The Scottish Government

« Previous | Contents | Next »

Listen

Chapter Three Results

Variations in attitudes to risk taking and understanding of risks across different social and demographic groupings of motorcyclists

Interview survey

3.1 The 124 interviewees were asked to compare themselves with other motorcycle riders of similar age and sex and to state how much more or less likely they were to be involved in an accident while riding on public roads in the next 12 months. No respondents thought they would be more likely than the others to be involved in an accident, and just over a third thought they would be as likely to have an accident. The other respondents all thought they would be less likely to be involved in an accident and 19% much less likely. This is a fairly typical finding, since most riders/drivers consider themselves at least as good a rider/driver as those of a similar age and sex.

3.2 Interviewees were then asked to consider riders of the same age, sex and experience as themselves and who rode the same number of miles per year, and state how many (out of 100) would have an accident or minor spill in the next 12 months. Responses ranged from none (n=3) to 85 with an average of just under 14. The median and modal values ¹ were both 10.. These did not change when outliers (values that it is hard to accept as the respondent's serious estimate of risk, i.e. zero or greater than 40) were excluded, but the average dropped to 12.

3.3 Interviewers had a table ² from which they could provide an estimate of the riders' accident risk, given their sex, age, experience and miles ridden per year. These estimates varied from 3 to 33 per 100 riders, with an average value of 7 and a median and mode of 6. The implication of this, when compared to the respondents' own estimates of risk from the earlier question, was that riders tended to judge the risk of being in an accident to be higher than that derived from the tables. Further information on the differences between the two measures of risk is given in paragraph 3.36.

3.4 The interviewees were asked if they were willing to believe the estimated accident risk taken from the tables, and then a series of follow-up questions. There were only 4% of interview respondents who were not willing to believe the estimated risk for them as derived from the tables; 96% agreed that they were aware that riding was this risky, and that they were prepared to accept the risk. Only 29% worried about the risk. Only 13% of respondents indicated a willingness to consider giving up riding if the risk of being killed was 25 times higher than when driving a car. Most interviewees wore protective gear which they believed would help reduce the risk of injury. Virtually all agreed that you had to accept some risk otherwise life would be too boring, see Table B1 of Appendix B for the full set of questions and responses.

3.5 These patterns of responses strongly suggest that respondents did not generally underestimate the statistical risk of riding, and that they were prepared to accept this risk. However, there were also indications that a substantial proportion of respondents (42%) felt that this statistical risk for riders like themselves did not actually apply to them because they were good riders.

3.6 Respondents were grouped by their age; the distribution is shown in Figure 3.1. The average scores on a number of questions relating to risk were computed and compared by age group.

Figure 3.1 - Age group of respondents

3.7 Younger riders (16-25 yrs) on average considered themselves about as likely to have an accident as others of the same age and sex. However, older riders were more optimistic when comparing themselves with their peers. For example riders aged 46-55 years considered themselves on average as 'less likely' to have an accident than others of the same age and sex. Riders' estimates of the absolute risks faced by others of the same age and sex also fell strongly with age. 18-25 year-olds estimated that 29 out of 100 riders of the same age and sex would have an accident or minor spill during the next 12 months. This fell to about 9 out of 100 for riders aged over 45. Further information on the differences between the age-groups is given in Table B3 of Appendix B, which summarises the results for the questions for which there was a statistically significant age effect.

3.8 Respondents were also grouped by their riding experience; the distribution is shown in Figure 3.2. The average values on a number of questions asking about understanding and acceptance of risk were computed and compared by experience. The statistically significant results are given in Table B5 of Appendix B, and the main ones are summarised below.

Figure 3.2 - Experience group of respondents

3.9 All the experience groups thought themselves (on average) less likely to be involved in an accident than other riders of their own age and sex, but this effect was stronger for very experienced riders. This difference is consistent with the well-established effect of experience on accident risk - an effect shown in the second row of Table B5. The experienced riders were more likely than the inexperienced to see themselves as avoiding some of the risk by being good riders, and they also worried less about the risk.

Self-completion questionnaire survey

3.10 The self-completion questionnaire also asked questions about different aspects of riders' perception of risk when riding, and their acceptance of risk. See Question 23 in Appendix D and Table B11 in Appendix B. Because it was not feasible in the self-completion questionnaire to present the respondent with an estimate of his or her personal statistical risk, the risk they were asked to consider in the self-completion questionnaire was that motorcyclists were 25-times more likely than car drivers to be killed in an accident.

Figure 3.3 -Awareness of motorcyclists' 25x excess fatality risk over car drivers

3.11 16% of respondents were not willing to believe that motorcyclists are 25 time more likely to be killed than car drivers. However, 66% indicated that they were aware of this and were prepared to accept it. However, only 29% worried about the risk of riding. 16% believed that they avoided some of this risk by being a good rider. Figure 3.3 presents some of the results in chart form.

3.12 About 84% of self-completion questionnaire respondents agreed that they wore protective gear which would help reduce the risk of injury. Many (64%) also agreed that you had to accept some risk otherwise life would be too boring, and only 3% (10 respondents) would even consider giving up riding if the risk of being killed was actually 25 times higher when riding than when driving. See Figure 3.4.

Figure 3.4 - Accept risk 'otherwise life too boring' and agree that 'would give up'

3.13 Respondents were grouped by their age and by their experience as riders; the distribution of respondents has already been shown in Figure 3.1 and Figure 3.2 respectively. A similar analysis to that carried out for the interview sample, comparing age and experience group averages for a number of questions about risk, was conducted on the self-completion questionnaire data.

3.14 Only the experience group analyses produced any statistically significant differences between group averages, see Table B14. All experience groups tended to believe that they were less likely to be involved in an accident than other riders of the same age and sex but this optimism was strongest for riders with 30 years of experience or more (see Figure 3.5). Also, the more experienced riders were less willing to believe that the risk of a fatal accident was 25 times higher for motorcyclists than car drivers. The less experienced riders worried more about the risk. Very few riders in any group said they would consider giving up riding if the fatality risk was really 25-times more than for car driving.

Figure 3.5 - Rider comparison with similar riders

3.15 Riders were asked to rate on a 7-point scale how safe or risky they thought motorcycling is. They were asked to rate the risk for riders in general and for themselves. The distribution of responses is shown in Figure 3.6. The estimate of risk for themselves was generally lower than for riders in general.

Figure 3.6 - Rider risk assessment of risk for riders in general and for themselves

Summary

3.16 Interview responses and self-completion questionnaire responses are broadly in agreement, and suggest that:

• Most respondents were willing to believe the statistical risk of motorcycling expressed in terms of estimated accident risk for their peers (interview survey) or in terms of the excess risk over car driving (self-completion survey). In fact, most said they were aware of this level of risk and were willing to accept it.
• There was in fact a tendency for riders to over-estimate the statistical risk of accidents for riders like themselves as compared to a computed risk from a model based on a previous survey.
• Few riders said they would consider giving up motorcycling if its fatality risk per mile really was 25 times higher than that of car driving.
• There was a strong tendency for riders to estimate their own risk as below that of their peers. A substantial proportion also thought that the statistical risk estimates did not apply to them because they were good riders.
• Highly experienced riders were more optimistic than less experienced riders when it came to comparing their own risk with that of others of the same age, sex, experience and annual mileage. In the interview sample only, older riders were more optimistic than younger riders when they compared themselves with their peers.
• Rider's estimates of risk levels for themselves and for their peers fell with increasing age and experience

3.17 The general picture, therefore, is that motorcyclists have a reasonably realistic view of the risks of motorcycling in terms of expected accidents per year, and in terms of the excess risk over car driving. They do not, as a group, appear to grossly underestimate this risk. This suggests that simply providing them with better information about statistical risk is unlikely to be very effective in changing their behaviour and improving safety (or dissuading them from riding), though providing a better understanding of what that risk means in practice might be effective. A substantial proportion of riders think that they avoid some of the statistical risk by being good riders. This may be true to an extent, though we do not know whether those who believe it are in fact those who are at relatively low risk. This optimistic comparison with peers, which is particularly strong for the older and more experienced riders, is something that has potential to be tackled by rider training and education.

Extent to which attitudes influence behaviour

Self-completion questionnaire survey

3.18 One of the most influential pieces of work relating to driving behaviour published in the last 15 years was that by Reason et al. (1990). Using a self-completion questionnaire (known as the Driver Behaviour Questionnaire - DBQ), Reason et al (1990) classified aberrant driving behaviours in terms of a system of errors and violations. Violations were defined as "deliberate deviations from those practices believed necessary to maintain the safe operation of a potentially hazardous system". Errors were broadly defined as the "failure of planned actions to achieve their intended consequences".

3.19 These findings have been replicated, with variations, in further studies in the UK. Sexton et al (2004b) developed a new version of the DBQ (the Motorcycle Rider Behaviour Questionnaire ( MRBQ)) specifically for motorcyclists, investigated its factor structure and its relation to motorcycle accidents. Previous research has found that people who score high on DBQ violations are statistically more likely to have been involved in accidents in the past (Parker, Reason, Manstead, & Stradling, 1995a) and to be involved in the future (Parker, West, Stradling, & Manstead, 1995b). There is also some evidence to suggest that self-reported driving errors are related to accidents ( e.g. Parker et al., 1995a).

3.20 Sexton et al (2004b) fitted two models of rider behaviour using statistical modelling techniques. In these models attitudes/motivations/perceptions and rider style influence rider behaviour, which in turn influence the likelihood of accident involvement. Age, sex and experience influenced both attitudes and behaviour, and may also have a direct influence on accidents. Accident risk was also directly influenced by the number of miles ridden in the past 12-months. The reported frequency of errors was the most important behavioural contribution to accident involvement (once the mileage effect had been taken into account). Traffic errors (mostly associated with failures of hazard perception or observational skills) were the most consistent predictors. Control errors (mainly to do with difficulties of control associated with high speed, or errors in speed selection) were also important in some analyses. However, these errors occur in a context that suggests they may be closely linked with riding styles involving carelessness, inattention and excessive speed - i.e. styles that might be termed 'violational'.

3.21 In the present study, a 24 item version of the MRBQ from Sexton et al. (2004b) was used in the self completion questionnaire, covering four of the behavioural factors identified by Sexton et al. Factor scores were computed for these four factors and provided the following measures of behaviour:

• TRAFFIC ERRORS - e.g. 'fail to notice that pedestrians are crossing when turning into a side street from a main road'; 'attempt to overtake someone that you hadn't noticed to be signalling a right turn'.
• SPEEDING - speeding behaviours - e.g. 'exceed the speed limit on a residential road'; 'race away from traffic lights with the intention of beating the driver/rider next to you'; 'open up the throttle and just go for it on country roads'.
• STUNT - performing stunts and other high risk behaviours - e.g. 'attempt to do, or actually do, a wheelie'; 'intentionally do a wheel spin'.
• CONTROL ERRORS - e.g. 'run wide when going round a corner'; 'brake or throttle back when going round a corner or bend'; 'find that you have difficulty in controlling the bike when riding at speed ( e.g. steering wobble)'.

3.22 Higher scores in these factors mean that the respondent indicated that the behaviour occurred more often. For example the question 'How often do you exceed the speed limit on a residential road' contributes to the SPEEDING behaviour measure, and a high score means that respondents reported doing this frequently or nearly all the time.

3.23 There are two errors factors (TRAFFIC ERRORS and CONTROL ERRORS). However, this distinction is not clear cut since as pointed out by Sexton et al (2004b) many of the error items are errors occurring in a violational context. The 'SAFETY' factor as used in Sexton et al (2004b) was not used in this study because it was in-part covered by other questions, and in-part because of space limitations.

3.24 Behavioural factors such as violations and errors might prove useful in terms of explaining why motorcycle accidents occur. To explain why people behave in these ways requires the psychological determinants of behaviour to be explored. Drivers' attitudes, motivations and perceptions about safety, for example, have been assessed in research studies in the past, and have proved to be useful predictors of driving behaviour ( e.g. Baughan et al 2005; Maycock & Forsyth, 1997; Parker et al, 1995b).

3.25 The self completion questionnaire asked a number of questions relating to risk, as reported earlier. These questions define measures of 'risk propensity', which may be helpful in comparing the behaviour measures. For the present analysis one of these, ("You have to accept some risk, if life is not to be too boring"), has been used as an indicator of rider attitude. The derived behaviour measures (scores on the four MRBQ factors) discussed above were compared for each of the response groups defined by this 'attitude' question and the results shown in Figure 3.7, see also Table B14.

Figure 3.7 - Rider behaviour measures by 'attitude' question

3.26 It is clear from Figure 3.7 that the more strongly riders agree that 'you have to accept some risk otherwise life would be too boring' the more likely are they to commit speeding behaviours. They also tend to make more traffic violation errors, control errors and pull stunts.

3.27 Based on the work of Schulz et al (1991) in Germany, Sexton et al (2004b) developed a Motorcycle Rider Motivation Questionnaire ( MRMQ) that in Britain identified three motivational factors in motorcycle riding.

3.28 In the present self-completion survey 14 items from the MRMQ were used to measure two of these factors:

• PLEASURE - the general pleasure derived from riding motorcycles
• SPEED - enjoy the dynamic aspects of performance from motorcycles

3.29 The question, "You have to accept some risk, if life is not to be too boring", has been used above as an indicator of rider attitude; the relationship between responses to it, and riders' scores on the two motivational factors from the MRMQ is shown in Figure 3.8 . It is clear that the more strongly riders agree that 'you have to accept some risk otherwise life would be too boring' the more pleasure they get from riding and the more they have a liking/motivation for speeding (a low score means more pleasure).

Figure 3.8 - Rider motivation and 'attitude' towards risk

Summary

3.30 The average values for each of the MRBQ behaviour factors and MMRQ motivational factors indicate that those riders who agree that "You have to accept some risk, if life is not to be too boring" tend to report 'riskier' behaviours and motivations than other riders. That is, riders who agree that they 'accept some risk otherwise life would be too boring' tend to report:

• More traffic errors
• More speeding behaviours
• More control errors
• Being more likely to do stunts when riding
• Getting more pleasure from their riding
• Being motivated by a liking of speed and the dynamic aspects of motorcycling

All the above relationships are statistically significant. It can thus be concluded that the attitude of the rider (as expressed in the statement considered) is reflected in their reported behaviours, i.e. those who accept 'more risk' tend to be riders who make more errors, speed more and get more pleasure from riding.

Variations between statistical risk assessments and motorcyclists assessments of risk

Interview survey

3.31 As was reported earlier, interviewees were asked to consider riders of the same age, sex and experience who rode the same number of miles per year, and state how many (out of 100) would have an accident or minor spill in the next 12 months.

Figure 3.9 - Rider's estimates of risk for riders like themselves in terms of age, sex, experience and annual mileage

3.32 The distribution of respondents' own risk assessments for riders like themselves in terms of age, sex, experience and annual mileage is shown in Figure 3.9.

3.33 Interviewers had a table from which they could provide an estimate of the riders' accident risk, given their age, experience and miles ridden per year. This table was derived from a statistical model, and enabled the interviewer to give a figure (out of 100 similar riders) based on actual accident history from a sample of motorcyclists who responded to a previous survey. The distribution of the statistical risk predictions for the same riders is shown in Figure 3.10.

Figure 3.10 - Risk estimates from statistical model

3.34 The difference in the distributions shown in Figures 3.9 and 3.10 implies that riders had a very strong tendency to judge the risk of being in an accident higher than that derived from the model, i.e. they tended to over-estimate the risk faced by riders like themselves, see Figure B1 in Annex B.

3.35 The finding that riders tend to over, rather than under-estimate their objective risk levels is consistent with the findings, reported above in paragraph 3.9, that riders are generally willing to believe the statistical risk estimates, and that most riders say they already know and have accepted that risks are at such levels. Paragraph 3.13 also showed that a substantial proportion of respondents, mainly the older and more experienced riders, felt that by being good riders they avoided some of the statistical risk for riders like themselves in terms of age, sex, experience and annual mileage.

Summary

3.36 Riders are aware that the risk of having an accident is fairly high, however they are not very good at estimating that risk. On average they over-estimate the risk to riders of the same age, sex and annual mileage as themselves derived from a statistical model (from Sexton et al, 2004b). However, riders tend to regard the risk to themselves as less than that for riders of the same age and sex, i.e. riders consider that they are safer than their peers. This finding is consistent with the hypothesis that a substantial proportion of riders think that they avoid some of the statistical risk by being 'better' riders than their peers.

Motorcyclists with attitudes which place them at risk

Interview survey

3.37 It seems reasonable to assume that riders with different attitudes may be at different levels of risk. Riders who realise that motorcycling is a risky activity and are aware of the risk, but consciously accept the risk may ride more defensively than riders who deny such risk exists for them. It was helpful to compare riders' responses to questions about risk, as perceived in a series of pictures of different road and traffic scenarios.

3.38 Using stimulus materials and a procedure developed by Broughton (Broughton & Stradling, 2005) interviewees were shown a series of 5 pictures depicting different road scenes (see Annex E). They were asked to imagine that they are riding the road in their normal manner and to rate the setting on four characteristics (on 5-point scales from 1 very low to 5 very high). The characteristics rated were:

• How enjoyable it would be
• The speed you would ride
• How much concentration you would need
• How risky it would be to ride the road.

3.39 For each picture, and overall, there were highly significant associations between ratings of enjoyment and speed, and between ratings of concentration and risk. The only association between ratings of speed and risk was a weak one for Picture5, and the only association between speed and concentration was a weak one for Picture4. A plausible hypothesis would be that, across the range of road settings represented in the 5 pictures, bikers vary speed to manipulate enjoyment and vary concentration in the face of varying perceived risk, but that rated enjoyment and rated risk are statistically independent, i.e. high enjoyment has little or nothing to do directly with the perceived risk of the setting. Further details of this analysis are reported in Annex B.

Self-completion questionnaire survey

3.40 We have already seen that, in the self-completion survey, riders who have a more favourable attitude towards accepting the risks of motorcycling tend to report more risky motorcycling behaviour patterns. This finding was explored further by using the k-cluster technique to define three rider groups based on responses to eight items about the risk of motorcycling (question 23). The k-cluster technique generates groups that differ from each other as much as possible with respect to their responses. Table A summarises the characteristics of the groups and Table B16 in Appendix B gives the group scores on each of the eight questionnaire items.

Table A - Rider Group characteristics (from self-completion questionnaire survey)

3.41 As a convenient shorthand, the three groups in Table A might be named as follows:

• RISK DENIERS (GROUP 1) - those who simply did not accept the risk of being killed in a crash is 25 times higher if riding as compared to being in a car, and did not worry about the risk, albeit they still accept that riding is a riskier activity than driving;
• OPTIMISTIC ACCEPTERS (GROUP 2) - those who agreed that there was a risk, but accepted the risk and did not worry about it and considered that it did not really apply to them as 'good' riders;
• REALISTIC ACCEPTERS (GROUP 3) - those who agreed and were aware that there was a risk, had not realised it was that high and considered that it did apply to them as riders.

3.42 . Respondents were asked to imagine three groups of 100 riders (typical riders; riders of the same age, sex, experience and amount and type of riding as themselves; riders exactly like themselves in every way). For each group, they were asked to estimate the number of accidents in 12 months with no injuries, minor injuries, and serious or fatal injuries. Responses to these questions for the three rider risk groups are summarised in Figure 3.11 and Figure 3.12 and Table B. Further information on responses to these questions is given in Table B10 of Annex B

Table B - Accident risk estimates by riders in the 3-rider groups (self-completion questionnaire)

3.43 It is noteworthy that the three groups differed in their estimates of risk for typical riders, as expected by their characteristics - i.e. the Risk Deniers gave lower estimates than the Optimistic Accepters, and the Realistic Accepters gave the highest estimates However, the differences were rather small. Somewhat bigger differences (in percentage terms) were apparent for the estimated risks for 'riders exactly like me' - so that, for example, the Realistic Acceptors estimated the risks each category of injury to be two to three times greater than did the Risk Deniers.

Figure 3.11 - Accident risk for 'riders like me'

Figure 3.12 - Accident risk for 'typical riders'

3.44 Table B17 in Annex B shows for each rider group the average scores on a range of demographic, behaviour, risk and attitude measures. Some of the figures have been reproduced in Table C. The average rider age in each group is very similar; however their motorcycle riding experience differs. The riders in the first group have most experience with an average of just over 18years (excluding periods with breaks longer than 12-months), the second group has about 18months less experience on average and the third group has the least experience with an average of just over 11years. The third group have, on average, smaller capacity motorcycles and ride fewest miles per year with an average on about 4,500 miles. The third group tend to rate their likelihood of being accident involved in the next 12-months as slightly higher than the other two groups and rate their skill level as slightly lower.

Table C Mean values for each group on a number of rider measures

* indicates that there was statistically significant difference between some of the group means on this variable

3.45 There were only 39 riders who were accident involved in the past 12-months (11%) but 64% of riders reported at least 1 near miss. An analysis of accident and near-miss involvement by rider group (from the self-completion questionnaire) indicated it is the first group of riders who actually report least near-misses - see Table D .

Table D - Characteristics of the 3-rider groups (self-completion questionnaire)

3.46 The third group tend to be 600c sport/sport-touring bike riders, whereas the first group are more likely to be on bigger bikes. All groups rated motorcycling as more risky than driving a car.

3.47 Table B17 in Annex B also shows that it is the second group of riders, the Optimistic Accepters, which have the 'riskiest profile'. Compared to the other groups, this group:

• Is slightly more sensation seeking (especially the intensity of the sensation)
• Is more likely to exhibit speeding behaviours
• Is more likely to pull stunts
• Likes speed more

The third group of riders (the Realistic Accepters) tended to report slightly fewer speed or stunt behaviours, were less strongly motivated by a liking for speed, and by the pleasure of motorcycling.

• Cycling
• Driving a car
• Rock climbing
• Skiing
• Motorcycling
• Hang-gliding
• Surfing

3.49 Respondents rated these activities in terms of three different constructs: the risk of having an accident while doing the activity, the seriousness of an accident during the activity, and the extent that personal skill can be used to avoid death or injury while engaged in the activity'. A low rating means that an activity was perceived as not risky, that the outcomes of accidents were not likely to be serious, and that the outcome is controllable by personal skill.

3.50 The average rating of these activities is shown in Figure 3.13, where the activities are ordered by average rating on the perceived risk. It shows that motorcycling was perceived as more risky than driving a car or cycling, about as risky as surfing (no significant difference between motorcycling and surfing average values), but less risky than skiing, rock climbing and hang-gliding. However, the outcomes of accidents were rated as more serious for motorcycling than skiing or surfing. Motorcycling accidents were rated as more subject to personal control than were accidents resulting from surfing, skiing, rock-climbing and hang-gliding but were not statistically significantly different from personal control when cycling.

Figure 3.13 - Activity average ratings (ordered by risk)

3.51 Examining these scores for the three rider groups, showed that all three groups ranked the risk of the non-motorcycling activities in the following order:

Hang gliding (highest risk)
Rock climbing
Skiing
Surfing
Cycling
Driving a car (lowest risk)

The Optimistic Accepters and the Realistic Accepters placed motorcycling between surfing and skiing in this list. However, the Risk Deniers placed motorcycling between driving a car and cycling. Further information is given in Annex B, Tables B18, B19 and B20, which show that the Risk Deniers also ranked motorcycling risks as more controllable in comparison to other activities than did the other groups.

Summary

3.52 Three rider groups were identified on the basis of responses to the self-completion questions about the relative risk of motorcycling and car driving. These were labelled as Risk Deniers, Optimistic Accepters and Realistic Accepters.

3.53 Estimated absolute risk levels for typical riders, expressed as the number of riders out of 100 who would have an accident in a year, varied between the groups. The Risk Deniers gave the lowest estimates, and the Realistic Accepters the highest. There were stronger differences between groups for the estimated risk of 'riders exactly like me'. Here, the Realistic Accepters estimated the risk to be two to three times greater than did the Risk Deniers. Optimistic Accepters tended to have the riskiest profiles in terms of sensation seeking, speeding behaviours, stunts and a liking for speed, though the failure of Risk Deniers to recognise the levels of risk involved in motorcycling might itself be considered a risk factor.

3.54 The Risk Deniers placed motorcycling between car driving and cycling in terms of risk of accident. The other two groups placed it between surfing and skiing. The Risk Deniers also tended to believe more strongly than the other groups that motorcycle risks can be controlled by rider skill.

« Previous | Contents | Next »