358 SAJEMS NS Vol 2 (1999) No 3 Risk Relievers in Mail-Order Catalogue Retailing: The Case of the High-Frequency Purchaser of Low-Involvement Products N S Terblanche Department of Business Management, University of Stellenbosch C Boshoff Department of Business Management, University of Port Elizabeth E vdMSmit Graduate School of Business, University of Stellenbosch ABSTRACT This article reports on the identification of how various risk relievers, available to consumers and mail-order catalogue organisations, influence general risk perception. Consumers who have purchased low-involvement products on a number of occasions by mail-order constitute the sample. The findings are that two of the major risk relievers significantly reduce, whilst another one increases, the general risk perceptions of mail-order customers when buying low- involvement goods. These findings are important to mail-order catalogue managers because they confirm that a variety of risk relievers need to be offered to ensure that perceived risks are properly addressed. JELM 31 INTRODUCTION Three reasons are put forward as to why shops traditionally served as primary distributors of retail products (Rosenberg & Hirschman, 1980), namely: customers were accustomed to purchasing at shops; few acceptable alternatives existed and the value of consumers' money exceeded the value of the time used for shopping. Retailers overcome a number of discrepancies for consumers. Typical of these are time, spatial, assortment and information gaps. Recent years have however witnessed a large increase in the volume of retailing transactions done at R ep ro du ce d by S ab in et G at ew ay u nd er li ce nc e gr an te d by th e P ub lis he r (d at ed 2 00 9) . SAJEMS NS Vol 2 (1999) No 3 359 shopping fonnats other than the conventional retail shops that overcome the above-mentioned discrepancies. The past two decades there has been an increase in the use of multiple channels for the distribution of products by manufacturers and intennediaries. McNair and May (1978) mention the role of technology in this context and argue that new fonns and applications of technology make mail- catalogue buying an acceptable shopping fonnat for an increasing number of consumers. Various previous studies have found that in-shop shopping is perceived as having less risks than non-shop or in-home shopping (Spence, Engel and Blackwell, 1970; Festervand, Snyder and Tsalikis, 1986; Hawes & Lumpkin, 1986; McCorkle, t 990). A risk reliever can be defined as a device or action that is initiated by a buyer or a seller, which is used to carry out a risk reduction strategy. A number of risk relievers such as infonnation seeking, store image and major brand image that are freely available to in-shop consumers, are however not available to mail-catalogue consumers. A consumer might use a specific risk reliever as a way to get a higher probability of purchase success or rely on another as a means of minimising the loss incurred in case of product failure. This article deals with how consumers who have purchased low- involvement products, perceive various risk relievers available to them. OBJECTIVES The primary objective of this study was to identify which risk relievers, that are available to consumers and mail-catalogue organisations, actually reduce observed general risk perceptions of consumers who have purchased low- involvement products in the past by mail-order. Insight into the impact of such risk relievers is useful to the design of product offerings and mail-order catalogue design and should, hopefully, enhance the sales volumes of direct marketers who can convert these findings into a lower risk associated with buying a product by mail-order. THEORETICAL CONSIDERATIONS Before discussing the risk relievers that are available to consumers and mail- catalogue organisations, it is necessary to consider the market in which maiJ- catalogue marketing takes place. R ep ro du ce d by S ab in et G at ew ay u nd er li ce nc e gr an te d by th e P ub lis he r (d at ed 2 00 9) . 360 SAJEMS NS Vol 2 (1999) No 3 The market for mail-catalogues Rapp and Collins (1990) identified the following trends in and characteristics of the marketplace that favour direct-order retailing : target markets that change continually because of changing demographics and lifestyles; demands on personal time; overcrowding by too many new products, services and stores; weakening of the impact of television advertising; decline in brand and store loyalty; and clutter, overkill and waste of advertising. The net effect of these trends is that it has become increasingly difficult to capture and keep the attention of the consumer. One way in which retailers succeed in achieving this, is to contact the consumer directly. Shopping in retail outlets will undoubtedly remain a vital social as well as a functional activity for a long time to come. There are, however, certain social and economic forces that make shopping at home attractive. Some of these forces are: the annoyance and wastefulness of having to contend with traffic and shopping crowds; the widely noted deterioration in the quality of service in many retail shops; an increase in the number of career and professional women; a greater emphasis on standardisation and branding of products (which reduces the risk involved in shopping at home); and the growing use of credit cards (Rapp & Collins, 1990; Darian, 1987). The value of the consumer's time has also increased tremendously over the past decade and any means that can give a consumer more shopping time flexibility is viewed as positive by consumers. Direct mail-order retailing thus offers consumers flexible shopping hours. Rosenberg and Hirschman (1980) also identified the willingness of consumers to change, and their acceptance of technology used to market products directly, as further reasons that will make shopping in the home attractive. Risk relievers in mail-order catalogue retailing Various perceived risks are attached to mail-order catalogue retailing, and the consumer selects whichever risk reliever appears best suited to the type of risk that is involved. Akaah and Korgaonkar (1988) undertook a conjoint analysis to investigate consumer preferences for risk relievers in direct order retailing. They found that a money-back guarantee ranked as the most important risk reliever. This was followed by the name of the manufacturer, the cost of the product, the reputation of the distributor, free sample/trial, endorsement by a trusted person, experience of the brand and the novelty of the product, respectively. It was also found that the lower the relative cost of an offering is, the greater will be the incentive to shop for it by means of direct-order retailing. It was reported by Hawes and Lumpkin (1986) that price/quality perception, personal experience, as well as R ep ro du ce d by S ab in et G at ew ay u nd er li ce nc e gr an te d by th e P ub lis he r (d at ed 2 00 9) . SAJEMS NS Vol 2 (1999) No 3 361 money-back guarantees are the most important risk reduction techniques used by in-home shopping respondents. Festervand, Snyder and Tsalikis (1986) and Jasper and Ouelette (1994) found that catalogue buyers with prior satisfactory purchase experience, perceived significantly less risk than catalogue buyers with prior unsatisfactory purchasing experience. The risk relievers available to consumers in direct order retailing are (Roselius, 1971; Hoover, Green and Saegert, 1978; Derbaix, 1983; Hawes & Lumpkin, 1986; Shimp & Bearden, 1986; Akaah & Korgaonkar, 1988; McCorkle, 1990; Jasper & Ouelette, 1994): money-back or other guarantees - typically a refund if the customer is not satisfied or replacement if the product does not function properly; endorsement by an expert or a public figure - designed to create confidence; samples a free miniature version of the product or a part of it is given to the customer; testing by an independent private institution the fact that a product was tested and found suitable by a respected laboratory or other institution serves to relieve risks; testing by a government institution similar to the above except that the testing institution is now part of or owned by government, for example the SABS or the CSIR; purchasing a well-known brand name - here the consumer relies on the known reputation of the brand to serve as a guide; information search the process by which a consumer consults family, friends or colleagues to gather information to make an informed decision; brand loyalty - a consumer buys only a specific brand (trade mark) that has proved satisfactory in the past; purchasing the most expensive model - this reliever works on the price/quality relationship to the effect that the more expensive a product is, the higher its quality should be and R ep ro du ce d by S ab in et G at ew ay u nd er li ce nc e gr an te d by th e P ub lis he r (d at ed 2 00 9) . 362 SAJEMS NS Vol 2 (1999) No 3 dealing only with big and reputable business finns - in the absence of other infonnation, the consumer relies on organisations known for their dependable image. The first six risk relievers above are initiated by the direct mail retailer, whilst the other four would be customer-initiated. All the above-mentioned risk relievers are included in this study as exogenous latent variables. It can be expected that an increase in or presence of risk relievers, would establish a positive relationship between the risk reliever and direct-order purchasing. In other words, the more effectively the risk relievers are used, the less the general risk perceptions of mail-order customers would be. SAMPLE AND DATA COLLECTION The population of the study was 57 823 customers of a South African mail-order organisation who had bought more than twice from the organisation in the 18 months preceding the study. A sample of2 500 respondents were drawn from the population. The figure of 2 500 respondents was based on an expected response rate of 20%. A response rate of 20% would have resulted in the return of 500 questionnaires. This was based on a preferred ratio of 15 respondents per item as a nonn which would have required 495 completed questionnaires to be returned to meet the desired cutoff point. It must, however, be pointed out that a 5 to I ratio is regarded as the minimum whilst a ratio of 10 to 1 is the more acceptable (Hair, Anderson, Tatham and Black, 1998: 98-9). The questionnaires were mailed to the respondents and a total of 422 questionnaires were returned, giving a response rate of 16,88%. Because of the relatively low response rate, it was necessary to estimate for non-response bias. Because the data were captured in the same sequence in which the questionnaires were received, it was possible to use the method called time trends extrapolation as suggested by Atmstrong and Overton (1977), to estimate for non-response bias. The assumption underlying this method is that the fourth quartile is the same as the non- respondents. It was therefore necessary to detennine whether the demographic characteristics of the fourth quartile differ from those of the first quartile. If no differences are found, quartiles one and four as well as the non-response are regarded as similar. The demographic characteristics of the first and fourth quartile were analysed to test for significant differences. A Chi-square goodness of fit test did not reveal any significant differences between the demographic variables of the first and the fourth quartiles. Table 1 contains the comparison of the first and fourth quartiles of the sample. R ep ro du ce d by S ab in et G at ew ay u nd er li ce nc e gr an te d by th e P ub lis he r (d at ed 2 00 9) . SAJEMS NS Vol 2 (1999) No 3 363 Table 1 Comparison of first and fourtb quartiles of tbe sample with regard to demographic characteristics by using a chi-square test Demographic Critical X2 df value Conclusion variable (001 level) Household size 2,62 4 13,28 No signific:ant difference ~respondent 3,81 4 13,28 No significant difference hold income 6,59 5 1509 No significant difference Cars in household 625 4 13,28 No significant difference Time spent on occu- pation/profession 0,45 3 11,36 No significant difference Education level 2,32 4 13,28 No significant difference Proximity of shops 0,93 1 6,64 No significant difference Physical disability 034 I 6,64 No significant difference Children 1,99 1 6,64 No significant difference Pre-school children 126 2 9,21 Nosignificant difference r------- Time spent on community/ welfare activities 3,04 4 13,28 No significant difference Shift work 264 1 664 No significant difference Time spent on fitness/sport 0,32 4 13,28 No significant difference Table I shows that when the first and fourth quartiles of the realised sample are compared in demographic terms, the two groups do not significantly differ from each other in any respect. As Armstrong and Overton (1977) argue that non- respondents are similar to the respondents of the fourth quartile, it may therefore be concluded that non-response in this study has minimal impact on the representativeness of the sample. The mail-catalogue organisation does not specialise in any particular merchandise, but offers a wide variety of products in 22 product categories, all of which may be characterised as low-involvement products. The low- involvement products referred to here have the usual characteristics of low price, little social concern and requiring very limited buying decision making. Typical products offered in the catalogue are cordless headphone sets for television, bathroom scales, sonic pest repellents and a range of apparatuses for exercising at home. R ep ro du ce d by S ab in et G at ew ay u nd er li ce nc e gr an te d by th e P ub lis he r (d at ed 2 00 9) . 364 SAJEMS NS Vol 2 (1999) No 3 THE MEASURING INSTRUMENT A self-administered questionnaire was used as measuring instrument. Thirty (30) items were self-generated to represent the ten (l0) risk relievers available to respondents that purchase via mail-order catalogue. Each of the ten risk relievers was thus measured by three items. General risk perceptions were also measured by three items. All the items used were subjected to an experience survey as suggested by Churchill (1995: 152-3). Two professors of consumer behaviour and five other individuals were asked for their opinions in respect of the suitability of the items to measure the risk relievers studied. Each of the items had to be evaluated on a six-point Likert-type scale ranging from "Agree completely" to "Disagree completely". SCALE PURIFICATION The scale purification process consisted of three distinct phases: an assessment of the underlying dimensionality (and thus also discriminant validity) using exploratory factor analysis, an assessment of reliability by calculating Cronbach alpha coefficients and, fmally, the theoretical model was subjected to empirical assessment by means of structural equation modelling. Dimensionality The first step was to assess whether the data do indeed contain 10 dimensions as suggested in the literature study. For this purpose, a Maximum Likelihood Exploratory Factor Analysis was conducted specifYing a Direct Quartinum oblique rotation of the original factor matrix (Jennrich & Sampson, 1966). The programme BMDP4M was used for the analysis. An oblique rotation was specified because the factors are correlated. As several of the 30 items did not load a significant extent (0,40) in several solutions or did not demonstrate sufficient discriminant validity by loading on more than one factor, these items were deleted as suggested by Churchill (1995). Several factor solutions had been considered and the most interpretable one to emerge was the eight-factor solution shown in Table 2. The factor analysis results suggest that the proposed instrument demonstrates a considerable degree of discriminant validity. R ep ro du ce d by S ab in et G at ew ay u nd er li ce nc e gr an te d by th e P ub lis he r (d at ed 2 00 9) . SAJEMS NS Vol 2 (1999) No 3 365 Table 2 Rotated Factor Loadings for the Eight Factor Solution Factor Factor Factor Factor Factor Factor Factor Factor 1 2 3 4 5 6 7 8 Expen Govt Ptest Samp Advice Loyal Expert Source ITEM Expenl 0,848 Expen2 0,810 Expen3 0,716 Govtl 0,817 Govt2 0,757 Govt3 0713 Ptestl 0,777 Ptest2 0,734 : Ptest3 0,730 Sampl 0794 Samp2 0690 Samp3 0,608 Advice 1 0,810 Advice2 0,536 Advice3 10,499 Brandl 0782 Brand2 0,684 Expertl 0,737 Expert2 0,675 ExpertJ 0,447 Source 1 0,893 Source2 0,575 Eigen- 1,943 1,860 1,728 1,601 1,340 1,336 1,278 1,244 values According to Table 2 twenty-three of the original thirty items loaded to a significant extent on one of eight clearly identifiable factors. These factors were termed: 1) Expen - purchasing of the most expensive model 2) Govt - testing of a product by a government testing institution 3) Ptest - testing of a product by a private testing institution 4) Samp - availability of samples for inspection 5) Advice - advice from family, friends and colleagues 6) Brand - well-known brands 7) Expert - endorsement by an expert R ep ro du ce d by S ab in et G at ew ay u nd er li ce nc e gr an te d by th e P ub lis he r (d at ed 2 00 9) . 366 SAJEMS NS Vol 2 (1999) No 3 8) Source - reliable source ReUability The remaining twenty three items were then, as suggested by Churchill (1995), subjected to a reliability analysis using the computer programme SAS PROC CORR (SAS Institute, 1988). Cronbach alpha values reported in Table 3 show that all the underlying dimensions of the construct "risk relievers in mail-order catalogue retailing" are measured by an instrument with sufficient reliability (a> 0,7) and that the Cronbach alpha of the entire instrument is 0,88. This figure exceeds the minimum ofO,7 suggested by Peterson (1994) and Nunnally (1978) and confirms the reliability ofthe instrument. Table 3 Internal Reliability Results DIMENSIONS a Most expensive product 0,84 Government testing 088 Private testing 085 Samples 0,84 Advice from family/friends/colleagues 0,74 Brand loyalty 0,81 Expert endorsement 0,82 Reliable source 0,81 Overall 0,88 Path analysis The proposed instrument to measure the influence of risk relievers on general risk perception (Table 4) was then subjected to a path factor analysis (structural equation modeling). The results, set out in Table 5, suggest that the 8-factor model in Figure 1 represents a reasonable fit to the data (Hair, Anderson, Tatham and Black, 1998: 656). All the indices reported in Table 5 meet or exceed the minimum acceptable standards suggested by Hair, et al. (1998). The path analysis also provides some evidence of the construct validity of the proposed instrument (Tull & Hawkins, 1993: 318). R ep ro du ce d by S ab in et G at ew ay u nd er li ce nc e gr an te d by th e P ub lis he r (d at ed 2 00 9) . SAJEMS NS V.ol2 (1999) N.o 3 367 Table 4 Items to Measure the Influence of Risk Relievers on General Risk Perception Most expensive model EXPENI EXPEN2 EXPEN3 Ordering the m.ost expensive m.odel .of a pr.oduct reduces the risk .of .ordering by mail-.order. It is safer t.o buy the m.ost expensive m.odel by mail-.order catal.ogue .of a pr.oduct .offered. A c.onsumer reduces the risks related t.o mail-.order by .ordering the m.ost expensive m.odel .of any product. Government testing GOVTl GOVT2 GOVT3 If a mail-.order catal.ogue c.ontains the end.orsement .of a testing instituti.on, like the SABS .or CSIR, in respect .of a product, then it makes the buying .ofthat pr.oduct safer. P.ositive results and/.or rec.ommendati.on by a testing instituti.on, such as the SABS .or CSIR, in respect .of a pr.oduct .offered by mail- .order catal.ogue make it safer t.o buy by mail-.order catal.ogue. It is better t.o buy a pr.oduct by mail-.order catal.ogue if it has been tested and approved by a testing instituti.on like the SABS .of CSIR. Private testing PTESTI PTEST2 PTEST3 Samples SAMPI SAMP2 SAMP3 P.ositive results and/.or rec.ommendati.ons by a private testing instituti.on in respect .of a product .offered by mail-.order catal.ogue make it safer t.o buy by mail-.order catal.ogue. If a mail-.order catal.ogue c.ontains the end.orsement .of a product by a private instituti.on, then it makes buying that pr.oduct safer. It is better t.o buy a product by mail-.order catal.ogue if it has been tested and appr.oved by a private testing instituti.on. A sample reduces the risk when buying by mail-.order catal.ogue. The availability .of samples is likely t.o increase the buying .of a pr.oduct by mail-.order catal.ogue. A c.onsumer is m.ore likely t.o buy a pr.oduct fr.om a catal.ogue if a sample .of the product is available bef.ore taking the buying decisi.on. R ep ro du ce d by S ab in et G at ew ay u nd er li ce nc e gr an te d by th e P ub lis he r (d at ed 2 00 9) . 368 SAJEMS NS Vol 2 (1999) No 3 Advice from family/friendS/colleagues ADVICE I Buying by mail-order catalogue is safer if one seeks advice from family, friends or colleagues before buying. ADVICE2 It is important to obtain advice from other people who have bought by mail-order before buying oneself by mail-order catalogue. ADVICE3 Advice from family, friends or colleagues makes buying products by mail-order catalogue less risky. Known brands BRAND I A consumer can reduce risk when buying products by mail-order catalogue if he/she buys only products with well-known brand names. BRAND2 When buying products by mail-order catalogue, it is safer to buy only products with brand names with which one has been satisfied in the past. Expert endorsement EXPERT I It is safer to buy a product by mail-order catalogue if it contains an endorsement by an expert. EXPERT2 An endorsement by an expert in respect of a product in a mail- order catalogue reduces risk. EXPERT3 It is more likely that a consumer will buy a product by mail-order catalogue if it contains a testimonial by an expert on the product. Reliable source SOURCE I It is less risky when one deals with an established and well-known mail-order organisation. SOURCE2 When a mail-order organisation is big and well known, then it makes transactions with such an organisation easier. R ep ro du ce d by S ab in et G at ew ay u nd er li ce nc e gr an te d by th e P ub lis he r (d at ed 2 00 9) . SAJEMS NS Vol 2 (1999) No 3 369 Figure 1 Empirical model : risk relievers All path coefficients are significant at p