Measuring Price and Quality Competition Among Conglomerates: Methodolgy and an Application to the Major Appliance Industry

ABSTRACT - A methodology is presented for developing price and quality scales for manufacturers competing head-on across many product lines. Using 20 years of data from Consumer Reports, the method is applied to firms competing in eight product classes in major appliances. The results indicate that Whirlpool and GE offer the highest value when their price and quality places are considered simultaneously. These and other outcomes in the present research appear to be substantiated by two major trade studies conducted recently by members of the appliance industry.


David J. Curry (1983) ,"Measuring Price and Quality Competition Among Conglomerates: Methodolgy and an Application to the Major Appliance Industry", in NA - Advances in Consumer Research Volume 10, eds. Richard P. Bagozzi and Alice M. Tybout, Ann Abor, MI : Association for Consumer Research, Pages: 400-405.

Advances in Consumer Research Volume 10, 1983      Pages 400-405


David J. Curry, University of Iowa

[The following people helped enter and verify the database for this study- Debra Castle, Barb Wilfred, and David Born and their efforts are deeply appreciated. Sue Helfers has also made valuable contributions to the integrity of the database, while special thanks are extended to Chris Jantz who did the majority of SAS programming necessary for the present study.]


A methodology is presented for developing price and quality scales for manufacturers competing head-on across many product lines. Using 20 years of data from Consumer Reports, the method is applied to firms competing in eight product classes in major appliances. The results indicate that Whirlpool and GE offer the highest value when their price and quality places are considered simultaneously. These and other outcomes in the present research appear to be substantiated by two major trade studies conducted recently by members of the appliance industry.


Although the product line approach is useful for structuring the decisions of middle-management, it may be insufficient for strategic planning in firms which compete across many lines. In larger, multiline firms, management should assess competition across an entire product portfolio for the following reasons: First, it is quite likely that consumers formulate opinions on a broadline or corporate level rather than line-by-line. For example with major appliances one could argue that it is more likely for consumers to feel that "manufacturer A is overpriced" than to make the more refined judgment that "X is overpriced in freezers but very competitive in gas ranges." Because interpurchase times are so long with durable products, each consumer has considerably more experience with the entire major appliance group than with individual categories within this group. Under these circumstances an image of the corporation as a whole is more likely to be formed than separate "line images." Secondly, formal, long-range, strategic plans typically require a focus which aggregates beyond product lines and which, as a consequence, entails assessing the moves and countermoves of major competitors at the corporate level. Assessing competition across a portfolio is usually difficult because even defining the appropriate set of competitors is troublesome. Companies which are represented in one line drop out of consideration in others where new competitors emerge. Clearly, however, there may exist a kernal or core set of competitors which remains relatively invariant as focus shifts through a product portfolio.


The purpose of this paper is to illustrate this "core set" concept and then measure the members of this set with respect to both their quality and their prices. The resulting quality and price scales will assess competition among members of the "core set" across all of the product lines in which they compete.

It is anticipated that this type of approach will find application in several areas of marketing management and strategic planning. Perhaps most importantly the methodology allows planners to work with a complete product value equation from the consumers view. Specifically, previous efforts to assess price competition across a product portfolio have been successful because price is relatively easy to measure. However, little headway has been made in estimating overall quality across lines and in many cases quality considerations are absent in formal planning discussions. The situation probably encourages management to evaluate options on an easy-to-measure attribute regardless of its inherent importance to sales and profit objectives. However, buyer evaluations of products are based on the effects of price and quality each acting alone as well as interacting: the main effect of price is less than half of the value equation.

The present methodology also represents a reasonable alternative to corporate image studies based on data from consumer surveys. Such studies (see Clevenger 1965; Spector 1961; Sims 1979; Singson 1975) often use multidimensional scaling techniques to decompose overall corporate similarity measures to reveal the separate dimensions of image. (See Green and Wind 1973.) However, these solutions are normally difficult to interpret due to their insensitivity to axis rotation and the need to appeal to additional, univariate, data for interpretation. In contrast the methodology used in the present report measures price and quality dimensions directly using data which is probably more reliable and valid than data from "one-shot" consumer surveys.

Finally, the results presented here may be useful for developing or substantiating certain advertising claims or as evidence in legal confrontations between the principals.


In order to fulfill the objectives cited above, the methodology is applied to firms selling major appliances. This particular product group was chosen for three reasons:

1. The product group has a well-documented history of competition; e.g. in the major appliance industry a small number of firms compete in eight, well-defined product lines.

2. There is considerable, practical, interest in the application of the methodology among firms selling major appliances. For example, a comprehensive survey on quality commissioned by Appliance Manufacturer showed that consumers judged appliance quality to be higher now than 10 years ago in every product group but one -- major appliances. [Appliance Manufacturer commissioned National Family Opinion Inc. to draw a representative sample of 2,000 households from their file of 100,000. The results of the study were based on 1,247 useable returns (62%). (See Appliance Manufacturer, April 1980, p. 35 for a more detailed explanation.)]

3. The database used for the present study consists of reports published over the twenty-year period (1961-1980) in Consumer Reports (CR) magazine. The Appliance Manufacturer survey indicated that Consumer Reports is ranked as the consumer's single most important source of information about quality. CR is ranked ahead of Good Housekeeping, friends and relatives, sales persons and endorsements in that order.


Competition in major appliances is defined by sales in eight sub-categories including: (1) trash compactors, (2) dishwashers, (3) garbage disposers, (4) clothes dryers, (5) freezers, (6) clothes washers, (7) ranges and (8) refrigerators. [These eight categories comprise the industry's operational definition of "major appliances" which is based on a long history of sales and product development. For example, products in these categories averaged more than eight years in their introductory stage and more than twenty-four in the growth stage of their product life cycles. (Quall et al. 1981). Newer categories such as "televisions" and "microwave ovens" are excluded from this scheme.] The industry has been marked by significant changes during the twenty-year period of study. Therefore, two separate sets of models were developed to account for the effects of mergers and acquisitions from 1961 through 1980. On the one hand the current ownership pattern is reflected in price and quality scales for the five broadline manufacturers who presently control the vast majority (at least 80%) of the production capacity in the eight categories listed above. In contrast, price and quality scales are also developed for the fifteen manufacturers who have been subsumed by the broadline conglomerates over the years. Clearly, the methodology utilized in the present report is not inherently dynamic. Rather the objective is to use separate, static, scales to summarize the levels of price and quality competition prevailing over the entire twenty-year time frame.

These two approaches represent the logical extremes in aggregation with the data currently available. Other aspects of the situation also make these two points-of-view worthwhile. For example although strategic and financial control has become more centralized in the industry, acquired companies do not necessarily mirror their parent company in management form or production output. In many cases an acquisition is designed to gain a particular production facility together with its labor force and management expertise. Under these circumstances the quality of the products manufactured at the acquired site would tend to be the same before and after the acquisition, changing only gradually through labor force attrition and replacement. Under these circumstances, analyses at both the ownership level as well as the individual manufacturer level are sensible.

The large number of acquisitions and mergers in the industry has resulted in the structure shown in Exhibit 1. This structure contains the five broadline manufacturers which compete in all or nearly all of the major appliance categories, three independent manufacturers and a group of product line specialists. The reader should note that the appliances sold by JC Penney, Montgomery Wards, Sears and Western Auto are manufactured elsewhere although there are long-standing relationships between certain manufacturers and certain retail chains; e.g., between JC Penney and GE and between Sears and Whirlpool.




The data for the present study were extracted from a much larger, computerized, database assembled by the author, a colleague and a number of research assistants. This database, called CONREP, contains information from more than 900 studies published by Consumer Reports magazine during the period January 1961 through December 1980. The entire database contains price, quality rank, manufacturer name, product category and other [See Riesz and Curry, (1982) for a more extensive description of CONREP.] information about nearly 14,000 brands. The data for the present study were extracted from CONREP according to the following procedures:

1. Every study in any of the eight major appliance categories was selected from CONREP.

2. This dataset was then searched to find studies where two or more of the fifteen major appliance manufacturers competed head-on. [A 16th manufacturer, Sanyo, was dropped from the analysis because it entered only a few studies. Also, if a manufacturer had two or more brands in a study then the average price and quality scores for this manufacturer were used.]

3. Steps 1 & 2 resulted in a database of 72 Consumer Reports studies ranging over the twenty-year analysis period.

4. Using each of these studies, separate PRICE and QUALITY dominance matrices were constructed. A typical entry in a quality dominance matrix is:

              0 if manufacturer i's quality rank exceeds manufacturer j's,

q(i,j) =   1 if j's exceeds i's

             .5 if i and j are tied in quality

Therefore a quality matrix for a single study contains a "1" wherever the column manufacturer's quality was rated better than the row manufacturer's

5. Finally, composite PRICE and QUALITY dominance matrices were constructed by adding their study-by-study counterparts. Note, the following relations hold for these two matrices;

Q(i,j) + Q(j,i) = FREQ(i,j) = P(i,j) + P(j,i),

i.e.; the frequency of times manufacturer's (i,j) met is given by the sum of the symmetric entries in either matrix.

6. These two dominance matrices were then transformed into proportions matrices by dividing each entry by the frequency of times the row and column manufacturers had met.

The two proportions matrices formed by the procedures described above represent the data input for constructing the price and quality scales reported here. These scales are output by the Law of Comparative Judgment proposed by Thurstone and subsequently modified by Gulliksen (1956a), Torgerson, (1958) and others. The model (see Torgerson 1958) postulates that each manufacturer has some true scale value on, say, a quality continuum and that the manufacturer's quality rank in any one Consumer Report study is a function of this true score plus some error. The effects contributing to error include sampling (the fact that CR's judgments are based on a sample of the manufacturer's output), fallibility in the CR test procedures and so on. [The procedures used by Consumers Union in rating quality and reporting prices are subject to a variety of problems. For example, until May, 1958, CU reported only list prices which often differed from their market counterparts. The market prices currently published in CR remain subject to sampling error and other biases as summarized by Morris (1971).] The power of the model is that even though the ordinal comparisons obtained from a single study do not contain information about the strength of one manufacturer's dominance over another such strength information is contained in the aggregate level data. For example if manufacturers i and j produce products nearly equal in quality then each should rank above the other about 50% of the time. Furthermore, in the aggregate level data the errors from individual studies will average out.

The Law of Comparative Judgment has been applied in a number of previous market studies (see Sims 1979; Wind 1973; Odesky 1967; Day 1965; Eastlack 1964) although it is typically applied to the rankings of paired-comparison data given by individuals. Furthermore, most previous studies in marketing fail to utilize the statistical theory available for the model. Because the goodness-of-fit to the data can be accurately tested, the model will be rejected if the data do not lend themselves to the unidimensional scale we seek.

Standard Price Scales

The price scale output by the LCJ is useful in that it measures the propensity of one manufacturer to position its price above or below competitors even if the absolute magnitude of the difference is small. The scale based on price dominance alone will, therefore, be referred to as a PRICE POSITIONING SCALE (PPS). However, the price's within a single study are clearly ratio-scaled and, therefore, price differences, the price origin and price ratios are meaningful. To capture the magnitude of price differences on a study by study basis a STANDARD PRICE SCALE (SPS) was also created according to the following procedures:

1. Prices from among the 5 broadline (or 15 individual) manufacturers included here were standardized within each study to mean = 50, standard deviation = 10.

2. The standard prices for each competitor were averaged over all studies involving that competitor. These average standard prices represent a price scale which may capture a different aspect of price competition than the PPS.

One would expect the PPS and SPS to be positively correlated. The magnitude of this correlation is-difficult to guess and depends on whether any or all of the manufacturers tend to position against one another and whether they do so consistently over individual categories and over time.


The results for the five broadline manufacturers are presented first because of their compactness and the fact that they represent the industry's current structure. The more detailed analysis for the fifteen individual manufacturers follows with emphasis on its added contributions.

A Standard Price Scale for the Broadline Manufacturers

Exhibit 2 shows the average standard prices for the five broadline manufacturers on a category by category basis and over all categories. There are a number of elements in this exhibit which bear highlighting. First, the different firms are represented with different frequencies in Consumer Report studies. These differential frequencies are a function of the number of product offerings by each as well as CR's manufacturer selection policy. The frequency of appearance ranged from a low of forty-five [see the #'s in ( ) in Exhibit 2] for the Raytheon corporation to a high of seventy for White Consolidated Industries. Secondly, the frequency of times each pair of firms met head on in the 72 studies also varies. There are ten pairs of competitors. Among these ten, products from Magic Chef and Raytheon met head-on in only 38 studies while at the other extreme General Electric and White Consolidated products met head-on in 66 studies. The mean frequency over these ten pairs was 51.



The mean standard prices over all categories ranged from a high of 62 (Raytheon) to a low of 45 (GE and Magic Chef.) These scale values may be interpreted as follows: In the 45 studies where it is represented, Raytheon's prices average [(62-50)/10] = 1.2 standard deviations above the mean price among Raytheon and any of the other four competitors in those studies. A look at the raw data verifies the high prices Raytheon charges. For example Raytheon's products were priced higher than those by any of the other four broadline manufacturers in more than 80% of the studies. Raytheon sometimes outpriced the field by 10-155. Furthermore, Raytheon maintained this high-pricing strategy in every product category save ranges where its products were priced at about the mean.

Price Position vs. Standard Price

Applying the Law of Comparative Judgment resulted in the price positioning scale shown in Exhibit 3. Because the LCJ provides an interval scale, it is permissible to transform the values in certain ways to aid in interpretation. [Conclusions from interval scales are invariant with respect to affine transformations, i.e. new scale = a(old scale) + b for a > 0. Such transformations do not affect product-moment correlations.] In the present case the MIN and MAX values among our five competitors are set to zero and 100 respectively. The reader should take care in realizing that the zero is arbitrary on an interval scale and avoid attributing certain algebraic properties to this value.

The results from the LCJ correspond very closely to the standard price values (also transformed for ease of comparison) previously estimated. In fact the correspondence is remarkable considering the different methodologies on which the two scales are based. The product/ moment correlation between the two price scales is t r = .992 (p < .001) and the absolute scale values themselves correspond very closely. This correspondence is both "bad news" and "good." The bad news is that our attempt to measure two different aspects of pricing policy has failed. The good news is that the result provides a very powerful check on the validity of the scale output by the LCJ and, therefore, increases our confidence in using an estimation process based on rank-order data alone.

Exhibit 3 also shows the quality scale output by the LCJ. The statistical test confirms that this unidimensional scale fits these data extremely well (see Exhibit 4; X = 4.61; df = 6 model can not be rejected at even a = .40) therefore, according to 72 CR studies over a 20-year period, Whirlpool products rank at the top in quality, followed by GE, White Consolidated, Raytheon and Magic Chef. Because the scale has interval properties, the differences between values are meaningful. For example, the gap between Magic Chef and Raytheon is 41 points which is considerably larger than any other difference between firms which are adjacent on the scale. However, Whirlpool dominates both White Consolidated and Raytheon by a margin which is even greater than Raytheon's dominance over Magic Chef.



Clearly, Whirlpool is the industry leader in quality when the data are aggregated to the level of these conglomerates. However, as we shall see, there is considerable variation in quality among products of the various company's held by a parent like Raytheon. In fact both Whirlpool and GE may be at an advantage in this broad analysis. Of the five firms included, only these two have remained out of the merger process during the study period. In essence neither has "contaminated" its quality by commingling its products with the products of other stand-alone manufacturers.

Price/Quality Correlations and Consumer Value

Perhaps the most interesting result in Exhibit 3 is the virtual lack of correlation between the prices charged by these manufacturers and the quality they provide. The estimated correlations are in fact not significantly different from zero; and it appears p may even be slightly negative.

This is a most disturbing result. Under these conditions consumers who infer high qualit& Manufacturing, which leads in dishwashers with a 45¦b share. Magic Chef is the leader in gas ranges with a 20% share. Thus the results presented here are consistent with current market shares which suggest that consumers are paying attention. The outcome also agrees with some previously published experimental results; e.g., see White and Cundiff (1978), Pincus and Waters (1975), and Szybillo and Jacoby (1974) among others.

Estimating One Firm's Dominance Over Another

Exhibit 4 provides a slightly different perspective on these scales as models-. The exhibit shows the observed proportion of times one manufacturer dominated another on either price or quality. A scale output by the LCJ can be used to predict these proportions for each pair and in fact these predictions form the basis for the test of a model's fit. A scan of both matrices shows a very good fit and the statistical tests confirm that neither model could be rejected at anything approaching standard a-levels.

(Exhibit 4 will be distributed at the conference)

Results For Individual Manufacturers

The nature of the data for the 15 individual manufacturers necessitated a slight change in the methodology. The reason is suggested in Exhibit 5 where the frequency of participation in CR studies is shown to be lower with these data. For example firms like Magic Chef (i.e., the Magic Chef brand name without including its acquisitions), McGraw-Edison, White-Westinghouse and Tappan are represented in relatively few studies e.g.; 9, 11, 16 and 20 respectively. This sparse representation resulted in some cases where two manufacturers did not meet head-on in any of the 72 CR studies.

(Exhibit 5 will be distributed at the conference)

Missing values necessitated the use of Gulliksen's (see Torgerson 1958 p. 174-179 and Kaiser and Serlin 1978) least-squares solution for the LCJ with an incomplete proportions matrix. However, each manufacturer still receives a scale value so that the model can be used to estimate the proportion of times one manufacturer's quality (or price) would have dominated another's had they met. For example, surprising as it may seem, Maytag and Amana did not face one another in any of the 72 studies. However, the model predicts that Maytag would be judged of higher quality than Amana about 64; of the time while having a higher price about 62% of the time.

Exhibit 5 shows the mean standard prices overall and by category. Maytag's prices tend to be highest averaging nearly two standard deviations above the mean. Amana's prices are next with an average of 67. Firm's with very low prices include Admiral, Philco-Ford, General Electric and Whirlpool. The exhibit is useful in determining who is competing in what product categories and each manufacturer's specific pricing strategy on a category by category basis.

The Scales

Exhibit 6 reports the two price scales and the quality scale for these firms. The most significant aspects of these results are as follows:

1. The two price scales again agree very closely (r = .98 p < .001). Therefore even with the significant amount of missing data the LCJ has apparently estimated a valid price scale.

2. On a company-by-company basis the price and quality scales are correlated quite highly (r = .66 p < .01). This result confirms our earlier suspicion that the aggregate results were masking considerable variation between manufacturer's held by the same conglomerate. For example, Raytheon owns both the Amana Corporation (2nd in quality) and McGraw-Edison (3rd from the bottom in quality). [A word of caution is due when interpreting the scales estimated by the LCJ. In both cases the goodness-of-fit test based on all 105 proportions was high enough to reject the model at about a = 01. However, these tests are very conservative because where a proportion is missing a fitted value is assessed against a "zero" entry Furthermore, the arcs" transformation used in the test is very sensitive to proportions near zero or one. For example 22 (20%) of the cells in the test for the quality scale were either missing or had 1/0 entries. In fact more than 66 points of the observed x2 = 166 (df = 91) were contributed by just 12 cells.]

(Exhibit 6 will be distributed at the conference)

Do Consumers Tradeoff Quality and Price

Exhibit 6 suggests that Maytag and Amana would receive high quality ratings if consumers were asked directly. However because the prices charged-by these two manufacturers are also considerably higher than those charged by Frigidare, Whirlpool and GE, we might suspect that consumers would be willing to trade some quality for a lower price. Fortunately, we have direct evidence to -bear on this issue. In addition to their 1980 study of quality, Appliance Manufacturer recently conducted a study on purchase intentions. (See "The Buying Consumer" April 1979). The relevant results of these two studies are reproduced in Exhibit 7. Maytag and Amana were mentioned as tops in quality in three categories (dryers, freezers and washers) yet in each of these cases the majority of consumers intended to buy Sears Kenmore Although alternative explanations may explain this switch, it is highly plausible that Sears represents affordable quality. The bulk of Sears major appliances are produced by Whirlpool as well as Frigidare of White Consolidated, a fact unknown to most consumers To quote the staff writers of Appliance Manufacturer (April 1980 p. 82):

(These results) may suggest that consumers understand the extra cost attached to "high quality" brands; they become more practical at purchase time. However, if they are willing to spend more for better quality in the future, as they say..., the response in this study suggests that mass merchandisers could lose share of market if they don't improve the quality of their products.


In this report we were able to summarize 20 years of Consumer Reports data on separate, unidimensional, scales for price and quality. These scales fit the data for the broadline manufacturers extremely well and, therefore, represent a parsimonious model for the relative position of each firm on these critical dimensions. The validity of the scaling method was also confirmed by a second price scale developed from observable, ratio-scaled data

These results can be used for strategic planning in a number of ways. For example, the price and quality scales implement many of the steps suggested by Oxenfeldt (1974-75) for firms interested in developing a better price-quality image Companies with a favorable position on one or both scales might also publish this information directly in their marketing communications. The approach also provides planners with a much more precise measurement of their firm's standing on the nebulous attribute "quality." Such information is not only useful for strategic marketing decisions but also for possible mergers designed to enhance a firm's quality standing in an industry. The method could also be applied to individual brands within a line to aid in brand deletion decisions

The present application represents a trial of the methodology. More experience is needed with other product groups to reveal additional strengths and weaknesses. Many other applications are possible using the present database because CONREP contains data on most consumer durables and nondurables since 1961. For example, plans include direct extensions to housewares (smaller appliances), televisions and other consumer electronics, cameras and numerous other, multi-line, product groups. Future applications will lead to a more refined version of the method which can be assimilated with confidence into the strategic planning process.




References will be available at the conference in a more complete version of this paper.



David J. Curry, University of Iowa


NA - Advances in Consumer Research Volume 10 | 1983

Share Proceeding

Featured papers

See More


Q7. Desire in Performed Consumption: Examining the Case of Korean Beauty Vlogging

Marie-Eve Jodoin, HEC Montreal, Canada
Marie-Agnès Parmentier, HEC Montreal, Canada

Read More


Ineffective Altruism: Giving Less When Donations Do More

Joshua Lewis, University of Pennsylvania, USA
Deborah Small, University of Pennsylvania, USA

Read More


The Impact of Anthropomorphized Cute Brands on Consumer Preferences for Distinctive and Majority-Endorsed Products

Marina Puzakova, Lehigh University
Nevena T Koukova, Lehigh University

Read More

Engage with Us

Becoming an Association for Consumer Research member is simple. Membership in ACR is relatively inexpensive, but brings significant benefits to its members.