UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Facets of anger : inter-relations and relations to driving behaviours Young, Sandra N. 2003-10-30

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata


831-ubc_2003-0544.pdf [ 4.05MB ]
JSON: 831-1.0091014.json
JSON-LD: 831-1.0091014-ld.json
RDF/XML (Pretty): 831-1.0091014-rdf.xml
RDF/JSON: 831-1.0091014-rdf.json
Turtle: 831-1.0091014-turtle.txt
N-Triples: 831-1.0091014-rdf-ntriples.txt
Original Record: 831-1.0091014-source.json
Full Text

Full Text

FACETS OF ANGER: INTER-RELATIONS AND RELATIONS TO DRIVING BEHAVIOURS by SANDRA N. YOUNG B.Sc, University of British Columbia, 1992 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER; OF ARTS in THE FACULTY OF GRADUATE STUDIES Department of Psychology We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA September 2003 © Sandra N. Young, 2003 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. The University of British Columbia Vancouver, Canada Date Octobt^ Q*j yj)QZ ABSTRACT This thesis examines the relationship of anger expression with other facets of the anger construct and applies these to account for individual differences in risky driving behaviors. Anger expression as measured by the Behavioral Anger Response Questionnaire is compared to measures of hostile attitude, trait driving anger, Type-A personality, and anxiety to further evaluate the construct validity for this new measure of anger expression as it relates to anger level at large and to angry driving. Driving-related anger is then related to traffic violations and motor vehicle accidents given that anger is know to contribute to risky driving and that the resulting traffic accidents are one of the leading causes of disability and death in our society. A large sample (N = 316) of active drivers of varying ages (range = 17-67 years) filled out a questionnaire package containing measures of driving anger, hostile attitude, Type-A personality, anger expression as measured by the Behavioral Anger Response Questionnaire, demographics, and driving behaviours. Analyses were first conducted to test the nomological network of general and driving related anger variables and differentiate anger level from anger expression. Driving anger was found to be related to hostile attitude (r = .33, p<.001), anxiety (r = .29, p<.00X), anger out (r = .33, p>00\), and rumination (r = .24, p<.00\). As for driving behaviours, more men reported receiving tickets for moving violations (X2(l) = 15.58, p<.001), and being involved in minor (X2(l) = 4.51, p<025), and major motor vehicle accidents (X2(l) = 5.95, p<.025) over the past five years, but not motor vehicle accidents overall (X2(l) = 3.56, p>.05). A greater number of participants under 30 years of age reported involvement in motor vehicle accidents as well (X2(l) = 11.77, p<.001). Next, driving anger related variables were used to test predictive models of who received tickets for moving violations and who has an accident history. After controlling for age, gender, and hours driven per week, none of the psychological variables or tested interactions of these variables with age or gender predicted the receipt of tickets for moving violations, or major motor vehicle accident involvement. The receipt of tickets for moving violations, however, along with an interaction between anger out, age and gender, predicted MVA involvement (X2(9) = 36.52, p<.001), for women under 30 years of age (odds = 1.47 for a lsd increase in anger out). IV TABLE OF CONTENTS Abstract 11 List of Tables vi List of Figures m Introduction 1 The Conceptualization of Anger 1 Anger Expression 4 Multidimensional Forms of Anger Expression 6 BARQ 6 Motor Vehicle Accidents 9 Person Factors That Contribute to Risky Driving Behaviours and Accidents 11 Anger On The Road 15 Psychological Variables and Driving Behaviour 17 Interactions Between Personality Variables and Demographic Variables 25 Summary and Hypotheses 27 Methods 30 Research Design 3ParticipantsMeasures 1 Procedure 33 Results 4 Power AnalysisOverview of Analyses 35 Relationships Among The Psychological Variables 37 Relationships Between age, gender, and the psychological variables (A-B) 38 Relationships Between Age, Gender, and Receipt of mTickets Over the Past Five Years (A-C) 39 Relationships Between Age, Gender, and Involvement in an MVA Over the Past Five Years (A-D)Relationship of Psychological Variables and Driving Variables (B-C, B-D, and C-D) 40 Discussion 46 Relationships Among the Psychological Variables 4Relationships Between age, gender, and BARQ anger out (A-B) 49 Relationships Between Age, Gender, and Receipt of mTickets Over the Past Five Years (A-C) 50 Relationships Between Age, Gender, and Involvement in an MVA Over the Past Five Years (A-D)Relationship of Psychological Variables and Driving Variables (B-C, B-D, and C-D) 50 Limitations 55 Summary and Conclusions 57 Footnotes 9 References 61 Tables 76 Figures 9VI LIST OF TABLES Table 1 Mean scores split by gender and age 76 Table 2 Overall correlations of psychological variables and sub-sample correlations (<30 years/>30 years and male/female) 77 Table 3 Overall correlations of psychological variables and sub-sample correlations (<30 years/>30 years and male/female) 78 Table 4 Overall correlations of psychological variables and sub-sample correlations (<30 years/>30 years and male/female) 79 Table 5 Overall correlations of psychological variables for individuals involved in an MVA split by fault/no fault 80 Table 6 Overall correlations of BARQ subscales for individuals involved in an MVA split by fault/no fault 81 Table 7 Frequencies of reported driving behaviours and chi-square statistics of the differences for gender and age 82 Table 8 Means of reported driving consequences and F statistics for significant differences between means of age and gender and interactions 83 Table 9 Means of personality variables overall and split by age and gender 84 Table 10 Means of BARQ subscales and F statistics for differences between means of age and gender 85 Table 11 Univariate Logistic regression for the outcome of receipt of one or more tickets for a moving violation 6 Table 12 Univariate logistic regression for the outcome of one or more MVAs over the past five years 87 Table 13 Univariate Logistic regression for the outcome of one or more minor MVAs reported over the past five years 88 Table 14 Univariate Logistic regression for the outcome of one or more major MVAs reported over the past five years 89 Table 15 Logistic regression for variables of interest, controlling for age, gender and hours driven per week 90 Table 16 . Fitting of the Logistic Regression Model for prediction of receiving one or more tickets for a moving violation over the past five years 91 Table 17 Logistic Regression model building for prediction of minor motor vehicle accidents 92 Table 17 continued Logistic regression: Model building for prediction of minor motor vehicle accidents 93 Table 18 Odds of involvement in one or more MVAs over the past five years with and without receipt of one or more tickets for a moving violation in the same period at three levels of BARQ Anger Out 94 Table 19 Change in odds of involvement in one or more MVAs over the past five years with and without receipt of one or more tickets for a moving violation in the same period at three levels of BARQ Anger Out 95 viii LIST OF FIGURES Figure 1 Diagram of the possible relationships among the variables 29 Figure 2 Frequency distribution of ages for this student/community sample 96 Figure 3 Odds ratio of MVA involvement for those having received a ticket in the past five years and those who have not (NT) as a function of reported anger out (holding hours per week constant at its mean: 10.12 HPW) 97 1 INTRODUCTION Anger is a common and frequent, but also often destructive, human emotion. Poor anger coping has been linked to increased CVD risk (Mittleman et al., 1995) via elevation in resting blood pressure (BP) and a noted worsening in blood lipid profiles. It is unclear whether it is the experience of anger or the method one uses to cope with anger which is related to the development of disease and negative health behaviors. A reliance on a dichotomy of anger expression, defined as anger-in and anger-out, has dominated research historically (Spielberger, Johnson, Russell, & Crane, 1985), but anger expression styles appear to be more complex than initially thought. This has led to the development of measures of anger expression which consider a larger number of dimensions than the original anger-in, anger-out dichotomy. One measure which has been developed to investigate a number of dimensions of anger expression is the Behavioural Anger Response Questionnaire (BARQ; Linden et al., 2003). The aim of this thesis is to further examine the construct validity of the BARQ as it relates to driving anger and also to test its predictive power to account for risky driving and accident rates. Previous research with the BARQ has exclusively examined its relationships to indices of cardiovascular health. As driving is thought to be a highly anger provoking experience and a very common experience in North America, I want to research this area to gain a better understanding of driving anger and accident risk. The Conceptualization of Anger In the past, hostility, anger, and anger expression have been used interchangeably leading to often inadequate measures of these constructs (Biaggio, Supplee, & Curtis, 1981; Spielberger et al., 1991, Martin, Watson, & Wan, 2000). Anger has been defined as an acute, normally transient emotional response, which depends on the appraisal of events and the assignment of meaning to them (Arnold, 1960). Anger can be defined as an emotional state varying in intensity from mild irritation to rage (Spielberger, Jacobs, Russell, & Crane, 1983). It is believed that most people, even those low in hostility, occasionally experience anger. Hostility has been described as more of a stable attitude rather than an emotion (Buss, 1961; Spielberger et al., 1983), often including cynicism, indignation, contempt, and resentment (Plutchik, 1980). Hostility is thought to cause individuals to be distrustful towards others, as well as increasing anger-proneness (Smith & Frohm, 1985). The distinction between anger and hostility parallels Spielberger's (1983) conceptualization of state and trait anger. Trait anger has been defined as the habitual feeling of anger, as is seen with hostility, while state anger is defined as the emotional experience of anger due to an external provocation. Hostility (anger proneness) and the emotional state of anger have been further differentiated from the expression of anger by Spielberger and colleagues (1983). The expression of anger refers to the behaviours that one employs in response to anger provocation. Spielberger (1983) conceptualized the main dimensions of anger expression as "anger-in" and "anger-out", with pronounced anger-in behaviour characterized by suppression of anger and passivity, while anger-out behaviour is associated with aggression (physical or verbal). Anger expression styles are thought to reflect characteristic ways of responding to, or coping with, the emotion of anger (Bishop & Quah, 1998; Harris, 1997; Mills, Kroner, & Forth, 1998; Musante, Treiber, Davis, Waller, & Thompson, 1999), although some situational variations in anger responding are expected within an individual. It is important to 3 note that individuals may display more than one style of anger expression, as anger-in and anger-out have been discovered to be unrelated, and individuals appear to express and suppress anger at similar levels (Deffenbacher, 1992). However, it has been argued that it is highly probable that individuals tend to display typical patterns of anger responding and rely heavily on a preferred style of responding to anger (Gentry, Chesney, Gary, Hall, & Harburg, 1982; Spielberger et al., 1983). Hostility has been proposed to encompass more than one construct, and this may help to explain previous inconsistencies in findings. Two forms of hostility have been proposed (Costa, McCrae, & Dembroski, 1989). Neurotic hostility tends to be related to the feelings of hostility (e.g., cynicism, mistrust, resentfulness, and anger proneness), while antagonistic hostility taps the expression of anger (e.g., uncooperativeness, irritability, and aggressive tendencies). These findings lead us to question some of the most commonly used instruments to measure hostility, such as the Buss Durkee Hostility Inventory (BDHI; Buss & Durkee, 1957), as well as the Cook-Medley Hostility Questionnaire (CMHQ; Cook & Medley, 1954). These instruments measure both the expression of as well as the attitude of hostility. A reportedly purer measure of hostile attitude is the Hostile Attitudes Scale (HAS; Arthur, Garfinkel, & Irvine, 1999) which is conceptualized to measure hostile cognitions as opposed to actions against others. It appears that further delineating the behavioural responses to anger and hostility may be critical to understanding the links between anger variables and health outcomes and health behaviours (Spielberger, Johnson, Russel, & Crane, 1985). It is important to discriminate between the transient experience of anger as well as the enduring trait of anger, but it is also important to identify how people characteristically express that anger. 4 Anger Expression Most research has focused on the anger-in and anger-out dimensions of anger expression. Originally it was hypothesized that anger suppression would lead to negative health outcomes, while the expression of anger would lead to positive health outcomes (Alexander, 1939). This hydraulic model postulated that the build up of emotions was akin to a build up of internal pressure, which could be deleterious if not released. Some findings support this model, linking anger-in to a number of negative health outcomes such as: resting BP, cardiovascular reactivity, cardiovascular and all-cause mortality (Jorgensen, Johnson, Kolodziej, & Schreer, 1996; Julius, Harburg, Cottington, & Johnson, 1986; MacDougall, Dembroski, Dimsdale, & Hackett, 1985; Spicer & Chamberlain, 1996; Steele & McGarvey, 1997; Thomas, 1997). Other findings have been contrary to this theory, revealing relationships between expressed anger, higher BP levels, and higher BP reactivity, but no relationships for anger suppression (Harburg, Gleiberman, Russell, & Cooper, 1991; Larkin, Semenchuk, Frazier, Suchday, & Taylor, 1998; Siegman & Snow, 1997). There may also be an effect of gender, as women who report anger-out tendencies display slower systolic BP recovery than those reporting anger-in tendencies (Lai & Linden, 1992). Some research has also revealed that extremes in expression of anger in either direction (e.g., both anger suppression and expression) have been related to negative health outcomes such as hypertension in men (Everson, Goldberg, Kaplan, Julkunen, & Salonen, 1998). This is in keeping with the social conflict model (Linden, 1993; Linden & Feurstein, 1981), which posits a curvilinear relationship between anger and health. People exhibiting extreme anger-in or extreme anger-out behaviour are thought to generate excessive repetitive social or intra-psychic conflict, and have been found to have the highest BPs (Linden & 5 Lamensdorf, 1990). The idea that less extreme forms of anger expression may be related to positive health outcomes has also been supported. More reflective styles of anger expression have been related to lower BPs (Gentry, Chesney, Gary, Hall, & Harburg, 1982), and treatment to decrease hostility by increasing constructive verbal expression has also resulted in reduction in BP (Davidson, MacGregor, Stuhr, & Gidron, 1999). Although many studies in this field have only considered men or have failed to examine gender differences, some research in support of gender differences has emerged. Recent research has suggested a relationship between anger-out and coronary heart disease for women only (Siegman, Townsend, Blumenthal, Sorkin, & Civeiek, 1998). As well, earlier research found relationships between suppressed anger and elevated systolic BP in men but not women (Dimsdale et al., 1986). Other research has established a relationship between anger-in tendencies and more rapid physiological recovery from anger provocation for women when compared to women reporting high anger-out (Lai & Linden, 1992). The previous patterns of results lend themselves to explanation with the social conflict model. This model suggests that conflict will be produced when an individual uses a form of anger expression that is socially unacceptable (Linden, 1993). In North American society women have traditionally been raised to remain non-aggressive (Hokanson, Willers, & Koropsak, 1968). This would support a hypothesis that the suppression of a behavioural anger response would not lead to dissonance in a woman's self perception. The reverse is true for men, as they report more physically and verbally aggressive responses than women (Harris, 1992b). There appears to be little consensus regarding anger expression styles, conceptualized as anger-in and anger-out, and how these may be related to health and health behaviours. This may be due to the fact that anger expression styles are less stable then once thought. 6 Perhaps these styles fluctuate to some degree according to situations (Linden et al., 1997). It is also possible that the conceptualization of anger expression as a dichotomy, composed of anger-in and anger-out, is too simplistic. This could explain the inconsistencies in findings, implying that anger expression should be investigated as a multidimensional construct. Multidimensional Forms of Anger Expression Stoney and Egebretson (1994) suggested that there may be multiple forms of anger expression that are not normally assessed in research, and urged researchers to investigate a more extensive view of the dimensions of anger expression. Studies using factor analysis to determine the dimensions of the anger expression construct, have found factors above and beyond the traditional anger-in and anger-out. These include three factors: verbal/adaptive, maladaptive/physical, anger experience/hostility (Riley & Treiber, 1989), four factors: aggressive overt hostility, alienated bitterness, introversion, and anxiety/depression (Friedman, Tucker, & Reise, 1995), and even eight factors: hostile anger expression, perceived control over anger expression, frequency of anger, ease of anger provocation, brooding, hostile outlook, cynicism, and sullenness (Miller, Jenkins, Kaplan, & Salonen, 1995). In sum, anger expression is a multidimensional construct. Research supports both positive and negative health outcomes associated with various anger response styles, but in the past no multidimensional measure of anger expression has been available. BARQ The perceived need to develop a comprehensive multidimensional tool for measuring anger expression styles led to the development of the Behavioural Anger Response Questionnaire (BARQ; Linden et al., 2003). This measure has good reliability and has been 7 supported with content, and construct validity. The questionnaire consists of 37 behavioural descriptions which are rated according to the frequency of use when feeling angry. This allows individuals to receive elevated scores on more than one subscale if their anger response repertoire is more flexible. One month test retest reliability reveals correlations ranging from 0.61 to 0.85, demonstrating strong temporal stability in how a person reports expressing their anger over time. The BARQ consists of six distinct anger response styles: Anger out, Assertion, Social Support Seeking, Anger Diffusion, Avoidance and Rumination. Anger out (BARQao) refers to an outward display of physical or verbal anger (e.g., "I make a sarcastic or critical remark to the person who annoyed me"). Assertion (BARQas) involves direct but not aggressive interaction with the person responsible for causing the angering event (e.g., "I let things cool off a little and then talk to the angering person about what happened"). Social Support Seeking (BARQss) refers to behaviours where the person actively or indirectly seeks out someone external to the event with whom to discuss his/her feelings (e.g., "I think about the problem for a while; later-for example in the evening-I discuss the incident with my spouse"). Anger Diffusion (BARQdif) is a direct, yet passive and nonviolent, method of dealing with anger, and involves occupying oneself with unrelated activities such as doing housework (e.g., "I worked off my frustration doing things like cleaning the house, organizing the office, or by doing garden work"). Avoidance (BARQav) involves efforts to ignore or forget the feelings of anger (e.g., "I convince myself that this is not worth getting upset about"). Finally, rumination (BARQrum) is an internal way of processing feelings of anger, involving repetitive thought patterns, mentally replaying the frustrating event but without outward physical manifestations of anger (e.g., " I think repeatedly about what I really would have liked to have done but did not"). 8 While the BARQ subscales measure extreme forms of anger expression similar to anger-in and anger-out, they also tap more intermediate forms of anger responding. These alternate forms of anger expression can be placed along a theoretical continuum between passivity and aggression. The only subscale which does not fit onto this continuum is rumination. Rumination has received less attention from anger researchers, but has been hypothesize to lengthen and even magnify angry feelings, while possibly acting as a moderator of aggressive behavior (Caprera, Barbaranelli, & Comrey, 1992). In one study it was found to moderate the effects of other anger expression styles on reactivity measures of blood pressure. Reported rumination when angry increased the negative effects of avoidance on blood pressure in men. As well, for women who had seen a benefit of assertiveness on their blood pressure, rumination when angry deleted this (Hogan & Linden, 2003). The multidimensional nature of the BARQ exhibits the potential to clarify the complex relationship between anger expression, health behaviors and health outcomes. Construct validity has been demonstrated for the BARQ in comparison to overall personality factors on the NEO-FFI (Costa & McCrae, 1992). The rumination subscale of the BARQ was found to correlate positively with the Neuroticism subscale of the NEO-FFI. As well, the BARQ subscale of direct anger-out correlated negatively with the NEO-FFI subscale of agreeableness and negatively with the NEO-FFI subscale of conscientiousness. Therefore, those who score high on Neuroticism are more likely to rely on the anger expression style of rumination. As well participants who reported low agreeableness and low conscientiousness were more likely to use an anger-out response style. Interestingly, this parallels research by Costa et al. (1989), in which neurotic hostility, characterized by the frequent and intense experience of anger without expression, has been related to Neuroticism, while antagonistic 9 hostility, characterized by stable, hostile, and irritable interpersonal orientation often accompanied by verbal and physical aggressive outbursts, has been related to low agreeableness. No studies have been conducted specifically examining how the BARQ is related to other measures of negative affect, especially those that make up the components of the construct of anger. This current investigation will begin to explore the nature of the relationship between the multiple anger response styles of the BARQ and indices of the anger construct including: hostility, trait driving anger, Type-A personality, and anxiety. Validity has been demonstrated for measures of BARQ anger expression in relation to health indicators such as BP (Hogan & Linden, 2003; Linden, Lenz, & Con, 2001). In order to extend previous research in the anger domain and to explore the predictive validity of BARQ subscales, analyses were conducted investigating the predictive validity of specific BARQ subscales for driving behaviours (e.g., traffic accidents, tickets for moving violations). As motor vehicle driving can be a highly anger provoking arena, it lends itself naturally to exploration of the anger construct. In the following sections I discuss the relevant literature in the area of driving and driving behaviours. Motor Vehicle Accidents Traffic accidents are one of the leading causes of disability and death in our society, especially amongst those under 40 years of age. It is estimated that Canadians spend up to 25 billion annually on costs related to traffic accidents (Health Canada web site, 2003). From 1987 - 1996 in British Columbia alone, in a population rising from 3 to 4 million people, 1,266,232 traffic accidents were reported to insurance companies causing 369,792 injuries and 5,537 fatalities (ICBC website, 1997). In North America driving is a frequent event that 10 touches most peoples' lives, and it is obvious that accidents can be extremely costly, both physically and monetarily. Numerous factors are known to contribute to motor vehicle accidents (MVAs). These include characteristics of the motor vehicle, the environment (e.g. road conditions and weather), as well as the person. It appears that the majority of accidents involve human factors. Police reports of 13,568 accidents in Indiana covering a 5 year period classified accidents according to cause and found that the cause of 70.7% of accidents was human error, 12.4% were due to environmental causes (e.g. road conditions), and 4.5% were due to vehicle conditions (Treat, 1980). Further subdivision of these categories distinguished improper lookout, excessive speed, inattention, and improper evasive action as components of human error. Are people aware of these risks? Guerin (1994) polled university students and community adults and discovered that there are five factors of perceived driving risk: environmental and road conditions, unexpected events, driver problems, necessary or unavoidable risks, and voluntary driving risks. It therefore appears that people are quite aware of the human factors that play a role in the riskiness of driving. Various estimates have placed the involvement of human factors in MVAs anywhere from 70-95%o (Treat, 1980; Beirness, 1993). As well, a literature review by Beirness (1993) recounted that although the obtained correlations of personality variables to driving variables were fairly low, personality appeared to account for about 10-20%> of the variance in car crashes and up to 35% of the variance in risky driving. Other factors mediate this relationship and behaviours such as alcohol use, age, and gender are often able to account for more of the variance in crash involvement than personality. Taking this into account personality still seems to explain a unique and substantial portion of the variance. Personality may also 11 influence a number of other factors about a person's driving including how often they drive and how they might respond in a particular traffic situation. This means that the way that an individual chooses to operate a vehicle is important in understanding who becomes involved in MVAs. As well, the major focus of traffic safety has been on road and vehicle improvement and that these have led to major gains in our safety, but we have not advanced as far in understanding how human behaviour affects driving (Rothengatter, 1997). By increasing the understanding of driving behaviour and individual differences which contribute to crash involvement, we may be able to intervene in ways that will decrease crashes, injury, and death, Person Factors that Contribute to Risky Driving Behaviour and Accidents Age and gender differences in rates of MVAs are well documented with males and younger individuals experiencing higher rates of MVAs (e.g., Tonkin, 1987; Pipkin & Thomason, 1989; Alexander, Kallail, Burdsal, Ege, & David, 1990; Maxim & Keane, 1992; Parker, Reason, Manstead, & Stradling, 1995; Hijar, Carrillo, Flores, Anaya, & Lopez, 2000; Norris, Matthews, & Riad, 2000). There are a number of differences between these groups that explain increased MVA involvement including (1) more time driving a vehicle, which leads to greater exposure to the possibility of traffic accidents for males (Maxim & Keane, 1992), (2) increased alcohol consumption among young people leading to greater chances for risky driving behaviour (Pipkin & Thomason, 1989), and (3) increased risky driving behaviours, such as higher driving speeds and tendency to disregard traffic rules (Norris, Matthews, & Riad, 2000). Other important person factors include proficiency in driving, and attention when driving. Large scale questionnaire studies have examined these constructs by means of the 12 Driver Behaviour Questionnaire (DBQ; Reason, Manstead, Stradling, Baxter, & Campbell, 1990) in order to examine individuals' self reported lapses, errors (mistakes), violations, and involvement in traffic accidents while driving. These categories are based on factor analysis and a model of human error called the Generic Error Mechanism System (GEMS; Reason, 1990). Errors are defined as the failure of a planned action to achieve its intended consequence without the intervention of chance. Errors are further broken down into lapses and mistakes. Lapses are actions that deviate from an adequate plan (e.g., forgetting where your car is parked, exiting on the wrong road), while Mistakes arise from an inadequate plan towards a desired goal (e.g., failing to notice pedestrians crossing, misjudging the speed of another vehicle when overtaking). Violations are the deliberate breach of a regulation or a social code of behaviour (e.g., speeding, tailgating), but if a violation occurs unintentionally, it is more likely an error. As well, a violation can occur with or without the intention to bring about harm. Errors, both lapses and mistakes, come about due to information-processing problems, while violations have a large motivational element (Parker, Reason, Manstead, & Stradling, 1995; Aberg & Rimmo, 1998). Results suggest little relation between lapses or errors and motor vehicle crash involvement (Parker, Reason, Manstead, & Stradling, 1995). As well, women report more lapses than men, violations were found to decline with age, while errors do not, and men report more violations than women (Reason, Manstead, Stradling, Baxter, & Campbell, 1990). Using a mail-out strategy with a volunteer panel in Great Britain revealed that after controlling for mileage, age, and gender, violations still predicted self reported accident involvement in the three years prior to the study (Parker et al., 1995). This relationship may not hold true for older people though as a similar study conducted with 49-90 year-olds did 13 support relationships of errors and lapses to more active accidents, and also found that lapses were related to more passive accidents (Parker, McDonald, Rabbitt, & Sutcliffe, 2000). Inconsistencies do exist though, as a more recent questionnaire study, conducted on 2763 Australian women (18-23 and 45-50 years of age), found that self-reported lapses predicted MVA involvement for both groups. Although this research replicated the finding that, violations were higher in younger women than older women (Dobson, Brown, Ball, Powers, & McFadden, 1999). Unfortunately most studies do not distinguish between aggressive violations and highway-code violations, and this may be an important distinction. Aggressive violations most likely are influenced by increased emotional arousal, which may have an effect on perception and information processing (Deffenbacher, Oetting, & Lynch, 1994), as well as expectations and preferred actions (Naatanen & Summala, 1976), the latter impact how the driver behaves and could increase accident risk (Mesken, Lajunen, & Summala, 2002). Risky driving behaviours have consistently been linked to MVA involvement (e.g., Jonah, 1986; Reason, Manstead, Stradling, Baxter, & Campbell, 1990; Parker, Reason, Manstead, & Stradling, 1992; West, French, Kemp, & Elander, 1993; Rajalin, 1994; Parker, Reason, Manstead, & Stradling, 1995; Cooper, 1997). It is often assumed that risk-taking is punished by police apprehension, but in general people are not caught and the risk taking is then negatively reinforced (Brewer, 2000). Risk taking has been measured using self-reported moving violations while driving (e.g. number of speeding tickets; Robertson & Baker 1975, Rajalin, 1994), self-reported propensity to speed or break the rules of the road (West, French, Kemp, & Elander, 1993; Parker, et al., 1995) and driving record investigation of moving violation tickets (e.g. Cooper, 1997). 14 A study conducted with insurance and police records in Canada, discovered that although previous tickets for traffic violations predicted MVA involvement, it did not matter what violation the person had received the tickets for. Those who had received tickets for speeding were not more likely (and in some cases less likely) to be involved in an MVA than those who had received tickets for other violations. Drivers with four or more tickets for excessive speeding had twice the overall predicted crash rate of those receiving tickets for other offences, and were at greater risk for serious injury and fatality due to the MVA (Cooper, 1997). In another study which examined driving records of those involved in fatal MVAs, it was discovered that traffic citations for three years prior to MVA predicted fatal MVAs, when compared to randomly selected controls (Rajalin, 1994). In sum, it is well documented that overall traffic violations predict MVA involvement. In one study examining a large group (n=l 160) of individuals (18-79 years of age), it was revealed that interpersonal violations while driving, (e.g., overtaking a slow driver on the inside, sounding the horn, indicating hostility to others, getting angry and giving chase) but not traffic violations (e.g., close following, speeding on highway, speeding in residential areas) predicted MVA involvement, but only for MVAs reported as another person's fault. They also found that self-reported driving errors predicted MVAs that people claimed as their own fault. They concluded that those who are able to report errors are probably more likely to report that an MVA was their own fault, while those who report violations are more likely to blame others for an MVA. No gender or age differences were presented (Mesken, Lajunen, & Summala, 2002). 15 Anger On the Road There are many interacting factors that make the driving situation such an anger-inducing arena. Factors that have been suggested include: traffic congestion and stress (Novaco, Stokols, Campbell, & Stokols 1989), lack of ability to adequately communicate with other drivers (Fong, Frost, & Stansfeld, 2001; Parkinson, 2001), anonymity of driving (Ellison, Govern, Petri, & Figler, 1995), and personological factors such as identifying with one's vehicle very strongly (Fong, Frost, & Stansfeld, 2001). Another study found that a flail 80% of the people reported negative affect while driving (e.g., aggressive behaviours, irritation, and anxiety). Inefficient coping responses were not found in low stress individuals, and drivers over 44 years of age were more relaxed in traffic jams and had fewer inappropriate behaviours and emotions (Gulian, Debney, Glendon, Davies, & Matthews, 1989). Another factor may be that it is very difficult to communicate or provide feedback to someone in another vehicle when driving, so it is often difficult to apologize for a mistake that has been made. These mistakes are often misconstrued as aggressive and intentional acts (Fong, Frost, & Stansfeld, 2001). Parkinson (2001) found that anger while driving as opposed to non-driving situations was more likely caused by communication difficulties, and involved a purer form of blame of others for the situation. Blame of others and its appraisal may be a central influence in the arousal of anger (Smith & Lazarus, 1993). This reduction in communication is a result of the lack of face to face communication between individuals in separate vehicles. With face to face communication you can obviously hear the other person much better, but you can also read the other person's expressions and body language; this would result in clearer communication. Individual's rated desire to communicate anger is one 16 of the best predictors of anger intensity in all types of situations, and communication difficulties have been rated as an influential cause of anger especially in driving (Parkinson, 2001). As well, many personological factors contribute to individuals responding in a hostile manner in their vehicles. "The car is an extension of personal space, often people's second most valuable possession, their main access to freedom, and a statement of self through the choice of vehicle, colour, make, model, and of course, the way they drive (Fong, Frost, & Stansfeld, 2001)." If individuals identify that extremely with their vehicles, it would help to explain driving aggression as a type of territorial response. Normally this anger may be contained, but other factors such as the anonymity of driving (Ellison, Govern, Petri, & Figler, 1995) have helped to provide a possible explanation for the expression of normally inhibited hostility and aggressive responding. Driving is often seen as a different type of arena and that this is where people who normally aren't aggressive become aggressive (Ellison, Govern, Petri, & Figler, 1995). Others find that the way we respond in our everyday lives is how we respond when driving. Brewer (2000) reported that in their sample, drivers tended to report similar off road behaviour to on road behaviour. A study conducted using drivers in the United Kingdom researched decision making style and MVA involvement (French, West, Elander, & Wilding, 1993). They suggested that the results supported a tendency for people to use parts of their overall general decision making style when they were driving. In sum, there are a number of factors in driving that lead it to be a more anger inducing arena. This may be the reason that individuals tend to report a great deal of driving 17 anger. This anger inducing arena may be a factor in increasing aggressive responding in those that use this form of anger expression in their everyday lives. Psychological Variables and Driving Behaviour Although a number of personality variables are associated with risky driving behaviours and motor vehicle crash involvement, there is conflicting evidence that personality is related to driving behaviour and motor vehicle crashes. In a review of the literature, Ranney (1994) reported inconsistent findings and few studies of importance. This may not be due to the lack of an association, but may be due to methodological or theoretical shortcomings. Despite these statements, a review of the literature on personality as it relates to driving behaviour and motor vehicle crash involvement, revealed three dimensions that were most strongly and consistently related to motor vehicle crash involvement: thrill seeking (or sensation seeking), impulsiveness, and hostility/aggression (Beirness, 1993). Self reported sensation seeking has often been related to risky driving behaviours, such as speeding, but not to crash involvement. This allows the conclusion that sensation seekers must drive fast, but that they often do it carefully (Burns & Wilde, 1995). This relationship appears to be a very strong one as a review of 38 studies investigating sensation seeking exhibited correlations of sensation seeking to risky driving behaviours ranging from .30-.40, with only four of the studies finding no relationship (Jonah, 1997). Impulsivity may not be orthogonal to sensation seeking, as it has been suggested that impulsivity is composed of sensation seeking and quick decision time (Loo, 1979). Sensation seeking is related to increased traffic convictions, while the fast decision time component of impulsivity correlated with MVAs (Loo, 1979). A number of other psychological personality variables 18 have been consistently researched in the area of driving behaviour, and these include Type-A behaviour, driving anger, and aggression. Type-A personality. Type-A personality (first put forward by Friedman & Rosenman, 1959) has been defined as a sense of time urgency/increased speed of behaviour, impatience, hostile/irritable, and competitive drive (Byrne, Rosenman, Schiller, & Chesney, 1985; Houston, Smith, & Zurawski, 1986; Perry, 1986). Due to these factors, Type-A individuals are thought to be in more of a rush when driving, and may drive more aggressively. Type-A personality has been related to more self reported MVAs and more self reported numbers of tickets for violations (Perry, 1986; Magnavita et al., 1997; Perry & Baldwin, 2000). Not all studies have found a relationship between Type-A personality and MVA involvement. A questionnaire study discovered that Type-A did not correlate with MVAs, and instead found that social deviance predicted MVAs (West, Elander, & French, 1993). As Type-A is known to be a complex construct, measurable by a number of standardized self report and interview measures; it is possible that the different measures the studies have used are partly responsible for conflicting results. It is known that measures of Type-A tend to correlate only modestly with one another (Byrne, Rosenman, Schiller, & Chesney, 1985). Driving anger. Driving anger has been postulated as a driving specific propensity towards anger that may be an important predictor of driving behaviour. In order to more fully explain personality as it relates to driving anger it has been conceptualized as a trait related to trait anger, but as a much narrower and more situation-based form of trait anger (Deffenbacher, Oetting, & Lynch, 1994). Where trait anger (Spielberger, Jacobs, Russell, & Crane, 1983) reflects a broad disposition to experience anger more frequently and intensely 19 across a number of situations, driving anger is that experienced more frequently and intensely in driving situations. The relationship of driving anger to trait anger is explained as analogous to measuring trait anxiety along with specific context based forms of anxiety, such as test anxiety. The more specific context-based measures, such as the Driving Anger Scale (DAS; Deffenbacher, Oetting, & Lynch, 1994) correlate moderately with trait anger, but are thought to be more predictive of behaviours and emotions in their specific contexts. The experience of driving anger is fairly prevalent but unfortunately some people may express this anger in inappropriate ways, either in the form of aggressive driving or at the extreme, road rage. A growing amount of popular literature has focused on road rage, while research has only fairly recently begun to examine the construct of driving anger (e.g. Deffenbacher Lynch, Oetting, & Yingling, 2001). With the large amount of media attention on road rage and aggressive driving it seems important to tackle the question of how driving anger and its related constructs affect driving behaviour. Interest in the construct of driving anger appears to be on the rise. Over the last decade there have been increasing numbers of reports of road rage and aggressive driving. The American Automobile Association (AAA) reported a 7% increase in road rage incidents per year from 1990-1996, resulting in an estimated 200 deaths and another 12,000 injuries (Mizell, Joint, & Connell, 1997). However it is unclear whether this reflects an actual increase in aggressive driving or an increase in media attention to this phenomenon. One study using mail-out questionnaire data with 2605 subjects (48% men, 52% women, mean age = 45) revealing that driving anger along with normlessness, and sensation seeking predicted risky driving behaviour, and risky driving behaviour predicted near MVAs 20 and crashes. In this study, the psychological variables were also related to crash outcomes, but risky driving influenced crashes the most (Iversen & Rundma, 2002). As well, clients being treated for aggressive driving have been found to be higher on trait driving anger as measured by the DAS (Deffenbacher, Huff, Lynch, Oetting, & Salvatore, 2000). They also report more frequent and intense anger, more aggressive and risky behaviour, and more minor MVAs and close calls. They do not report more major MVAs or moving violations. As well, driving anger does not appear to vary between men and women. In this study it was suggested that people experiencing high rates of driving anger will experience higher rates of crashes and violations, as this anger may elicit aggressive responding on the road (aggression hypothesis). Increased anger may elicit increased levels of impulsive and risky behaviour (risky behaviour hypothesis), and these angry emotional and cognitive processes may disrupt information processing and interfere with safe driving (negative outcomes hypothesis). A further study using undergraduate students (median age =19 years) measured on the DAS found that trait driving anger did not correlate with crashes or moving violations, but did correlate with aggressive behaviour on the road in that population (Deffenbacher Lynch, Oetting, & Yingling, 2001). There is little evidence of any gender differences in how angry people become when driving, but these may exist for what makes people angry. In an Australian community sample comprised of bus drivers and non bus drivers, women were likely to report more anger than men about illegal driving and traffic obstruction (Brewer, 2000). As well, although it may be intuitively expected, exposure differences to the driving situation have not been found to affect driving anger. In the same study, bus drivers and non-bus drivers tended to experience the same amount of driving anger, with bus drivers engaging in less aggressive 21 driving. This may be due to extensive driver training, and competency testing among bus drivers, or even possibly increased driving experience (Brewer, 2000). Anger is often seen as having only negative connotations (see DeAngelis, 2003). It is assumed that everyone will get angry at some point or other, it may be how we express that anger that may be more important and for some problematic. Anger calls our attention to something that needs to be changed; we can deal with that anger-in numerous ways (Tafrate, Cassinov, & Dunden, 2002). In general, driving anger appears to be associated with risky driving behaviours and violations and does not differ for men and women, but the full relationship between anger and driving behaviours has not been elucidated. Aggression, is defined as a tendency to express anger either verbally (e.g., raised voice, rudeness) or physically (e.g., slamming doors, destroying objects) (Johnson & Spielberger, 1992; Smith, 1992). Aggressive driving occurs when individuals express their anger wile driving. The Larson's Driver Stress Profile (Blanchard, Barton, & Malta, 2000) was developed using drivers identified for treatment for aggressive driving, and it defines aggressive driving as behaviours meant to compete with or punish other drivers. It also measures the frequency with which driver's experience anger and impatience when driving. The Larson's Driver Stress Profile has been seen to correlate significantly with trait driving anger, trait anger, anger-out, and Type-A behaviour. Linking aggression to accidents is not a new idea. Taxi drivers' self reported histories of aggressive behaviour in childhood (e.g. fights, bullying, and juvenile court appearances) have been correlated with more frequent traffic accidents (Tillman & Hobbs, 1949). Student pilots' descriptions of themselves are also related to accident involvement. Those pilots involved in accidents were more likely to be uncooperative, less responsive, and less socially 22 effective (Cowley, 1956). Another study examining a group of industrial workers found that those people who had experienced a high number of accidents had a history that was characterized by high aggression and low tolerance for discipline at home, work, or socially (Wong & Hobbs, 1949). Although these studies did not use standardized measures their findings are of interest in developing further hypotheses. There are a number of studies examining the relationship between aggression while driving and MVAs, but inconsistencies abound. A number of studies have found support for aggression as a significant predictor of MVA involvement (see Beirness, 1983; Selzer & Vinokur, 1974; Wells-Parker, Cheminski, & Hallberg, 2002). One study examined a number of personality and stress variables, and found that aggression was the only personality variable significantly related to MVA involvement, but stress measures were found to correlate more highly with MVA involvement (Selzer & Vinokur, 1974). Unfortunately they failed to control for important demographic variables that are known to interact with both the outcome and predictor. A prospective study following 500 drivers found a relationship between young age, high hostility combined with low self esteem, high job stress, prior MVA, and self reported tendency to speed and disregard traffic rules predicted MVA involvement (Norris, Matthews, & Riad, 2000). In another study, a telephone survey of 1382 US adult drivers found relationships between angry threatening driving (e.g., cutting people off, chasing cars, making threatening moves on the road, and tailgating), and MVA involvement and tickets for moving violations over the past year, after controlling for age and gender. Verbal frustration (e.g., complain, say things under your breath, give the person a dirty look), did not predict MVA involvement or traffic tickets (Wells-Parker, Cheminski, & Hallberg, 2002). 23 Other studies have found no relation between the tendency to report aggression while driving and MVA involvement. In a university and community sample MVAs correlated negatively with age, and self reported aggression while driving did not correlate with MVAs, although they did not control for age or gender of driver (Furnham & Saipe, 1993). Self reported aggressiveness on the road also did not differentiate MVA involved individuals from those with no MVA experience in another study, but males had reported higher aggressiveness than females (Wilson & Greensmith, 1983). The Driving Anger Expression Inventory (DAXI; Deffenbacher, Lynch, Deffenbacher, & Oetting, 2001) is a 49-item scale which measures aggressive anger expression while driving. In a group of 272 university students, this measure was not correlated with trait anxiety, moving violations, close calls, or MVAs. It is important to point out differences between road rage and aggressive driving. Road rage is defined by Rathbone and Huckabee (1999) as uncontrollable anger that leads to violence or the threat of violence on the road. Aggressive driving is more often less extreme and may result in traffic offences such as tailgating, abrupt lane changes, yelling at another driver, and/or speeding (Rathbone & Huckabee, 1999), and it is possible that these aggressive driving behaviours put individuals at greater risk for MVAs. People may not feel that mild forms of aggression are dangerous, as they do not appear as frightening as incidents of road rage (Novaco, 1991). Given the speed involved in driving and the destructive potential of the vehicle, even mild aggression can not be ignored on the road as a potential source of danger (Hennessy & Wiesenthal, 1999). It has been hypothesized that aggressive driving may not only put the individual at risk but may also increase anger or aggressive behaviour in other drivers, leading to a back 24 and forth escalation of anger between individuals with may further increase aggression (Novaco, Stokols, Campbell, & Stokols, 1979; Deffenbacher, Huff, Lynch, Oetting, & Salvatore, 2000). It seems important to be able to identify people who may be at risk for developing driving anger that would lead to aggressive, risky, dangerous driving behaviours in order to develop interventions that could target these individuals and help them to deal effectively with these situations. The relationship between aggressive driving and MVAs has not been fully delineated, and research in the field is often without theoretical direction. The Frustration-Aggression Hypothesis has been postulated in this relationship (Dollard, Doob, Miller, Mowrer, & Sears, 1939), suggesting that the frustration of the driving environment (e.g., congested roads, traffic jams) leads one to aggress on other drivers who appear to be frustrating you. This hypothesis has gained little support as not all people who are frustrated experience aggression (Lajunen, Parker, & Summala, 1999). It has been postulated that the motor vehicle is unique in its ability to provoke aggressive responses on the road because people think of the motor vehicle as a special territory that needs to be protected. Aggression is enacted as a defensive response to others encroaching on this territory, and as young males often do not own any other property they are more likely to defend their cars, thus expressing more aggression on the road (Marsh & Collett, 1986). Counter to this, those who drive their "dream cars" are found to be less likely to engage in risky behaviours, although we would expect them to be more territorial (Hemenway & Solnick, 1994). The relationship between general aggressiveness and aggression while driving has only recently begun to be revealed. One study measured general aggression as well as self reported aggressive driving and found that both verbal aggression and physical 25 aggressiveness were related to aggressive driving behaviour, with the link from verbal aggression being mediated by driving anger (Lajunen & Parker, 2001). The lack of a driving anger mediation with physical aggressiveness supports the idea that aggression (behaviour) is not always preceded by anger (emotion), and that some may use aggression as a social problem solving strategy (Berkowitz, 1993). In sum, aggression on the road has been inconsistently linked to MVA involvement with negative findings for some studies. This may be due to methodological inconsistencies, as a number of studies are correlational in nature, and fail to control for relevant demographics such as age, gender, and exposure. As well, it appears likely that if age and gender are related to aggressiveness, as well as to MVAs, there may be interactions between these variables and aggressive responding may not predict MVAs for all of these groups. Interactions Between Personality Variables and Demographic Variables People tend to search for main effects of variables, as they are more generalizable, but it is often informative to explore how variables behave in different populations. Although a number of studies state that there is most likely an interaction between personality factors that predict crash involvement and other driving behaviours or demographic factors, they often fail to test interaction models. Many studies have revealed that there are age and gender differences in driving behaviours such as MVA involvement, as well as amount of time driving. Yet interactions between these variables and personality variables are not as often tested. A study examining Canadian statistics for gender, age and the risk of violent death over the period of 1950 - 1986 found that males and females continue to have differing risks of violent death over that time period with males having greater risk. These gender ratios for MVA death appear to be slowly narrowing so that women are creeping up on the men with 26 the ratio dropping consistently from 3.03:1 in 1950-54 to 2.41:1 in 1982-86. They report that this is probably due to increased exposure of women to driving over that time period. Analyzing relationships with age, they found that the young and the old are disproportionately at risk, with the maximum risk lying between 15 and 30, and among those over 65. This pattern is identical for both men and women, and is not unexpected as unlike other causes of death, MVAs tend to involve more than one victim and the victims tend to be within the same age group and of mixed gender (Maxim & Keane, 1992). Gender and age are related to differences in both aggressiveness and accident rates. It is possible that a confounder may lead to an interactional model, so this needs to be tested. Gender and age differences in aggression are well documented. Bettencourt and Miller (1996) reported that men were generally more aggressive than women physically, but women tended to be equal to men if the aggression was verbal or written. This could be why women high on aggression do not tend to come to the eyes of the authorities as often as men. These gender differences were not as strong when people were provoked. They believe that the gender differences are due to women's fear that they will be retaliated against. This means that women may be just as aggressive as men in situations where they are highly provoked and there is little chance of retaliation. Females may actually be more indirectly aggressive than men, harming a target through the use of other individuals or social structures (Bjorkqvist, Osterman, & Kaukianen, 1992). The explanation offered for this is one of sociocultural-learning (Hokanson, Willers, & Koropsak, 1968, Linden, 1993; Linden & Feurstein, 1981). As mentioned previously, women tend to be raised to be "peacemakers" and to avoid overt expressions of aggression. This then makes it acceptable for women not to be aggressive. 27 Summary and Hypotheses In summary, this review of the literature suggests that a relationship exists among known facets of the anger construct (hostility, anger, & anger expression), but that inconsistencies exist when these anger-related variables are used to predict health behaviours and health outcomes, particularly MVAs. In response to these inconsistencies, a multidimensional scale of anger expression has been developed (BARQ) but it has so far only been related to cardiovascular disease risk and not to any other situations. The driving arena seems to be characterized by a relatively large amount of naturally ocurring anger and a fairly large amount of research exists examining the relationship between psychological risk factors and driving behaviours such as MVA risk; hence, this area appears to be ideal for studying predictive validity of the BARQ. Findings in the area of MVA involvement prediction have been inconsistent and in general fail to examine interactions among the demographic risk factors which are known to be related to the outcome variables and the predictor variables. Aggression appears to fall out as one of the stronger predictors of MVA involvement, and as BARQ anger-out is meant to assess this construct, the final purpose of the present paper is to analyze relationships between BARQ anger-out and demographic risk factors in predicting MVAs and moving violations. I predict that a general tendency to use aggression when angry will predict involvement in MVAs. The objectives are: 1. The psychological variables will be investigated to determine their relationships through the use of correlational analyses. The BARQ subscales will be investigated to see how they relate each other and to the other variables measured. BARQ anger-out as a measure of aggression is expected to correlate moderately with the other anger factors: Driving anger and Hostility. BARQ rumination, which may be a moderator of 28 the anger experience, is expected to correlate with BARQ anger-out and anger-in as has been seen previously, as well as the anger factors of Driving anger, Hostility. As rumination is specifically tied to anxiety, a relationship is expected there as well. Specifically the relationship between the anger variables: BARQ anger-out, Hostility and Driving anger will be investigated. The relationships among the variables will also be explored to determine differences which may exist between those who report accidents as their fault versus those who report accidents as someone else's fault. 2. For the second and primary objective of this thesis, a pathway model has been constructed (see figure 1 at end and below) delineating the relationships which are hypothesized based on previous research. Relationships between classes of variables, clustered into blocks A-E are explored one by one and in an interactive model. For blocks A-B: age and gender are predicted to be related to BARQ anger-out and mean differences will be tested. As well, mean differences of all psychological personality variables will be explored. A-C: Age and gender are predicted to be related to receipt of tickets for moving violations (mTickets) and this will be tested both with frequency analysis and logistic regression. A-D: Age and gender are predicted to be related to involvement in an MVA and this will be tested both with frequency analysis and logistic regression. B-C: BARQ anger-out is predicted to be related to receipt of mTickets and this will be tested using logistic regression. B-D: BARQ anger out is predicted to be related to involvement in an MVA and this relationship will be tested using logistic regression analysis. This relationship is expected to be moderated by both age and gender i.e. an interaction model will be tested. It is expected that for young males who are generally high in MVAs and aggressiveness, 29 aggressiveness will not predict MVAs as it will not be able to differentiate between high and low MVA groups. BARQ Rumination is also hypothesized to be involved in this relationship between BARQ anger-out and MVA involvement, as those with a higher tendency to ruminate about an angering event, may experience and express that anger for a longer time. C-D: Receipt of mTickets is predicted to be related to involvement in an MVA and this will be tested with logistic regression analysis. Demographics: Age, gender A. B. Personality: Aggressive responding when angry, Anxiety, Hostile attitude, Driving anger, Competitive Drive, and Speed/impatience Rumination? C. D. MVA involvement Risky Driving (mTickets) Figure 1. Diagram of the possible relationships among the variables. B-C and B-D: Anxiety, Hostile attitude, Driving anger, Competitive Drive, and Speed/impatience and their relationship to MVA involvement and receipt of mTickets will be explored with logistic regression analyses. The Competitive drive and Speed/impatience subscales of the Framingham Type-A personality scale are thought to be a relevant way to simplify the construct of Type-A personality. As Driving Anger has shown no gender or age differences, no interaction models will be tested with it. An interactional model will, however, be tested for the relationship between age and gender in predicting the receipt of mTickets, as age and gender have been related to both anxiety levels and receipt of mTickets in the past. It is hypothesized that most people become angry while driving. Driving anger has been seen before to predict risky driving, so it is hypothesized that it will predict receipt of mTickets. METHODS Research Design A one sample observational design was used for this study, and all measures are self-report questionnaires of demographics, psychological variables, and driving behaviours. Participants University students and a community sample were recruited. University psychology students were offered a portion of their undergraduate class credit in exchange for participation, and were recruited using posters displayed in visible locations on campus. As well, research assistants approached undergraduate psychology classes to announce the opportunity to participate in the study. In order to increase the range of ages in our sample, all students were offered the opportunity to obtain further credit if they attempted to recruit a driver who was 35 years or older to complete a second questionnaire package. 400 questionnaire packages were handed out and 319 subjects with valid driver's licenses completed the questionnaire packages and returned them over one month (November, 2003). Three subjects were removed due to inconsistencies in responding1, leaving 316 subjects 31 (122 (38.6%) male and 194 (61.4%) female, mean age = 30.59, age range = 17-67 years). The ethnic composition of the sample was: 35.7% North American/European, 52.5% East Asian, 7.6% South Asian, and 4.1% Other. The mean number of years with a license = 11.96, SD = 11.92, and mean number ofhours driven perweek= 10.13, SD=7.57(mean number hours driven per week for commuting = 5.26, SD = 4.23, work = 1.31, SD = 4.36, and other activities = 3.96, SD = 3.47). The mean number of years of reporting driving behaviours for the sample = 3.94, SD = 1.34, as 43% of the sample had a license for less than five years. Measures Questionnaires were handed out as part of a larger study, used to develop a driving simulation, which is not discussed here. Demographics were collected both to describe the participants in the sample and as independent variables to test hypotheses. These include: age, gender, ethnicity, number of years with a license, number ofhours driving per week for commuting, work and other activities. Personality variables measured include Driving Anger (DAS), Anger Expression (BARQ), Competitive drive and Speed/impatience from Type-A personality (FTAS), Hostile Attitude (HAS), and Trait Anxiety (STAI). Driving behaviours were all measured using self report and include: number of mTickets received over the past five years, number of MVAs involved in over the past five years (major and minor), number of close calls (almost being involved in a MVA) over the past five years, personal rating of driving skill as compared to others on a ten centimeter scale, and personal rating of driving speed as compared to others on a ten centimeter scale. 32 Trait driving anger. The 14-item short form of the DAS (Deffenbacher, Oetting, & Lynch, 1994) is answered with a Likert response scale. It has no subscales and correlates .95 with the original DAS long form. It has a coefficient alpha of internal reliability = .80 as compared to .90 for the original 3 3-item scale. The short form was used in order to reduce the length of the questionnaire package. This scale although originally developed using undergraduate students, was subsequently validated on older drivers (Deffenbacher, Huff, Lynch, Oetting, & Salvatore, 2000). The DAS has been found to correlates moderately with trait anger and trait anxiety. In our sample, all items of the DAS loaded well and exhibited an overall alpha reliability = .88. Trait anxiety. The 20-item trait portion of the State-Trait Anxiety Inventory (STAI; Spielberger, Gorsuch, & Lushene, 1970) was used to assess trait anxiety. Items are rated on a 4 point Likert-type scale (l=almost never - 4=almost always). STAI Alpha coefficient reliabilities range from .89 - .90, and test-retest reliabilities range from .86 (2 weeks) - .66 (3 months). The STAI correlates with many measures of anxiety and has been widely validated. In our sample all items were consistent with each other and revealed an overall alpha reliability = .90. BARQ. The 37-item Behavioural Anger Response Questionnaire (BARQ; Linden, et al., 2003) was used to assess anger expression styles. It consists of 6 subscales: Direct Anger Out, Social Support Seeking, Rumination, Avoidance, Diffusion, and Assertiveness. 6 items load onto each subscale, with the exception of Direct Anger Out which contains 7 items. Internal consistencies for the subscales ranged from .72-.86 across community, student, male, female, Asian, and Caucasian subgroups, with a combined groups internal consistency of .85. One month test-retest stability ranges from .51 for Avoidance to .85 for Direct Anger-Out. 33 Adequate construct and concurrent validity has been obtained. In our sample all items demonstrated good reliability, with the following alpha reliabilities for the subscales: anger-out = .75, assertiveness = .86, social-support seeking = .80, rumination = .73, avoidance = .70, and diffusion = .52. Type-A. The Framingham Scale (FTAS; Haynes et al., 1978) is a 10-item scale and was used to measure two factors of Type-A personality; Competitive Drive, and Speed/Impatience (Houston, Smith, & Zurawsaki, 1986). The Competitive Drive subscale has been related to increased blood pressure reactivity during an interpersonal stressor and includes 2 items, while the Speed/Impatience subscale has been related to anxiety and includes 4 items. Reliability analyses for our sample revealed poor alpha reliability for the competitive drive items = .58, that decreased to .40 for the speed/impatience subscale. Due to this finding, less emphasis has been placed on analyses using these measures. Hostile Attitude. The Hostile Attitude Scale (HAS; Arthur, Garfmkel, & Irvine, 1999) was used to measure hostile cognitions. This measure was developed to identify those individuals who may not express overt anger or aggression, but still have hostile cognitions. This measure of hostility is different from other forms in that it does not measure hostile behaviours. Psychometric properties of the HAS include Cronbach's alpha of internal consistency = .80, adequate convergent validity with other self report measures of hostility, and predictive validity of angiogram outcome. Reliability analysis for the items in our sample uncovered alpha reliability = .79. Procedure All participants were directed to come to the lab where a research assistant offered them a package of questionnaires. If the package was accepted they were given verbal 34 instructions to fill them out as completely as possible and return them to the lab. Consent was obtained by including a letter at the front of the package describing the nature of the study and where to return the questionnaires. The research assistant handing out the questionnaires, informed participants that by returning the questionnaires they were agreeing to consent to the study. The entire questionnaire package was piloted on 10 lab members (research assistants and graduate students) to determine clarity of all questions, as well as to detect spelling and other errors. Errors found during piloting were corrected prior to handing out all questionnaires. Pilot questionnaire data were not be included in the final analysis. In order to deter students from filling out a second questionnaire themselves they were asked to identify the name and phone number of the individual who filled out their second questionnaire package. In keeping with attempts to maximize validity of responding, individual questionnaires were reviewed and any with salient inconsistencies in responding to reverse scored items or aberrant responses to open ended questions were removed from the analysis. Social Desirability of responding was not assessed because this is a confidential questionnaire study, and individuals' responses are not connected to their identities. Hence, I expect that they responded in a largely honest manner. In keeping with attempts to maximize validity of responding, individual questionnaires were reviewed and any with salient inconsistencies in responding to reverse scored items or aberrant responses to open ended questions were removed from the analysis (3 were removed, see footnote 1.). RESULTS Power Analysis 35 Power analysis was conducted to be able to detect effect sizes of .20 while maintaining a power of about 80. This power analysis led me to choose a target sample size of 300 (Glass & Hopkins, 1996). Overview of Analyses Overall means are presented in table 1. First, correlational analyses were run for hostile attitude, driving anger, anxiety, Type-A personality and the BARQ subscales. Separate correlations were run for men and women, as well as for younger and older individuals, although differences were not predicted. Mean differences were analyzed for driving variables and psychological variables to determine differences between men and women, and those 30 years of age and over and under 30. Self reported (1) crash involvement over the past five years, and (2) number of mTickets received over the past five years was treated as the outcome variables for the second part of the analysis. Frequencies of MVA involvement and receipt of mTickets was analyzed to determine age and gender differences. As MVA involvement is a fairly rare occurrence, a number of participants reported no MVA involvement (MVA yes = 155, no = 161; MVA mean = 0.8, mode = 0.0) and the distribution of crash involvement is highly positively skewed. In order to further analyze crash involvement it has been converted into a binary outcome variable with individuals being labeled as "involved in crash in past five years" or "not involved in crash in past five years" (as done in Mesken, Lajunen, & Summala, 2002). As a number of individuals reported having had their license for less than five years, the average number of years of reporting accident involvement was 3.94 years (SD = 1.34 years), with 54.7% having had a license for five years or more. Separate analyses were conducted for overall MVA involvement, minor 36 MVA involvement, and major MVA involvement as different types of MVAs may have differing mechanisms behind them (Summala, 1996). As the outcome variable is dichotomous, logistic regression analyses2 were conducted for the effects of hostile attitude, trait anxiety, driving anger, anger out, rumination, competitive drive, and speed/impatience. Number of hours driven per week was entered in the logistic regression equation, as it allows exposure to be controlled for. Logistic regression analyses were conducted to investigate the interactions between gender, age, BARQ anger-out and BARQ rumination because we know from previous studies that anger-out differs by gender and age, and we believe that rumination moderates anger-out. Logistic regression analysis was chosen as opposed to discriminant function analysis as logistic regression is considered a more robust test with fewer assumptions (Hosmer & Lemeshow, 2000). Discriminant function analysis assumes that group sizes are equal and that data are normally distributed; neither of these conditions were the case for this data set. Number of mTickets received is also very negatively skewed (moving violation yes = 130, no = 186; moving violation mean = 0.8, mode = 0.0), and was converted to a dichotomous outcome variable. Logistic regression was conducted using the personality variables: trait anxiety, driving anger, anger-out, rumination, Type-A competitive drive and speed/impatience, and the interaction between age and sex with anxiety while controlling for number of hours driving per week. Although self reported ethnicity data were collected, data were not analyzed according to ethnicity as acculturation measures were not included. T-tests were carried out to determine if the reported number of hours driven per week differed by age or gender. Men reported driving more hours per week with a mean of 12.27 37 (SD = 9.5J) hours versus a mean of 8.73 hours (SD = 5.66) for women (t = 4.14, p<.001). Those individuals aged 30 or over also reported driving a greater number ofhours per week (mean = 12.28, SD = 9.73) than individuals under 30 years of age (mean = 8.72, SD = 5.51; t = 4.17, p<. 001). Relationships Among the Psychological Variables As a large number of correlations were conducted, alpha for significance was set at p <. 001 in order to control Type-I error. Strict application of the Bonferroni correction would require p to be set at .0008. A second reason for a .001 criterion was to use a cutoff equivalent to correlations conducted in previous validation studies of the BARQ (Linden et al., 2003). Correlations are presented split by age and gender, but none are discussed further as no consistent differences were found. Driving anger was found to be related to greater hostile attitude (r = .33, p<.001), higher trait anxiety (r = .29, p<001), and more speed and impatience (r = .25, p<.001). Hostile attitude was related to greater anxiety(r = .31, p<001), competitive drive(r = .20, p<.001), and speed/impatience(/ = .21, p<001). Speed and impatience was also found to be related to greater anxiety (r = .39, p<.001), and higher competitive drive (r = .37, p<.001) (see table 2.). Of the BARQ subscales (table 3.) only anger out and rumination were found significantly correlate with other personality variables. Anger out was related to higher hostile attitude (r = .49, p<.001), more driving anger (r = .33, p<.001), greater anxiety (r = .21, p<.001), and more speed/impatience (r = .27, p<.001). Rumination was positively related to hostile attitude (r = .28, p<.001), driving anger (r = .24, p<.001), anxiety (r = .52, p<001), and speed and impatience (r = .25, p<.001). 38 Correlations were also conducted for those who report being involved in an MVA, split by those reporting one or more accidents as their fault ("at fault") as opposed to those reporting no accidents as their fault ("not at fault") (see tables 5 & 6.). For those "at fault": hostility was positively related to driving anger (r = .47, p<.00J) and anger out (r =. 53, p<.00J), and driving anger and anger out were positively related (r = .41, p<.001). As well, anxiety was positively related to rumination (r = .57, p<.001). Of the BARQ subscales only rumination correlated significantly with anger out (r = .34, /?<.001). For those "not at fault" hostility correlated with anger out (r = .53, p<.001), anxiety (r = .39, p<.001), and rumination (r = 44, p<.001). As well, anxiety was positively related to rumination (r = .55, p<. 001). Of the BARQ subscales, diffusion was significantly correlated to Avoidance (r = .40,p<001). Overall inter-correlations of the BARQ subscales were also investigated (see table 4). Avoidance was negatively related to anger out (r = -.34, p<.001). Rumination was positively related to anger out (r = .29, p<.001), and to social support (r = .27, p<.001). Avoidance correlated positively with diffusion (r = .23, p<.001). A weaker relationship was found for assertiveness and diffusion (r = .20, p<.001), as this relationship was not found for any of the subgroups of age or gender. Relationship Between Age, Gender, and the Psychological Variables (A-B) Age by gender, 2x2 ANOVAs3 were conducted to examine the psychological variables as they differed across men and women, and older and younger age groups. Due to the multiple ANOVA and Chi-square analyses run, a somewhat conservative level of alpha was used to determine significance (alpha = .025). A strict Bonferroni (Bland & Altman, 1995) correction was not used as this is thought by some to be too conservative and to cause 39 difficulties with Type-II error (Feise, 2002). Given the novelty of this work, an overly conservative approach was considered undesirable. But a .025 level of alpha was chosen because not all tests were driven by specific hypotheses. Exploring the means of the personality variables (table 9.), men (F(3, 310) = 9.48, p<.001), and younger subjects (F(3,310) = 77.25, p<.01) reported greater hostile attitude, but the interaction between age and gender just missed significance (F(3,310) = 5.01, p = .026). Higher driving anger was reported by younger subjects (F(3,311) = 6.96, p<.01), but men and women reported equal amounts of driving anger and no age by gender interaction was found. Older subjects reported significantly lower anxiety than younger subjects. (F(3,316) = 21.00, p<. 001). No significant differences were found for the competitive drive or the speed/impatience subscales of the FT AS. Investigating the BARQ anger expression variables (see table 10.), younger subjects reported greater use of anger out (F(3,3J6j = 6.21, p<.025), rumination (F(3,316) =31.35, p<.001), and social support {F(3,3i6) =11.24, p<.001) than older subjects. Men reported greater use of assertion than females (F(3,316) =5.61, p<.025). Relationship Between Age, Gender, and Receipt of mTickets Over the Past Five Years (A-C) Percent of subjects reporting receipt of mTickets were split by gender and age, and were analyzed using Chi square analyses (see table 7.). This was done in order to examine overall frequency differences between men and women, and between the two age groups: Frequency of tickets for moving violations: More men reported receiving mTickets than women (X2 (1) = 15.58, p< .001). Relationship Between Age. Gender, and Involvement in an MVA Over the Past Five Years (A-D) 40 Percent of subjects reporting involvement in an MVA were split by gender and age, and were analyzed using Chi square analyses (see table 7.). This was done in order to examine overall frequency differences between men and women, and between the two age groups: Frequency of MVA: Fewer older (>30 years) participants reported MVAs than younger participants (<30 years) (X2 (1) = 11.77, p< .001). Frequency of Minor MVA: More men reported minor MVAs than women (X2 (1) = 4.51, p< .05). More participants under 30 years of age reported minor MVAs than older participants (X2 (1) = 12.06, p< . 001). Frequency of Major MVA: More men reported major MVAs than women (X2 (1) = 5.95, p< .025). More participants under the age of 30 reported major MVAs than participants >30 years (X2 (1) = 4.53, p< .05). Relationship of Psychological Variables and Driving Variables (B-C, B-D, and C-D) Logistic Regression4. All personality variables were found to be normally distributed or slightly positively skewed. Age was found to be highly bimodal in nature (figure 2.) and so was dichotomized at a value of 30 (under 30, n = 194; 30 and over, n = 122) which split the groups well, allowing us to conduct comparisons. A second reason for choosing 30 years of age to split the data is that 30 is the age when MVA rates tend to decrease for Canadians (Maxim, 1992). Age was included as a measure of driving experience. Number of years with a license is highly correlated with age (r=.92, p<001) and so is not included in the logistic regression in order to avoid multicollinearity (Hosmer & Lemeshow, 2000). All variables of interest were analyzed using univariate logistic regressions in order to assess their fit for entry into the model. Variables that were not hypothesized to fit in a 41 specific model were also analyzed to rule out their association with that outcome variable. Any variables found to have significance at less than .10 were considered for input into the initial model (Hosmer & Lemeshow, 2000). Significance was determined by chi-square tests of-21og likelihood functions as well as 95% confidence intervals of the odds (being significantly different from 1. At odds = 1, the probability of something occurring equals the probability of it not occurring.) Gender, age, BARQ rumination and STAI were found to predict mTickets (see table 11.). Gender, age, receipt of ticket(s) for a moving violation, DAS and BARQ anger-out were found to predict MVAs overall (see table 12). Minor MVAs were also predicted by these, as well as hostile attitude (see table 13.). Major MVAs were predicted by age, gender, and mTickets received (see table 14.). Logistic regression model building for prediction of MVAs and mTickets. As I wanted to determine if each of the personality variables predicted outcome over and above the important demographic variables of gender, age, and number of hours driving per week, separate logistic regression analyses were run controlling for these (see table 15.) for MVA, minor MVA, major MVA, and mTickets. After controlling for age, gender, and hours driven per week none of the psychological variables predicted receipt of mTickets or major MVA. As age had been found to be related to anxiety and gender was found to be related to receipt of tickets, a model was built to investigate these interactions, but none were found to be significant (see table 16.). As well, anger-out no longer predicted MVAs. For Minor MVAs, anger-out significantly predicted outcome above age, gender, and hours driven per week, and as driving anger just missed significance at p<10 so a larger model including these was tested. 42 Significance for the larger models was set at alpha = .05. Although four logistic regressions were run, alpha was not reduced using a Bonferroni correction as the analysis of the interactions is exploratory and hypothesis driven, and I wanted to avoid being too conservative making a type-II error. Prediction of minor MVA involvement in the past five years. Rumination was not found to be a significant predictor of MVAs, but it is thought to act as a modifier of other anger expression variables so the first model tested included rumination as a modifier of the age by gender by anger out interaction. As our hypothesis consisted of one initial four-way interaction a forced entry method was used for all variables (Jaccard, 2001). Jaccard (2001) suggests the use of a Hierarchically Well Formulated Model in logistic regression just as would be used to fit an interaction term in linear regression. This indicates that proper fit of the four way interaction into the model requires inclusion of all lower order interactions and all univariate variables participating in the interactions. Age and gender are thought to be related to both the outcome of minor MVA involvement (A-D) and the anger-out predictor (A-B), so they are included as part of the interaction. Number of years with a license is excluded from the model due to a correlation above .90 with age. The dichotomous variable of receipt of one or more mTickets over the past five years was included in order to test the hypothesis that personality variables predicted MVA involvement over and above mTickets which is a known predictor of MVAs (C-D). Number of hours driving per week (HPW) was not significantly related to the outcome variable, but it was still included in order to theoretically control for exposure to the driving situation. HPW was mean centered so that all analyses were run holding HPW constant at its mean (10.12 hours per week). 43 Omnibus chi-square and chi-square of the change in the log-likelihood statistic for each step were used to build the model. Significance of individual terms in the model examined both the change in the log likelihood of the model upon removal of a variable as determined by the difference in the model chi-square statistic and significance of predicted odds using 95% confidence intervals calculated from the Standard Error (SE) of the log odds. Wald statistics for individual terms were not used as they may behave in an aberrant manner, often failing to reject the null hypothesis in the face of a significant result (Hauck & Donner, 1977, also see discussion in Hosmer & Lemeshow, 2000). A preliminary full model was first run including: age, gender, HPW, received ticket, DAS, BARQao, BARQrum, BARQaoXGender, BARQaoXAge, AgeXGender, BARQrumXGender, BARQrumXAge, BARQrumXBARQao, AgeXGenderXBARQrum, AgeXBARQrumXBARQao, GenderXBARQrumXBARQao, BARQaoXAgeXGender, and AgeXGenderXBARQrumXBARQao (see table 17). In order to create as parsimonious a model as possible the model was tested for possible removals. Hours driving per week, was not significant in the model but was not considered for removal as theoretically it controls for exposure, which is considered an important mediator variable. In the preliminary model, the four-way interaction of AgeXGenderXBARQrumXBARQao was not significant (X2 (1) = 0.51, p>.10). All interactions with rumination were tested and removed form the model as they were not found to be significant. Next, BARQrum was found not to add significantly to the model (X2 (1) = . 78, p>. 10), and so was removed from the model. Driving anger became non-significant as tested by 90% confidence interval and change in chi-square upon removal from the model (X2(l)= 1.49, p>.10), therefore it was removed. Finally the three-way interaction between age, gender, and anger-out was tested for significance using both a 90% 44 confidence interval and change in model chi-square upon removal from the model and was found to predict minor MVA (X2 (1) = 3.79, p=05). Received mTicket was significant using a 95% confidence interval so was retained. All other variables were subordinate to and involved in the three-way interaction and were not considered for removal. The overall model was found to be a good fit as determined by overall final model chi-square (X2(9) = 36.52, p<.00J). Hosmer and Lemeshow Goodness of Fit was acceptable (X2 (8) = 6.33, p> .JO), and examination of standardized residuals indicated no residuals outside of the acceptable range. Those who had previously received an mTicket, had higher odds of having been involved in a minor MVA over the past five years (odds ratio = 1.89, p<01). Note that mTickets were not only for speeding, but include all types of tickets for moving violations. No effect sizes are reported as no good measure of effect size has been demonstrated for use with logistic regression models, and estimates of effect sizes from well fitted logistic regression analyses are found to be low when compared to effect sizes typically encountered with good linear regression models (Hosmer & Lemeshow, 2000). Analysis of the age, gender. BARQ anger out interaction in predicting minor MVA involvement. A significant three-way interaction was found for gender by age by anger-out (Odds = 1.24, p = .05), and this was further examined (see figure 3.). When interactions are included in a logistic regression model, coefficients (B) for variables in the equation are no longer representative of main effects, but represent log odds for that variable at levels of the other variables in the equation as run with indicator coding. Separate logistic regressions were run, with each level of each of the dichotomous variables taking a turn as the indicator for that run (2 variables with 2 levels each led to 4 logistic 45 regressions). Separate calculations were also made for three levels of aggression; mean, and one standard deviation above and below the mean. Anger out significantly predicted involvement in a minor MVA for women under 30 years of age (odds = 1.47 for a one SD increase in anger out, p<.05), but did not predict minor MVA involvement for men under 30 years of age (odds = 1.22 for a one SD increase in anger out, p>.10; see table 14.). A tendency was also found for anger out to predict minor MVAs for men >30 years of age (odds = 1.69 for a one SD increase in anger out, p<. 10) but this was not significant at .05. Although the BARQao multiplier for older women appears to be sloping downward (figure 3.) as aggression increases, this slope was not significant (B = 0.93, p>.05). Odds Ratios of the odds BARQao coefficients were constructed (see table 18.) to explore the three-way interaction (Jaccard, 2001). The slope for older women was found to be significantly lower than, and trended in the opposite direction to that of younger men (Odds ratio = 1.12, p < .05), and was found to approach a similar significant difference when compared to older men (Odds ratio = 1.17, p < .10), and to younger women (Odds ratio = 1.19,p< .05). Examining the table of odds ratios (see table 19.) comparing all groups at the three levels of aggression, it appears that as aggression increases from one standard deviation below the mean (lo) to one standard deviation above the mean (high), younger women's odds of having an MVA in comparison to the odds of an older woman increase from an Odds Ratio of 1.21 (not significantly different from 1.0) to and Odds Ratio of 5.13 (p < .01). The series of odds ratios for older men in comparison to older women reveal no significant difference at lo and mean aggression, but approach significance at hi aggression (OR = 3.59, 46 p < .10). The series of odds ratios comparing younger men to older women reveal steadily increasing odds ratios as aggressive responding increases. For young women, there is a significant increase in odds of MVA for every one unit increase in BARQao (BARQao multiplier; Odds = 1.08, p < .05). As well there is a tendency for the same relationship to exist for older men (Odds = 1.11, p < .10), but this was not significant. DISCUSSION The objective of this present thesis was to explore the relationship between factors associated with anger, including anger expression, and apply these to account for individual differences in risky driving behaviors. Anger expression as measured by the Behavioral Anger Response Questionnaire was compared to measures of hostile attitude, trait driving anger, Type-A personality, and anxiety in order to further validate this new measure of anger expression in the realm of motor vehicle driving. As anger is known to impact risky driving and motor vehicle accidents, driving-related and overall anger was then used to attempt to predict traffic violations and motor vehicle accidents. The Type-A personality subscales were found to have quite low reliability, so analyses did not centre on these variables. Moderate relationships were found among the psychological variables, and differences were found based on age and gender. As well, frequencies of receipt of mTickets and involvement in MVAs were found to differ among men and women, and younger and older participants. After controlling for age, gender and hours driven per week, none of the psychological variables predicted receipt of one or more mTickets over the past five years or major MVA involvement over the past five years. Minor MVA involvement was related to BARQ anger-out and receipt of a ticket over the past five years, and BARQ anger out was found to interact with age and gender in predicting minor MVA involvement. 47 Relationships Among the Psychological Variables The three constructs of anger; hostile attitude, driving anger, and anger out showed a positive yet moderate inter-correlations. As was expected, anxiety and rumination were also moderately related to each other and to the three anger constructs. Correlations among the three anger constructs were greater in magnitude than those of the anger constructs with anxiety or rumination, suggesting that they are not measuring the same constructs as anxiety and rumination, but that they are related. The moderate correlations among the anger constructs lends support to the impression that these three factors, although related, are not measuring the same construct. If they had been measuring the same phenomenon they would have had higher correlations. As well, they would have behaved in exactly the same manner in all other analyses. For example, although younger subjects reported greater hostile attitude, driving anger, and anger out, men and women reported equal driving anger, and anger out, while men reported greater hostile attitude. Anger out appears to be related to both driving anger and hostile attitude, but moderate correlations lend credence to the idea that not all individuals who become angry while driving are also frequent aggressive responders in their every day lives. This relationship has been seen before with other measures of hostility, state anger, and aggression (Pope, Smith, & Rhodewalt, 1990). BARQ Rumination was most highly related to anxiety, which is to be expected as rumination is a known characteristic of anxiety. BARQ rumination was less strongly related to the three anger variables, and there is very little research on rumination in the area of anger. Some rumination is expected to extend angry feelings and increase the anger experience and to moderate other forms of anger expression (Hogan & Linden, 2003). Low to moderate correlations of rumination with the anger constructs denote that some individuals may be ruminating when angry and this may 48 either be a response to the social unacceptability of aggressive behaviour, or that rumination amplifies or extends the feelings of anger in some individuals (Linden et al., 2003). This may affect those individuals who are high on any one of the three facets of anger, causing greater hostile attitude, greater driving anger, or greater aggressive responding when angry. An interesting finding was that for individuals involved in an MVA over the past five years, those that reported the accident as not their fault reported greater hostile attitude, greater anxiety and greater rumination. Driving anger was not related here, but greater anger out was related to greater hostile attitude. For those who claimed the accident as their fault, driving anger, hostile attitude, and anger out were correlated, but anxiety and rumination were not related to these anger variables. This relationship may signal that those who are higher on anxiety may be less likely to report an accident as their own fault, or possibly that those who are higher on anxiety show less driving anger and are more likely to be involved in an accident that is not their fault (Lerner & Keltner, 2001). None of the other BARQ subscales were found to be consistently and significantly related to the personality variables included. It appears that in this sample both higher driving anger and hostile attitude overlap with greater use of BARQ anger out and BARQ rumination as measured by self report. Looking at the relationships among the BARQ subscales, the use of rumination was found to be greater for both those who more often used anger-out and social support seeking when angry. This relationship may exist as social support seeking requires the individual to keep the angering event in their mind until they are able to share it with another individual. This requires rumination or thinking about the angering event sometimes long after it has come to pass. As discussed previously, anger out on the other hand may also be moderately 49 related to rumination as this style of expression may require the person to think about their actions or the person may continue to think about the angering event (Linden et al., 2003). Greater use of avoidance was related to less use of anger out, and this moderate correlation supports the suggestion that in general those using anger out when angry may be less likely to hold anger in. This relationship lends support to the hypothesis that these anger expression variables exist on a continuum with anger in and anger out anchoring the ends of the continuum. This goes against the idea that those high in anger use both anger in and anger out at equal levels (Deffenbacher, 1992). Among those reporting accident involvement, relationships among the personality variables differed between those who reported the accident as their fault versus those who reported the accident as not their fault. For both groups anxiety was related to rumination as expected. For those reporting one or more accidents as their fault, hostility, driving anger, and anger-out were positively inter-related. As well, those reporting greater rumination, reported greater anger-out. For those reporting accidents as not their fault, hostility was related to anger out, anxiety, and rumination. Those who reported being at fault for accidents hostility was related to driving anger, but not to anxiety. For those not at fault, hostility and anger out were not related to driving anger, but hostility was related to anxiety and rumination. Those hostile individuals not at fault for accidents may experience less driving anger and may experience more anxiety. This may lead them to drive more carefully and take less risk (Lerner & Keltner, 2001). Relationship Between Age. Gender, and BARQ anger-out (A-B) Those under 30 years of age reported greater use of anger out, but no gender differences were found in the means between the two groups. Younger individuals are often 50 found to show higher levels of aggression or expressed anger. It is well documented that gender differences in aggression exist between men and women. Men tend to be more physically aggressive than women, but it is thought that women may be equally as aggressive verbally or in writing, and may be more indirectly aggressive than men (Bettencourt & Miller, 1996; Bjorkqvist, Osterman, & Kaukianen, 1992). It is thought that through sociocultural learning women are raised as "peacemakers" and that it is unacceptable for women to be aggressive (Hokanson, Willers, & Koropsak, 1968, Linden, 1993; Linden & Feurstein, 1981). It appears that in our sample, men and women reported equal amounts of anger-out behaviour when feeling angry. Relationship Between Age. Gender, and Receipt of mTickets Over the Past Five Years (A-C) A greater number of men reported receiving mTickets than women. This is often explained by the fact that men drive more often than women, and this was the case for our sample as well. Men and older individuals reported driving significantly more per week, but no age differences were reported for receipt of mTickets over the past five years. Relationship Between Age, Gender and Involvement in an MVA Over the Past Five Years (AJ3) Participants under 30 years of age and men reported more MVAs overall, minor MVAs and major MVAs than participants 30 years of age or over and women. As men and older individuals reported driving significantly more hours per week, both would be expected to report more MVA involvement. This holds true for men, but not older individuals. Those 30 years of age and over reported less accident involvement even though they had greater exposure to the road. This is may be due to the fact that those 30 years and over are more 51 careful in their vehicles and take less risks than younger individuals, or just that they have more experience driving (Matthews & Norris, 2002). Relationship of Psychological Variables and Driving Variables (B-C, B-D, and C-D) In our sample we expected that driving anger would predict the receipt of mTickets, as a proxy for risky driving (Deffenbacher, Huff, Lynch, Oetting, & Salvatore, 2000), but this was not the case. It is often assumed that driving anger leads to risky driving which is then punished more often by receipt of tickets. In actuality not all people who commit driving violations receive tickets, and not all driving violations are committed due to emotionality. Most individuals who speed or run a light do so for reasons other than anger. It may be important to distinguish between aggressive violations which appear to be on the rise and highway-code violations (Lajunen et al., 1998). Aggressive violations are more likely paired with increased physiological arousal that could influence perception and information processing while driving (Deffenbacher et al., 1994). This could theoretically lead to a greater risk of an MVA both your fault and not your fault. There was a non-significant tendency for driving anger to predict receipt of mTickets, and as driving anger has been seen to be related to receipt of mTickets in the past, this may be a power issue in our data set. It also may be explained by the idea that those who receive tickets more often, may be speeding more often, or in a hurry, and may have a greater tendency to feel impeded by other drivers, which may lead to greater driving anger in those individuals (Deffenbacher et al., 1994). Receipt of a ticket in the past five years and BARQ anger-out moderated by age and gender, predicted minor MVA involvement, but did not predict major MVA involvement or MVA involvement overall. Those who had received a ticket for a moving violation in the 52 past five years had almost twice the odds of being involved in a minor MVA (odds ratio = 1.89, p<01). Receipt of tickets for moving violations was conceptualized as a proxy measure of risky driving, and has often been found to be related to MVA involvement (e.g., Jonah, 1986; Reason, Manstead, Stradling, Baxter, & Campbell, 1990; Parker, Reason, Manstead, & Stradling, 1992; West, French, Kemp, & Elander, 1993; Rajalin, 1994; Parker, Reason, Manstead, & Stradling, 1995; Cooper, 1997). In our model, aggression significantly predicted an increase in odds as anger-out increased but not for all four groups as split by age and gender. This interaction was what I was searching for but relationships were not specifically predicted. In our sample, aggression significantly predicted an increase in odds as anger-out increased for women less than 30 years of age. A similar relationship was seen in older men at p< 10, but this relationship was not seen for men less than 30 years or women 30 years and over. Odds of having a minor MVA increased more quickly as anger-out increased for younger women, older men, and younger men, in contrast to older women (see figure 3.). At low levels of anger out (1 standard deviation below the mean) younger men have significantly higher odds of an MVA as compared to younger women and older men, but this difference disappears at high anger-out. At high levels of anger-out (+1 standard deviation above the mean), younger men and younger females have significantly higher odds of having an MVA than older individuals at all levels of aggression. Anger-out did not predict MVA involvement for young men, and this may be due to the fact that young males are known to be not only more aggressive overall, but also involved in more MVAs. Young drivers may also take more risks while driving which may contribute to greater MVA involvement (Matthews & Norris, 2002). Due to this fact, BARQ anger-out 53 did not differentiate those who were involved in accidents from those who were not for males less than 30 years of age. Aggressive responding when angry did predict MVA involvement for women less than 30 years of age. Women in general are thought to be less aggressive than men, but it is well known that younger drivers are involved in more MVAs and are more aggressive overall. Young drivers are more likely to assume there is lower risk in driving situations by rating high skill activities as less hazardous than they really are (Elander, West, & French, 1993). It appears that for our sample of female drivers under 30 years of age, those who use anger-out more often to express their anger, more often become involved in minor MVAs. This may signal that they use anger-out to express their anger on the road, and that becoming aggressive on the road somehow puts them at greater risk for minor MVA involvement, although no causality was tested with this self report questionnaire model. Why was this trend for anger out to predict MVA involvement not seen in women 30 years and over? Men and women most likely become just as angry, but express it differently. Men express more physical aggression, and more verbal aggression (Harris, 1992). As well, women who are normally aggressive responders when angry may not be as aggressive in the car. They may be fearful of retaliation or have less confidence behind the wheel as they often drive less than men. Additionally, anxiety was more strongly related to hostile attitude, driving anger, and anger out for women and those 30 years and over. These individuals may be experiencing more anxiety as well as more anger. The anxiety may lead to less risky behaviour or aggression on the road. It has been seen that angry people show a stronger sense of control and certainty than fearful people. Fearful people assess risk as higher and are more 54 risk aversive, whereas angry people assessed risk as lower and were more risk seeking (Lerner & Keltner, 2001). For these drivers, aggression may not be intended to lead to an MVA (Brewer, 2000). In the past aggression has been seen to be related to minor, but not major MVA involvement, and it is postulated that the mechanism behind involvement in a minor MVA is different from that of a major MVA (Summala, 1996). Perhaps those who are responding aggressively when angry in the automobile are experiencing greater physiological arousal and this is what is interfering with their ability to negotiate traffic successfully (Deffenbacher, Oetting, & Lynch, 1994). This increased arousal may have an effect on perception and information processing, as well as expectations and preferred actions (Naatanen & Summala, 1976), the latter impacting how the driver behaves and possibly increasing accident risk (Mesken, Lajunen, & Summala, 2002). Drivers with good records, less MVAs and mTickets, have been shown to have less arousal in a driving situation as measured by galvanic skin response (Brown & Huffman, 1972). As well, one study of unmarried male drivers found that MVAs were related to impulsiveness and number of miles driven (exposure). Confidence tended to increase with age and older drivers in this sample took more chances when angry. They assumed that lack of confidence may restrict the impulsiveness of the younger drivers (Schuman & Pelz, 1967). This may be true for some, possibly more so for females than males. Driving anger came close, but did not predict MVA involvement, and so was not retained in the model. It may be that it does not matter how angry you get when you are driving, but how you express that anger. We may all respond differently when angry depending upon our personality and the situation. This data supports the idea that for younger 55 women and possibly older men, those who become aggressive in their everyday lives have a greater risk of minor MVA involvement. This may be due to the fact that they are also aggressive on the road and this interferes with their ability to drive successfully. Clinical significance is not reported with the data in my study, as no adequate effect size has been developed for use with logistic regression. Some that have been suggested give numbers that are much lower than that which is obtained using linear regression, even with a very good fitting model (Hosmer & Lemeshow, 2000). One comparison that I can make is that of amount of log likelihood ratio explained by parts of the model. Age, gender, and hours per week driven are considered consistent and strong predictors of MVA and in our model they explain 20.34 of the -2LL. Addition of aggression and all of the interaction terms resulted in a further 10.2 of the -2LL being explained, half the amount of the well known predictors (Hosmer & Lemeshow, 2000). Limitations There are a number of limitations to this current study. All findings were based on data collected from self report questionnaires. Self reports on personality measures possess unique threats to validity. Self reported crash involvement and speeding tickets were measured as opposed to measures which may not be biased by reporting errors and memory bias. I endeavored to find a measure of validity for the self reported data I collected on MVAs and receipt of mTickets for moving violations and was unable to find any. The subject of interest for this study was all MVAs that a person had been involved in, including even the most minor MVAs. The Insurance Corporation of British Columbia (ICBC), and most other jurisdictions collect information regarding traffic collisions from both police reports and insurance reports. Both of which exclude many traffic collisions due to non-reporting of 56 collisions below a certain threshold and non-reporting of low severity collisions. As well, most reported statistics only include MVAs in which there has been a physical injury or fatality. ICBC estimates that this non-reporting results in the exclusion of thousands of traffic collisions from its estimates (ICBC, 2001). One study examining this discrepancy revealed that self report data of crashes and mTickets were much higher than that which was recorded in their state records (Arthur et al., 2001). They suggested that self report data may include all types of accidents, even those that were not reportable, and may be more useful if the researcher is interested in studying all crashes as opposed to only more serious crashes. One study using self report of crash involvement and receipt of tickets found extremely high test-retest reliability of r = .96 (Arthur & Doverspike, 1992). It is for this reason that self reported crash involvement and receipt of mTickets was used as an outcome variable. A further study cross validated self reports of driving behaviours (MVAs and moving violations received) and found support for the validity of self report (Parker et al., 1995). As well, in support of validity in our sample, half of the MVAs reported were claimed as the reporting individual's fault. Self report data would be seen as less valid if more than half of all participants claimed that their accidents were not their fault. A number of studies make the observation that insurance and police records are often thought of as complete, but not all MVAs are necessarily reported (Burns & Wilde, 1995; Cooper 1997). Overtime, the MVA damage amount required before the police need to be contacted fluctuates, changing reporting. As well, not all MVAs are reported to insurance companies as some people would rather not report, MVAs may not meet a certain dollar value or injury threshold, or the crash may occur in an other jurisdiction (McGuire, 1973; Smith 1976). 57 Involvement in an MVA and receipt of a ticket for a moving violation were determined by asking people to report their records for the past 5 years. Some of the drivers had not possessed a license for that long. Estimates for younger individuals therefore are lower than those for older individuals. I did not use a multiplicative correction factor as we felt this would inflate MVA rates and possibly lend liberal bias to the data analysis (Parker et al., 1995). I felt that because MVAs are such a rare event, this would not terribly bias MVA numbers reported. As well, as this was a self report, questionnaire, non-experimental study, directions of the relationships found can not be determined, and causality can not be concluded. Directions of relationships in the logistic regression analyses were chosen according to previous research and hypotheses, and could very well have been exchanged to test opposite direction hypotheses. Summary and Conclusions Overall, further construct validity was obtained for the BARQ subscale of anger out by relating it to hostility, driving anger, and anxiety. None of the psychological variables were found to predict receipt of a ticket for a moving violation, over and above that predicted by age, gender and number ofhours driving per week. Both receipt of a ticket for a moving violation and BARQ anger out interacting with age and gender predicted involvement in a minor motor vehicle accident, but not a major MVA. Further analysis of the interaction revealed that BARQ anger out predicted minor MVA involvement for women under the age of 30 only. Future research should consider the relationship between driving anger and aggressive responding along with physiological arousal during driving simulation and actual 58 driving situations. Postulations that physiological arousal may interfere with safe driving (Brown & Huffman, 1972; Deffenbacher, 1994) need to be further examined. It may be that although self reported driving anger differences do not exist between men and women, actual anger experiences or physiological arousal differences do exist while driving. Further research in this area could also examine the role that anger expression plays in driving, especially looking at gender differences in aggressive responding on and off the road. As well, future research should look at the flexibility of individuals anger expression repertoires both in driving and in other arenas. FOOTNOTES 1 One subject was removed due to aberrant responses on open ended questions. A second respondent was removed due to inconsistencies in responding on the reverse scored items of the STAI. A third respondent was removed due to an unusually high MVA rate of 15 accidents over the past five years. 21 have decided to use logistic regression as opposed to linear regression because our outcome variable is dichotomous. An assumption of linear regression is that the relationship between the variables is linear and with a dichotomous outcome variable this assumption is usually violated (Berry, 1993): Violations of assumptions may lead to invalid results. Linear regression could still be used if we were to transform our dichotomous data using a logarithmic function, but as this is one of the principles behind logistic regression, in order to overcome the violation of the assumption of linearity, the logistic transformation is unnecessary and can often be confusing. As well, as our research question principally examines which factors differentiate the two groups, logistic regression is best able to do this in comparison to analysis of variance. A further benefit of using logistic regression is that continuous and/or categorical predictors can be easily used (Hosmer & Lemeshow, 2000). 3 As well, Bartlett's Test was conducted for each ANOVA and if found to be significant the data were examined for Behren's Fischer problem. If positive, a Welch's modification of ANOVA was conducted to decrease bias (Glass & Hopkins, 1996). 60 4 Logistic regression produces two summary statistics, the goodness of fit statistic and the log likelihood statistic. The goodness of fit statistic is distributed as chi-square and is a measure of how well the data fit the model. The log likelihood statistic is analogous to the error sum of squares in linear regression and is normally reported as -21og likelihood (-2LL) as this follows a chi-square description, making it possible to compare values to tabled chi-square determining probabilities of obtaining values by chance alone. 61 REFERENCES Aberg, L., & Rimmo, P. A. (1998), Dimensions of aberrant driver behaviour. Ergonomics, 41, 39-56. Alexander, F. (1939). Emotional factors in essential hypertension. Psychosomatic Medicine, 1, 175-179. Alexander, E. A., Kallail, K. J., Burdsal, J. P., & Ege, D. L. (1990). Multifactorial causes of adolescent driver accidents: Investigation of time as a major variable. Journal of Adolescent Health Care, 11, 413-417. Arnold, M. B. (1960). Emotion and personality. New York: Columbia University Press. Arthur, W., & Doverspike, D. (1992). Locus of control and auditory selective attention as predictors of driving accident involvement: A comparative longitudinal investigation. Journal of Safety Research, 23, 73-80. Arthur, H. M., Garfinkel, P. E., & Irvine, J. (1999). Development and testing of a new hostility scale. Canadian Journal of Cardiology, 15, 539-544. Arthur, W., Jr., Tubre, T., Day, E. A., Sheehan, M. K., Sanchez-Ku, M. L., Paul, D., Paulus, L., & Archuleta, K. (2001). Motor vehicle crash involvement and moving violations: Convergence of self-report and archival data. Human Factors, 43, 1-11. Beirness, D. J. (1993). Do we really drive as we live? The role of personality factors in road crashes. Alcohol, Drugs and Driving, 9, 129-143. Berkowitz, L. (1993). Aggression: It's causes consequences, and control. New York: McGraw-Hill. Berry, W. D. (1993). Understanding regression assumptions. Sage university paper series on quantitative applications in the social sciences, 07-092. Newbury Park, CA: Sage. Bettencourt, B. A., & Miller, N. (1996). Gender differences in aggression as a function of provocation: A meta-analysis. Psychological Bulletin, 119, 422-447. Biaggio, M. K., Supplee, K., & Curtis, N. (1981). Reliability and validity of four anger scales. Journal of Personality Assessment, 45, 639-648. Bishop, G. D., & Quah, S. (1998). Reliability and validity of measures of anger/hostility in Singapore: Cook & Medley Ho Scale, STAXI and Buss-Durkee Hostility Inventory. Personality and Individual Differences, 6, 867-878. Bjorkqvist, K., Osterman, K., & Kaukianen, A., (1992). The development of direct and indirect aggressive strategies in males and females. In K. Bjorkqvist & P. Niemela (Eds.), Of mice and women: Aspects offemale aggression (pp. 51-64). New York: Academic Press. Blanchard, E. B., Barton, K. A., & Malta, L. (2000). Psychometric properties of a measure of aggressive driving: The Larson Driver's Stress Profde. Psychological Reports, 87, 881-892. Bland, J. M., & Altman, D. G. (1995). Multiple significance tests: the Bonferroni method. British Medical Journal, 310, 170. Brewer, A. M. (2000). Road rage: What, who, when, where, and how? Transport Reviews, 20, 49-64. Brown, J. D., & Huffman, W. J. (1972). Psychophysiological measures of drivers under actual driving conditions. Journal of Safety Research, 4, 172-178. Burns, P. C, & Wilde, G. J. S. (1995). Risk taking in male taxi drivers: Relationships among personality, observational data and driver records. Personality & Individual Differences, 18, 267-278. 63 Buss, A. (1961). The Psychology of Aggression. New York: Wiley. Buss, A. H., & Durkee, A. (1957). An inventory for assessing different kinds of hostility. Journal of Consulting Psychology, 21, 1061-1070. Byrne, D. G, Rosenman, R. H, Schiller, E., & Chesney, M. A. (1985). Consistency and variation among instruments purporting to measure the Type A behavior pattern. Psychosomatic Medicine, 47, 242-261. Caprara, G. V., Barbaranelli, C, & Comrey, A. L. (1992). A personological approach to the study of aggression. Personality and Individual Differences, 13, 77-84. Cook, W. W., & Medley, D. M. (1954). Proposed hostility and pharisaic-virtue scales for the MMPI. Journal ofApplied Psychology, 238, 414-418. Cooper, P. J. (1997). The relationship between speeding behaviour (as measured by violation convictions) and crash involvement. Journal of Safety Research, 28, 83-95. Costa, P. T., Jr., & McCrae, R. R. (1992). RevisedNEO Personality Inventory (NEO PI-RTM) and NEO Five Factor Inventory (NEQ-FFI) professional manual. Odessa, FL: Psychological Assessment Resources. Costa, P.T., Jr., McCrae, R. R.. & Dembroski, T. M. (1989). Agreeableness versus antagonism: Explication of a potential risk factor for CHD. In A. Siegman & T. M. Dembroski (Eds.), In search of Coronary Prone Behavior (pp.41-63). Hillsdale, NJ: Erlbaum. Cowley, J. J. (1956). Aggression and accidents. Journal of the National Institute of Personnel Research. South African Council Scientific & Industrial Research, 6, 144-152. Davidson, K., MacGregor, M., Stuhr, J., & Gidron. Y. (1999). Increasing constructive anger behavior decreases resting blood pressure: A secondary analysis of a randomized 64 controlled hostility intervention. International Journal of Behavioral Medicine, 6, 268-278. DeAngelis, T. (2003). When anger's a plus, APA Monitor on Psychology, 34, 44-45. Deffenbacher, J. L. (1992). Trait anger: Theory, findings, and implications. In C. D. Spielberger, J. N. Butcher, et al. (Eds.) Advances in Personality Assessment Vol. 9, (pp. 177-201). Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. Deffenbacher, J. L., Huff, M. E., Lynch, R. S., Oetting, E. R., & Salvatore, N. F. (2000). Characteristics and treatment of high anger drivers. Journal of Counseling Psychology, 47, 5-17. Deffenbacher, J. L., Lynch, R. S., Deffenbacher, D. M., & Oetting, E. R. (2001). Further evidence of reliability and validity for the Driving Anger Expression Inventory. Psychological Reports, 89, 535-540. Deffenbacher, J. L., Lynch, R. S., Oetting, E. R., & Yingling, D. A. (2001). Driving anger: Correlates and a test of state-trait theory. Personality & Individual Differences, 3, 1321-1331 Deffenbacher, J. L., Oetting, E. R., & Lynch, R. S. (1994). Development of a driving anger scale. Psychological Reports, 74, 83-91. Dimsdale, J. E., Pierce, C, Schoenfeld, D., Brown, A., Zusman, R., & Graham, R. (1986). Suppressed anger and blood pressure: The effects of race, sex, social class, obesity, and age. Psychosomatic Medicine, 48, 430-436. Dobson, A., Brown, W., Ball, J., Powers, J., & McFadden, M. (1999). Women drivers' behaviour, socio-demographic characteristics and accidents. Accident Analysis & Prevention, 31, 525-535. 65 Dollard, J., Doob, L., Miller, N. Mowrer, O., & Sears, R. (1939). Frustration and Aggression. New Haven: Yale University Press. Ellison, P. A., Govern, J. M., Petri, H.L., & Figler, M. H. (1995). Anonymity and aggressive driving behaviour: A field study. Journal of Social Behavior and Personality, 10, 265-272. Everson, S. A., Goldberg, D. E., Kaplan, G. A., Julkunen, J., & Salonen, J. T. (1998). Anger expression and incident hypertension. Psychosomatic Medicine, 60, 730-735. Feise, R. J. (2002). Do multiple outcome measures require p-value adjustment? BioMed Central Medical Research Methodology, 2, 8-12. Fong, G, Frost, D., & Stansfeld, S. (2001). Road rage: A psychiatric phenomenon? Social Psychiatry and Psychiatric Epidemiology, 36, 277-286. French, D. J., West, R. J., Elander, J., & Wilding, J. M. (1993). Decision-making style, driving style, and self-reported involvement in road traffic accidents. Ergonomics, 36, 627-644. Friedman, H. S., Tucker, J. S., & Reise, S. P. (1995). Personality dimensions and measures potentially relevant to health: A focus on hostility. Annals of Behavioral Medicine, 17, 245-253. Furnham, A., & Saipe, J. (1993). Personality correlates of convicted drivers. Personality & Individual Differences, 14, 161-170. Gentry, W. D., Chesney, A. P., Gary, H. E., Jr., Hall, R. P., & Harburg, E. (1982). Habitual anger-coping styles: I. Effect on mean blood pressure and risk for essential hypertension. Psychosomatic Medicine, 44, 195-202. 66 Glass, G. V., & Hopkins, K.D. (1996). Statistical methods in educations and psychology. Needham Heights, MA: Allyn & Bacon. Guerin, B. (1994). What do people think about the risks of driving? Implications for traffic safety interventions. Journal of Applied Social Psychology, 24, 994-1021. Gulian, E., Debney, L. M., Glendon, A. I., Davies, D. R, & Matthews, G. (1989). Coping with driving stress. In: F. McGuigan, W. E. Sime, & J. M. Wallace (Eds), Stress and tension control, Vol. 3 (pp. 173-186). New York: Plenum Press. Harburg, E., Gleiberman, L., Russell, M., & Cooper, M. L. (1991). Anger-coping styles and blood pressure in Black and White males: Buffalo, New York. Psychosomatic Medicine, 53, 153-164. Harris, M. (1992b). Sex and ethnic differences in past aggressive behavior. Journal of Family Violence, 7, 85-102. Harris, J. A. (1997). A further evaluation of the Aggression Questionnaire: Issues of reliability and validity. Behavior Research and Therapy, 35, 1047-1053. Hauck, W. W., & Donner, A. (1977). Wald's test as applied to hypotheses in logit analysis. Journal of the American Statistical Association, 72,851-853. Haynes, S. G, Levine, S., Scotch, N., Feinleib, M., & Kannel, W. B. (1978). The relationship of psychosocial factors to coronary heart disease in the Framingham study: I. Methods and risk factors. American Journal of Epidemiology, 107, 362-383. Health Canada. (2003). Safety and injury. Retrieved February, 15th, 2003, from: http://www.hc-sc.gc.ca/english/lifestyles/injury.html Hemenway, D., & Solnick, S. J. (1994). Fuzzy dice, dream cars, and indecent gestures: Correlates of driver behavior. Accident Analysis & Prevention, 25, 185-195. 67 Hennessy, D. A., & Wiesenthal, D. L. (1999). Traffic congestion, driver stress, and driver aggression. Aggressive Behavior, Vol. 25(6), 409-423. Hijar, M, Carrilo, C, Flores, M., Anaya, R., Lopez, V. (2000). Risk factors in highway traffic accidents: A case control study. Accident Analysis and Prevention, 32, 703-709. Hogan, B. E., & Linden W. (2003). Anger response styles and blood pressure: At least don't ruminate about it. Annals of Behavioral Medicine, in press. Hokanson, J., Willers, K., & Koropsak, E. (1968). The modification of autonomic responses during aggressive interchange. Journal of Personality. 36, 386-404. Hosmer, D. W., & Lemeshow, S. (2000). Applied Logistic Regression. 2nd ed.. Wiley series in probability and statistics. New York, NY: John Wiley and Sons Inc.. Houston, B. K., Smith, T. W., & Zurawski, R. M. (1986). Principal dimensions of the Framingham Type A scale: Differential relationships to cardiovascular reactivity and anxiety. Journal of Human Stress, 12, 105-112. ICBC (2001). Traffic collision statistics. Retrieved on January, 20, 2003. from: B.C.%20Traffic%20Collision%20Statistics%202001.pdf# Iversen, H., & Rundmo, T. (2002). Personality, risky driving and accident involvement among Norwegian drivers. Personality & Individual Differences, 33, 1251-1263. Johnson, E. H., & Spielberger, C. D. (1992). Assessment of the experience, expression, and control of anger in hypertension research. In E. H. Johnson, W. D. Gentry, & S. Julius, (Eds.) Personality, Elevated Blood Pressure, and Essential Hypertension, (pp. 3-24). Washington, DC: Taylor & Francis. 68 Jonah, B. A. (1986). Accident risk and risk-taking behaviour among young drivers. Accident Analysis & Prevention, 18, 255-271. Jonah, B. A. (1997). Sensation seeking and risky driving: A review and synthesis of the literature. Accident Analysis & Prevention, 29, 651-665. Jorgensen, R. S., Johnson, B. T., Kolodziej, M. E., & Schreer, G. E. (1996). Elevated blood pressure and personality: A meta-analytic review. Psychological Bulletin, 120, 293-320. Julius, M., Harburg, E., Cottington, E. M., & Johnson, E. H. (1994). Anger-coping types, blood pressure, and all-cause mortality: A follow-up in Tecumseh, Michigan (1971-1983). In A. Steptoe, & J. Wardle (Eds.), Psychosocial Processes and Health: A reader (pp. 215-234). Cambridge, Great Britain: Cambridge University Press. Lai, J. Y., & Linden, W. (1992). Gender, anger expression style, and opportunity for anger release determines cardiovascular reaction to and recovery from anger provocation. Psychosomatic Medicine, 54, 297-310. Lajunen, T., & Parker, D. (2001). Are aggressive people aggressive drivers? A study of the relationship between self-reported general aggressiveness, driver anger and aggressive driving. Accident Analysis & Prevention, 33, 243-255. Lajunen, T., Parker, D., & Summala, H. (1999). Does traffic congestion increase driver aggression? Transportation Research Part F: Traffic Psychology & Behaviour, 2, 225-236. Larkin, K. T., Semenchuk, E. M., Frazer, N. L., Suchday, S., Taylor, R. L. (1998). Cardiovascular and behavioral response to social confrontation: Measuring real-life stress in the laboratory. Annals of Behavioral Medicine, 20, 294-301. Lerner, J. S., & Keltner, D. (2001). Fear, anger, and risk. Journal of Personality & Social Psychology, 81, 146-159. Linden, W. (1993). Sex differences and the social conflict model of anger expression. Paper presented at the 101st Annual Convention of the American Psychological Association, August, Toronto, Ontario, Canada. Linden, W. & Feuerstein, M. (1981). Essential hypertension and social coping behavior. Journal of Human Stress, 7, 28-34. Linden, W., Hogan, B. E., Rutledge, T., Chawla, A., Lenz, J., & Leung, D. (2003). There is more to anger coping than "in" or "out". Emotion, 3, 12-29. Linden, W., & Lamensdorf L. (1990). Hostile affect and casual blood pressure. Psychology andHealth, 4, 343-349. Linden, W., Lenz, J. W., & Con, A. (2001). Individualized stress management for essential hypertension: A randomized trial. Archives of Internal Medicine, 161, 1071-1080. Linden, W., Leung,'D., Chawla, A., Stossel, C, Rutledge, T., & Tanco, S. (1997). Social determinants of experienced anger. Journal of Behavioral Medicine, 20, 415-4432. Loo, R. (1979). Role of primary personality factors in the perception of traffic signs and driver violations and accidents. Accident Analysis & Prevention, 11, 345-352. MacDougall, J. M., Dembroski, T. M., Dimsdale, J. E., & Hackett, T. P. (1985). Components of Type A, hostility, and anger-in: Further relationships to angiographic findings. Health Psychology, 4, 137-152. Magnavita, N., Roberto, N., Sani, L., Carbone, A., Di-Giuseppe, L., & Sacco, A. (1997). Type A behaviour pattern and traffic accidents. British Journal of Medical Psychology, 70, 103-107. Marsh, P., & Collett, P. (1987). The car as a weapon. Etc, 44, 146-151. Martin, R., Watson, D., & Wan, C. K. (2000). A three-factor model of trait anger: Dimensions of affect, behavior, and cognition. Journal of Personality, 68, 869-897. Maxim, P. S., & Keane, C. (1992). Gender, age, and the risk of violent death in Canada, 1950-1986. Canadian Review of Sociology & Anthropology, 29, 329-345. Mesken, J., Lajunen, T., & Summala, H. (2002). Interpersonal violations, speeding violations, and their relation, to accident involvement. Ergonomics, 45, 469-483. Miller, T. Q., Jenkins, C. D., Kaplan, G. A., & Salonen, J. T. (1995). Are all hostility scales alike? Factor structure and covariation among measures of hostility. Journal of Applied Social Psychology, 25, 1142-1168. Mills, J. F., Kroner, D. G, & Forth, A. E. (1998). Novaco anger scale: Reliability and validity within an adult criminal sample. Assessment. 5. 237-248. Mittleman, M. A., Maclure, M., Sherwood, J. B., Mulry, R. P., Tofler, G. H., Jacobs, S. C, Friedman, R., Benson, FL, & Muller, J. E. (1995). Triggering of acute myocardial infarction onset by episodes of anger. Determinants of Myocardial Infarction Onset Study Investigators. Circulation, 92, 1720-1725. Mizell, L., Joint, M., & Connell, D. (1997). Aggressive driving: Three studies. Retrieved from American Automobile Association Foundation for Traffic Safety, March 1, 2002, from http://www.aaafoundation.org/resources/index.cfm?button=agdrtext. Musante, L., Treiber, F. A, Davis, H. C, Waller, J. L., & Thompson, W. O. (1999). Assessment of self-reported anger expression in youth. Assessment, 6, 225-233. Naatanan, R., & Summala, H. (1976). Road User Behaviour and Traffic Accidents, Amsterdam: North Holland. 71 Norris, F. H., Matthews, B. A., & Riad, J. K. (2000). Characterological, situational, and behavioral risk factors for motor vehicle accidents: A prospective examination. Accident Analysis & Prevention, 32, 505-515. Novaco, R. W. (1991). Aggression on roadways. In R. Baenninger (Ed.), Targets of violence and aggression (pp. 253-326). Novaco, R. W., Stokols, D., Campbell, J., & Stokols, J. (1979). Transportation, stress, and community psychology. American Journal of Community Psychology, 7, 361-380. Parker, D., Manstead, A. S., Stradling, S. G., & Reason, J. T. (1992). Determinants of intention to commit driving violations. Accident Analysis & Prevention, 24, 117-131. Parker, D., McDonald, L., Rabbitt, P., & Sutcliffe, P. (2000). Elderly drivers and their accidents: The Aging Driver Questionnaire. Accident Analysis & Prevention, 32, 751-759. Parker, D., Reason, J. T., Manstead, A.S.R., & Stradling, S. G. (1995). Driving errors, driving violations, and accident involvement. Ergonomics, 38, 1036-1048. Parkinson, B. (2001). Anger on and off the road. British Journal of Psychology, 92, 507-527. Perry, A. R. (1986). Type A behavior pattern and motor vehicle drivers' behavior. Perceptual & Motor Skills, 63, 875-878. Perry, A. R., & Baldwin, D. A. (2000). Further evidence of associations of type A personality scores and driving-related attitudes and behaviors. Perceptual & Motor Skills, 91, 147-154. Pipkin, N. L., Walker, L. G., & Thomason, M. H. (1989). Alcohol and vehicular injuries in adolescents. Journal of Adolescent Health Care, 10, 119-121. Plutchik, R. (1980). Emotion: A psychoevolutionarv synthesis. New York: Harper & Row. Pope, M. K., Smith, T. W., & Rhodewalt, F. (1990). Cognitive, behavioral, and affective correlates of the Cook and Medley Hostility scale. Journal of Personality Assessment, 54, 501-514. Rajalin, S. (1994). The connection between risky driving and involvement in fatal accidents. Accident Analysis & Prevention, 26, 555-562. Ranney, T. A. (1994). Models of driving behavior: A review of their evolution. Accident Analysis & Prevention, 26, 733-750. Rathbone, D. B., & Huckabee, J. C. (1999). Controlling road rage: A literature review and pilot study. Retrieved from American Automobile Association Foundation for Traffic Safety. March 1, 2002, from http://www.aaafoundation.org/resources/index.cfm?button=roadrage. Reason, J. T., (1990). Human error. Cambridge: Cambridge University Press. Reason, J. T., Manstead, A. S., Stradling, S. G., Baxter, J. S., & Campbell (1990). Errors and violations on the roads: A real distinction? Ergonomics, 33, 77-84. Riley, W. T., & Treiber, F. A. (1989). The validity of multidimensional self-report anger; and hostility measures. Journal of Clinical Psychology, 45, 397-404. Rothengatter, T. (1997). Psychological aspects of road user behaviour. Applied Psychology : An International Review, 46, 223-234. Schuman, S. H, Pelz, D. C, & Ehrlich, N. J. (1967). Young male drivers: Impulse expression, accidents, and violations. JAMA: Journal of the American Medical Association, 200, 1026-1030. Selzer, M. L., & Vinokur, A. (1974). Life events, subjective stress, and traffic accidents. American Journal of Psychiatry, 131, 903-906. Siegman, A. W., & Snow, S. C. (1997). The outward expression of anger, the inward experience of anger and CVR: The role of vocal expression. Journal of Behavioral Medicine, 20, 29-45. Siegman, A. W., Townsend, S. T., Blumenthal, R. S., Sorkin, J. D., & Civeiek, A. C. (1998). Dimensions of anger and CHD in men and women: Self-ratings vs. spouse ratings. Journal of Behavioral Medicine, 21, 315-336. Smith, T. W. (1992). Hostility and health: Current status of a psychosomatic hypothesis. Health Psychology, 11, 139-150. Smith, T. W., & Frohm, K. D. (1985). What's so unhealthy about hostility? Construct validity and psychosocial correlates of the Cook Medley Ho scale. Health Psychology, 4, 503-520 Smith, C. A., & Lazarus, R. S. (1993). Appraisal components, core relational themes, and the emotions. Cognition and Emotion, 7, 233-269. Smith, T. W., Sanders, J D., & Alexander, J. F. (1990). What does the Cook and Medley Hostility scale measure? Affect, behavior, and attributions in the marital context. Journal of Personality and Social Psychology, 58, 699-708. Spicer, J., & Chamberlain, K. (1996). Cynical hostility, anger, and resting blood pressure. Journal of Psychosomatic Research, 40, 359-368. Spielberger, C. D., Crane, R. S., Reams, W. D., Pellegrin, K. L., Rickman, RL., & Johnson, E. H. (1991). Anger and anxiety in essential hypertension: In C. D. Spielberger & I. G. Sarason (Eds.), Stress and emotion (Vol. 14, pp. 266-283). New York: Hemisphere/Taylor & Francis. Spielberger, C. D., Gorsuch, R., & Lushene, R. (1970). Manual for the State-Trait Anxiety Inventory (Self evaluation questionnaire). Palo Alto, CA: Consulting Psychologists Press. Spielberger, C. D., Jacobs, G., Russell, S., & Crane, R. S. (1983). Assessment of anger: The State-Trait Anger Scale. In: J. N. Butcher, & C. D. Spielberger (Eds.), Advances in personality assessment, Vol. 2, (pp. 161-189). London: Lawrence Erlbaum. Spielberger, C. D., Johnson, E. H., Russell, S. F., & Crane, R. J. (1985). The expression and experience of anger: Construction and validation of an anger expression scale. In Chesney, M., & Rosenhan, R. (Eds.) Anger and Hostility in Cardiovascular and Behavioral Disorder, (pp. 5-30). New York: McGraw Hill. Steele, M. S., & McGarvey, S. T. (1997). Anger expression, age, and blood pressure in modernizing Samoan adults. Psychosomatic Medicine, 59, 632-637'. Stoney, C, & Engebretson, T. O. (1994). Anger and hostility: Potential mediators of the gender difference in coronary heart disease. In A.W. Siegman, T. W. Smith, et al. (Eds). Anger, Hostility, and the Heart, (pp. 215-237). Hillsdale, NJ: Lawrence Erlbaum Associates. Tafrate, R. C, Cassinov, H., & Dunden, L. (2002). Anger episodes in high and low trait anger community adults. Journal of Clinical Psychology, 58, 1573-1590. Tillman, W. A., & Hobbs, G. E. (1949). The accident-prone automobile driver. A study of the psychiatric and social background. American Journal of psychiatry, 106, 321-331. Thomas, S. P. (1997). Angry? Let's talk about it! Applied Nursing Research, 10, 80-85. Tonkin, R. S. (1987). Adolescent risk-taking behavior. Journal of Adolescent Health Care, 8, 213-220. Treat (1980). A study of precrash factors involved in traffic accidents. HSRIResearch Review, 10, 1-35. Wells-Parker, E., Ceminsky, J., & Hallberg, V. (2002) An exploratory study of the relationship between road rage and crash experience in a representative sample of US drivers. Accident Analysis & Prevention, 34, 271-278. West, R. J., French, D., Kemp, R., & Elander, J. (1993). Direct observation of driving, self reports of driver behaviour, and accident involvement. Ergonomics, 36, 557-567. West, R. J., Elander, J., & French, D., (1993). Mild social deviance, Type-A behaviour pattern and decision-making style as predictors of self-reported driving style and traffic accident risk. British Journal of Psychology, 84, 207-219. Wilson, T., & Greensmith, J. (1983). Multivariate analysis of the relationship between drivometer variables and drivers' accident, sex, and exposure status. Human Factors, 25, 303-312. Wong, W. A., & Hobbs, G. E. (1949). Personal factors in industrial accidents: A study of accident proneness in an industrial group. Industrial Medicine, 18, 291-294. 76 Table 1. Mean scores split by gender and age. Overall Female Male <30 years >30 years Variable N = 316 N= 194 N= 122 N= 194 N=122 Age 30.6 (13.8) 29.5 (13.5) 32.4(14.2) 20.2 (2.3) 47.1 (6.5) #years with license 12.0 (11.9) 10.5 (11.4) 14.2 (12.5) 3.8 (2.2) 25.0 (9.2) #hours driving/week 10.1 (7.6) 8.7 (5.7) 12.3 (9.5) 8.7 (5.4) 12.3 (9.7) # mTickets received 0.8 (1.5) 0.7(1.6) 1.1(1.4) 0.8 (1.6) 0.8 (1.4) #MVAs 0.8 (1.2) 0.7(1.1) 1.0(1.3) 0.9(1.2) 0.6 (1.0) # minor MVAs 0.4 (0.5) 0.4 (0.5) 0.5 (0.5) 0.5 (0.5) 0.3 (0.5) # major MVAs 0.1 (0.3) 0.1 (0.3) 0.2 (0.4) 0.2 (0.4) 0.1 (0.3) HAS 70.1 (13.7) 68.3 (13.6) 73.1 (13.4) 72.2(12.7) 66.9 (14.6) DAS 44.0 (10.4) 44.0(10.6) 43.9 (10.3) 45.2 (9.9) 42.1 (11.0) STAI 42.2 (9.4) 42.9 (9.5) 41.2(9.2) 44.2 (9.5) 39.1 (8.4) Competitive Drive 1.0 (.5) 1.0 (.5) 1-1(5) 1.0 (.5) 1.0 (.4) Speed/Impatience 1.8 (1.0) 1.8 (1.0) 1.9(1.0) 1.8 (1.0) 1.8 (.9) BARQ av 17.5 (4.2) 17.2 (4.3) 17.9 (4.2) 17.4 (4.3) 17.6 (4.2) BARQao 16.2 (5.0) 15.8 (4.9) 17.0(5.1) 16.7 (5.1) 15.5 (4.8) BARQ rum 17.7 (4.8) 18.2 (4.7) 17.0 (4.7) 18.9 (4.7) 15.9 (4.2) BARQ ss 19.3 (4.9) 19.7 (4.9) 18.6(5.0) 20.1 (5.0) 18.1 (4.8) BARQ diff 14.0 (3.9) 14.1 (3.9) 13.9(3.9) 13.9 (3.9) 14.2 (3.8) BARQ as 16.6(5.2) 17.0 (5.4) 15.8 (5.0) 16.2 (5.4) 17.1 (4.9) Note: numbers in brackets refer to Standard Deviations 77 Table 2. Overall correlations of psychological variables and sub-sample correlations (<30 years/>30 years and male/female). 1. 2. 3. 4. 1. HAS 1.0 DAS .38* 1.0 (age: <30/>30) .35*736* (gender: male/female) .43*735* STAI .31*/ .29* 1.0 .20/.40* .217.35* .29735* .17/.37* 4. Competitive drive .20* .16 .02 1.0 .17/.18 .15/.12 -.01/-.05 .18/.20 .16/.17 -.01/.05 Speed/impatience .21* .25* .39* .37* .17/.24 .25*723 .37*740* .34*740* .177.24 .157.31* .34*7.43* .41*735* *p<.001 78 Table 3. Overall correlations of psychological variables and sub-sample correlations (<30 years/>30 years and male/female). HAS DAS STAI BARQ subscales: Competitive Speed/ Drive impatience Anger out .49* .33* .21* .14 .27* (age: <30/>30) .54*/.40* .37*/.25 .11/33* .09/. 18 . 227.34* (gender: male/female) .46*/.49* .32*/.34* .14/.27* .19/. 10 .267.28* Avoidance -.06 -.09 -.11 -.08 -.16 -.09/.00 -.14/-.02 -.08/-.15 .00/-.18 -.13/-.20 -.11/-.04 .01/-. 15 .02/-. 18 -.11/-.06 -.07/.22 Rumination .28* .24* .52* .10 .25* .30*/.17 .23/.17 .53*/.37* .06/.03 .33*/.10 .30*/.32* .19/.28* .47*/.54* .05/.14 .18/.30* Social support .09 .08 .08 05 -.06 .06/.05 .09/-.02 .04/.03 .04/.00 .00/-.20 .15/.10 .09/.07 .10/.06 -.04/.13 -.15/.00 Diffusion -.06 -.01 .00 .06 .01 -.03/-.09 .02/-.03 -.04/. 11 .17/-.08 .08/-.07 -.11/-.02 .00/-.02 .01/-.01 .02/. 10 .01/.02 Assertiveness -.07 .04 -.18 .07 -.01 -.07/-.05 .08/.02 -.27*/.04 .10/.06 .00/-.02 -.05/-.06 -.06/. 10 -.23/-. 17 .12/.05 -.09/.04 *p< .001 79 Table 4. Overall correlations of psychological variables and sub-sample correlations (<30 years/>30 years and male/female). 1. 2. 3. 4. 5. BARQ subscales: 1. Anger out 1.0 2. Avoidance -.34* 1.0 (age: <30/>30) -.29*/-.42* (gender: male/female) -.44*/-.29* 3. Rumination .29* -12 1.0 .26*727 -.15/-.04 .30*721 .06/-.21 4. Social support -.03 .17 .27* 1.0 -.02/-. 10 .16/.23 .21/.25 .06/-.07 .20/. 17 .27726* 5. Diffusion .04 .23* .04 .16 1.0 .05/.05 .22722 -.02720 .20/. 12 .117.01 .19/.25* .10700 .19713 6. Assertiveness -.08 -.02 -.08 .13 .20* .00/-.22 -.15/.21 -.15713 .09/.25 .18722 -.18/-.01 .14/-. 10 -.19/-.03 .14710 .14722 *p <.001 80 Table 5. Overall correlations of psychological variables for individuals involved in an MVA split by fault/no fault. 1. 2. 3. 4. 5. 1. HAS 1.0 2. DAS .47*728 1.0 3. STAI .19739* .33703 1.0 4. Competitive drive -.03710 .04723 .00706 1.0 5. Speed/impatience -.01/-.11 .11715 .18/-.17 .39*731 1.0 6. BARQ ao .53*753* .41*732 .13713 -.01709 -.07/-.04 7. BARQ av -.03/-.01 -.15/-.03 .00/.03 .10/.01 .18702 8. BARQ rum .24744* .31720 .57*755* -.01/.03 .12/-. 10 9. BARQ ss .09717 .07/. 00 .09716 .08727 .24/. 11 10. BARQ diff -.14/-.05 .02/-.20 .00/-.03 -.01/-.01 .04/-.02 11. BARQ as .08/-. 12 -.01/-.03 -.16/-.25 .07707 .01701 *p <.001 81 Table 6. Overall correlations of BARQ subscales for individuals involved in an MVA split by fault/no fault. 1. 2. 3. 4. 5. BARQ subscales: 1. Anger out 1.0 2. Avoidance -.27/-.24 1.0 3. Rumination .34*730 -.02/-.06 1.0 4. Social support .06700 .24724 .24726 1.0 5. Diffusion .31/-.01 -.10740* .14/-.06 .23717 1.0 6. Assertiveness -.01/-.24 -.08/-. 10 -.15/-. 18 .16705 .10712 *p <.001 82 Table 7. Frequencies of reported driving behaviours and chi-square statistics of the differences for gender and age. Gender Age Variable Overall n=316 Female n=194 Male n=122 <30years n=194 >30years n=122 Ticket received No yes 186 (58.9%) 131(67.5%) 55 (45.1%) 120(61.9%) 66 (54.1%) 130(41.1%) 63 (32.5%) 67(54.9%) 74 (38.1%) 56 (45.9%) 15.58*** 1.86 MVA No Yes 161(50.9%) 107 (55.2%) 54 (44.3%) 84 (43.3%) 77 (63.1%) 155 (49.1%) 87 (44.8%) 68 (55.7%) 110(56.7%) 45 (36.9%) 3.56 11 77*** Minor MVA No Yes 179 (56.6%) 137 (43.4%) 119(61.3%) 75 (38.7%) 60 (49.2%) 62 (50.8%) 95 (49.0%) 99 (51.0%) 84 (68.9%) 38 (31.1%) 4.5V 12.06* Major MVA No Yes 277 (87.9%) 177(91.2%) 100 (82.0%) 164 (84.5%) 113(92.6%) 39(12.3%) 17(8.8%) 22(18.0%) 30(15.5%) 9(7.4%) 5.95< 4.53" *p< 025 **p< .01 ***p<001 83 Table 8. Means of reported driving consequences and F statistics for significant differences between means of age, gender and interactions. Variable Overall n=316 Gender Female n=194 Male n=122 Age <30years n=194 >30years n=122 Gender XAge # mTickets received .82(1.53) .67(1.57) 1.06(1.44) .80(1.58) .85 (1.45) F=4.62 F=0.01 F=0.01 #MVAs .81 (1.17) .68(1.10) 1.02(1.25) .92(1.24) .63 (1.03) F=7.]9** F=6.04* F=0.04 # minor MVAs .43 (.50) .39 (.49) .51 (.50) F =5.70* .51 (.50) .31 (.46) F =15.03*** F=0.53 # major MVAs .12 (.33) .09 (.28) .18 (.39) F=6.32* .15 (.36) .07 (.26) F=6.47* F=0.70 Note: numbers in brackets refer to Standard Deviations *p<.025 **p<.01 ***p<.001 84 Table 9. Means of personality variables overall and F statistics for differences split by age and gender and interactions. Variable Overall n=316 Gender Female n=194 Male n=122 Age <30 years n=194 >30 years n=122 Gender XAge HAS 70.12 (13.68) 68.27 (13.59) 73.07 (13.36) F=9.48* 72.16 (12.71) 66.92 (14.56) F =17.52* F=5.01 DAS 43.98 (10.44) 44.03 (10.55) 43.90 (10.32) F=0.01 45.18 (9.93) 42.08 (10.98) F = 6.96* F=0.27 STAI 42.24 42.92 41.15 44.21 39.11 (9.39) (9.49) (9.15) (9.48) (8.35) F=1.25 F = 21.00*** F=0.07 Competitive drive 1.00 (.49) .97 (.50) 1.06 (.47) F = 2.31 .99 (.51) 1.02 (.45) F=0.J5 F=0.01 Speed/impatience 1.81 (.98) 1.76 (1.00) 1.88 (.96) 1.81 (1.03) 1.80 (.91) F= 1.70 F=0.01 F= 1.96 Note: numbers in brackets refer to Standard Deviations *p < .025 **p< .01 ***p<001 85 Table 10. Means of BARQ subscales and F statistics for differences between means of age and gender. Variable Overall n=316 Gender Female n=194 Male n=122 Age <30years n=194 >30years n=122 Gender Xage Avoidance 17.47 (4.24) 17.22 (4.28) 17.86 (4.16) F = 1.10 17.37 (4.29) 17.62 (4.17) F = 0.04 F=1.00 Anger out 16.25 (5.01) 15.79 (4.86) 16.98 (5.12) F = 4.61 16.73 (5.12) 15.48 (4.74) F=6.21* F = 0.43 Rumination 17.71 (4.75) 18.16 (4.74) 16.98 (4.68) F=3.32 18.86 (4.71) 15.87 (4.20) F =31.35*** F = 0.80 Support seeking 19.29 (4.91) 19.74 (4.91) 18.57 (5.00) F = 3.14 20.05 (4.96) 18.07 (4.76) F = 11.24** F = 0.28 Diffusion 14.00 (3.89) 14.09 (3.88) 13.86 (3.91) F = 0.47 13.90 (3.93) 14.16 (3.83) F = 0.23 F = 0.51 Assertion 16.55 (5.23) 16.98 (5.35) 15.85 (4.96) F =5.61* 16.21 (5.42) 17.09 (4.87) F=1.74 F = 3.44 "p < .025 **p < .01 ***p < .001 Note: numbers in brackets refer to Standard Deviations 86 Table 11. Univariate Logistic regression for the outcome of receipt of one or more tickets for a moving violation. Variable B SE Exp (B) 95% CI Models2 Wald Age -.32 .23 .73 .46- 1.15 1.86 1.86 Years with license .02 .01 1.02 1.003 - 1.04 5.02* 4.99* Gender -.93 .24 .40 .25 - 63 15.55*** 15.25*** Hours per week .03 .02 1.03 .99- 1.06 3.10 2.93 HAS .01 .01 1.01 .99- 1.02 .35 .35 DAS -.002 .01 1.0 .98- 1.02 .05 .05 Competitive drive -.28 .24 .76 .48- 1.21 1.38 1.38 Speed/impatience -.14 .12 .87 .69- 1.10 1.34 1.34 STAI -.03 .01 .97 .95- 1.00 4.70* 4.59* BARQao .02 .02 1.02 .98- 1.07 1.03 1.03 BARQav -.01 .03 .99 .94- 1.04 .12 .12 BARQrum -.05 .02 .95 .90- 1.00 4.62* 4.52* BARQss -.02 .02 .98 .93- 1.02 1.0 1.0 BARQdiff -.002 .03 1.0 .94- 1.06 .003 .003 BARQas .03 .02 1.03 .98- 1.07 1.55 1.54 *p<05 **p<.01 ***p<.001 87 Table 12. Univariate Logistic regression for the outcome of one or more MVAs over the past five years. Variable B SE Exp (B) 95% CI Model X2 Wald Age .81 .24 2.24 1.41-3.57 11 87*** 11.58*** Sex -.44 .23 .65 .41 - 1.02 3.56 3.54 Hours per week .01 .02 1.01 .98-1.04 .60 .59 mTicket received -.54 .23 .58 .37 - .92 5.49* 5.44* HAS .01 .01 1.01 1.0-1.03 1.95 1.93 DAS .02 .01 1.02 1.003- 1.05 5.09* 4.98* Competitive drive -.09 .23 .70 .58 - 1.44 .15 .15 Speed/impatience .07 .12 1.07 .85 - 1.34 .32 .32 STAI .02 .01 1.02 .99-1.04 1.75 1.74 BARQao .05 .02 1.06 1.01 - 1.10 5.54* 5.41* BARQav -.02 .03 .98 .94-1.04 .33 .33 BARQrum .02 .02 1.02 .97-1.07 .56 .56 BARQss .02 .02 1.02 .97-1.06 .41 .41 BARQdiff .02 .03 1.02 .96-1.08 .44 .44 BARQas -.02 .02 .98 .94-1.03 .53 .53 *p<.05 **p<01 ***p<001 88 Table 13. Univariate Logistic regression for the outcome of one or more minor MVAs reported over the past five years. Variable B SE Exp (B) 95% CI Model X2 Wald Age .83 .24 2.30 1.43-3.71 12 26*** 11.83*** Sex -.49 .23 .61 .39 - .96 4.50* 4.48* Hours per week .01 .02 1.01 .98-1.04 .71 .70 mTicket received -.62 .23 .54 .34-.85 7 14** 7.21** HAS .02 .01 1.02 1.001 - 1.04 4.51* 4.41* DAS .03 .01 1.03 1.004- 1.05 5.37* 5.25* Competitive drive -.20 .24 .82 .52-1.30 .73 .72 Speed/impatience .01 .12 1.01 .80-1.26 .003 .003 STAI .01 .01 1.01 .98-1.03 .40 .40 BARQag .07 .02 1.07 1.02-1.12 g 9** 8.69** BARQav -.03 .03 .97 .92-1.02 1.53 1.52 BARQ rum .01 .02 1.02 .97-1.06 .34 .34 BARQss .02 .02 1.02 .97-1.07 .59 .59 BARQdiff .02 .03 1.02 .97-1.09 .72 .72 BARQas -.02 .02 .98 .94-1.03 .61 .61 *p<.05 **p<01 ***p<.001 89 Table 14. Univariate Logistic regression for the outcome of one or more major MVAs reported over the past five years. Variable B SE Exp (B) 95% CI Model X2 Wald Age .83 .40 2.30 1.05 -5.02 4.82* 4.34* Sex -.83 .35 .44 .22 - .86 5.78* 5.73* Hours per week -.02 .03 .98 .93 - 1.03 .49 .44 mTicket received -.59 .34 .56 .28- 1.09 2.91 2.92 HAS .02 .01 1.02 .99- 1.04 1.62 1.62 DAS -.001 0.02 1.00 .97- 1.03 .001 .001 Comp drive .19 .35 1.21 .61 - 2.41 .31 .31 Speed/imp .04 .18 1.04 .74- 1.47 .06 .06 STAI .01 .02 1.01 .98- 1.05 .40 .40 BARQao .01 0.03 1.01 .94- 1.08 .04 .05 BARQav .05 .04 1.05 .97- 1.14 1.36 1.34 BARQrum .04 .04 1.04 .97- 1.12 1.28 1.28 BARQss .01 .04 1.01 .94- 1.08 .09 .09 BARQdiff -.06 .05 .94 .86- 1.03 1.78 1.74 BARQas .02 .03 1.02 .96- 1.09 .34 .33 *p<05 **p<.01 ***p<.001 90 Table 15. Logistic regression for variables of interest, controlling for age, gender and hours driven per week. Variable B SE Exp (B) 95% CI StepX2 Model X2 Outcome mTickets STAI -.02 .01 0.98 0.95 - 1.00 3.00 20.03*** BARQrum -.04 .03 .96 .91 - 1.01 2.17 19.20*** MVA mTicket received -.54 .24 .59 .36 - .94 4.85* 23.45*** DAS .02 .01 1.02 .997- 1.04 2.87 19.90*** BARQao .04 .02 1.04 0.99- 1.09 2.48 21.08*** Minor mTicket received -.62 .25 .54 .33 - .87 6.38* 26.72*** MVA HAS .01 .01 1.01 .99- 1.03 1.07 20.70*** DAS .02 .01 1.02 .997- 1.04 3.08 21.62*** BARQao .05 .02 1.05 1.005- 1.11 4.86* 25.19*** Major mTicket received -.53 .36 .59 .29- 1.19 2.20 14.63** MVA *p<05 **p<01 ***p<001 91 Table 16. Fitting of the Logistic Regression Model for prediction of receiving one or more tickets for a moving violation over the past five years. Model 1 Model 2 Final Model Variables Constant .56 (.33) .49 (.32) -.04 B(SE) (.28) Gender -1.33*** -1 23*** _ gy** (.43) (.41) C24) Age -.59 -.53 .19 (.42) (.41) (.25) HPW .01 .01 .01 (.02) (.02) (.02) STAI .010 -.01 (.04) (.03) GenderXSTAI -.01 .02 (.05) (.03) AgeXSTAI -.07 -.04 (.05) (.03) AgeXGender .72 .64 (.53) (.52) AgeXGenderXSTAI .05 (.06) 26.66** 25.00** 17.10*** Model X2(df) (8) (7) (3) *p<0.05 **p<0.01 ***p<0.001 Table 17. Logistic Regression model building for prediction of minor motor vehicle accidents Model 1 Model 2 Variables B (SE) B (SE) Constant -.72 (.44) -.46 (.33) Gender (male) -.01 (.53) -.24 (.44) Age 1.49** (.51) 1.25** (.43) HPW .02 (.02) .02 (.02) Received Ticket -.64* (.26) -.62* (.26) DAS .02 (.01) .02 (.02) AO .16 (.08) .10 (.06) RUM -.08 (.08) -.02 (.03) AgeXAO -.13 (.10) -.06 (.08) GenderXAO -.24 (.11) -.19* (.09) AgeXGender -.59* (.63) -.29 (.54) AgeXRUM .07 (.10) GenderXRUM .07 (.11) AOXRUM .02 (.02) AgeXGenderXAO .26 (.13)* .22* (.11) AgeXGenderXRUM -.07 (.14) AgeXRUMXAO -.02 (.02) GenderXRUMXAO -.01 (.02) AgeXGenderXAOXRUM .02 (.03) Model X2(df) 41.58*** (18) 38.78*** (11) *p<0.05 **p<0.01 ***p<0.001 93 Note: AO refers to BARQ Anger out, RUM refers to BARQ Rumination Table 17. continued Logistic regression: Model building for prediction of minor motor vehicle accidents. Model 3 Model 4 Model 5 Final Model Variables B(SE) B(SE) Constant -.40 (.33) -.44 (.32) -.45 (.31) -.44 (.32) Gender -.27 (.44) -.21 (.44) -.13 (.42) -.21 (.44) Age 1.18** (.42) 1.23** (.42) 1.17** (.41) 1.23** (.42) HPW .02 (.02) .02 (.02) .02 (.02) .02 (.02) Received Ticket -.64* (.25) -.64* (.25) -.63* (.25) -.64* (.25) DAS .02 (.01) AO .10 (.06) .11 (.06) .04 (.05) .11 (.06) AgeXAO -.06 (.08) -.06 (.08) .05 (.05) -.06 (.08) GenderXAO -.19* (.09) -.18* (.09) -.04 (.05) -.18* (.09) AgeXGender -.28 (.54) -.30 (.54) -.34 (.53) -.30 (.54) AgeXGenderXAO .22* (.11) .21* (.11) .21* (.11) Model X2 (df) 38.00***(10) 36.52*** (9) 3.79* (1) 32.73*** (8) 36.52*** (9) *p<0.05 **p<0.01 ***p<0.001 Note: AO = BARQ Anger Out, RUM = BARQ Rumination 94 Table 18. Odds of involvement in one or more MVAs over the past five years with and without receipt of one or more tickets for a moving violation in the same period at three levels of BARQ Anger out. With mTickets Without mTickets Aggression multiplier -1SD Mean +1SD -1SD Mean +1SD Female > 30 years 0.93 .74 .51* .35 29** 26*** 18** Male > 30 years 1.11 .38* .65 1.11 .20** .34** .57 Female < 30 years 1.08* .88 1.30 1.90* .46 ** .68 .99 Male < 30 years 1.04 1.35 2.29** 2.80** .97 1.19 1.46 *p< 05 **p< 01 ***p< 001 95 Table 19. Change in odds of involvement in one or more MVAs over the past five years with and without receipt of one or more tickets for a moving violation in the same period at three levels of BARQ Anger Out. Change in Odds of MVA Comparison -1 SD Mean +1 SD Interaction Male < 30 years / Female > 30 years 2.87** 4 74** 8.04** 1.12* Male < 30 years / Male > 30 years 4.43* 3.53** 2.21 0.94 Male < 30 years / Female < 30 years 2.39 1.92 1.54 0.96 Female < 30 years / Female > 30 years 1.21 2.54** 5.13** 1.17 Female < 30 years / Male > 30 years 1.87** 1.62** 1.42* 0.97 Male > 30 years / Female > 30 years 1.21 1.27 3.59 * 1.19 *p< 05 **p<01 200 Figure 2. Frequency distribution of ages for this student/community sample. 97 2.5 ~ 2 m OL X ^ 15 CO T3 T3 ° 1 0.5 / / / Female >30 years / Ma/e >30 years / Female <=30 years / Male <=30 years Female >30 years NT — - Male >30 years NT - - • Female <=30 years NT — - Male <=30 years NT ^ ^ «-* *"*" ——_____ i i •1 SD Mean +1 SD BARQ anger out Figure 3. Odds of MVA involvement for those having received a ticket in the past five years and those who have not (NT) as a function of reported anger out (holding hours per week constant at its mean: 10.12 HPW). 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items