Price, brand name, and product composition characteristics as determinants of perceived quality

Journal of Applied Psychology1971, Vol. 55, No. 6, 570-579

PRICE, BRAND NAME, AND PRODUCT COMPOSITIONCHARACTERISTICS AS DETERMINANTS OF

PERCEIVED QUALITY*

JACOB JACOBY,2 JERRY C. OLSON, AND RAFAEL A. HADDOCK

Purdue University

While price and brand image have both been found to be determiners of productquality perception, the potency of these two cues has never been directly compared.Moreover, those studies which found price to be a determiner of perceived qualitymanipulated only price information, without permitting actual compositioncharacteristics to vary across brands. A 2 X 2 X 2 X 3 factorial experiment,using 136 adult male beer drinkers, and four test beers, examined the effects ofprice, composition differences, and brand image cues on the perception of beerquality. Price was found to serve as an indicant of product quality when it was theonly cue available but not when embedded in a multicue setting. Brand image hada stronger effect upon quality perception, particularly for brands with strongpositive images. In addition, it was found that neither price nor brand name hadsignificant effects on perceived quality except when product composition character-istics were allowed to vary between product samples. Last, in contrast to earlierfindings, the data suggest that beer drinkers possess at least some ability todistinguish among different brands of beer on the basis of composition (i.e., tasteand aroma) cues alone.

As the literature on perceived risk (cf. Cox,1967) clearly demonstrates, consumer purchasedecisions are frequently made under conditionsof varying uncertainty regarding the productand its attributes. To reduce such uncertainty,consumers seek and process information re-garding the product and generally attempt toform accurate impressions of it. Given thatproducts may be viewed "as an array of cues,"the "consumer's task in evaluating a productis to use cues from the array as the basis formaking judgments about the product (Cox,1962, p. 413)."

One impression usually of considerable im-portance to the consumer is the product's(or brand's) quality. Cues relevant to formingimpressions of quality include (a) price; (b)product composition characteristics such astaste, aroma, color, style, and size; (c) packag-ing; (d) brand, manufacturer (i.e., corporate),and store image; (e) advertising; (f) word-of-mouth reports; and (g) past purchase experi-

JThe authors gratefully acknowledge the criticalreading and helpful comments provided by HowardFromkin.

The authors wish to thank William H. Campbell,Kevin C. Peterson, and Raymond C. Stokes for theirassistance in conducting the experiment.

2 Requests for reprints should be sent to JacobJacoby, Department of Psychology, Purdue University,Lafayette, Indiana 47907.

ence. Of these cues, greatest attention thusfar has been directed toward price. Shapiro(1968) noted that "price is a powerful piece ofinformation for the consumer.... Since priceis concrete and measurable, the consumer viewsit with much confidence. He trusts it more thanmost cues concerned with quality (p. 20, 22)."

The first conceptual treatment of the price-quality relationship is usually credited toScitovsky (1967), and subsequent empiricalinvestigations (Gabor & Granger, 1966;Leavitt, 1954; McConnell, 1968a, 1968b; Tull,Boring, & Gonsior, 1964) have found that,with other cues held constant, price is indeeda determiner of perceived quality. Illustrativeof this approach is McConnell's work. Labelingdifferent bottles of the same brand of beerM, L, P, and associating different prices withthese labels ($1.30, $1.20, and 99j£ per six-pack, respectively), he was able to demonstratethat the high-priced brand was perceived tobe of higher quality than the medium-pricedbrand (p — .06). There was no difference inperceived quality, however, between themedium- and low-priced brands.

As Enis and Stafford (1969) noted, ". . . itseems fairly clear that if the only informationconsumers are given about product quality isprice information, they will indeed judgequality by price (p. 340)," Combining price

570

DETERMINANTS OF PERCEIVED QUALITY 571

and store image cues in one experiment, theseinvestigators still found price to strongly affectquality perception (Stafford & Enis, 1969).

It is noteworthy that nowhere, in all thosestudies that placed 5s in choice situations in-volving real products, have actual differences inproduct composition been allowed to vary.Investigators have given 5s identical samplesof beer, or carpet swatches cut from the sameroll, and simply attributed different labels,prices, and images to them. In contrast, andin consonance with real-world situations whereproduct differences can and do vary, the pres-ent investigation allowed actual physical dif-ferences to vary, while also manipulating priceand brand image cues. Both product composi-tion differences and price may be consideredobjective attributes of the product, whilebrand image may be defined as the subjective,emotional cluster of meaning and symbolsthat the consumer attributes to particularbrands. Usually brand names are sufficient toevoke brand images.

Several considerations suggested beer as theproduct to be studied. First, a replication ofMcConnell's (1968a, 1968b) often cited find-ings, in which price was the only cue to vary,was considered desirable. A second purpose wasto examine whether beer drinkers could dis-criminate among different brands on the basisof taste and aroma cues alone, given a rangeencompassing substantial differences in productcomposition. Earlier work had indicated thatconsumers are unable to distinguish amongdifferent brands of beer in "blind" taste tests(Allison & Uhl, 1964). Reexamination ofsimilar data from cigarettes and cola studies,however, yielded evidence to suggest someability to discriminate among various brands inthese product classes (Myers & Reynolds,1967, p. 18). Therefore, examination for similarabilities with respect to beer seemed worth-while.

With price absent, and given limited (orin-) ability to discriminate among differentbeers on the basis of taste and aroma cues,what does determine perception of productquality? Several investigators (Cox, 1962,Froman, 1953; White, 1966) have suggestedthat nonobjective criteria, especially varioustypes of images (e.g., brand, store, corporate,product, and salesman), are frequently used by

consumers to evaluate alternative brands.Empirical support was provided by a secondpart of the Allison and Uhl (1964) study inwhich 5s received properly labeled bottles ofbeer and were again asked to make the samejudgments regarding several product attributesincluding overall quality. The results of thismanipulation clearly indicated that

labels, and their associations, did influence their evalua-tions. In other words, product distinctions or differ-ences, in the minds of the participants, arose primarilythrough their receptiveness to the various firms'marketing efforts rather than through perceived physi-cal product differences (p. 39).

Given that price and brand image have eachbeen demonstrated to affect perception of beerquality, the present experiment compared therelative efficacy of these two cues.

In summary, the general purpose of thisstudy was to examine the unique and inter-action effects of three cues—price, compositiondifferences, and brand image—on perceptionof brand quality. Specific foci of interest were:(a) to determine whether the price-qualityrelationship could be replicated in a morecomplex, multicue environment; (i) to ex-amine the relative efficacy of price and brandimage as determinants of perceived quality;and (c) to determine whether beer drinkers areable to discriminate among different brands ofbeer on the basis of only taste and aroma cues,given substantially different beer samples.

METHOD

Subjects

The Ss, 136 male volunteer students who answeredan advertisement in the campus newspaper requestingparticipants for a "beer taste-testing study," were allover 21 yr. old, and each stated that he was a frequentbeer drinker and had consumed at least 12 oz. of beerduring the preceding week. Potential 5s called atelephone answering service in response to the ad, andthose that met the age and usage criteria were randomlyassigned to one of the eight experimental groups.

Design

A 2 X 2 X 2 X^3 factorial experiment with 17 5sper cell was conducted which consisted of two levels ofprice (present, absent), brand name (present, absent),and composition differences across beer samples(present, absent) as between-S factors, and three beersamples as within-5 factors. The dependent variable ofinterest was perceived overall quality, and repeatedmeasures were taken over three beer samples for each S.

572 J. JACOBY, J. C. OLSON, AND R. A. HADDOCK

Instruments

Each S received a questionnaire booklet that gatheredrelevant information concerning his beer-drinkinghabits and familiarity with various brands of beer,explained and administered the appropriate manipu-lations, and collected five dependent measures (per-ceptions of aroma, bitterness, strength, smoothness,and overall quality) for each of the three beer samples.The first four dependent measures were included todisguise the purpose of the experiment.

There was a graphic 0-100-point scale containing 10equally spaced, numbered graduations and havingendpoints labeled "worst beer" and "best beer" for eachdependent variable. Written and verbal instructionsdirected 5s to evaluate the beer samples served to themin relation to the best and worst beers they had evertasted. Evaluations were recorded by placing threearrows at appropriate points on each scale and labelingeach with the name or letter of the beer sample beingrated.

Procedure

Each of the eight cells was run separately. The 5swere seated at a long table and separated by table-toppartitions to cut off between-5 communications. A"credible-appearing" source in his early 40s, who servedas E, introduced the experiment as a marketing studyfunded by a national brewer for the purpose of develop-ing reliable instruments to measure beer taste per-ceptions. The 5s were told that college students werelarge consumers of beer, and therefore the manufacturerwas very much interested in their responses.

The 5s were then given the questionnaire bookletand completed the first two pages regarding their beer-drinking habits and general familiarity with beer brands.Upon completion, E set up the appropriate manipula-tions and, using a blackboard, demonstrated how touse the graphic rating scales.

Four different brands of beer were used. A nationallydistributed ultrapremium beer (costing $1.60 per warmsix-pack), a nationally distributed popular-priced beer(costing $1.35), and a regionally distributed inexpensivebeer (costing 99(4) were used for the 68 5s in the four"composition differences present" cells. These beerswere labeled Y, X, and Z, respectively, in the "brandname absent" cells. The three brands were selected fromamong 20 different brands on the basis of a pretest thatindicated that a comparable group of 5s could, relativeto other beers, accurately estimate the actual cost perwarm six-pack of these brands, with minimal variancein these estimates across 5s. These selection criteriahad two purposes: (a) to minimize differential cross-5familiarity with respect to the beers, a variable likelyto influence preference ratings (cf. Zajonc, 1968); and(b) to cover as broad a range of domestic beers (interms of price, brand image, and probable compositiondifferences) as possible. Another nationally distributedpopular-priced beer, whose price was also easily andaccurately estimated, was used for all three beer samplespresented to the 68 5s in the four "composition differ-ences absent" cells (see Table 1).

For those cells in which neither price nor brand

names were provided, 5s were simply told they wereevaluating three beers generally available in the localarea and here designated X, Y, and Z. When pricewas given, the beers were listed in the questionnairebooklet with their actual store prices (per six-pack)appearing next to them. The same procedures were usedwhen brand name (used to evoke brand image) waspresent except that, instead of X, Y, and Z, the actualbrand names of the beer samples were provided.

For the four cells in which the same brand of beerwas used for all three beer samples, directions wereidentical to those given when the beer samples differed.That is, 5s in those cells were also led to believe thatthey were tasting three brands of beer readily availablein the local area.

Beer samples were given to each 5 in plastic beige-colored cups that tended to disguise color cues. Asappropriate, the cups were marked either with brandnames or letters, and the accompanying instructions(in the experimental booklet) did or did not presentinformation. The 5s were told to sample the beers inany order but to complete each dependent measurerating scale for all three beers before proceeding to thenext scale, for example, complete the aroma ratingsfor all three beers, then proceed to the bitterness scale,etc. Because of university regulations, 2 oz. of beerwere served in each cup, making a total of 6 oz. per 5.

Precautions were taken to insure that the beer sam-ples were quite cold and had little foam. The threesamples for each 5 were poured at the same time (outof sight of 5s) and immediately taken to him. Com-ments elicited at the conclusion of the experiment in-dicated that the state each sample arrived in was com-parable to normal drinking conditions.

RESULTS

Two general assumptions should be met forthe repeated-measures analysis of variancemodel to hold: (a) homogeneity of the variance-covariance matrixes and (b) symmetry of thepooled variance-covariance matrix (see Winer,1962, pp. 369-374). Tests conducted for boththese assumptions were not statistically signifi-cant (X2 = 51.80, df = 42, and X2 = 5.69,df = 7, respectively, both ps > .10), andtherefore the model was assumed appropriatefor these data.

Table 1 presents the means and standarddeviations of perceived quality ratings foreach experimental cell. Simple inspection ofthese data reveals at least two items of in-terest. First, there was substantially greatervariability of interbeer sample quality ratingswhen composition differences were presentthan when the samples were of the same beer.In fact, the range of the overall beer samplemeans was twice as great when compositiondifferences were present (70.88-37.58 = 33.30

DETERMINANTS OF PERCEIVED QUALITY 573

TABLE 1

MEANS AND STANDARD DEVIATIONS OT OVERALL QUALITY RATINGS

Condition

Brand name presentPrice present

XSD

Price absentXSD

Brand name absentPrice present

XSD

Price absentXSD

OverallXSD

Composition differences present(samples of different beers)

BeerZ

(regionalinexpensive)

33.4717.82

34.1817.23

34.4718.08

48.1822.54

37.5819.60

BeerX

(popularpriced)

65.7123.65

64.8815.81

66.1217.90

55.1222.99

62.9620.42

BeerY

(ultra popularpriced

premium)

79.8213.21

71.4115.37

67.6522.84

64.6516.10

70.8817.82

Composition differences absent(samples of same popular-priced beer)

BeerZ

(purportedregional

inexpensive)

50.7619.49

50.1219.99

55.3521.08

51.4721.66

51.9320.21

BeerX

(purportedpopularpriced)

64.1816.47

56.8823.33

54.6521.79

68.5320.38

61.0620.95

BeerY

(purportedultra-

premium)

76.1817.96

71.0613.65

68.4121.25

57.1819.30

68.2119.17

Note.—n = 17 per cell.

vs. 68.21-51.93 = 16.28), with the regional in-expensive beer's low ratings accounting formost of this difference. Second, the brand name(BrN) absent, price (P) present, compositionaldifferences (CD) absent cell replicated Mc-Connell's (1968a, 1968b) findings, that is,the high-priced brand was perceived to be ofhigher quality than the medium- and low-priced brands (t = 1.85, df = 16, p < .05 andt = 1.81, df = 16, p < .05, respectively), withno significant difference between the latter two.

Results of a four-way ANOVA based on thedata presented in Table 1 are summarized inTable 2. A strong beer samples (BS) maineffect was obtained (F = 57.99, df = 2, 256,p < .0001), and a Newman-Keuls analysisindicated that all differences between pairs ofmeans for the three beers (%t = 44.75,Xx = 62.01, Xy = 69.54) were significant be-yond the .01 level. These differences were allin the expected direction, with the ultrapre-mium beer (Y) receiving the highest meanquality rating and the regional inexpensivebeer (Z) receiving the lowest mean rating.

Two two-way interactions were significantin the overall ANOVA, CD X BS and BrNX BS, and are shown in Figure 1 (Parts a andb, respectively).

TABLE 2

ANALYSIS OF VARIANCE OF OVERALL QUALITYRATINGS FOR COMPLETE DESIGN

Source

Between 5sPrice (P)Brand name (BrN)Composition differences (CD)P XBrNP XCDBrN X CDP X BrN X CDError (between)

Within 5sBeer samples (BS)P XBSBrN X BSCD XBSP X BrN X BSP X CD X BSBrN X CD X BSP X BrN X CD X BSError (within)

df

1111111

128

22222222

256

MS

378.55511.88

1083.88285.0021.20

.069.12

371.06

21972.33762.67

2015.993143.13

87.90756.71234.74

1461.18378.88

F

1.021.382.92

<1<1<1<1<1

57.99***2.015.32**8.30**

<12.00

<13.86*

* p < .025.**p < .01.

<.0001.

574 J. JACOBY, J. C. OLSON, AND R. A. HADDOCK

80

>• 75H

< 7°

O 65

< 60

£ 55

Q 50>

0 45

J5 40

1 35

2 30

la. Ib.

o—o Composition Diff. present o—o Brand Name present

- A- -̂ Composition Diff. absent A--A Brand Name absent's

Z X Y 'I 2 X Y

Beer Sample Beer Sample

COMPOSITION DIFFERENCES BRAND NAME BYBY BEER SAMPLES BEER SAMPLES

INTERACTION INTERACTION

FIG. 1. Two interactions with beer samples from overallanalysis of variance.

Analysis of variance of the simple maineffects indicate that the effect of compositiondifferences (see Part a, Figure 1) was signifi-cant only for Beer Z (F = 18.62, df = 1, 384,p < .01), and the effect of brand name (seePart b, Figure 1) was significant only for BeerY (F = 9.30, df = 1, 384, p < .01). It appearsthat, for the three brands of beer used, pres-ence of the true compositional characteristicsdoes not significantly affect one's perceptionof their quality, except for the low-pricedregional beer (Z), in which case presentationof the actual beer depressed the perceivedquality ratings. One might say that Beer Ztastes worse than its image (or at least worsethan the fourth beer used in CD absentconditions). Similarly, it appears that only thehigh-priced ultrapremium beer's (Y) image (asmediated by brand name) was strong enoughto significantly increase perceptions of quality.As might be expected, Beer Z seems to havea somewhat negative image, since presentationof its brand name depressed perceived qualityratings, although not to a significant extent(F = 2.48, df = 1, 384, p > .20). Brand namehad no significant effect on Beer X.

The four-way interaction, P X BrN X CDX BS, was also significant (p < .025) and is

shown in Figure 2. The increase in variabilityunder CD present conditions is clearly ob-vious. To clarify interpretation, simpler levelsof analysis were sought by analyzing four three-way ANOVAs.

By dividing the study into two three-wayanalyses—one in which CD are absent and onein which they are present—one obtains a de-sign similar to most of the previously conductedquality perception research in the former andan extension of such studies by allowing inter-brand composition differences to vary as theydo in a natural setting in the latter. Results ofthese two analyses are presented in Table 3.

The only significant effect for CD absentwas the_BS main effect. All intermean differ-ences (X, = 51.93, Xx = 61.06, Xy = 68.21)were significant beyond the .05 level using aNewman-Keuls analysis. A logical and par-simonious explanation for these differences isthat they are due to demand characteristics(cf. Orne, 1962) created by the experimentalprocedures. Given that these 5s (under CDabsent) were told that they were evaluatingthree different beers and that 5s in three outof these four cells received additional infor-mation (i.e., brand name and/or price) sup-

so

75

70

65

60

55

50I

45

I 40

3PRICE AND BRANDNAME

A BRAND NAME ONLY

//.-a PRICE ONLY

yNO CUES

V NO CUES

I/

Z X Y

Beer Samples

COMPOSITIONDIFFERENCES

PRESENT

Z X Y

Beer Samples

COMPOSITIONDIFFERENCES

ABSENT

FIG. 2. Four-way interaction: Price X Brand X BeerSample X Composition Differences.

DETERMINANTS OF PERCEIVED QUALITY 575

TABLE 3

ANALYSIS op VARIANCE OF OVERALL QUALITY RATINGS FOR Two LEVELSOP COMPOSITION DIFFERENCES

Source

Between 5sPrice (P)Brand name (BrN)P X B r NError (between)

Within 5sBeer samples (BS)P X B SBrN X BSP X BrN X BSError (within)

df

111

64

2222

128

Composition differences absent

MS

289.46261.57189.04369.48

4527.66559.37946.51955.69406.65

F

<1<1<1

11.13***1.382.332.35

Composition differences present

MS

110.29250.3796.08

372.64

20587.80960.01

1301.21593.39351.11

F

<1<1<1

58.64***2.73*3.71**1.69

* p te .07.** p < .03.

***p < .0001.

porting these cues, the average 5 tended to ratethe quality of his three beer samples in a wayconsistent with these demand characteristics.In other words, he rated the supposed ultra-premium, high-priced beer (Y) relatively high,the popular-priced beer (X) lower, and theregional inexpensive beer (Z) lowest. In con-trast to earlier studies, however, price exertedno effect upon quality perception.

For the ANOVA for CD present (see Table3), the main effect due to BS was again signif-icant (F = 58.64, df = 2, 128, p < .0001), anda Newman-Keuls analysis revealed that differ-ences in mean quality ratings between Beers Zand X, and Beers Z and Y were significantbeyond the .01 level, with the difference be-tween Beers _X and Y significant beyond the.05 level (X, = 37.57, X* = 62.96, Xy =70.88). This main effect was much strongerin this instance than when CD was absent,indicating the amount of increased variabilitydue to the presence of actual compositiondifferences (e.g., taste, carbonation, bitterness,etc.) between the beer samples above that dueto the demand characteristics.

The BrN X BS interaction was also signifi-cant (p < .03) under CD-present conditions.Again, only the simple, simple main effect ofbrand name for the high-priced, ultrapre-mium beer (Y) was significant (F = 4.26,df = 1, 192, p < .05), indicating that brand

image positively affected quality perceptions ofthis beer but had no significant effect on theother two beers.

Although the P X BS interaction (at CDpresent) approached but did not achieve acustomary level of statistical significance(F = 2.73, df = 2, 128, p^ .07), it wasexamined further since it was of a priori im-portance. Subsequent ANOVA of the simple,simple effects of price at the three levels of BSfailed to reveal any statistically significanteffects (all ^s > .20). In other words, knowl-edge of price (at this level of analysis) had nosignificant effect on perceived quality ratingsfor any of the three real beers.

A last set of three-way analysis that, withits relevance to prior studies, seems to be mostgeneralizable is for both levels of BrN (i.e.,BrN present and BrN absent). Results ofthese two ANOVAs are presented in Table4. A significant main effect of BS occurs underboth levels of BrN (both ps < .0001). Asmight be expected, this main effect is strongerwhen the beer samples are identified by theirbrand names (F = 60.83 vs. F - 12.26,df = 2, 128), as evidenced by the_ intermeandifferences for BrN present (X,"= 42.13,Xx = 62.91, Xy = 74.62) compared to thosefor BrN absent (Xx = 47.31, XK = 61.10,Xy = 64.47). All intermean differences withinboth sets are significant beyond the .01 level

576 J. JACOBY, J. C. OLSON, AND R. A. HADDOCK

TABLE 4

ANALYSIS OF VARIANCE OF OVERALL QUALITY RATINGS FORTwo LEVELS OF BRAND NAME

Source

Between 5sPrice (P)Composition differences (CD)P X C DError (between)

Within 5sBeer samples (BS)P X BSCD X BSP X CD X BSError

df

111

64

2222

128

Brand name absent

MS

3.31533.82

1.26362.70

5581.78651.68

1078.722116.50445.16

F

<11.47<1

12.26***1.432.374.56*

Brand name present

MS

660.42550.1229.06

379.41

18406.53198.89

2299.15101.39302.60

F

1.741.45<1

60.83***<1

7.60**<1

*p <.02.** p < .002.**p < .0001.

using a Newman-Keuls analysis, except thenonsignificant X-Y difference for BrN absent.

A different perspective on the interestingphenomonon is reflected in the two analysesshown in Table 4. CD do not significantly(i.e., with a p < .05) interact with BS unlessanother cue (e.g., BrN or P) is present. Thesignificant P X CD X BS interaction (p < .02)at BrN absent and the CD X BS interaction(p < .002) at BrN present both provideevidence to support this statement. To bemore specific, the CD X BS interaction is notsignificant when BrN is absent (p = .095),but under BrN present conditions, the CDX BS interaction is significant beyond the .002level. Probing of this latter interaction revealsthat the effect of CD is again significant onlyfor Beer Z (F = 14.73, df = 1, 98, p < .01).

Likewise, probing of the three-way inter-action, P X CD X BS, reveals a significantCD X BS interaction only when price cuesare present (F = 5.04, df = 2, 64, p < .009).The CD X BS interaction when no cues werepresent (i.e., neither P nor BrN) was notsignificant (F = 2.01, df = 2, 64, p^ .14).Therefore, the results of this experiment in-dicate that composition differences have asignificant effect on perceived quality ratingsonly when supported by another cue or, statedconversely, cues such as brand name have aneffect only when product composition differ-

ences are allowed to vary (e.g., see interactionunder CD present, Table 3).

A summary of the basic findings includesthe following: Composition differences didexert a significant effect upon quality percep-tion, but only for Beer Z, in which case usingthe "real" beer depressed quality ratings belowthose obtained when using the "fake" beer.Similarly, brand image exerted a significanteffect upon quality perception, but only forBeer Y, the most expensive and status-orientedbeer. Price, except in isolation, did not exertany significant effect upon quality perception.In other words, for the three brands of beerand the sample of 5s tested, brand name andactual composition characteristics were moreimportant determiners of perceived qualitythan was price. Last, there was substantiallygreater variability in quality ratings whencomposition differences were allowed to vary,thereby suggesting that beer drinkers can makesome distinctions among different brands onthe basis of composition (i.e., taste and aroma)cues.

DISCUSSION

In light of previous research on the price-quality relationship, one of the most strikingaspects of the obtained results was the in-ability of the price cue (when combined withother cues) to significantly influence percep-

DETERMINANTS OF PERCEIVED QUALITY 577

tion of product quality. While McConnell's(1968a, 1968b) findings were replicated, itwould appear that when price is presented inconjunction with other cues potentially indi-cative of product quality, it is used to a muchlesser extent than when presented alone.It should be noted, however, that price wasnot manipulated as in earlier studies (i.e., byassigning different price levels to the samebeer). Inasmuch as the purpose of the presentresearch was primarily to determine theeffect of price relative to other cues in a multi-cue setting, not to examine whether differentprices would affect quality perception, theappropriate manipulation was simply thepresence or absence of price.

Comparing the relative strength of price andbrand-image cues, the results indicate that,while the effect of price was not significant,awareness of the brand name did influencequality perception. A priori, one would expectknowledge of the brand name to increase per-ception of quality for the ultrapremium beerand decrease it for the inexpensive regionalbeer, and this was precisely the trend observed(see Table 1).

It would therefore seem as if brand imageis a more important cue than price in deter-mining quality perception of beer. The price-brand name distinction is somewhat con-founded. However, especially for establishedproducts and knowledgeable 5s, knowing onlythat the brand name provides more informa-tion than simply knowing the price. Again,this is reflected by the data (see Table 1).The presence of brand name alone producedresults more similar to the brand name-pricepresent condition than did the presence ofprice alone. This confound was virtually in-sured by the pretest criteria used to select theexperimental brands.

The variables that affect quality perceptionseem to manifest themselves primarily throughinteraction with other variables. Of the fourvariables considered, only one (beer samples)was strong enough to consistently affect qualityperception across a variety of situations. Theeffects of the other variables were moderatedby the situation in which they were introduced.This tendency toward interactive effects maylimit the number of variables that can bestudied in a single experiment, since higher

order interactions become exceedingly difficultto interpret.

Considering the nature of these interactionsmore closely, one of the most interesting, andpotentially important, results of the presentstudy is the lack of Cue X Product Sampleinteractions when differences in brand com-position were not present, that is, when thesamples were identical. Significant Cue XProduct Sample interactions occurred onlywhen actual composition differences existedbetween the three samples, that is, when theywere really different beers.

If this rather strong result is reliable, it hasimportant implications for future research.Since past experimental Investigations ofquality perception have manipulated othercues while holding actual differences betweenthe samples constant (i.e., nonexistent), theymay be using an oversimplified representationof reality and, therefore, missing importantand complex relationships between the cuesand the brands themselves. Consequently,relationships obtained under these conditionsmay have little or no generalizability to realworld considerations.

One of the more interesting findings was theevidence to indicate that beer drinkers arecapable of discriminating quality differencesin the expected direction, given only taste andaroma cues. If one assumes that, at least forthe brands studied, differences in price andimage reflected actual differences in quality,then perceived quality corresponded to ob-jective quality for these 5s.

Although it did tend to confound some of theresults, another interesting finding was a"letter effect" which evidenced itself in theprice absent- brand name absent- compositiondifferences absent cell. In some respects, thiscorresponds to Tucker's (1964) investigationin which housewives selected one out of fourloaves of bread, differentiated only by theletters H, L, M, and P, over 12 trials. Despitethe fact that all four loaves were identical,were taken from a single oven on the day ofdelivery, and, except for the letter labels, werepackaged identically, half the housewives inthe study manifested brand loyal behavior(i.e., selected the same brand on three or moresuccessive trials) by the end of the experi ment.

Such findings may be explained by the fact

578 J. JACOBY, J. C. OLSON, AND R. A. HADDOCK

that people tend to use cues systematicallyeven when the cues possess no real predictiveutility whatsoever (Dudycha & Naylor, 1966).In other words, "if a decision maker is placedin a situation where none of the cues availableto him is of any value, he will still tend topick out and use some of them as if they didhave value [Blum & Naylor, 1968, p. 459]."This would apply not only to such trivial cuedifferences as that which exist between differ-ent letters of the alphabet, but also to suchstrongly differentiated cues as price and brandimage. The implication here is that consumersmay use cues to form impressions of productquality even when these cues may have littleor no relationship to actual (i.e., objectivelyascertainable) product quality. Some tentativesupport for this conjecture is provided byGreen (1966), who found that a large propor-tion of consumers do not differentiate betweenrelevant and irrelevant information.

A possible limitation of the current investi-gation is the type of 5s used. It could beargued that the screening procedures insuredthe use of 5s who were relatively "expert"consumers of the product. This would not onlyrestrict the generality of our results, but mayin part explain 5s' tendency not to use priceas a quality indicator. Both Scitovsky (1967)and Shapiro (1968) have hypothesized thatconsumers lacking experience with the productare more prone than expert consumers to useprice in forming impressions of productquality.

Another possible limitation was the use of afourth brand of beer in the composition-differences-absent cells. In retrospect, it mighthave been more appropriate to have randomlyassigned each of the 68 5s in these four cellsto one of the three experimental beers.

In conclusion, to the extent one can gen-eralize from this experiment, product, andsample, the major findings of this investigationwere as follows:

1. Price does exert an effect upon the per-ception of product quality when it is the onlycue permitted to vary.

2. Price does not operate as an indicant ofproduct quality when embedded in a multicuesetting.

3. Brand image (as mediated by brand name)does affect perception of quality, especiallyfor brands with strong positive images.

4. Given substantial differences in productcomposition, and on the basis of taste andaroma cues alone, beer drinkers are able todiscriminate quality differences in the ex-pected direction across different brands ofbeer.

5. Consumers may use cues systematicallyeven when these cues possess little or no pre-dictive power.

Clearly, attempts should be made to re-plicate these findings with different types of5s (e.g., nonexperts), incorporating differentcues and treatment combinations (e.g., usingseveral beers, each at different price levels, orpouring ultrapremium beer into cups bearinga regional inexpensive beer brand name) andusing different types of products. Given theimportance of quality perception in consumerdecision processes,3 such effort seems welljustified.

3 For example, results of a recent study indicated thatthe extent to which housewives perceive differences toexist across the various brands comprising a productclass is strongly related to brand loyalty: the greaterthe perceived quality differences, the greater the likeli-hood of strong brand loyalty (Jacoby, 1970).

REFERENCES

ALLISON, R. I., & UHL, K. P. Influence of beer brandidentification on taste perception. Journal of Market-ing Research, 1964, 1, 36-39.

BLUM, M. L., & NAYLOR, J. C. Industrial psychology.New York: Harper and Row, 1968.

Cox, D. F. The measurement of information value:A study in consumer decision-making. In WilliamS. Decker (Ed.), Emerging concepts in Marketing.Chicago: American Marketing Association, 1962.

Cox, D. F. (Ed.) Risk taking and information handlingin consumer behavior. Cambridge: Harvard UniversityPress, 1967.

DDDYCHA, L., & NAYLOR, J. C. Characteristics of thehuman inference process in complex choice behaviorsituations. Organizational Behavior and Human Per-formance, 1966, 1, 110-128.

ENIS, B. M., & STAFFORD, J. E. Consumer's perceptionof product quality as a function of various infor-mational inputs. In Philip R. McDonald (Ed.),Marketing involvement in society and the economy.Chicago: American Marketing Association, 1969.

FROMAN, R. Marketing research. In J. H. Westing

DETERMINANTS OF PERCEIVED QUALITY 579

(Ed.), Readings in marketing. Englewood Cliffs,New jersey: Prentice Hall, 1953.

GARBOR, A., & GRANGER, C. W. J. Price as an indicatorof quality: Report on an inquiry. Economica, 1966,33, 43-70.

GREEN, P. E. Consumer use of information. In JosephNewman (Ed.), On knowing the consumer. New York:Wiley, 1966.

JACOBY, J. An attitudinal model of multi-brand loyalty:Preliminary results and promotional strategies.Paper presented at the Spring Conference on Re-search Methodology, American Marketing Associa-tion, New York Chapter, New York, May 14, 1970.

LEAVITT, H. J. A note on some experimental findingsabout the meaning of price. Journal of Business,1964, 27, 205-210.

McCoNNELL, J. D. The price-quality relationship inan experimental setting. Journal of Marketing Re-search, 1968, 5, 300-303. (a)

McCoNNELL, J. D. The effect of pricing in an experi-mental setting. Journal of Applied Psychology, 1968,52, 331-334. (b)

MYERS, A. H., & REYNOLDS, W. H. Consumer behaviorand marketing management. Boston: HoughtonMifflin, 1967.

ORNE, M. T. On the social psychology of the psycho-logical experiment: With particular reference to de-

mand characteristics and their implications. A meri-can Psychologist, 1962, 17, 776-783.

SCITOVSKY, R. Some consequences of the habit ofjudging quality by price. The Review of EconomicStudies, 1944-45, 12, 32. Reprinted in Perry Bliss(Ed.), Marketing and the behavioral sciences (2nd ed.).Boston: Allyn & Bacon, 1967.

SHAPIRO, B. P. The psychology of pricing. HarvardBusiness Review, 1968, 41, 14-16, 18, 20, 22, 24-25,160.

STAFFORD, J. E., & ENIS, B. M. The price quality re-lationship : An extension. Journal of Marketing Re-search, 1969, 6, 45fr458.

TUCKER, W. T. The development of brand loyalty.Journal of Marketing Research, 1964, 1, 32-35.

TULL, D. S., BORING, R. A., & GONSIOR, M. H. A noteon the relationship of price and imputed quality.Journal of Business, 1964, 37, 186-191.

WHITE, I. S. The perception of value in products. InJoseph Newman (Ed.), On knowing the consumer.New York: Wiley, 1966,

WINER, B. J. Statistical principles in experimental de-sign. New York: McGraw-Hill, 1962.

ZAJONC, R. B. Attitudinal effects of mere exposure.Journal of Personality and Social Psychology, 1968,9, (2, Pt. 2).

(Received August 27, 1970)