Retailing to experienced and inexperienced consumers: a perceived risk approach.
As public policy makers press the case for more freedom of information and the right to be informed, retailers must also become more aware of how consumers use that information.
Consumers have many strategies to deal with the information required for purchase, but one theoretical framework which has stood the test of time is perceived risk theory. The theory allows the examination of consumer information acquisition and usage, as well as other strategies such as brand loyalty, by setting the information in a context of risk reduction, thus giving a focus and purpose for consumers' activities. This article considers the information acquisition and processing within a perceived risk framework and attempts to provide some data on this in relation to 24 products and services. Arguably, consumers with greater purchase experience will use less information and perhaps different types of information than inexperienced purchasers. That being the case, it is important for retailers to know how experienced consumers use information because it reflects the "value" of that information.
RISK REDUCTION, INFORMATION SEARCH AND PERCEIVED RISK
One of the earliest theoretical predictions about search behaviour derives from the simple utility equation: search will continue as long as the perceived value of the information exceeds the cost of obtaining this information. In other words, search, or risk reduction, equals the value of the information minus its costs |2,3,4~ and the value of the information in turn is likely to be a function of the value of the product/service to be purchased. Thus one will spend more effort on obtaining information on a high-value than a low-value product. The costs of obtaining information or using risk reducing strategies will also vary from product to product. This will also be true for the usefulness or value of the strategy or information. For example, the cost of asking someone about the purchase of a chocolate bar is likely to be less than asking someone about the purchase of a stacking hi-fi.
Ross|5, p. 49~ speculates that: "The farther along the 'learning curve' the consumer is, the less inclined he will be to seek information". Similarly, Arndt|6, p. 10~ cites literature which suggests that: "the greater the number of times a given goal-object has been bought in the past, the lower the consumer propensity to search for external information."
Jacoby et al.|7~ tested the hypothesis that the amount of information acquired will be related negatively to indices of past purchasing experience. Contrary to prediction, statistics indicating high amounts of past experience were correlated positively with research behaviour: "Consumption frequency showed an opposite tendency. Heavy users looked at few brand alternatives (p |is less than~ 0.01), but at a greater number of information dimensions (p |is less than~ 0.01)"|7, p. 539~. Despite being based on a relatively small sample of mainly female students (n = 60), the results are provocative since the methodology used was one based on behavioural processes which has numerous advantages over self-report measures.
Other studies have confirmed a negative relationship between search and product usage or purchase experience: Newmann and Staelin|8~ (appliances and cars); Claxton et al.|9~ (furniture and appliances); Moore and Lehmann|10~ (nondurable). Newmann and Staelin|11~ have found that internal search decreases with satisfactory experience with a product class and Bucklin|12~ found a slight tendency for people to shop less while purchasing non-food items when they knew all the features that they wanted. Other researchers have found no significant relationship between experience and external search|9,13~. In some studies, knowledge has been assessed instead of experience or purchase frequency. Katona and Mueller|14~ found no relationship between knowledge and the amount of search for several electrical appliances; while Beatty and Smith|15~ found that product class knowledge was negatively associated with total search effort.
The not insubstantial amount of research has produced conflicting evidence for the effect of purchase frequency or purchase experience of information search and risk reduction. An examination of the studies to compare methodologies, instruments, samples, products, dates and assumptions revealed little insight to explain the discrepancies of findings. This is primarily due to the fact that so many intervening factors could account for the conflicting findings, that in order to achieve any insight a large number of studies needs to have been conducted from which the effects or particular intervening factors can be discerned. However, one possible suggestion was forthcoming. When products are relatively low in value, or are purchased frequently, the amount of information required to make a satisfactory purchase may be very small; so small in fact that the differences in frequency of purchase or product class knowledge make very little difference to the amount of information search. If this is true, then it leads to the suggestion that for high-value products the amount of search will show a much greater variation with different levels of purchase experience. By value we are principally referring to financial value, although it is acknowledged that other factors such as product complexity and fashionability may also affect risk perception and reduction. The inherent assumption is that these factors will be reflected in the financial value placed on the product service. This threshold idea has been suggested by Gemunden|16~ to explain why the relationship between perceived risk and information search can sometimes not be found. He suggests that this is because the product being studied is below some perceived risk threshold above which search is stimulated, but below which it is not. Evidence to support the notion that as risk increases, risk reduction changes, comes from Boze|17~ who found that, as perceived risk increases in the purchase of legal services, consumers are more likely to shop around to compare alternatives and ask friends and relations for advice.
Interestingly, the effect of buy-frequency on perceived risk has received little attention (e.g. see|18~), although several studies have considered how perceived risk changes over repeated purchases. Sheth and Venkatesan|19, p. 310~ conclude, "It would appear that perceived risk with the two components of adverse consequences and uncertainty seems to increase in the beginning and then decline only over time."
The suggested explanation for their findings was that perhaps the consumer realizes the adverse consequences of a brand only after sensitizing him/herself to the brand by either prior information or actual experience. They make the analogy to a child burning his/her hand by playing with matches and only then realizing it is a dangerous alternative.
Other authors have used different measures of purchase frequency. For example, Asembri|20~ hoped to determine the impact of product holding time on consumer risk perception and risk acceptance. Asembri's study suggests that while overall risk may change, changes in the components of overall risk, e.g. financial risk, time risk, psychological risk, may be more obvious. It was therefore considered worthwhile to measure these component losses in the present study and examine the changes resulting from different buy-frequencies. They may be expected to follow the same pattern as overall risk, exhibiting no change with low-value products and some change with high-value products. The interesting observations are made when the risk-search threshold is met. Do some of the losses change more than others? For example, is financial loss more or less influenced by purchase experience than psychological loss? Answering such questions would form part of the objectives of the study.
Cunningham|21~ examined the length the product had been in use and risk perception. He found a weak relationship between the two, which suggested that the longer the product had been in use the lower the perceived risk. He also considered rate of usage, but results which suggested that the higher the perceived risk the higher the usage rate, contrary to expectation, were statistically non-significant. Since only one product, fabric softener, was used in the study, the results are not generalizable.
From the previous discussion several hypotheses can be developed:
H1: The effect of experience on the difference in information usage/risk reduction strategies will be greater, the more valuable the product.
Although Cox concluded that, "there is no direct evidence yet that high perceived risk consumers seek greater amounts of information..."|22, p. 613~, later investigators have revealed the expected significant positive relationship|23,24~. Evidence can also be found elsewhere to support the threshold notion. Sheth and Venkatesan|19~ divided a sample of 104 female undergraduate volunteers into high- and low-risk groups. In the repeated purchase of hair spray, the high-risk group engaged in more pre-purchase deliberation, spent more time collecting information, except in the first two weeks, and sought significantly more information from personal sources. On the other hand, Jacoby et al.|24~ found no significant association between perceived risk and information search for cold breakfast cereals. If we can accept that hair spray may be a riskier purchase than cold breakfast cereals, these results are supportive of the notion of a risk-search threshold.
Given Gemunden's|16~ suggestion that perceived risk remained below a "threshold" above which search/risk reduction is activated, it was considered worthwhile to control for this variable in the study. The hypothesis which could be tested is:
H2: Perceived risk will not vary significantly with purchase experience for low-value products, but will vary with high-value products.
The rationale is simple. For high-value offerings a degree of perceived risk is stimulated which is above the search threshold. As information search progresses, it acts as a risk reducer, reducing the level of perceived risk. No such mechanism, however, should operate below the risk threshold. Since we have no idea at which risk/value level this threshold exists in absolute terms, a large number of different value offerings will need to be considered, hence the 24 offerings chosen. It is likely that this threshold will vary by individual, according to the individual's ability to bear risk, and each individual will begin their search actively at a different risk threshold. In this study, we are less concerned about these individual differences and more concerned with testing the general hypothesis. As a result, we shall be losing some accuracy of measurement on two counts. First, because the products increase in value/perceived risk in a stepwise function and not a graduation, we shall only be able to measure the effect on search at each step. Those respondents whose threshold is somewhere between the two risk/value levels will be grouped together. Second, because we are averaging the results from the entire sample, there is unlikely to be a clear-cut point at which the threshold can be identified. Respondents with very low thresholds will be averaged with respondents with very high thresholds. What we should see then in the statistics is a gradual movement towards significance as more and more respondents' thresholds are reached. This movement may take place over three to five products. Nevertheless, because of the wide range of offerings, we should be able to test the hypothesis satisfactorily.
The literature regarding the effect of purchase experience or buy-frequency on perceived risk is scant and provides little guidance as to the effect of the buy-frequency, especially over a range of products. The present study set out to provide some further evidence to answer the question, what effect does buy-frequency have on perceived risk? It is reasonably logical to suppose that the greater the frequency of purchase the less risk will be perceived. One might therefore test the hypothesis that:
H3: As buy-frequency increases perceived risk will decrease
In the light of previous discussion, this hypothesis is only likely to hold true for high-value products and services which are above the "risk threshold".
An additional complication is that for some offerings it is not inconceivable that they will always have a residual amount of perceived risk which cannot be eliminated by further search or risk reduction. This suggestion leads us into the importance of assessing services in this study. Services have been shown to be riskier than products|25,26,27~ and this has been attributed to the fundamental properties of services. These properties of inseparability, heterogeneity, intangibility and perishability engender risk, but it is the variation in service provision, heterogeneity, which is of concern to us in establishing a minimum level of risk which cannot be eliminated. Thus, as consumers become more experienced with products, they can be virtually certain that their next purchase of the same product will perform as well as the last. Given that the product had performed satisfactorily and there were no adverse consequences, the product can be repurchased almost without uncertainty. For services, however, the situation is different. Inherent variability in the service means that the consumer can never be certain that the service will be performed satisfactorily. There is, therefore, likely to be a residual level of risk for services which no amount, or perhaps only very large amounts, of experience can reduce.
Aside from the residual level, heterogeneity is also likely to reduce the rate at which increasing experience is translated in decreasing risk perception, because much more experience is needed in order to gain sufficient information about how and why the service varies. In general, risk perceptions in services are likely to be reduced much more slowly given the same number of purchase experiences than in products. We would therefore expect that, for the services examined in the study, buy-frequency or purchase experience would have less of an effect on risk perception. The hypothesis can be written:
H4: Perceived risk in high-value services will exhibit less variation as a result of changes in buy-frequency than will perceived risk in high-value products.
In order to test all four hypotheses it was considered necessary to have a large range of products/services, since risk can be markedly task/product-specific|28~. If the phenomena being observed are a part of the general perceived risk theory, then it should be observable over many products. Given that there may be slight product variations, a better indication might be gleaned from examining groups of products. Such groupings also ensured that the minimum ANOVA sample size requirement was comfortably met. The six services clearly form a classification of their own. The 18 products were divided into foodstuffs, convenience goods and shopping goods.
To help facilitate comparability with other studies, a basic two-component model of risk was used, which included uncertainty about the purchase and seriousness of consequences if the purchase failed, a la Cunningham|21~. Many other studies have found this measure useful|e.g. 29-32~. The two components were measured on four-point scales and combined additively to give an overall risk measure. An additive as opposed to a multiplicative model was used because the limited evidence which exists on testing various models is slightly in favour of the additive model|33,34~. While a four-point scale may appear restrictive, measures of uncertainty and consequences on three-point scales have shown some convergent and discriminate validity|35~. Having identified the various possible loss types from the literature|36,37~, four were chosen to be rated for each product. These were financial, time, physical and psychosocial. The reasons why performance loss was not measured, and psychological and social losses were combined, are given by Mitchell and Greatorex|38~. The subjects were not naive and had been given a class on perceived risk prior to administration of the survey. Subjects were asked to rate each product or service on the four loss types using a 1-4 importance scale. Risk relievers were essentially those used by other researchers, e.g. Roselius|36~, Derbaix|39~ and Guseman|25~. Risk relievers had to be usable by all the sample and be applicable to each product. This meant that for some clearly useful product-specific risk relievers such as the BSI mark were not included in the list. The following 14 were chosen:
(1) Taking the advice of family and friends.
(2) Purchasing the same brand of the product or using the same supplier of the service that you purchased/used before.
(3) Shopping around to compare what is on offer.
(4) Reading consumer guides.
(5) Taking the advice of the sales assistant.
(6) Choosing a more expensive product/service.
(7) Reading advertising material about the product/service.
(8) Choosing a cheaper product/service.
(9) Ensuring the product/service has some form of guarantee.
(10) Favouring the products/services which are endorsed by a celebrity.
(11) Choosing a product/service which is subject to some sales promotion.
(12) Using the image of the product/service as a guide.
(13) Choosing brand/supplier of the product/service which is well-known or popular.
(14) Trying the product/service before purchase.
The strategies were rated for each of the 24 products/services on a 1-5 usefulness scale. Subjects were given definitions of some of the terms used in the questions, e.g. consumer guides, advertising material, guarantee, image. In order to make strategies more meaningful, the descriptions used were adapted to fit the offering without losing the essence of the strategy. For example, reading advertising material about the service of a hotel was written as reading the hotel's brochure. Questionnaire assessment of attitudes to search rather than behaviour was considered acceptable, since the literature shows a strong positive relationship between attitudes towards search and actual search behaviour|40,41,42~.
The 24 products/services used in the study were chosen to represent a range of values, buy frequencies and riskiness. The Appendix shows the ranges for the buy-frequencies and risk ratings for each product. The most frequently purchased item was chocolate, being bought on average 134.95 times a year. The least frequently purchased items were jointly stacking system and portable TV, being purchased on average only 0.21 times per year (or approximately once every five years). The offerings also had to be within the experience of the sample population and had to be purchasable by both sexes. The categorizing of purchasers into high-, medium- and low-frequency purchasers was done on a product-by-product basis, bearing in mind the sample size requirements for successful ANOVA application. Where feasible, care was taken to ensure each grouping contained a sample of at least 30. Some of the 24 frequency distributions, for example those of fast-food restaurant and casual shirt/blouse, were approximately normal. For these, the modal frequency values were assigned to the medium-frequency purchaser group, and two remaining distribution portions forming the low- and high-frequency purchaser groups. It was considered acceptable to divide the frequency distribution simply into approximate thirds only for products/services with linear buyer frequency value increments. Attention was also paid to the increment steps in buy-frequency behaviour. Some products, tinned pilchards and sausages, for example, exhibited natural breaks in buy-frequency, and were grouped accordingly. For eight of the products/services, for example jeans, instant coffee, batteries, no medium frequency purchaser group existed.
The undisguised questionnaire was administered to a convenience sample of 100 undergraduate students aged between 19 and 21, with approximately equal number of males and females. To reduce variation in interpretation, subjects were given the definitions of losses, products and some relievers to read before answering the questionnaire. Because of the large data capture requirements, five booklets each containing six products were given at weekly intervals to the sample. As well as reducing respondent fatigue, it allowed a test re-test measure of reliability. A two-week interval was chosen|43~ and six products were rated on two occasions. Reliability coefficients for the two administrations were mainly above 0.65 which gives some indication of the instrument's reliability. The necessity for repeated access to a population which could be informed of perceived risk and who were intelligent enough to cope with various definitions and an undisguised questionnaire were the main reasons for choosing a student sample. A student sample also had the benefit of affording maximum comparability with other studies of perceived risk which have also used students (e.g.|7,28,37,44-47~). In addition, we had no reason to suspect that the key hypotheses being tested were sample-specific. Given these considerations, the cost, speed and convenience advantages of the population seemed to make the choice of student sample acceptable. However, the generalizability of the results to other populations is limited, particularly for products which are clearly dependent upon income.
RESULTS AND DISCUSSION
H1: The effect of experience on the difference in information usage/risk reduction strategies will be greater the more valuable the product.
There are difficulties involved in ranking all the products in order of value; a necessary prerequisite for the consideration of hypothesis one. Offerings were originally chosen because they provided a range of values across each product classification. Simply ordering the individual items within each product category would yield overall value-rank inconsistencies, e.g. matches would be deemed more valuable than a bottle of wine. Therefore, products were grouped into a low-value classification, comprising of foodstuffs and convenience goods, and a high-value classification, consisting of shopping goods.
Both classifications were split into low, medium and high frequency purchasers and the mean risk relieving score (i.e. an aggregate of all 14 individual risk relievers), in addition to the individual risk reliever scores, was computed. Summary results are displayed in Table I.
In order to accept the first hypothesis we expect the probability of significance (p) to decrease as product classification value increases. This expectation is borne out when considering the aggregated RRS averages; however, in examining all 14 risk-reducing strategies individually, it appears that on many occasions their usage does not support acceptance of the TABULAR DATA OMITTED hypothesis. In fact, in six out of 14 strategies greater variation was seen for low-value products. Why should this be the case? The explanation could have something to do with the choice of products which are somehow atypical of low-value or high-value products. Although possible, this is unlikely. More likely is the differential effect experience has on various risk-reducing strategies. Clearly, if such findings were repeated, there would seem to be the suggestion that some RRS change in their usefulness and others simply do not. This explanation presupposes that we are comparing like with like, i.e. that high-frequency purchases of hi-fis are somehow equivalent to high-frequency purchases of matches (a low-value product). While the data for each individual product allowed division into high-, medium- and low-frequency purchasers, the different nature of the products concerned allowed little comparability between products in any absolute sense. As a result of some of the high-value products having very low purchase frequencies, the high-frequency purchase group may not have been very high at all and perhaps not high enough to notice the changes we were expecting. For example, stacking hi-fi is bought on average 0.21 times a year and the maximum number of times for the sample is five per year, whereas a bottle of wine is purchased on average 35 times a year and has a maximum of 208 times per annum. The greater the difference in buy-frequency the greater will be its effect on risk perception and reduction. Although low-value products have lower risk which should subdue the effects of buy-frequency, they have a very large range of buy-frequencies which acts to enhance the effect of buy-frequency. Changes in risk perception and reduction for high-value products should be greater because they are above the risk threshold, but they may be less than is seen in some low-value products simply because the change in buy-frequency has not been sufficient. This latter explanation appears more likely and is the one preferred by the authors.
TABLE I. Average Risk-relieving Strategy Usage for Low-, Medium- and High-frequency Purchasers for Low- and High-value Products Aggregate risk reliever average(a) p L M H Low-value products 0.072 2.72 2.62 2.73 High-value products 0.018 3.33 3.21 3.37 p = probability of significance L = low-frequency purchasers M = medium-frequency purchasers H = high-frequency purchasers Note: a 1 = not very useful 5 = very useful
From Table II we can see that buy-frequency does not have a consistent effect. Some strategies increase in usefulness, e.g. prepurchase trial for high-value goods; some strategies decrease in usefulness, e.g. more expensive choice for high-value products; and some strategies decrease and increase in usefulness, e.g. advice of family and friends for low-value products. Since no consistent pattern can be discerned, we can suggest each individual buy-frequency group must be considered separately.
In keeping with our ideas that high-frequency purchasers will use different strategies from low-frequency purchasers, we can suggest that as purchase frequency increases consumers may begin to give more emphasis to some strategies than others. For example, low-frequency purchasers might prefer strategies A, B and C; medium-frequency purchasers C, D and E; and high-frequency purchasers F, G and H. There may be some strategies which show an increasing usefulness across the three purchase-group types, for example: endorsement by a celebrity and special offers (low-value products); brand loyalty and prepurchase product trial (high-value products). Other strategies change their usefulness status depending on the purchase frequency status of the purchasers and the product. For example, in the case of low-value products, reading advertising material is rated least useful by medium-frequency purchasers and rated most useful by low-frequency buyers. Conversely, brand loyalty is judged to be most useful to medium-frequency buyers and least useful to low-frequency purchasers. This individual group perspective must be taken, since the evidence suggests that there is no consistent effect of buy-frequency on risk reducing strategies.
H2: Perceived risk will not vary significantly with purchase experience for low-value products, but will vary with high-value products.
Analysis of the hypotheses was initially carried out on a product-by-product basis. In the case of the second hypothesis, discrepancies in p-values from low-value products such as tinned pilchards or batteries (p=0.967 and p=0.983, respectively) to high-value products such as coat (p=0.194) were observed. However, none of these probabilities was significant at the 0.05 level and hence the products were grouped in order to exaggerate differences. Analysing the second hypothesis required the data from the three product classifications of, in ascending value order, food items (comprising chocolate, fresh apples, wine, sausages, instant coffee and tinned pilchards), convenience goods (matches, shampoo, deodorant, socks and batteries) and shopping goods (casual shirt, jeans, coat, track suit, tennis racquet, portable TV and stacking system). Table III (ignoring the service row data) shows the product classifications and average risk scores for low, medium and high buy-frequency purchasers.
Initially, considering overall perceived risk, we observe little difference in low and high buy-frequency groups for either food items or convenience goods (p=0.676, p=0.654, respectively), yet for shopping goods the two buy-frequency groups experience less risk than low buy-frequency groups. Clearly, therefore, the threshold referred to earlier may lie at a product value somewhere between convenience goods and shopping goods. Examination of overall risk differentials between low and high frequency groups on an individual product basis may help us ascertain this threshold value more precisely.
However, the hypothesized threshold where changes in buyer experience would begin to have a significant effect on risk perception was not as distinct as expected. Although at the product classification level the p-values and risk changes observed fitted with our hypothesized direction, at the individual product level there was no distinct threshold above which changes were significant and below which they were not. Given that changes were seen at the product classification level, it could be that the specific sample of products chosen were to blame for a less discernible effect at the individual product level. Second, the reliance on only one measure of risk leaves no scope for the reduction of measurement error inherent within any rating scale. This will contaminate results to some degree. A more comprehensive multivariate measure would have helped to overcome this problem and facilitated the assessment of the validity of this reason. Third, the subjectivity present in the division of the buyers into low, medium and high frequency groups may confound the accurate measurement of the phenomenon. Fourth, respondents were often imprecise in their statement of buy-frequencies. Generalizing to reference points of weekly, monthly and annual purchases may be an additional contaminator. Fifth, on an individual product basis, although every effort to the contrary was made, some frequency groupings consisted of an inadequate sample size for misleading findings to be avoided. Finally, no TABULAR DATA OMITTED distinct threshold could be observed, perhaps because no distinct threshold actually exists. This possibility cannot be discounted, although the failure to discover it on this occasion, given the above observations, does not precipitate the conclusion that it does not exist at all.
Considering the four components of overall perceived risk, the above trend is largely supported, i.e. perceived risk varies little between low- and high-frequency purchaser groups for low-value goods, but varies more for high-value goods. However, for food items there is a statistically significant difference in financial risk between low- and medium-frequency groupings. Why should this be the case? One possible explanation is that food items are characterized by a large range of buy-frequencies (from 104 to 900 purchases per year) and so, although these low-value products may be perceived as only very slightly more risky for low-frequency purchasers as opposed to high-frequency purchasers, when the difference in frequency is very great, i.e. 804 times p.a., significant risk discrepancies may result even for low-value goods. Although significant variations are also seen for higher risk products, these variations occur with far smaller changes in buy-frequency.
For convenience goods, there is a significant difference in physical risk between low- and medium- and high- and medium-frequency groupings. This is interesting and, notwithstanding possible measurement errors, the authors are unable to find a satisfactory explanation for the results. These two slightly anomalous results, however, do help in answering some of the questions posed earlier; namely, do losses change independently when the risk threshold is passed? Clearly the answer is "yes". The effect of buy-frequency on convenience goods is significantly greater for financial and physical loss. The significance of this finding is as yet unclear, but further investigation in this area may prove more illuminating.
When considering shopping goods and services variations in risk perceptions are much more significant. The only loss to be slightly out of line with expectation is physical loss. This low physical loss significance (p = 0.470) for shopping goods could be attributed to the safe nature of these products, whose high value is partially a function of stringent quality tests and regulations. Despite minor exceptions, there is strong foundation for accepting the hypothesis that perceived risk will not vary significantly with purchase experience for low-value products, but will vary with high-value products.
H3: As buy-frequency increases perceived risk will decrease.
When evaluating hypothesis three, all four item classifications were considered. From Table III we see that for three classifications overall perceived risk decreases from the low buy-frequency group to the high buy-frequency group, and remains constant in the case of convenience goods. The third hypothesis is hence supported for overall perceived risk. The theoretical relationship between perceived risk and buy-frequency has been confirmed by these empirical findings. As anticipated, perceived risk decreases as buy-frequency increases. However, when we consider the individual losses there are several results which contradict the hypothesized trend. For food items, notice that in the medium buy-frequency group, overall risk is slightly less than for the high buy-frequency group. An examination of the multiple ranges Scheffe test found, however, no significant difference between medium- and high-frequency purchasers at the 0.05 level, and hence no grounds to reject the original hypothesis. Although statistically insignificant, the low medium-frequency buyer risk ratings are nonetheless curious. This observation is particularly salient for shopping goods where frequent purchasers of shopping goods such as coat, tennis racquet, track suit, will have more psychosocial loss and physical loss. High-frequency purchasers of such items are likely to be very fashion-conscious and buy more through compliance with changing trends than through necessity. Hence high-frequency purchasing of these items may be characterized by greater psychological perceived risk than for low-frequency purchasing. High psychological risk may sensitize the consumer to other risk factors and hence distort the overall risk rating away from theoretical predictions.
H4: Perceived risk in high-value services will exhibit less variation as a result of changes in buy-frequency than will perceived risk in high-value products.
The evidence supporting the acceptance of the fourth hypothesis is strong. Products/services were grouped into high-value products (composed of convenience goods and shopping goods) and high-value services (composed of hotel and meal in a restaurant). Although the variation in overall perceived risk for high-value services was not insubstantial (p = 0.203), for high-value products the differential was pronounced (p = 0.000). For the components of risk, only in the case of physical risk is the variation in risk for changes in buy-frequency less for high-value products than for high-value services. A possible explanation for this was cited when discussing the second hypothesis.
This article has highlighted a need to understand consumers' choice processes with a view to including this type of data in any changes which retailers make in order to assist consumers when TABULAR DATA OMITTED purchasing. Buy-frequency does not have a consistent effect on risk-reducing strategies -- some increase in usefulness and some decrease with increasing frequency of purchase -- but this effect is greater with high-value products.
One practical application of the results is in the transference of knowledge from high-frequency to low-frequency purchasers. Through trial and error, high-frequency purchasers will have amassed knowledge and expertise in purchasing different products. They will be in a better position to decide on the usefulness of various risk relievers than will low-frequency purchasers. Knowing which strategies are most useful for high-frequency purchasers might allow retailers to tailor their marketing mixes more precisely. For example, the top three risk-relieving strategies of low-frequency users were brand loyalty, choosing a more expensive product and the advice of the sales assistant (low-value products); choosing a more expensive brand, prepurchase product trial and the advice of family/friends (high-value products); and, for high-frequency users, brand loyalty, choosing a more expensive product and the advice of the sales assistant (low-value products); prepurchase product trial, choosing a more expensive product and choosing a well-known product (high-value products).
Given this information, companies may wish to use it to change their communication mixes if usage segmentation has been shown to be useful in their market. Knowing which strategies are most useful to high- or low-frequency segments could be valuable in tailoring the communication mix to appeal to a particular usage segment. Similarly, retailers may wish to know not only which strategies and considered most useful, but also those which are viewed as least useful. Resources can then be withdrawn from inefficient mediums. For example, the bottom three strategies for high-frequency purchases are shopping around, special offers and celebrity endorsements (low-value products); celebrity endorsements, special offers and product image (high-value products).
Allied to the point of examining how high-frequency purchasers differ in the possibility of using the information to target them as opinion leaders. For some goods where information search costs are high, high-frequency purchasers would conceivably be acting as opinion leaders. By exploring the strategies they employ it may be possible for retailers to use the results to influence their attitudes and perceptions.
There is one problem with using data such as this to redesign communication mix strategies. That a strategy fares badly as a risk reducer could reflect: its inherent weakness; poor implementation of the strategy; or under-utilization of the strategy. The prescriptive nature of any recommendations based on the data can also be criticized for not encouraging the innovation of novel methods of reducing risk.
The results presented under the second hypothesis also have certain practical implications. The fact that buyer experience has little or no effect on risk perception for low-value products should alert consumers and retailers to the importance of risk-reducing strategies which are not experientially based. Even low-value products can have some perceived risk associated with their purchase and, if it cannot be reduced by experience with the product, then this highlights the importance of using other strategies. This is particularly disappointing for retailers of low-value fast-moving consumer goods because purchase experience is something which they have to their advantage when it comes to helping consumers in reducing the perceived risks, since most of the low-value items have relatively short repurchase cycles. It is ironic then that this "naturally occurring" risk reducer should have little or no effect. It has the greatest effect in the purchase of high-value goods. These are bought infrequently and hence the strategy is less useful in assisting the consumer in the pursuit of risk reduction.
As expected, purchase frequency or buyer experience is a much less useful way of reducing risk for services than for products because of the heterogeneity involved in producing and consuming services. Service retailers therefore need to be more aware than their product retailing counterparts about the need to develop risk-reducing strategies which are not experientially based.
The results also indicate that, for high-value products and services, buy-frequency cannot be ignored as an intervening variable when assessing purchase risk. Implications for questionnaire design, future attempts to measure risk and risk reduction, and to interpret any data collected, appear to be that the number of times a consumer has purchased an item before can significantly affect their perceptions and attitudes. Further research might care to use several of the intervening variables, e.g. age, recency of purchase, etc., in a multifactorial experiment in order to gauge their relative importance and the interaction effects.
1. Day, G.S., "Assessing the Effects of Information Disclosure Requirements", Journal of Marketing, Vol. 40, April 1976, pp. 42-52.
2. Engel, J.F., Kollat, D.T. and Blackwell, R.D., Consumer Behaviour, 2nd ed., Holt Rheinhart & Winston, New York, NY, 1973.
3. Lanzetta, J.T., "Information Acquisition in Decision Making", in Harvey O.J. (Ed.), Motivation and Social Interaction: Cognitive Determinants, Ronald Press, New York, NY, 1963, pp. 239-65.
4. Lanzetta, J.T. and Kanareff, VT., "Information Cost, Amount of Payoff and Level of Aspiration as Determinants of Information Seeking in Decision Making", Behavioural Science, Vol. 7, 1962, pp. 459-73.
5. Ross, I., "Applications of Consumer Information to Public Decisions", in Sheth J.N. and Wright, P.L. (Eds), Marketing Analysis for Societal Problems, University of Illinois, Urbana-Champaign, IL, 1974, pp. 42-76.
6. Arndt, J., Consumer Search Behaviour, University Press, Oslo, Norway, 1972.
7. Jacoby, J., Chestnut, R.W. and Fisher, W.A., "A Behavioural Process Approach to Information Acquisition in Non-durable Purchasing", Journal of Marketing Research, Vol. 15 No. 3, 1978, pp. 532-44.
8. Newmann, J.W. and Staelin, R., "Prepurchase Information Seeking for New Cars and Major Household Appliances", Journal of Marketing Research, Vol. 9, 1972, pp. 249-57.
9. Claxton, J.O., Fry, J.N. and Portis, B., "A Taxonomy of Prepurchase Information-gathering Patterns", Journal of Consumer Research, Vol. 1, December 1974, pp. 35-42.
10. Moore, W.L. and Lehmann, D.R., "Individual Differences in Search Behaviour for a Non-durable", Journal of Consumer Research, Vol. 7, 1980, pp. 296-307.
11. Newmann, J.W. and Staelin, R., "Multivariate Analysis of Differences in Buyer Decision Time", Journal of Marketing Research, Vol. 8, 1971, pp. 192-8.
12. Bucklin, L.P., "Testing Propensities to Shop", Journal of Marketing, Vol. 30, 1966, pp. 22-7.
13. Bennett, P.C. and Mandell, R.M., "Prepurchase Information Seeking Behaviour of New Car Purchasers -- The Learning Hypothesis", Journal of Marketing Research, Vol. 6 November 1969, pp. 430-33.
14. Katona, G. and Mueller, E., "A Study of Purchase Decisions", in Clark, L.H. (Ed.), Consumer Behaviour: The Dynamics of Consumer Reaction, New York University Press, New York, NY, Vol. 1, 1955, pp. 30-87.
15. Beatty, S.E. and Smith, S.M., "External Search Effort: An Investigation across Several Product Categories", Journal of Consumer Research, Vol. 14, June 1987, pp. 83-95.
16. Gemunden, H.G., "Perceived Risk and Information Search. A Systematic Meta-Analysis of the Empirical Evidence", International Journal of Research in Marketing, Vol. 2 No. 2, 1985, pp. 79-100.
17. Boze, B.V., "Selection of Legal Services: An Investigation of Perceived Risk", Journal of Professional Services Marketing, Vol. 3 No. 2, 1987, pp. 287-97.
18. Kogan, N. and Wallach, M., Risk-Taking; A Study in Cognition and Personality, Holt, Rhinehart & Winston, New York, NY, 1964.
19. Sheth, J.N. and Venkatesan, M., "Risk Reduction Processes in Repetitive Consumer Behaviour", Journal of Marketing Research, Vol. 5, August 1968, pp. 307-10.
20. Asembri, C.A., "The Effects of Consumers' Planned Products Holding Time on Risk Perception and Acceptability", unpublished PhD thesis, City University of New York, NY, 1986.
21. Cunningham, S.M., "The Major Dimensions of Perceived Risk", in Cox, D.F. (Ed.), Risk Taking and Information Handling in Consumer Behaviour, Graduate School of Business Administration, Harvard University Press, Boston, MA, 1967, pp. 82-108.
22. Cox, D.F., "Risk Taking and Information Handling in Consumer Behaviour", in Cox, D.F. (Ed.), Risk Taking and Information Handling in Consumer Behaviour, Graduate School of Business Administration, Harvard University Press, Boston, MA, 1967, pp. 604-40.
23. Swan, J.E., "Search Behaviour Related to Expectations concerning Brand Performance", Journal of Applied Psychology, Vol. 56, August 1972, pp. 332-5.
24. Weigl, K.C., "Perceived Risk and Information Search in a Gift-buying Situation", Dissertation, Purdue University, 1975.
25. Guseman, D.S., "Risk Perception and Risk Reduction in Consumer Services", in Donelly J.H. and George W.R. (Eds), Proceedings of American Marketing Association, 1981, pp. 200-4.
26. Lewis, W.F., "An Empirical Investigation of the Conceptual Relationship between Services and Products in Terms of Perceived Risk", PhD dissertation, University of Cincinnati, OH, 1976.
27. Mitchell, V-W. and Greatorex, M., "Risk Perception and Reduction in the Purchase of Consumer Services", The Service Industries Journal, Vol. 13, No. 4, October 1993, pp. 179-200.
28. Slovic, P., "Convergent Validation of Risk-taking Measures", Journal of Abnormal and Social Psychology, Vol. 65, 1962, pp. 68-71.
29. Dash, J.F., Schiffman, L.G. and Berenson, C., "Risk and Personality-related Dimensions of Store Choice", Journal of Marketing, Vol. 40, 1976, pp. 32-9.
30. Yavas, U. and Tuncalp, S., "Saudia Arabia: Perceived Risk in Buying 'Made in Germany' Label", Management International Review, Vol. 25 No. 4, 1985, pp. 58-65.
31. Toh, R. and Heeren, S.C., "Perceived Risk in Generic Grocery Products and Risk Reduction Strategies of Consumers", Akron Business & Economic Review, Vol. 13, 1982, pp. 43-8.
32. Hoover, R.J., Green, R.T. and Saegert, J., "A Cross-national Study of Perceived Risk", Journal of Marketing, Vol. 42 No. 3, 1978, pp. 102-08.
33. Bettman, J.R., "Perceived Risk and Its Components: A Model and Empirical Test", Journal of Marketing Research, Vol. 10, May 1973, pp. 184-90.
34. Horton, R.L., "The Structure of Decision Risk: Some Further Progress", Journal of the Academy of Marketing Science, Vol. 4 No. 4, 1976, pp. 694-706.
35. Lumpkin, J.R. and Massey, P.K., "Convergent and Discriminant Validity of Alternative Perceived Risk Scales", Proceedings of the Southern American Marketing Association, 1983, pp. 153-67.
36. Roselius, T., "Consumer Rankings of Risk Reduction Methods", Journal of Marketing, Vol. 35, 1971, pp. 56-61.
37. Kaplan, L.B., Szybillo, G.J. and Jacoby, J., "Components of Perceived Risk in Product Purchase: A Cross-validation", Journal of Applied Psychology, Vol. 59 No. 3, 1974, pp. 287-91.
38. Mitchell, V-W. and Greatorex, M., "Measuring Perceived Risk and its Components across Product Categories", Proceedings of the 19th European Marketing Academy Conference, Innsbruck, 1990, pp. 153-67.
39. Derbaix, C., "Perceived Risk and Risk Relievers: An Empirical Investigation", Journal of Economic Psychology, Vol. 3, 1983, pp. 19-38.
40. Duncan, C.P. and Olshavsky, R.W., "External Search: The Role of Consumer Beliefs", Journal of Marketing Research, Vol. 19, February 1982, pp. 332-43.
41. Kiel, G.C. and Layton R.A., "Dimension of Consumer Information Seeking", Journal of Marketing Research, Vol. 18, May 1981, pp. 233-9.
42. Punj, G.N. and Staelin, R., "A Model of Consumer Information Search Behaviour for New Automobiles", Journal of Consumer Research, Vol. 9, March 1983, pp. 336-80.
43. Peter, J.P., "Construct Validity: A Review of Basic Issues and Marketing Practices", Journal of Marketing Research, Vol. 18, 1981. pp. 133-45.
44. Perry, M. and Hamm, C.B., "Canonical Analysis of Relations between Socioeconomic Risk and Personal Influence in Purchase Decisions", Journal of Marketing Research, Vol. 6, 1969, pp. 351-4.
45. Folkes, V.S., "The Availability Heuristic and Perceived Risk", Journal of Consumer Research, Vol. 15 No. 1, 1988, pp. 13-23.
46. Kahn, B.E. and Sarin, R.K., "Modelling Ambiguity in Decisions under Uncertainty", Journal of Consumer Research, Vol. 15 No. 2, 1988, pp. 265-72.
47. Lantos, G.P., "The Influences of Inherent Risk and Information Acquisition on Consumer Risk Reduction Strategies", Journal of the Academy of Marketing Science, Vol. 11 No. 4, 1983, pp. 358-81.
TABULAR DATA OMITTED
|Printer friendly Cite/link Email Feedback|
|Author:||Mitchell, V-W.; Prince, G.S.|
|Publication:||International Journal of Retail & Distribution Management|
|Date:||Sep 1, 1993|
|Previous Article:||Complaint behaviour, price paid and the store patronized.|
|Next Article:||Male grocery shoppers' attitudes and demographics.|