|
a b s t r a c t 摘要
本研究的核心目的是检查风险意识，风险意识，省心，一方面风险承受能力和安全性，并降低风险需求的优先级之间的关联在另一方面交通工具。结果是基于秋冬季2004年（N = 510）期间从挪威人口登记获得挪威公众中随机选出代表性样本进行了自我完成问卷调查。答复率为51％，与普通人群的统计数据比较表明，该样品充分代表挪威人口的性别，年龄，受教育程度和受伤的经验方面。三个变量（风险意识，省心和安全的优先级）为降低风险的需求显著预测。也有担心和风险承受能力之间的负相关性。相反，对风险的认识以前的研究，目前的研究没有发现假设支持，一般的风险认知是一个显著的预测，无论是安全性的优先级，也不是降低风险的需求。然而，由于预期的感知风险是紧密联系在一起的受访者担心有关。担心是安全的优先级，以及为降低风险需求的显著预测。个人的担心并没有对风险管控作为一般需求的担心同样的效果。The core aim of the present study is to examine the associations between risk awareness, risk perception, worry and risk tolerance on the one hand and priority of safety and demand for risk mitigation in transport on the other hand. The results are based on a self-completion questionnaire survey carried out in a randomly selected representative sample of the Norwegian public obtained from the Norwegian population registry during the autumn and winter 2004 (n = 510). The response rate was 51% and comparisons with general population statistics showed that the sample was adequately representative of the Norwegian population in regard of gender, age, education and previous injury experience. Three variables (risk awareness, worry and priority of safety) were significant predictors of demand for risk mitigation. There was also a negative association between worry and risk tolerance. Contrary to previous research on risk perception, the present study did not find support for the hypothesis that general risk perception was a significant predictor, neither of priority of safety, nor of demand for risk mitigation. However, as expected perceived risk was strongly associated with the respondents’ worry. Worry was a significant predictor of priority of safety as well as for demands for risk mitigation. Personal worry did not have the same effect on demand for risk mitigation as general worry.
1. Introduction 介绍
Demand for risk mitigation is the demand from the public towards policy makers and experts to reduce a specific risk source (Moen & Rundmo, 2004). Risk mitigation is defined as the steps policymakers and experts take to reduce a risk source. In addition, safety behaviour and health/accident protection behaviour may be categorised as individual-level risk mitigation. The present paper focuses on demands for risk mitigation from the public on policymakers and experts related to reducing hazards and risks in transport. This includes public (plane, train, bus, boat and ferry) as well as private transportation modes (car, motorcycle, scooter, bike and as a pedestrian). Compared to private transportation it is often more difficult to replace risk behaviour with safe behaviour or health protection behaviour by individual-level decisions in public transportation. However, also when e.g. driving own car, the driver is at the mercy of all other road users and their unpredictable behaviour. Still the driver may be able to avoid an accident by individual-level action in a risky situation, whereas a passenger would not have the same ability to act. Consequently, the driver is more in control compared to the passenger. This may also cause differences in demand for risk mitigation placed on policy makers and experts in regard of public versus private transportation. On the other hand, it could be argued that passengers in public transportation e.g. may have the option of sitting in the location where there is the lowest statistical likelihood of injury in the event of a crash. In civil aviation the passengers have the choice of different airlines, and thereby 1369-8478/$ - see front matter 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.trf.2013.04.004 [email protected] (T. Nordfj.rn). Imprint logoJournal logo
they may make an informed decision about which company to use, based upon their relative safety track records. However, when a decision has been made, the passengers in public transportation do not have the same ability to control the danger by individual behaviour as the driver, pilot, ferry or passenger boat officer, etc. Passengers of public transportation are at the mercy of these operators and, consequently, the perception of control may be considered as lower in public compared to private transport. Several studies have shown that risk perception is associated with health protection behaviour as well as demands for risk mitigation (Borcherding, Rohrman, & Eppel, 1986; Marris, Langford, Saunderson, & O’Riordan, 1997; Rohrman, 1994; Sj.berg, 1999). Risk perception is understood to be the risk we envisage, which results from how we assess the probability of a particular incidence or event and how concerned we are in relation to the risk (Risk Research Committee, 1980). Fischhoff, Slovic, Lichteinstein, Read, and Combs (1978, 2000) found support for the hypothesis that perceived risk and public demand of risk mitigation measures are positively correlated. Thereby, it may be expected that when a perceived risk is ‘‘too high’’ and health protection behaviour at an individual level alone is unsuited to reduce the risk, a demand of mitigating risk increases upon policymakers and experts. However, several characteristics may influence perceived risk and there is no simple way to decide whether a risk is perceived to be ‘‘too high’’. Factors which are important for risk perception include cognitive evaluations related to the probability of an event or incident as well as the evaluation of consequences or severity if this event or incident takes place. The psychometric paradigm in risk perception (Fischhoff, Slovic, Lichteinstein, Read, & Combs, 1978, 2000; Slovic, 1987, 2000) hypothesised that nine general properties of activities or technologies are important for the subjective risk judgement (Fischhoff et al., 2000). These are voluntariness of risk, immediacy of effect, knowledge about the risk by the person who is exposed to the potentially-hazardous risk source, knowledge about the risk in science, control over the risk, newness, chronic/catastrophic potentials, common/dread potentials, and severity of consequences. The most important predictors of risk perception seemed to be ‘‘novelty’’ and ‘‘dread’’ factors (Fischhoff et al., 2000). Slovic (1999) and Palm (1999) refer to several studies which showed that probability as well as the consequences of an accident are important predictors of health protection behaviour, e.g. the health belief model (Janz & Becker, 1984; Rosenstock, 1966), subjective expected utility theory (Edwards, 1961) and prospect theory (Kahneman & Tversky, 1979). A great deal of research has upheld the prediction power of the health belief model (Becker & Rosenstock, 1984; Curry & Emmons, 1994) and several other empirical studies have shown that general risk perception was significantly associated with various health protection behaviours (Brun, 1992; Kraus & Slovic, 1988; Marris, Langford, Saunderson, & O’Riordan, 1997; Rohrman, 1994; Slovic, MacGregor, & Kraus, 1987; Slovic & Monahan, 1995). The consequentialist perspective as well as the risk-asfeeling perspective also hypothesise that cognitive evaluations of risk, including probability judgements, are important determinants for decisions and behaviour (Loewenstein, Weber, Hsee, & Welch, 2001). Sj.berg (1999) showed that demand for risk mitigation was most strongly related to the expected severity of the consequences, while general risk perception was more strongly related to the probability of harm. Probability had a weaker relation to demand for risk mitigation. These results indicate that demands for risk reduction primarily are driven by the severity of consequences, not by the probability of risk. Accordingly, Moen and Rundmo (2005) also found support for the conclusion that the subjective judgement of probability of an accident or injury was the most significant predictor of risk perception. The same study showed that probability judgements contributed significantly to explaining perceived risk in public transport. Assessment of consequences was an insignificant factor for explaining risk perception. Thus, the conclusions of empirical research of the role of probability and consequences for risk mitigation seem to be contradictory in terms of the role of these two factors. However, all the studies referred to above seem to agree that general risk perception may be an important predictor variable of health protection behaviour as well as demand for risk mitigation independent of controversies about the relative importance of probability and consequence judgements. In addition to the role of the general risk perception, worry may also be a relevant determinant of demand for health protection behaviour. It seems reasonable that especially thinking about the severity and dreadfulness of the consequences may be associated with worry. Worry may also be hypothesised to be positively associated with demand for risk mitigation. Emotion or affectivity, e.g. worry, is also recognised as an important predictor of perceived risk. In the psychometric perspective dread is said to constitute the affective factor of perceived risk (see Sj.berg, 2004). Myers, Henderson-King, and Henderson- King (1997) showed that worry was related to personal action to reduce risk as well as a desire for risk regulation. Worry has been found to be influenced by cognitive judgements of risk and has been reported as the most significant predictor variable in explaining differences in individual desire for action and priority of risk reduction measures (Baron, Hershey, & Kunreuter, 2005). Accordingly, the present article hypothesises that general risk perception may be a significant predictor of affectivity, i.e. transport-related worry when thinking about the risk associated with specific transportation modes, and that general risk perception (including probability and consequences) as well as worry are significant predictors of demand for risk mitigation. Whether or not a risk is judged to be ‘‘too high’’ may also be associated with a general tendency to perceive all risks as large or small. This has been called risk awareness as well as ‘‘risk amplification–attenuation’’ (Sj.berg, 2004). If risk is rated to be high in one domain it is more likely to be rated as high in another domain. Sj.berg (2004) found that risk amplification– attenuation was a significant predictor variable of personal as well as general risk perception. Rundmo and Moen (2005) showed that perceived risk was significantly associated across various types of transport. In addition, Rundmo (1999) showed that risk behaviour as well as health protection behaviour in one area, e.g. health behaviour, is positively associated with risk and safety behaviour in another area, e.g. consumer and environmental behaviour. There is also some evidence 184 T. Rundmo, T. Nordfj.rn / Transportation Research Part F 20 (2013) 183–192 #p#分页标题#e#
showing that occupational risk taking is positively associated with risky driving in traffic and risk taking behaviour in the leisure time (Rundmo, Ulleberg, & Bjerke, 1997). In both these studies perceived risk significantly predicted the various types of risk behaviour. Accordingly, the present study hypothesises that risk awareness significantly predicts risk perception and that risk awareness is directly associated with a demand for risk mitigation in transport. In addition to the magnitude of perceived risk and risk awareness, acceptability or tolerance may influence decisions about health protection behaviour and demand for risk mitigation. In an influential paper Starr (1969) concluded that the public was willing to accept risks by voluntary activities that are about 1000 times greater than the public would tolerate from involuntary activities. It was also concluded that the acceptability of risk was approximately proportional to the third power (cube) of the benefits and that the acceptable level of risk is contrariwise related to the number of exposed persons. Fischhoff et al. (2000) showed in a study carried out among 76 respondents that many of the 30 risky activities and hazards were judged to be unacceptably high by the respondents. It could be argued that risks are never accepted, however, they may be tolerated, partly because the activities involved have some important benefits. Transport, whether public or private, is often motivated by the beneficial consequences and may consequently be more tolerated. Risk tolerance means that we tolerate, however not accept, that we are resigned to their existence and decide to ignore them. Risk tolerance, therefore, has replaced the risk acceptance concept, which should be avoided due to its weaknesses (Sj.berg, 1999). Additionally, Fischhoff et al. (2000) found a significant, however, not very strong, relationship between risk benefits and risk tolerance. Risk perception was negatively associated with tolerance, especially the two factors severity of consequences and dread. As mentioned above, the measurement of the dread factor is also intending to capture affectivity or emotion related to perceived risk. Accordingly, it seems reasonable to expect risk perception as well as worry of hazards in transport to be associated with a demand for risk mitigation. A high demand for risk mitigation is expected when the risk is perceived to be high and the tolerance is low. The greater the risk is perceived and the more worry it causes the less tolerance may be expected. Risk perception, worry and risk tolerance may, in addition to a demand for risk mitigation, also be associated with priority of safety. A higher priority of safety also entails greater demand for risk reduction measures. Several studies investigating occupational safety have shown that management and employee priorities of safety related to commitment and involvement of safety in these groups are important for employee risk and accident prevention behaviour (Cohen, 1977; Rundmo, 1996; Smith, Cohen, Cohen, & Cleveland, 1978). Rundmo (1996) also showed that management priority of safety and management commitment and involvement in safety promotion were associated with risk perception and employees’ accident protection behaviour. Empirical studies of the role of priority of safety in risk mitigation are scarce and to the authors’ knowledge no studies have related such priorities to demands for risk mitigation in transport. However, studies carried out to study occupational safety may indicate that the more safety is prioritised when choosing between various types of transportation modes, the greater is also the demand for risk mitigation. From previous empirical research it is well known that there are differences in risk perception due to gender, education and age. Women perceive more risk than men (Byrnes, Miller, & Schafer, 1999; DeJoy, 1992; Drottz Sj.berg, 1991; Drottz- Sj.berg, Rayanseon, & Martyushou, 1993; Flynn, Slovic, & Mertz, 1994; Glendon, Dorn, Davis, Matthews, & Taylor, 1996) and may as such be more risk sensitive than men. There is also some evidence supporting that a higher level of education relates to lower risk judgements (Kraus, Malmfors, & Slovic, 2000). The same differences have also been found when lay people were compared to ‘‘experts’’ (Rundmo & Moen, 2005). Young people assess dangers to be lower compared to elderly people and younger individuals tend to take more risk (Cohn, Macfarlane, Yanwz, & Imai, 1995; Glendon et al., 1996; Jonah & Dawson, 1987; Matthews & Moran, 1986; Sicard, Joouve, Courderc, & Blin, 2001). In addition, previous injury experience may be associated with risk perception and accident involvement (Rundmo, 1995). Because the evidence-base is solid the present study did not specifically test whether gender, age, education and previous injury experience influenced risk perception and risk awareness. A heuristic working model aimed to explain priority of safety and demand for risk mitigation is presented in Fig. 1. The present study aims at testing the fit of this model.
2. Methods 方法
2.1. Sample
The results are based on a self-completion questionnaire survey carried out during the autumn and winter 2004 in a randomly selected representative sample of the Norwegian population obtained from the Norwegian population registry (n = 510). This registry contains records of all living individuals in Norway. The sampling was carried out by an electronic selection of household addresses conducted by a Norwegian Firm which has access to the registry. The response rate was 51%. Visual comparisons of the present sample and statistics from the general Norwegian population (Norwegian Statistics, 2004) in regard of gender, age and education showed that our sample did not diverge substantially from the population statistics. The respondents were on average 43 years of age and 43% were educated at a college or university. A total of 57% had a high school degree and the remaining had completed primary/secondary school. A total of 51.8% of the respondents were women. T. Rundmo, T. Nordfj.rn / Transportation Research Part F 20 (2013) 183–192 185
2.2. Questionnaire
The questionnaire was developed by an expert group of safety researchers at the Norwegian University of Science and Technology. Relevant literature related to transport safety was screened in order to ensure that relevant items were included (e.g. Sj.berg, 1999; Slovic, 2000; Ulleberg & Rundmo, 2003). Information from Statistics Norway (2004) was used to examine the most commonly used modes of transportation in Norway. A new measurement instrument was established to examine risk awareness. The respondents were asked to judge how probable it was to experience an injury or accident due a total of 28 potentially-hazardous activities. The list of activities included natural, technological, foot-related, life-style related and terror-related hazards. The list was based on the hazard list established by Fischhoff et al. (1978). Some additional items considered particularly relevant in a Norwegian risk context were also added, e.g. avalanches. A seven point evaluation scale ranging from ‘‘very probable’’ to ‘‘not at all probable’’ was applied on all the indicators. This scale has been deployed in a number of previous studies (e.g. Lund & Rundmo, 2009; Nordfj.rn, J.rgensen, & Rundmo, 2011). The scale of risk perception was established on the basis of the theoretical assumption that risk perception is the probability severity of consequences of undesirable events. When risk perception is measured it is important to deploy items specifically related to the hazards in question. Based on transport statistics in Norway (Statistics Norway, 2004) we identified 10 different transportation modes which were included into the questionnaire. These modes included the most common public transportation modes (plane, train, bus, boat and ferry) and private transportation modes (car, motorcycle, scooter, bike and as a pedestrian). The respondents were asked to judge the risk of experiencing an accident or injury by using each of these 10 modes on a seven point evaluation scale. The scale ranged from ‘‘very great risk’’ to ‘‘no risk at all’’. The scale has also been deployed in previous empirical accounts (e.g. Moen & Rundmo, 2006; Nordfj.rn & Rundmo, 2010). The same list of 10 transportation modes was applied to measure worry, risk tolerance and demand for risk mitigation (see also Moen, 2008; Nordfj.rn & Rundmo, 2010). Seven point evaluation scales were also applied for these measures. The measure of worry used a scale which ranged from ‘‘very worried’’ to ‘‘not at all worried’’ and for the risk tolerance list the scale ranged from ‘‘absolutely tolerable’’ to ‘‘not at all tolerable’’. A total of six indicators were included to measure priority of safety (Moen, 2008). A seven point Likert type evaluation scale was applied ranging from ‘‘fully agree’’ to ‘‘not at all agree’’. The seven-point scale measuring the demand for risk mitigation ranged from ‘‘very important’’ to find countermeasures to ‘‘not at all important’’. The respondents assessed their worry when thinking about the risk of an average Norwegian when using each of the transportation modes (general worry). In addition, they were asked about their worry when thinking about the possibility for themselves personally to experience a transport accident or injury (personal worry). The respondents were also asked about their judgements regarding the importance of injury and accident countermeasures implemented by authorities (general demand for risk mitigation). It also intended to measure the importance that the respondents personally conducted risk reduction activities (personal demand for risk mitigation). In addition, the questionnaire asked about the respondents’ gender, age education and previous injury experience. #p#分页标题#e#
2.3. Statistical analysis
To determine the dimensionality of risk awareness as well as priority of safety confirmatory factor analyses were applied. The confirmatory factor analyses were carried out by the hypothesis that the factor structures would be similar to those Image of Fig. 1Fig. 1. Heuristic working model of the study.
reported in exploratory factor analysis in previous studies which utilised the same instruments (e.g. Nordfj.rn & Rundmo, 2010; Nordfj.rn et al., 2011; Rundmo and Moen, 2006). Furthermore, Cronbach’s a was used to determine the internal consistency of the indices. Structural Equation Modelling Made Simple (STREAMS) offers a consistent interface to the LISREL program (J.reskog & S.rbom, 1993) and was used as a graphical support layer to this program (Gustafsson & Stahl, 2000). Various fit indicators were applied to examine the fit of the models. This included the comparative fit index (CFI), Critical N (CN), goodness-of-fit index (GFI), the adjusted goodness-of-fit index (AGFI), and the root mean square error of approximation (RMSEA). Traditionally, a CFI, GFI and AGFI of .90 or above and a RMSEA of .07 or less have been considered to indicate a good fit between the model and the data. The Critical N should be above 200 to demonstrate a good model fit. The reliability was assessed for all the dimensions or sub-elements/indices. According to Nunnally’s (1978)criteria, alpha reliability obtained for scales should equal or exceed .70. However, the alpha coefficient is sensitive to the number of test items, which implies that it is easier to obtain a high value with many compared to few items. This should be considered when evaluating the homogeneity of the items. In addition, item analysis was carried out for each of the indices. The average corrected interitem correlations should be above 0.30 to constitute a reliable scale (Hair, Anderson, Tatham, & Black, 1998). In addition to the confirmatory factor analysis STREAMS was used to examine path models with latent variables to predict priority of safety and demand for risk mitigation. The same fit statistics were used to assess model-data in confirmatory factor models (measurement models) and the path models (structural models).
3. Results 结果
3.1. Factor structure of the measurement instruments
A confirmatory factor analysis showed that risk awareness segmented into five dimensions. The first dimension intended to measure risk perception related to natural hazards and included five indicators measuring the respondents’ risk judgement related to the probability of avalanche (snow), which happens frequently in Norway in the wintertime due to many high and steep mountains, earth falls, floods, fires and storms, Cronbach’s a = .903, Mean Corrected Item-Total Correlation = .75. The next dimension measured lifestyle-related hazards and included four indicators measuring the probability of sickness or an health injury due to lack of exercise, unhealthy eating habits, smoking and alcohol, Cronbach’s a = .854, Mean Corrected Item-Total Correlation = .67. The third dimension was entitled technological hazards and included six indicators measuring the subjective assessment of the probability of an health injury due to the hole in the ozone layer, radiation from cellular phones and mobile transmitters, radiation from power lines, radioactive radiation and radon, Cronbach’s a = .865, Mean Corrected Item-Total Correlation = .63. The fourth dimension was food-related hazards which contained the following four items: remains of eradicates in food, chemical additives in food, contaminated food, and preservatives in food, Cronbach’s a = .898, Mean Corrected Item-Total Correlation = .74. The last dimension was risk judgements which contained a total of three items related to terrorism, war and hijacking Cronbach’s a = .821, Mean Corrected Item-Total Correlation = .64. In addition, six indicators intended to measure triviality hazards, sexuality-related hazards, i.e. sexual assault and sexual harassment, and occupational hazards. This dimension also showed a satisfactory internal consistency, Cronbach’s a > .700, Mean Corrected Item-Total Correlation > .30. However, it was excluded from the confirmatory factor analysis because it was weakly associated with risk perception in transport. The fit of the five factor model to the data was satisfactory, RMSEA = 0.060, CFI = 0.98, CN = 204.45, GFI = 0.90, AGFI = 0.87. Six indicators were used to measure the respondents’ priority of safety in transport. The indicators intended to measure two dimensions. The first dimension measured priority of safety when using various modes of transportation and the second dimension measured the importance of emphasising safety and to follow the regulations. A confirmatory factor analysis was carried out to test the fit of the model and the fit was satisfactory, RMSEA = 0.043, CFI = 0.99, CN = 636.8, GFI = 0.99, AGFI = 0.97. The first dimension consisted of the following four indicators: ‘‘When I choose a transportation mode I prioritise safety above all’’, ‘‘I don’t risk my life and health by riding with an unsafe transportation mode’’, ‘‘I will exit at the first possibility if safety regulations are violated when I use public transportation modes like bus or railroad’’ and ‘‘to choose a safe transportation mode is important to me’’, Cronbach’s a = .716, Mean Corrected Item-Total Correlation = .64. The second dimension consisted of the two indicators: ‘‘I follow safety rules when I use modes of transportation’’ and ‘‘It is important for me to emphasise safety’’, Cronbach’s a = .719, Mean Corrected Item-Total Correlation = .58. An identical set of 10 indicators, i.e. transportation modes, measured general risk perception in transport, transport-related worry and demand for risk mitigation. The items were divided into two dimensions measuring five public as well as five private transportation modes. The internal consistency of the dimensions was found to be satisfactory. For general risk perception the figures were the following: Cronbach’s a = .885, Mean Corrected Item-Total Correlation = .72, for public transportation and Cronbach’s a = .848, Mean Corrected Item-Total Correlation = .65 for private transportation. Identical figures for worry when thinking about the risks for an average Norwegian were also satisfactory, Cronbach’s a = .918, Mean Corrected Item-Total Correlation = .80 (public transportation), and Cronbach’s a = .891, Mean Corrected Item-Total Correlation = .73 (private transportation). When thinking about personal risk the identical figures for worry were as follows: Cronbach’s a = .909, Mean Corrected Item-Total Correlation = .78, and Cronbach’s a = .880, Mean Corrected Item-Total Correlation = .71 for public and private transportation, respectively. For demands for risk mitigation directed at the Norwegian policy makers and experts the internal consistency was also satisfactory, Cronbach’s a = .946, Mean Corrected T. Rundmo, T. Nordfj.rn / Transportation Research Part F 20 (2013) 183–192 187
Item-Total Correlation = .85 (public transportation) and for the importance of implementing countermeasures for the respondent personally, Cronbach’s a = .921, Mean Corrected Item-Total Correlation = .79 (private transportation). The respondents were also asked about how important it was for them personally that the risk caused by each of the 10 transportation modes was reduced (personal demand for risk mitigation). The internal consistency of the indices was as follows: Cronbach’s a = .967, Mean Corrected Item-Total Correlation = .91 (public transportation) and Cronbach’s a = .920, Mean Corrected Item-Total Correlation = .79 (private transportation).
3.2. Model testing
The next step was to examine the heuristic working model presented in Fig. 1. This model indented to predict priority of safety as well as demand for risk mitigation. In the analysis the dimensions of the predictor variables as well as the criteria listed above were entered as directly measured variables. This was carried out to avoid that this model became too complex and too preserve parsimony.
 Fig. 2 shows the results of a SEM-analysis aimed at examining the determinants of demand for risk mitigation on Norwegian policy makers and experts. The model had good fit to the data, RMSEA = 0.056, CFI = 0.94, CN = 229.26, GFI = 0.92, AGFI = 0.90. Risk awareness was found to be a significant predictor of risk perception in transport, b = .39. It was also a small, however, significant direct association between risk awareness and priority of safety, b = .07. The most important predictor variable was priority of safety, b = .39. Worry and risk awareness were also significant predictor variables, b-values of .22 and .21, respectively. These three variables alone explained 35% of the variance in demand for risk mitigation, R 2 = .35, e5 = .65. There was also a positive association between priority of safety and demand for risk mitigation, i.e. the more priority the more demands. Risk awareness had the same ‘‘effect’’, b = .21. General risk perception was strongly associated with worry, b = .78, and worry was directly as well as indirectly related to demand for risk mitigation due to the associations between this variable and priority of safety, b = .22. We also tested the same model with worry set to predict risk perception. However, this caused a non-significant path coefficient (b = .07 and worse fit than the model described above, RMSEA = 0.083, CFI = 0.81, CN = 173.24, GFI = 0.68, AGFI = 0.59. An additional model using the two dimensions of worry as manifest variables of risk perception and risk perception as a direct predictor of demand for risk mitigation was also tested. Although there was a relation between risk perception and demand for risk mitigation in this model (b = .23), the factor loadings of worry in private (.23) and public transport (.27) were relatively weak. This model also had poorer fit than the initial working model, RMSEA = 0.086, Image of Fig. 2Fig. 2. Determinants of priority of safety and demand for risk mitigation in transport – demand on Norwegian policy makers and experts. #p#分页标题#e#
CFI = 0.80, CN = 171.15, GFI = 0.64, AGFI = 0.57. This indicates that worry does not segment together with risk perception and should be treated as a separate, but related, construct. In an additional analysis worry was excluded from the analysis and risk perception was entered as a direct predictor variable with priority of safety and demand for risk mitigation on Norwegian policy makers and experts as criteria variables. This model was based on the theoretical assumption that risk perception is an independent direct predictor of demand for risk mitigation (e.g. Janz & Becker, 1984; Kahneman & Tversky, 1979). With this exception the model was identical to the model presented in Fig. 2. Perceived risk was significantly associated with both the criteria variables, however the real association was weaker compared to a model including worry. Consequently, risk perception was included as a predictor of worry and these two components were treated as separate constructs in further analyses. Based on theory and previous research we also expected risk perception to be directly associated with a demand for risk mitigation. However, this was not supported by an additional SEM-analysis where a model which used general risk perception as a direct predictor variable of priority of safety as well as demand for risk mitigation on Norwegian policy makers and experts. This model was identical to the one presented in Fig. 2 with the exception that risk perception was entered as a direct predictor of priority of safety and demands for risk mitigation on Norwegian policy makers and experts. However, the results showed that this did not cause an increase in explained variance (b < .10 in both cases) resulting in no increase in the percentage of explained variance in these two criterion variables. This model also explained 33% of the variance in priority of safety, e4 = .67 and 35% in demand for risk mitigation, e5 = .65. In both cases the contribution to an increased explained variance was less than 1% for both the predictor variables. Thus, the results of the present study did not yield any support to the idea that general risk perception is an important predictor, neither of priority of safety, nor on demand for risk mitigation. The association seems primarily to be indirect through the influence of perceived risk on worry. First, demographic variables (gender, age and education) as well as injury experience were entered as predictors of priority of safety and demand for risk mitigation on Norwegian policy makers and experts. The b-values of the associations between demand for risk mitigation and gender were 0.09. Identical figures for education were 0.00, for previous injury experience 0.02 and for age 0.03. Because they did not contribute to any significant increase in explained variance these associations were excluded from the model. However, gender, education, the respondents’ previous injury experience and age were significant predictor variables of priority of safety and an additional model was tested without the associations between these variables and demand for risk mitigation on Norwegian policy makers and experts (Fig. 2). As expected, female respondents gave more priority to safety than men in transport. Priority of safety also increased with educational level and Image of Fig. 3Fig. 3. Determinants of priority of safety and demand for risk mitigation in transport – personal demand for risk mitigation.
those who had experienced an accident or injury themselves also prioritised safety more compared to those who had no such experience. Gender is a dichotomous variable and so was injury experience. This most probably caused an underestimation of the real contribution of these predictors. This may to some extent also be true for education. Priority of safety also increased with age. Gender and age were the strongest predictor variables, b-values of .25 and .37 respectively. Risk awareness and the four demographic variables explained 33% of the variance in priority of safety, R 2 = .33, e5 = .67. There was a negative association between worry and risk tolerance, b = .12, i.e. the more often the respondents tolerated the risk the less worried where they also when thinking about them and the more risk the respondents tolerated the less priority did they give to safety, b = .15. The fit of the model to the data was satisfactory, RMSEA = 0.056, CFI = 0.94, CN = 229.26, GFI = 0.92, AGFI = 0.90. The model presented in Fig. 2 aimed at explaining demands on the Norwegian policy makers and experts for risk mitigation (general demand for risk mitigation). In the results presented in Fig. 2 the respondents assessed their worry when thinking about the risk of an average Norwegian by using each of the transportation modes. In addition, the respondents were asked about their worry when thinking about the possibility that they themselves were victimised in a transport accident (personal worry). A SEM-model identical to the one presented above was tested to investigate predictors of personal demand for risk mitigation (Fig. 3). An identical percentage of explained variance in priority of safety as in the previous figure was found, e4 = .67. However, the model in Fig. 3 was not equally successful in explaining demand for risk mitigation e5 = .71. Priority of safety was the most significant predictor variable for personal demands of risk mitigation, b = .41. In addition, risk awareness was significantly associated with personal demands of risk mitigation, b = .26. Risk perception had a strong indirect relation to personal demands of risk mitigation through worry (b = .95). As expected, age b = .35, worry b = .29, gender b = .14 education b = .26, and injury experience b = .10, also significantly affected priority of safety. However, this model explained a somewhat lower percentage of variance in demand for risk mitigation e5 = .71. It is interesting to note that contrary to what was the case for general worry, personal worry was not a significant predictor of personal demand for risk mitigation demand for risk mitigation, i.e. did not predict how important it was for the respondent personally that countermeasures aimed to improve safety were implemented, b = .03. All other predictor variables loaded equally well on this criterion variable compared to the model predicting demand for risk mitigation on the Norwegian policy makers and experts. The fit of the model was also judged to be satisfactory, however, a bit weaker compared to the previous model, RMSEA = 0.068, CFI = 0.92, CN = 176.06, GFI = 0.90, AGFI = 0.87.
4. Discussion 讨论
Several previous studies have shown that perceived risk is a significant predictor variable of health protection behaviour as well as demand for risk mitigation (Brun, 1992; Kraus & Slovic, 1988; Marris, Langford, Saunderson, & O’Riordan, 1997; Rohrman, 1994; Slovic et al., 1987; Slovic & Monahan, 1995; see also Slovic, 1999). However, there has been a debate whether it is the probability assessments or the judgement of the severity of consequences which is most important for perceived risk and demand for risk mitigation (Sj.berg, 1999; Slovic, 1999). The research in this area seems to agree that perceived risk is an important determinant for risk and safety behaviours as well as risk decisions and demands for risk mitigation. However, the present study showed that there was a weak association between perceived risk and demand for risk mitigation. In opposition to theory and previous research, an implication is that it may not be efficient to focus on risk perception when communicating risks related to transport. In the present study the respondents were asked to assess their worry when thinking about the potential danger related to 10 transportation modes. Risk perception was found to be important for this type of worry and it seems reasonable to hypothesise that perceived risk may cause worry and concern. Although the present design does not allow for causal interpretations, this was also supported by the models tested in the present study. The data also indicated support for the assumption that worry is not an integrated part of risk perception, but a separate emotional construct. The authors argue that risk perception is a cognitive construct, while worry is related to emotional reactions to cognitions about risk. On the other hand, the present study did not intend to measure background mood or pre-cognitive general worry. Studies carried out previously have shown that such emotions may precede cognitive judgements and may as such not always be a consequence of such judgements (Zajonc, 1980, see also e.g. Rundmo & Sj.berg, 1998). However, longitudinal efforts related to other risks than transport also supported the assumption that cognitive evaluations predict worry over time, while the opposite idea did not receive support in the empirical data (Kobbeltved, Brun, & Eid, 2005). It seems to be the case that when factors associated with risk perception are entered into the models, the relative importance of risk perception for risk behaviour/health protection behaviour, decisions as well as demands for risk mitigation becomes insignificant. This may indicate that the same set of background variables is important for people’s perceptions and experiences as well as their behaviour and decisions. Therefore, they are also significantly associated. However, model testing showed that risk perception does not seem to be a direct causal factor, neither in relation to safety priorities nor in demand for risk mitigation. The results of the present study indicate that perceived risk primarily has an indirect effect on priority of safety and demand for risk mitigation because it may cause worry. Worry was in turn found to be a relatively strong predictor of the respondents’ priority of safety as well as of their demand on policy makers and experts to implement safety measures to reduce the risk in transport. As expected, increased worry was also associated with a lower risk tolerance, 190 T. Rundmo, T. Nordfj.rn / Transportation Research Part F 20 (2013) 183–192 #p#分页标题#e#
while an increased risk tolerance was associated with lower priorities of safety. An implication is that countermeasures could focus on emotions such as worry instead of cognitions regarding transport safety (i.e. risk perception) in order to directly influence priorities of safety, risk tolerance and policy demands such as demands for risk mitigation in transport. Theoretical approaches which suggest a direct association between risk perception and policy demands may not have applicability when generalised to the transport domain. The risk-as-feeling model distinguishes between anticipated and anticipatory worry. Related to model tests based on cross-sectional data this distinction may cause problems. When respondents are asked to ‘‘anticipate’’ their worry ‘‘when ‘‘thinking about’’ the probability and the severity of consequences of a negative outcome, e.g. when using a specific transportation mode, it seems reasonable to assume that they have to take risk perception into account. Such worry may be a result of perceived risk, which is based on some evaluation of the probability of a negative outcome as well as the seriousness of consequences if something should happen. Thus, this type of worry could be a reaction to and ‘‘effect’’ of the cognitive evaluations of probability and consequences. It is based on a cognitive evaluation of probability and consequences and on a general perception of risk ‘‘shaped’’ by intuitive judgements of probability and seriousness of consequences. The present study distinguished between general (when thinking about the risks for an average Norwegian) and personal worry (when thinking about the risk for the respondent himself or herself) as well as demand of risk mitigation placed on policy makers and experts as well as personal demands, i.e. how important it was for the single respondent that countermeasures to reduce risks in transport were implemented. General worry was a significant predictor of demands on policy makers and experts, i.e. on Norwegian Transport Safety Authorities, to implement safety measures to reduce transport risks. However, there was a weak association between personal worry and the personal importance for the single respondents of such safety measures. In fact, general worry was a better predictor of personal demands for risk mitigation. These findings could have important implications for how risk should be communicated to the public in order to influence policy demands. From the risk perception literature it is well known that general risk, i.e. for the population at large, is perceived as larger than personal risk. This may contribute to the explanation for why general worry predicted demands for risk mitigation better than personal worry. Risk is perceived to be lower for the respondent himself or herself compared to the average citizen and the relation between worry and policy demands is not necessarily fully linear. People do, for instance, not quit smoking if their worry for health damages is quite low. Thus, an hypothesis which may be interesting to investigate further is that when worry is below a certain level, this could cause no or very low demands for risk mitigation and that worry has to be at a certain level to be associated with demand for risk mitigation as well as health protection behaviour. Some limitations of the present study warrant discussion. Related to the differences between the cognitive and emotional components of risk, Loewenstein et al. (2001) suggested that these components diverge due to two reasons: (1) emotions respond to probabilities and outcomes differently than cognitive evaluations; (2) emotions are influenced by situational variables that play only a minor role in cognitive evaluations. Hence, one could argue that cognitive evaluation will be more important when the individuals have time to assess the risk as opposed to a more immediate threat where emotions may play a more important role. As such the importance of worry may be even more important than a questionnaire assessment can reflect, because cognitions may interfere with emotions when respondents report their worry on a self-report measure. Studies to come could aim to examine these possibilities by controlled experimental designs. Low response rates and non-response bias are important issues related to questionnaire surveys. Although we cannot exclude the possibility that respondents and non-respondents diverged significantly in the present study, a comparison of the present sample and population data reflected few differences in demographics. Moreover, the sample contained a large number of individuals from a randomly selected proportion of the Norwegian population. Iversen (2004) also compared nonrespondents and respondents in a similar survey of the Norwegian public and did not find substantial differences between respondents and non-respondents.
References 文献
Baron, J., Hershey, J. C., & Kunreuter, H. (2005). Determinants of priority for risk reduction: The role of worry. <http://www.sas.upenn.edu/-baron/prot.htm> Retrieved 10.03.05. Becker, M. H., & Rosenstock, I. M. (1984). Compliance with medical advice. In A. Steptoe & A. Mathews (Eds.), Health care and human behavior (pp. 22–40). London: Academic Press Borcherding, K., Rohrman, B., & Eppel, T. (1986). A psychological study on the cognitive structure of risk evaluations. In B. Brehmer, H. Jungeman, H. Lourens, & G. Sevon (Eds.), New directions in research on decision making (pp. 245–262). Amsterdam: North-Holland. Brun, W. (1992). Cognitive components in risk perception: Natural versus manmade risks. Journal of Behavioral Decision Making, 5(2), 117–132. Byrnes, J. P., Miller, D. C., & Schafer, W. D. (1999). Gender differences in risk taking: A meta-analysis. Psychological Bulletin, 125(3), 364–383. Cohen, A. (1977). Factors in successful occupational safety programs. Journal of Safety Research, 9(4), 168–178. Cohn, L. D., Macfarlane, S., Yanwz, C., & Imai, W. (1995). Risk perception – Differences between adolescents and adults. Health Psychology, 14(3), 217–222. Curry, S. I., & Emmons, K. M. (1994). Theoretical models for predicting and improving compliance with breast cancer screening. Annals of Behavioural Medicine, 16(4), 302–316. DeJoy, D. M. (1992). An examination of gender differences in traffic accident risk perception. Accident Analysis and Prevention, 24(3), 237–246. Drottz Sj.berg, B. M. (1991). Risk conceptions, reactions and communication. In B. M. Drottz Sj.berg & L. Sj.berg (Eds.), Rhizikon. Studies of risk and hazards 1: Perception of risk. Studies of risk attitudes, perceptions and definitions (pp. 217–232). Stockholm: Center for Risk Research, Stockholm School of Economics Drottz-Sj.berg, B. M., Rayanseon, C. M. & Martyushou, A. N. (1993). Perceived risk and risk attitudes in southern russia in the aftermath of the chernobyl accident. Stockholm: Rhizikon Research Report no. 13. Center for Risk Research, Stockholm School of Economics. Edwards, W. (1961). Behavioural decision theory. In P. R. Farnsworth, O. McNemar, & Q. McNemar (Eds.). Annual review of psychology (Vol. 12, pp. 473–498). Annual Reviews: Palo Alto, CA
Fischhoff, B., Slovic, P., Lichteinstein, S., Read, S., & Combs, B. (1978). How safe is safe enough? A psychometric study of attitudes towards technological risks and benefits. Policy Science, 9(16), 127–152. Fischhoff, B., Slovic, P., Lichteinstein, S., Read, S., & Combs, B. (2000). How safe is safe enough? A psychometric study of attitudes towards technological risks and benefits. In P. Slovic (Ed.), Perception of risk (pp. 80–104). NY: McGraw-Hill. Flynn, J., Slovic, P., & Mertz, C. K. (1994). Gender, race and environmental health risks. Risk Analysis, 14(6), 1101–1108. Glendon, I. A., Dorn, L., Davis, D. R., Matthews, G., & Taylor, R. G. (1996). Age and gender differences in perceived accident likelihood and driver competencies. Risk Analysis, 16(6), 755–762. Gustafsson, J., & Stahl, P.A. (2000). STREAMS User’. Vsion 2.5 f Windows. M.lndal: Multivariate Ware. Hair, J. F., Anderson, R. E., Tatham, R. L., & Black, W. C. (1998). Multivariate data analysis. New York: Prentice-Hall, Upper Saddle River. Iversen, H. (2004). Risk behaviour in traffic. A study investigating relationships between attitudes, personality, stress and behaviour. Trondheim: NTNU. Janz, N. K., & Becker, M. H. (1984). The health belief model: A decade later. Health Education Quarterly, 11(1), 1–47. Jonah, B. A., & Dawson, N. E. (1987). Youth and risk: Age differences in risky driving, risk perception, and risk utility. Alcohol, Drugs and Driving, 3(3–4), 13–39 Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decisions under risk. Ergonomics, 47(2), 263–291. Kobbeltved, T., Brun, W., & Eid, J. (2005). Risk as feelings or risk and feelings? A cross-lagged panel analysis. Journal of risk research, 8(5), 417–437. Kraus, N. N., Malmfors, T., & Slovic, P. (2000). Intuitive toxicology: Experts and lay judgements of chemical risks. In P. Slovic (Ed.), Perception of risk (pp. 285–315). NY: McGraw-Hill Kraus, N. N., & Slovic, P. (1988). Taxonomic analysis of perceived risk: Modelling individual and group perceptions with homogenous hazard domains. Risk Analysis, 8(3), 435–455. Loewenstein, G. F., Weber, E. U., Hsee, C. K., & Welch, N. (2001). Risk as feelings. Psychological Bulletin, 127(2), 267–286. Lund, I. O., & Rundmo, T. (2009). Cross-cultural comparisons of traffic safety, risk perceptions, attitudes and behaviour. Safety science, 47, 547–553. Marris, C., Langford, I., Saunderson, T., & O’Riordan, T. (1997). Exploring the ‘‘psychometric paradigm’’: Comparisons between aggregate and individual analysis. Risk Analysis, 17(3), 303–312. Matthews, M. L., & Moran, A. R. (1986). Age differences in male drivers perception of accident risk: The role of perceived driving abilities. Accident Analysis and Prevention, 18(4), 299–313. Moen, B. E. (2008). Risk perception, priority of safety, and demand for risk mitigation in transport. Doctoral thesis, Norwegian University of Science of Technology, Department of Psychology, Trondheim, NO. Moen, B. E. & Rundmo, T. (2004). Explaining demand for risk mitigation. Trondheim: Rotunde Publ. no. 86 (ISDN 82-7892-026-5). Moen, B. E. & Rundmo, T. (2005). Worrying about transport risks. In: T. Rundmo (Ed.). Risk judgement and safety in transport (pp. 5–16). Trondheim: Rotunde Publ., no. 87. Moen, B. E., & Rundmo, T. (2006). Perception of transport risk in the Norwegian public. Risk Management, 8, 43–60. Myers, J. R., Henderson-King, D. H., & Henderson-King, E. I. (1997). Facing technological risks: The importance of individual differences. Journal of Research in Personality, 31(1), 1–20. Nordfj.rn, T., J.rgensen, S. H., & Rundmo, T. (2011). A cross-cultural comparison of road traffic risk perceptions, attitudes towards traffic safety and driver behaviour. Journal of Risk Research, 14, 657–684. Nordfj.rn, T., & Rundmo, T. (2010). Differences in risk perception, priorities, worry and demand for risk mitigation related to transport among Norwegians in 2004 and 2008. Safety Science, 48, 357–364. Norwegian Statistics (2004). Fatalities due to transport Accidents. Oslo: Norwegian Statistics. Nunnally, J. C. (1978). Psychometric Theory (2 Ed). N.Y.: McGraw-Hill. Palm, R. (1999). Perceived risk and earthquake insurance purchase decision: A commentary on a paper by Lennart Sj.berg. Journal of Risk Research, 2(4), 289–294 . Accidents in Norway. How do we perceive and handle risk? Oslo: Norwegian Research Council. Rohrman, B. (1994). Risk perception of different societal groups: Australian findings and cross-national comparisons. Australian Journal of Psychology, 46(3), 150–163 Rosenstock, I. M. (1966). Why people use health services. Millbank Memorial Fund Quarterly, 44, 94. Rundmo, T. (1995). Perceived risk, safety status, and job stress among injured and noninjured employees on offshore petroleum platforms. Journal of Safety Research, 26(2), 87–97. Rundmo, T. (1996). Associations between risk perception and safety. Safety Science, 24(3), 197–209. Rundmo, T. (1999). Perceived risk, health and consumer behaviour. Journal of Risk Research, 2(3), 187–200. Rundmo, T., & Sj.berg, L. (1998). Risk perception by offshore personnel during bad weather conditions. Risk Analysis, 18(1), 111–118. Rundmo, T., Ulleberg, P., & Bjerke, W. (1997). Safety attitudes and risk behaviour – Employee assessments. In B. M. Drottz Sj.berg (Ed.), Proceedings. New risk frontiers. 10th anniversary the society for risk analysis Europe (pp. 556–564). Stockholm: Center for Risk research, Stockholm School of Economics. Rundmo, T., & Moen, B.E. (2005). Demand for Risk Mitigation in Transport. In T. Rundmo, & B.E. Moen (Eds.), Risk Judgement and Safety in Transport. (pp. 31–45). Trondheim: Rotunde Publ. no. 87. Rundmo, T., & Moen, B. E. (2006). Risk perception and demand for risk mitigation among experts, politicians and lay people in Norway. Journal of Risk Research, 9(6), 623–640. Sicard, B., Joouve, E., Courderc, H., & Blin, O. (2001). Age and risk-taking in Frenc Naval crew. Aviation Space and Environmental Medicine, 72(1), 235–252. Sj.berg, L. (1999). Consequences of perceived risk: Demand for risk mitigation. Journal of Risk Research, 2(2), 129–149. Sj.berg, L. (2004). Explaining individual risk perception: The case of nuclear waste. Risk Management: An International Journal, 6(1), 51–64. Slovic, P. (1987). Perception of risk. Science, 236(4799), 280–285. Slovic, P. (1999). Comment: Are trivial risks the greatest of all? Journal of Risk Research, 2(4), 281–288. Slovic, P. (2000). Perception of risk. In P. Slovic (Ed.), Perception of risk (pp. 220–231). NY: McGraw-Hill. Slovic, P., MacGregor, D. G., & Kraus, N. N. (1987). Perception of risk from automobile safety defects. Accident Analysis and Prevention, 19(5), 359–373. Slovic, P., & Monahan, J. (1995). Probability, danger, and coercion: A study of risk perception and decision making in mental health law. Law and Human Behavior, 19(1), 49–65. Smith, M. J., Cohen, H. H., Cohen, A., & Cleveland, R. J. (1978). Characteristics of successful safety programs. Journal of Safety Research, 10(1), 5–15. Starr, C. (1969). Social benefits versus technological risk: What is our society willing to pay for safety? Science, 165(899), 1232–1238. Statistics Norway (2004). Statistics by subjects. Derived 15.06.12. Ulleberg, P., & Rundmo, T. (2003). Personality, attitudes and risk perception as predictors of risky driving behaviour among young drivers. Safety Science, 41, 427–443 Zajonc, R. B. (1980). Feeling and thinking. Preferences need no inferences. American Psychologist, 35(2), 151–175. 192 T. Rundmo, T. Nordfj.rn / Transportation Research Part F 20 (2013) 183–192 #p#分页标题#e#
|
 |
|||
| 网站地图 |
