L., Hoyes, T. W., Davies, D. R., Glendon, A. I., & Taylor, R. G. (1998). Driver stress and performance on a driving simulator. Human Factors, 40, 136–149. McKenna, F. P. (1983). Accident proneness: A conceptual analysis. Accident Analysis & Prevention, 15, 65–71. Nauts, S., Langner, O., Huijsmans, I., Vonk, R., & Wigboldus, D. H. J. (2014). Forming impressions of personality: A replication and review of Asch’s (1946) evidence for a primacy-of-warmth effect in impression formation. Social Psychology, 45, 153–163. Nisbett, R., & Ross, L. (1980). Human inferences: Strategies and shortcomings of social judgment.Englewood Cliffs, NJ: Prentice Hall. Richter, L., & Kruglanski, A. W. (1998). Seizing on the latest: Motivationally driven recency effects in impression formation. Journal of Experimental Social Psychology, 34, 313–329. Robinson, E. S., & Brown, M. A. (1926). Effects of serial position on memorization. The American Journal of Psychology, 37, 538–552. Roseborough, J. (2014). Retaliatory aggressive driving: A justice perspective (unpublished doctoral dissertation).Toronto Canada: York University. Roseborough, J., Wiesenthal, D., Flett, G., & Cribbie, R. (2011). The belief in an unjust world and narcissism-the influence of personality on perceptions of injustice, driving anger and aggressive driving. Proceedings of the XXI Canadian multidisciplinary road safety conference.Canada: Transport Canada and Canadian Association of Road Safety Professionals. Rothe, J. P. (1994). Beyond traffic safety.New Brunswick, NJ: Transaction Publishers. Rumar, K. (1988). Collective risk but individual safety. Ergonomics, 31, 507–518. Stewart, A. E. (2005). Attributions of responsibility for motor vehicle crashes. Accident Analysis and Prevention, 37, 681–688. Suhr, J., Demireva, P., & Heffner, K. (2008). The relation of salivary cortisol to patterns of performance on a word list learning task in healthy older adults. Psychoneuroendocrinology, 33, 1293–1296. Summala, H. (2005). Traffic psychology theories: Towards understanding driving behaviour and safety efforts. In G. Underwood (Ed.), Traffic and transport psychology: Theory and application (pp. 383–394). New York: Elsevier. Svenson, O., Fischhoff, B., & MacGregor, D. (1985). Perceived driving safety and seatbelt usage. Accident Analysis & Prevention, 17, 119–135. Walster, E. (1966). Assignment of responsibility for an accident. Journal of Personality and Social Psychology, 3, 73–79. Walton, D., & Bathurst, J. (1998). An exploration of the perceptions of the average driver’s speed compared to perceived driver safety and driving skill. Accident Analysis & Prevention, 30, 821–830. Weiner, B. (1972). Attribution theory, achievement motivation, and the educational process. Review of Educational Research, 42, 203–215. Weiner, B. (1986). An attributional theory of motivation and emotion.New York, NY: Springer-Verlag. Wickens, C. M., Wiesenthal, D. L., Flora, D. B., & Flett, G. L. (2011). Understanding driver anger and aggression: Attributional theory in the driving environment. Journal of Experimental Psychology: Applied, 17, 354–370. Wiesenthal, D. L., Hennessy, D. A., & Gibson, P. M. (2000). The Driving Vengeance Questionnaire (DVQ): The development of a scale to measure deviant drivers’ attitudes. Violence and Victims, 15, 115–136.