Car Rental Insurance Tips

Rental experience and likelihood to purchase rental car insurance among young adults

Post By: Insurance Top Stories

Several studies have demonstrated an inverse relationship between experience in an activity and perceived risk for the individual in the activity. That is, as experience increases, perceived risk decreases. This phenomenon has been labeled risk acculturation. The present study aims to examine prior rental car experience for its effect on estimated odds of an accident and likelihood to purchase rental car insurance.

Car Rental Insurance Tips

Introduction
Several studies support the existence of a general phenomenon of risk acculturation, a diminishment of perceived risk in an activity as one’s experience in the activity increases. This was first reported in a longitudinal study of skydivers as participants progressed from novice to more experienced divers (Celsiet al., 1993). Initially terrified of jumping out of an airplane, divers gradually began to think of the activity as normal as they acquired experience. A similar inverse relationship between experience and perceived risk was reported for participants during a 14-week scuba diving course (Morgan and Stevens, 2008) and for a cross-sectional study of experience and perceived risk in mountain biking (Creyer et al., 2003). Halpern-Felsheret al.(2001) explored the generalizability of risk acculturation, finding support for a variety of risky behaviors, such as drunk driving and unprotected sex.

In contrast to the above, some investigators have failed to demonstrate risk acculturation. Barnett and Breakwell (2001) examined the relationship between experience in an activity and risk perception, finding no significant correlation between the two for a variety of activities (skiing, smoking, driving, having unprotected sex, traveling by plane, or drinking alcohol). In one of the few investigations to study risk acculturation within a business context, Pireset al.(2004) report that frequency of internet purchases over the previous 12 months was not correlated with perceived risk. In summary, it might be concluded that the literature on the effect of experience on risk perception is equivocal.

The present study assumes the existence of risk acculturation and examines its effects within a rental car insurance purchase scenario. This context is intuitively appealing because respondents have already self-segregated into those that have (and do not have) experience in renting a vehicle, and there is an opportunity to measure risk perception and risk reduction behavior (likelihood of insurance purchase) all in one contextual unit. Also, a sample of young consumers of narrow age range is particularly appropriate for this study as it minimizes any effect of age on the dependent variables yet allows easy acquisition of subjects who do not have experience in renting a vehicle. The experience factor and the phenomenon of risk acculturation have apparently not been studied before in an insurance context, and so the results are believed to be a contribution to knowledge in this area.


Background and hypotheses

Perceived risk
A number of factors influence risk perception (Sjoberg, 2000) including: real (actuarial) risk, subjective heuristics and biases, whether the risk target is the rater or an ‘‘other’’ person, and a variety of individual factors including risk sensitivity. It should be emphasized that consumer evaluation of risk is subjective. For example, rare (but spectacular) risks are often overestimated (Slovic, 1987). However, for many common hazards, people are able to give risk estimates that are strongly correlated to statistical data (Sjoberg, 2000).

Historically, researchers have viewed risk evaluation from a rationalistic rather than affective paradigm (Finucaneet al., 2000). However, mood appears to have an important influence on risk judgment. Johnson and Tversky (1983) found that people in an induced positive mood gave low risk estimates while people in a negative mood gave a more pessimistic judgment, a conclusion also supported by more recent research (Finucaneet al.,2000; Hogarthet al., 2007). An additional factor is the risk target. In general, people tend to rate their own chance of a stated misfortune as being less than that of an average ‘‘other’’ at-risk person. This phenomenon has been labeled the ‘‘optimistic’’ bias due to the impossibility that everyone’s risks are lower than those of everyone else (Whalenet al., 1994).

Risk and insurance
The nature of insurance is that it is a risk-shifting strategy. Through the payment of premiums, consumers shift risk to well-capitalized companies that are better equipped to handle the consequences of a loss. Sjoberg (1999) noted that demand for risk reduction (insurance) has been assumed to be strongly predicted by the level of perceived risk. In a rare empirical study of this relationship, he investigated level of perceived risk and demand for homeowners’ insurance, finding that level of perceived risk was mostly related to the probability of harm but that demand for risk reduction was related to the expected severity of the harm. Palm (1999) criticized Sjoberg’s study, saying that mortgage lenders require homeowners’ insurance commensurate with the value of the home and that this confounded the interpretation of the dependent variable (amount of home insurance purchased). Using data from the voluntary purchase of earthquake insurance in California, Palm (1999) found evidence that both perceived likelihood of harm and possible severity of consequences were good predictors of insurance purchase.

As stated earlier, the effect of past experience on insurance purchase decisions has apparently not been investigated. The current study hopes to contribute to knowledge in this area.


Experience and risk perception
In an ethnographic study of skydivers, Celsiet al.(1993) repeatedly interviewed participants over time, reporting a relationship between experience and perceived risk. As subjects gained experience in skydiving, their fear began to subside, and participants were able to be more relaxed during jumps. Over time, participants underwent a transformation whereby the high-risk activity, once seen as extraordinary, became viewed as normal. This phenomenon was labeled risk acculturation. Additional research studying a variety of risk behaviors supports the existence of risk acculturation, finding that people who engage in a particular risk behavior (compared to people who do not) estimate their chance of harm as less likely (Halpern-Felsher et al., 2001).

It may be noted that in the two longitudinal studies which have documented risk acculturation (Celsiet al., 1993; Morgan and Stevens, 2008), the inverse relationship between experience and risk was associated with an increase in competence for the activity. Specifically, in these studies, participants were learning to skydive and scuba dive, and risk perception was accessed over the course of the training, a time in which subjects were probably progressing in self-efficacy for the activity. In contrast, some studies have failed to demonstrate a phenomenon of risk acculturation (Pireset al., 2004; Barnett and Breakwell, 2001), but in these studies, the risk to the participant for some of the listed activities may not have been ameliorated by any skill level of the participant. For example, what skill could the subject have developed to diminish the risk of internet purchase (as in the Pireset al., 2004, study) or the risk of being a passenger in an airplane (as in the Barnett and Breakwell, 2001, investigation)? Therefore, failure to demonstrate risk acculturation in these studies might be due to differences in the activities studied.

In the present study, it is assumed that prior experience in renting a car and driving in a strange city will result in an enhancement in self-efficacy for the activity such that an accident is now perceived as less likely and therefore the need for any additional insurance is diminished. This leads to a set of hypotheses:

H1. Compared to subjects who have not rented a car before, people who have rented a car before will estimate the odds of an accident (or other loss) involving their rental vehicle to be significantly less likely.
H2. Compared to subjects who have not rented a car before, people who have rented before will be significantly less likely to purchase rental car insurance.

Methods
Data were gathered by pencil and paper questionnaire from a convenience sample of junior and senior level undergraduate students enrolled in business courses at a large public university in the southwest USA. Questionnaires were distributed, completed and collected during regularly scheduled class periods. Students did not receive extra-credit for participation.

Scenario
A car rental scenario was chosen as the context within which to measure risk perception and risk reduction behavior. Driving a car is associated with certain risks. There is a risk of physical damage to the rented vehicle through collision or vandalism as well as possible theft of the car while it is in the care of the renter. Also, there is a risk of bodily injury to pedestrians and the drivers of other vehicles and their passengers as well as physical damage to the property of these people. Car rental agencies sell insurance which allow renters to manage these two principal types of risk: the collision damage waiver (CDW) and supplemental liability insurance (SLI), respectively. Thus, a car rental scenario offered the possibility of measuring risk perception (odds of an accident or other loss, magnitude of such an accident or loss were it to occur), risk-associated variables, and risk-reduction (likelihood of purchase of CDW and SLI) all in one contextual unit.


Questionnaire
The full questionnaire, except for introduction and concluding demographic items, is reproduced in the Appendix (Figure A1). The first question is a multiple-choice item addressing past risk behavior: ‘‘How would your friends describe you when it comes to taking risks?’’ There were five response choices in ascending order of risk. Subject responses were coded 1 through 5, respective to the sequence of choices.

All remaining items on the questionnaire related to a car rental scenario. Subjects were asked to imagine that they had been offered a job interview in a distant, never-before visited city and that they had reserved a rental car in order to drive around town and get a feel for what it would be like to live there. In the scenario, subjects fly to the city then arrive at the rental agency desk to pick up their Chevrolet Malibu (a mid-sized sedan) when they are asked to pause and answer a few questions.

The scenario paragraph was immediately followed by four items addressing, in sequence:

1. odds of an accident or other loss to the rental vehicle;
2. the magnitude of the potential loss referred to in the preceding question;
3. self-reported knowledge of whether their existing personal auto insurance would or would not extend to a rented vehicle; and
4. the value the subject places on the convenience and hassle reduction offered by the insurance the agent is trying to sell.

Responses for all four items were 1 through 9 numerical scales.

At this point, the questionnaire returned to the scenario and presented insurance purchase options:

1. the collision damage waiver at $19.99 per day; and
2. liability insurance at $9.99 per day.

Each purchase option was followed by a 1 through 9 numerical scale with which to indicate the likelihood of purchase of the insurance.

The three remaining items in the questionnaire addressed, in sequence:

1. the desirability of actually renting a car and driving around a city in which they were having a job interview;
2. the importance to the subject that things in their life be predictable and clear-cut; and
3. the estimated level of control the subject felt over an auto accident, theft, or vandalism of the rental vehicle in the scenario.

Response choices for these three items were all 1 through 9 numerical scales.

Measures
It may be noted that constructs in the present study are measured with single-scale items. There are advantages in using single-item measures (Jordan and Turner, 2008):

• single-item measures reduce subject fatigue;
• increase the likelihood of completed questionnaires; and
• may decrease respondent confusion when faced by a set of seemingly repetitive questions.

Indeed, subjects may believe that different numbered questions require different responses, or they may simply resent being asked questions that are so similar. An additional rationale for single-item measures occurs when there is a lack of ambiguity (from the perspective of respondents) on what they are being asked to rate. Gardneret al.(1998) noted that when the construct to be measured is uncomplicated and unidimensional, then it may be appropriate to consider a scaling format other than the standard multi-item scale.

Perhaps the ultimate question to be asked in the consideration of single-item versus multiple-item measures is whether using good single-item measures in place of multiple-item measures would change empirical findings and theoretic tests. Writing in a prestigious methodology journal, Bergkvist and Rossiter (2007) demonstrate that single-item and multiple-item measures have equal predictive validity, and they suggest that theoretic tests and empirical findings would be unchanged if good single-item measures were substituted for multiple-item measures. For all of the reasons noted, the present study has adopted single-item measures.


Results

Sample
A total of 102 completed questionnaires were returned from respondents aged 19-24, and this constituted the final sample. The sample was evenly divided between men and women. Average age was 21.37 years with a standard deviation of 1.19 years. A total of 30 subjects (17 of 51 males and 13 of 51 females) reported having rented a car before. All subjects reported holding a current, valid driver’s license.

Ten subjects reported having worked in an office that sells insurance. A preliminaryt-test screen was conducted, comparing people who had and had not worked in an insurance office on all of the metric dependent variables in this study. No significant (p,0.05) differences were found, and so these ten respondents were retained in the sample.


Manipulation check
One of the items in the questionnaire asked subjects if renting a car and driving around the city in which they might be offered a job is something they would want to do in this situation. The purpose of this question was to determine if respondents were really able to project themselves into the car rental scenario. The scenario manipulation is assumed to be successful if the mean of this variable is greater than the theoretic mid-point of the scale. For the sample as a whole, the mean was 6.70 which is significantly greater than 5 (the theoretic neutral point of the scale) by one-sample t-test (t¼8.17, df¼101, p,0.001). This suggests that respondents generally thought the scenario was realistic.


Experience effect
A comparison of subjects who have and have not rented a car before on dependent variables in this study is shown in Table I. The groups were found to significantly differ (p,0.05) on the estimated odds of an accident or other loss involving the rental vehicle and likelihood to purchase both CDW and SLI. Compared to novice renters, experienced renters reported the odds of an accident as less likely and they were less likely to purchase both types of insurance.

Table I Profile of Experience Groups On Measured Variables

Hypotheses tests
The hypotheses proposed that experienced car renters (as opposed to novices) would perceive the odds of an accident involving their rental vehicle as less likely and that they would be less likely to purchase both types of insurance coverage. Looking at Table I, it may be noted that experienced renters (compared to novice renters) perceive the odds of an accident as very remote (1 chance in 104,629 versus 1 chance in 2,441) and this difference is significant (p¼0.002). Thus,H1is upheld. Also, compared to novices, experienced renters are less likely to purchase a CDW (means of 3.63 and 4.92, respectively,p¼0.041) as well as SLI (means of 2.97 and 5.13, respectively,p,0.001). Therefore,H2is supported for both types of insurance.

Purchase of insurance would, of course, be affected by knowledge that your existing auto insurance would apply to a rental car. Whether the respondent’s existing personal auto insurance would actually extend to coverage of a rented vehicle is assumed to be a variable that is roughly evenly distributed among the two experience groups. As shown in Table I the two experience groups do not significantly differ on knowledge of insurance coverage.


Regression
Additional analyses were conducted to test the ability of experience, and other variables, to predict likelihood of purchase of CDW and SL. Results are shown in Table II. For CDW, the overall model was significant (F¼2.07,p¼0.046) accounting for 15.1 percent of variance in the dependent variable. For SLI, model results were also significant (F¼2.97,p¼0.005), and the predicting variables explained 20.3 percent of variance in the dependent variable. Of the variables in the regression equation, prior rental car experience is the only significant predictor for both CDW and SLI purchase. Interpretation of the effects of variables in Table II is straightforward because multicollinearity among them was found to be very low (highest tolerance value is 0.853). The findings suggest that prior rental experience has a significant unique predictive ability for purchase of both CDW and SLI that is not explained by the other variables in the equation.


Mediation analysis
Additional analyses were undertaken to examine the possibility that odds of an accident or knowledge of insurance coverage might mediate the effect of experience on likelihood of insurance purchase. Following the mediation test procedure of Baron and Kenny (1986), this involves a series of three regression equations. First, experience was shown to significantly predict odds of an accident (b¼20.297,p¼0.002). Second, experience was shown to significantly predict CDW purchase (b¼0.203,p¼0.041). Finally, CDW purchase was regressed on both experience and odds of an accident. For mediation to occur, odds of an accident should significantly predict CDW purchase in the third equation, but this did not occur (b¼0.095,p¼0.356) and the effect of experience on CDW is even stronger than in the second equation (b¼0.231,p¼0.027). The conclusion is that odds of an accident does not mediate the effect of experience on CDW purchase.

Table II Results of Regression Analysis

The series of regression equations was repeated using SLI purchase as the target variable and odds of an accident as a possible mediator. As before, experience is a significant predictor of odds of an accident. Second, experience significantly predicted SLI purchase (b¼0.347,p,0.001). Finally, SLI was regressed on both experience and odds of an accident, but odds of an accident was not a significant predictor in the third equation (b¼0.107,p¼0.263). The conclusion is that odds of an accident does not mediate the effect of experience on SLI purchase likelihood.

The mediation procedure was repeated, testing knowledge of existing insurance coverage as a possible mediator of the effect of experience on insurance purchase. Here, however, experience was not found to significantly predict knowledge of insurance coverage (b¼20.168, p¼0.092) and knowledge, in association with experience, was not a significant predictor of either CDW purchase (b¼20.060,p¼0.546) or SLI purchase (b¼20.107,p¼0.263). No evidence was found to support knowledge as a mediator of the effect of experience on likelihood of insurance purchase


Risk acculturation effect
Before leaving the Results section, a critical point must be addressed. The dramatic difference in perceived likelihood of an accident between novice and experienced renters (Table I) is referred to here as a risk acculturation effect – a lowering of the perceived likelihood of an accident due to experience. However, it could alternatively be argued that people who have rented a car before have self-selected into that group because they are adventurous and have a high tolerance for risk in their personal lives. The question is whether the difference in Table I among experience groups for odds of an accident is due to an inherent risk-tolerance difference that happens to define members of the groups or whether the difference is due to experience.

To address this issue, the means for the first question in the questionnaire (taking risk in one’s personal life) may be examined. It is seen that the means for the experience and inexperienced groups are very similar (2.97 and 2.83, respectively) and the difference between the means is non-significant (p¼0.497). Therefore, it is concluded that the experience groups do not significantly differ on risk-taking in their personal lives and that the group difference in perceived odds of an accident is not the result of self-selection into experience groups based on inherent risk tolerance.

Although results in this study appear to demonstrate a risk acculturation effect, the variable associated with this effect (odds of an accident) was not a significant predictor of insurance purchase nor did it mediate the effect of experience on insurance purchase.


Discussion
The present study supports the existence of a risk acculturation phenomenon for the consumer behavior of renting a vehicle and driving in a strange city. That is, subjects with prior rental car experience estimated the odds of an accident involving their rented vehicle to be significantly more remote than did subjects who had never rented a car before, and this difference was not associated with any difference in reported past risk behavior for the two groups. The results of the current study are in agreement with other investigations which have reported a general phenomenon of risk acculturation for a variety of activities (Halpern-Felsheret al., 2001; Morgan and Stevens, 2008; Celsi et al., 1993).

Prior rental car experience had the effect of significantly diminishing the likelihood of purchase of a collision damage waiver and supplemental liability insurance. An experience effect on insurance purchase has apparently not been demonstrated before in the literature, and so this is believed to be a contribution of this study. The effect of experience on insurance purchase was not mediated through a change in perceived odds of an accident involving the rental vehicle or knowledge of existing insurance coverage. Indeed, the results of regression analyses suggest that experience has a significant, unique, predictive ability for insurance purchase that is independent of estimated odds of an accident, magnitude of an accident, knowledge of existing coverage, or perceived control over a possible accident involving the rented vehicle. In this regard, the results of the present study are inconsistent with those of Palm (1999) who found that perceived likelihood of harm and possible severity of consequences were good predictors of earthquake insurance purchase.


Managerial implications
In the present study, prior experience in renting a car had a potent effect, exerting a strong negative influence on purchase of both types of rental insurance. The more experienced members of the sample appear to have reached the same conclusion as personal finance columnists who, in general, advise against the purchase of insurance for rental cars (Bogdanich, 1984; Solheim, 2003). This could have negative implications for car rental companies, as it suggests that their best customers (repeat renters) will be increasingly unlikely to purchase the insurance they sell.

It may be noted that 20.6 percent of the sample in this study rated themselves at the lowest level of knowledge of auto insurance coverage and whether that coverage extended to a rented vehicle (a rating of 1 on a 1 to 9 scale). This is perhaps understandable given the youthful sample for this study, but even veteran drivers appear to be confused about the extent to which a car owner’s existing insurance will cover rental cars. A survey conducted by the Progressive group of companies revealed that 25 percent of drivers purchased a collision damage waiver because they were uncertain if a rental vehicle would be covered by their current insurance (Insurance Advocate, 2004). A Consumer Reportsarticle advises renters to not assume that a rental vehicle will be covered under their personal auto insurance, noting that only 60 percent of policies cover rentals (Consumer Reports, 2000). Given the level of confusion, auto insurance companies should perhaps more clearly communicate rental car coverage to their policyholders. This could take the form of printing the existence of coverage and its limits on the insurance cards carried by policyholders in their wallets.

A final point is that rental car companies should perhaps make their insurance policies available online for potential buyers to read prior to the day of rental pick-up. Car renters are often tired and distracted on the day of rental pick-up, and making the policy available beforehand would increase transparency and probably reduce confusion regarding coverage. For example, the rental insurance contract can contain some surprising clauses. As a case in point, coverage is typically voided if you drive off-road, defined as a surface other than concrete or asphalt (Consumer Reports Travel Letter, 2000). Many car renters driving on gravel roads (common in rural areas in the USA) are probably unaware that they have no coverage while on these roads.


Limitations
The present study has several limitations. First, demonstration of a risk acculturation effect is best done through longitudinal study rather than the cross-sectional methodology of the current investigation. Second, although a sample of young adults facilitated finding subjects without experience in renting a vehicle, the findings may not generalize to the more mature general population of auto drivers. Third, data were collected on a relatively small number of independent and dependent variables; the findings that can be discovered and the conclusions to be drawn are limited by the number of variables investigated.


References
Barnett, J. and Breakwell, G.M. (2001), ‘‘Risk perception and experience; hazard personality profiles and individual differences’’,Risk Analysis, Vol. 21 No. 1, pp. 171-7.
Baron, R.M. and Kenny, D.A. (1986), ‘‘The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations’’,Journal of Personality and Social Psychology, Vol. 51 No. 6, pp. 1173-82.
Bergkvist, L. and Rossiter, J.R. (2007), ‘‘The predictive validity of multiple-item versus single-item measures of the same constructs’’,Journal of Marketing Research, Vol. 44 No. 2, pp. 175-84.
Bogdanich, W. (1984), ‘‘Purchases of flight, rental-car insurance are often unnecessary, many experts say’’,The Wall Street Journal (Eastern Edition), 30 July, p. 1.
Celsi, R.L., Rose, R.L. and Leigh, T.W. (1993), ‘‘An exploration of high-risk leisure consumption through skydiving’’,Journal of Consumer Research, Vol. 20 No. 1, pp. 1-23.
Consumer Reports(2000), ‘‘Navigating the fast turns of car-rental insurance’’,Consumer Reports Travel Letter, April, pp. 1-11.
Creyer, E.H., Ross, W.T. and Evers, D. (2003), ‘‘Risky recreation: an exploration of factors influencing the likelihood of participation and the effects of experience’’, Leisure Studies, Vol. 22 No. 3, pp. 239-53.
Finucane, M.L., Alhakami, A., Slovic, P. and Johnson, S.M. (2000), ‘‘The affect heuristic in judgment of risks and benefits’’,Journal of Behavioral Decision Making, Vol. 13 No. 1, pp. 1-17.
Gardner, D., Cummings, L., Dunham, R. and Pierce, J. (1998), ‘‘Single-item versus multiple-item measurement scales: an empirical comparison’’,Educational and Psychological Measurement, Vol. 58 No. 6, pp. 898-915.
Halpern-Felsher, B.L., Millstein, S.G., Ellen, J.M., Adler, N.E., Tschann, J.M. and Biehl, M. (2001), ‘‘The role of behavioral experience in judging risks’’, Health Psychology, Vol. 20 No. 2, pp. 120-6.
Hogarth, R.M., Portell, M. and Cuxart, A. (2007), ‘‘What risks do people perceive in everyday life? A perspective gained from the experience sampling method’’,Risk Analysis, Vol. 27 No. 6, pp. 1427-39.
Insurance Advocate (2004), ‘‘Progressive surveys discover dilemma of auto renters when faced with damage waiver option’’,Insurance Advocate,31May,p.8.
Johnson, E.J. and Tversky, A. (1983), ‘‘Affect, generalization, and the perception of risk’’,Journal of Personality and Social Psychology, Vol. 45 No. 1, pp. 20-31.
Jordan, J. and Turner, B. (2008), ‘‘The feasibility of single-item measures for organizational justice’’, Measurement in Physical Education, Vol. 12 No. 4, pp. 237-57.
Morgan, C. and Stevens, C.A. (2008), ‘‘Changes in perceptions of risk and competence among beginning scuba divers’’,Journal of Risk Research, Vol. 11 No. 8, pp. 951-66.
Palm, R. (1999), ‘‘Perceived risk and the earthquake insurance purchase decision: a commentary on a paper by Lennart Sjoberg’’,Journal of Risk Research, Vol. 2 No. 4, pp. 289-94.
Pires, G., Stanton, J. and Eckford, A. (2004), ‘‘Influences on the perceived risk of purchasing online’’, Journal of Consumer Behaviour, Vol. 4 No. 2, pp. 118-31.
Sjoberg, L. (1999), ‘‘Consequences of perceived risk: demand for mitigation’’,Journal of Risk Research, Vol. 2 No. 2, pp. 129-49.
Sjoberg, L. (2000), ‘‘Factors in risk perception’’,Risk Analysis, Vol. 20 No. 1, pp. 1-11.
Slovic, P. (1987), ‘‘Perception of risk’’,Science, Vol. 236 No. 4799, pp. 280-5.
Solheim, M.K. (2003), ‘‘Don’t waiver’’,Kiplinger’s Personal Finance, July, p. 84.
Whalen, C.K., Henker, B., O’Neil, R., Hollingshead, J., Holman, A. and Moore, B. (1994), ‘‘Optimism in children’s judgments of health and environmental risks’’,Health Psychology, Vol. 13 No. 4, pp. 319-25.



To cite this document:
Dwane H. Dean, (2010) "Rental experience and likelihood to purchase rental car insurance among young adults", Young Consumers, Vol. 11 Issue: 3, pp.215-225, doi: 10.1108/17473611011074287