Measurement of Milk Container Preferences Mark Speece Douglas L. MacLachlan ABSTRACI'. Consumer preference for container types for fluid milk was investigated for gallon and half gallon containers using conjoint anal is. Consumers clearly preferred plastic jugs to paper- board for gal r" on containers, which explains the nearly complete cap- ture of the gallon milk market by plastic. They prefer paperboard for half gallon containers, which corresponds to the failure of plastic to penetrate that size container very well. Relative prices of container types are adjusted for the utility consumers see in the containers. Utilitv adiusted orices of ~ l a s t i c eallon containers make them very compktitiGe relaiive to afternative ackaging materials. Prices , i f ~lastic half eallon containers make t R em less com~etitive once ut~li- iies are accounted for. INTRODUCTION The battle for market share between plastic and paperboard in the fluid milk packaging industry is still hotly contested. Plastic has captured nearly all of the gallon container market, but paperboard continues to be-used for around four-fifths of'half gallon milk con- Mark W. S~eece has doctorates in Marketing from the Universitv of Washinn- ton and in ~ i d h l e East Geography from the ~nGersity of Arizona. ~e is currenzy Lecturer, Department of Marketing, Chinese University of Hong Kong, Shatin, M, Hong Kong. Douglas L. MacLachlan has a PhD in Marketing and MA in Statistics from the University of California, Berkeley. He was formerly Chairman and is currently Professor, Department of Marketing and International Business, University of Washington, Mackenzie Hall DJ-10, Seattle, WA 98195. Journal of International Food & Agribusiness Marketing, Vol. 3(1) 1991 Q 1991 by The Haworth Press. Inc. All rights reserved. 43
Transcript
Measurement of Milk Container
Preferences Mark Speece
Douglas L. MacLachlan
ABSTRACI'. Consumer preference for container types for fluid milk was investigated for gallon and half gallon containers using conjoint anal is. Consumers clearly preferred plastic jugs to paper- board for gal r" on containers, which explains the nearly complete cap- ture of the gallon milk market by plastic. They prefer paperboard for half gallon containers, which corresponds to the failure of plastic to penetrate that size container very well. Relative prices of container types are adjusted for the utility consumers see in the containers. Utilitv adiusted orices of ~ las t i c eallon containers make them very compktitiGe relaiive to afternative ackaging materials. Prices , i f ~ las t ic half eallon containers make t R em less com~etitive once ut~li- iies are accounted for.
INTRODUCTION
The battle for market share between plastic and paperboard in the fluid milk packaging industry is still hotly contested. Plastic has captured nearly all of the gallon container market, but paperboard continues to be-used for around four-fifths of'half gallon milk con-
Mark W. S~eece has doctorates in Marketing from the Universitv of Washinn- ton and in ~ i d h l e East Geography from the ~nGersity of Arizona. ~e is currenzy Lecturer, Department of Marketing, Chinese University of Hong Kong, Shatin, M, Hong Kong. Douglas L. MacLachlan has a PhD in Marketing and MA in Statistics from the University of California, Berkeley. He was formerly Chairman and is currently Professor, Department of Marketing and International Business, University of Washington, Mackenzie Hall DJ-10, Seattle, WA 98195.
Journal of International Food & Agribusiness Marketing, Vol. 3(1) 1991 Q 1991 by The Haworth Press. Inc. All rights reserved. 43
44 Journal of Intemafional Food & Agribusiness Marketing
tainers, and nearly all of sizes smaller than half gallons. To a large extent, performance of these two packaging materials in different sizes relates to the utilities consumers attach to various container characteristics. But consumers are also highly price sensitive in a market like milk, which is essentially a commodity, and a few cents difference in the cost of the package may well influence acceptance of the packaging material.
Packaging decisions in such a situation should be supported by good knowledge of consumer preferences. However, the issue of consumer preferences for packaging materials in fluid milk markets has received almost no attention in published sources. This study demonstrates a methodology for turning consumer perceptions of milk containers into estimates of relative utilities consumers see for different packaging types. The progression of milk packaging rnate- rials from glass to paperboard to plastic represents new generations of technology. Presumably, new generations improve upon the old, and therefore, consumers are likely to derive additional utility from new package types. But if the new generation is not seen by con- sumers as having higher utility, it is unlikely to make much head- way. This is particularly true if consumers do not think they are getting good value for the prices they must pay for the new technol- O D .
We can use measured utilities to calculate a "utility adjusted" price, so that the relative price competitiveness of the packaging materials can be judged. Most consumers in industrial countries are not strictly price conscious in the sense that they always want the lowest price, but are rather value conscious. They balance utility (i.e., benefit to be received) against price (i.e., the sacrifice ex- changed for the benefit). We encountered the concept of utility ad- justed pricing as used here to measure value in an article by Stem, Ayers, and Shapanka (1975) where they introduced it as part of a technological forecasting model. The latter authors compared use of glass and plastic bottles in various industries, including dairy, but not specifically fluid milk. Their model was applied by Stephens (1977) to plastic and paperboard fluid milk containers. Otherwise, to our knowledge, the concept of utility adjusted pricing has not been used to study competitiveness of different generations of tech- nology in packaging or other areas of marketing. Only occasionally
Mark Speece and Douglas L. Machchlan 45
have researchers even used relative utilities (without prices) to help model competition across generations of technology (e.g., Stover 1978; Silverman 1981).
PACKAGING MATERIALS lN THE FLAYID MILK INDUSTRY
We investigated consumer attitudes toward three packaging ma- terials: glass, paperboard, and plastic. Glass milk bottles were in- vented in the 1880s, and the initial use of bottles in the U.S. was in one very specialized market niche: boiled bottled milk for infants. Before their invention, milk was distributed in bulk. Milk route drivers poured from large cans into a measuring can, then repoured into consumers' containers. This somewhat unsanitary method of delivering raw milk still accounted for the majority of retail milk deliveries as late as 1910, although during the first decade of the Twentieth Century sealed glass bottles began to appear in larger cities. After the turn of the Century, several technical developments such as the soaker bottle washer and the automatic bottling machine helped accelerate development of mass markets for glass-bottled milk. Improvements in bottling technology continued in the second decade of the Twentieth Century, and glass bottles soon became the standard container for handling retail milk in the U.S. (Roadhouse and Henderson 1950; Henderson 1956, 1971; Harper and Hall 1976).
Paperboard cartons, initially usually coated with paraffin, and later with plastic, were first used commercially as containers in 1929. There was little interest at first and they were not immedi- ately adopted for milk packaging. By 1940, paperboard had only captured about 4.8 percent of the fluid milk market, mostly in smaller sizes. Paperboard cartons became more popular for larger sizes in the late 1940s and early 1950s, as milk distributors and consumers eventually saw numerous advantages. Most importantly, paperboard was lighter weight and did not need to be returned, which eliminated deposits from consumers and extensive handling by distributors. Even a quart glass bottle still weighed nearly 18 ounces after lightweight bottles were developed in 1940. Gallon bottles weighed over three pounds. By contrast, paperboard cartons
46 Journal of lnrernafional Food & Agn3usiness MarkPling
(and later plastic bottles) weigh only a few ounces even for half gallon or gallon sizes (Roadhouse and Henderson 1950; Henderson 1956, 1971; USFTC 1973; Harper and Hall 1976; Modern Packag- ing, March 1965).
Plastic milk bottles, manufactured with a blow molding process from high-density polyethylene (HDPE) resin, were introduced into milk packaging in 1964. Originally, both returnable and non-re- turnable types of plastic bottles were developed. Ultimately, non- returnable jugs became much more popular with consumers and returnables nearly disappeared. Neither consumers nor distributors liked having to deal with the additional handling of returnables, factors which had contributed heavily to the demise of glass milk bottles in earlier decades. By the mid-1980s, plastic jugs had almost completely captured the market for gallon containers (Figure 1). They weigh only a few ounces, and are much more sturdy in this size than gallon paper cartons (Modem Packaging, March 1965,
FIGURE 1. Market Share by Package Type for Gallon Containers
YeaK Glass + P a p r 0 Plastic
Mark Speece and Douglas L. Machchlan 47
February 1967, April 1978; Chemical Week, July 9, 1966; Food Processing, October 1974;.Plastics World, August 1988; Pulp & Paper, April 1986; data on market shares was obtained from a man- ufacturer of paperboard cartons and supplemented with data from USDA 1987, and Stephens 1977).
This success in gallon containers has been the key reason that plastic has been so strong in fluid milk packaging. Plastic jugs have not achieved nearly as much popularity in half gallon containers. They have gained steadily since 1964, but had achieved less than 20 percent of the half gallon market by 1987; paper cartons retained an 80 percent market share (Figure 2). Plastic milk containers gener- ally cost more'than paper, but they do not have much differential advantage over paper cartons for half gallon sized containers. Half gallon paper cartons are much more sturdy than gallon ones. In quarts and smaller sizes, plastic has made hardly any impact at all. But the past several decades have seen a major shift in consumer
FIGURE 2. Market Share by Package Type for Half Gallon Containers 100
90
80
70
c 60 6 $. 50 - e
40
30
20
10
0 1965 1970 1975 . 1980 1985
Year Glass + Paper . o Plastic
48 Journal of lntemational Food & Agribusiness Marketing
preference toward gallon sizes. Gallons accounted for thirteen per- cent of fluid milk sales in 1963, against 56 percent for half gallons. By 1987, gallons accounted for 62 percent, and half gallons 21 percent. Smaller sizes accounted for 31 percent and 17 percent of volume sales at the beginning and end of this period, respectively (USDA 1987).
Paperboard producers have nearly abandoned gallon markets, but have vigorously resisted loss of market share to plastic in other sizes. They have also attempted to draw consumers back from gal- lons to half gallons. The introduction of double pack half gallon paperboard cartons was part of this strategy. So were extensive ad- vertising campaigns of the mid to late 1980s, which presented the paperboard industry's assertion that direct exposure to light could damage milk's nutritional content and adversely affect flavor. Pa- perboard manufacturers have also used smaller sizes to partially subsidize prices in markets where paper faced serious competition from plastic. For example, profit margins dropped to only about two percent on half gallon paperboard blanks, compared to 10 to 15 percent on blanks of smaller sizes. With shortages and steep price hikes for HDPE in 1987 and 1988, adoption of plastic half gallons actually reversed, and paperboard began to regain market share (Modern Packaging, March 1965, February 1967; Paperboard Packaging, October 1965; Plastics World, February 1981, May 1981, August 1988, November 1988; Pulp & Paper, April 1986, October 1987.)
METHODOLOGY
To illustrate the metholology for determining consumer utilities for milk containers, consumer responses were gathered in Fair- banks, Alaska, where one of the authors of this article taught. Al- though Alaska is a somewhat atypical U.S. state, much of the popu- lation of Fairbanks is from elsewhere in the Lower 48. The trade literature cited above provided no evidence of geographical differ- ences in how people perceived relative benefits of the three packag- ing materials. Care was taken to screen out respondents who lived very far outside the city. Besides possible personality differences among those who prefer the relative isolation of rural life in interior
Mark Speece and Douglas L. MacLachlan 49
Alaska, there was the consideration that much housing in rural areas does not have electricity. Strictly speaking, then, results reported here concern Alaska only, but surveys taken in other areas of the U.S. are likely to show similar results.
Conjoint analysis is a method developed by mathematical psy- chologists that has received extensive application in marketing re- search (Green and Srinivasan 1978). The method allows researchers to assess the utilities that people attach to the various attributes or components that compose a product (or other concept). By asking people to indicate relative preference for various bundles of attri- bute combinations, analysts are able to infer the utility people attach to the attribute levels, assuming some rule being used for combin- ing the attribute level partworths. The feature of the method that has inspired so much attention is that easily obtained ordinal judgments of relative preference can be converted into interval-scaled partworths for attribute levels that are comparable across attributes (no matter what the measurement scales of those attributes). The various conjoint data collection and analysis procedures have been extensively examined in terms of their reliability and validity (e-g., Reibstein, Bateson, and Boulding 1988), which has given research- ers confidence in their usefulness. To validly apply conjoint analy- sis, it is important to select attributes that are reasonably representa- tive of those actually used by people in making comparative judgments of the product or concept under investigation.
A pilot questionnaire was administered to 25 shoppers at local grocery stores in Fairbanks when they were observed buying milk. Its purpose was to get an idea of attribute importance so that a lengthy preliminary list of possible package attributes could be pruned. Respondents were asked to rate the importance of a set of nine container attributes on a scale of 1 (not important at all) to 7 (very important), and to list the two most and two least important attributes to them when shopping for each size of milk. The attri- butes were adapted from those used by Stephens (1977) and Stern, Ayres, and Shapanka (1975). Review of the trade literature indi- cated that these attributes represented quite thoroughly factors con- sumers looked at (Chemical Week, July 9, 1966, June 14, 1972; Modern Packaging, March 1965; Paperboard Packaging, October
50 Journal of international Food & Agribusiness Marke!ing
1965; Pulp & Paper, April 1986, July 1986). Respondents were also drawn into discussion about what they thought each attribute in the attribute list meant, and asked to point out any other attributes which did not appear on the list but which they thought were impor- tant in their decision. Attributes were generally clear to people and the only additional attributes mentioned more than once were price and pull date.
Summary statistics of responses are presented in Table 1. Possi- bility of leakage, possibility of breakage, and flavor protection were regarded the most important attributes as measured by mean rat- ings, frequencies that attributes were rated very important (7 on the scale), and frequencies that attributes were cited as most important. Patterns for gallon and half gallon containers were similar. A sec- ond group of attributes was almost as important as shown in the mean ratings: ease of opening, ease of pouring, pourabilityldrip- ping, and recyclability. They did not show up very often in the list of most important attributes, and were cited more often in the list of least important. Light weight and disposability were regarded as least important. Paired t-tests showed that most means of attribute ratings within each of these three groups of attributes were signifi- cantly different (at .10 significance levels) from means in the other two groups, but Scheffk tests of multiple comparisons generally showed significance only between the two lowest rated attributes and the three highest.
Factor analysis was used to combine attributes into relevant di- mensions to reduce the number of attributes to a manageable size for a conjoint analysis study. Such use of factor analysis to reduce the attribute list has been suggested by, among others, MacLachlan, Mulhern, and Shocker (1988). Factor analysis of the nine attributes in the pilot yielded four factors with eigenvalues greater than one. These factors were somewhat difficult to interpret, since patterns were not consistent across gallons and half gallons. Forcing five factors resulted in more consistency by size, and factors with inter- pretable dimensions emerged (Table 2). Therefore, a five-feature conjoint design was used in the survey questionnaire: package de- sign (4 levels), package sturdiness (3 levels), nutrition protection (2 levels), convenience (2 levels), and weight (2 levels). Levels of the five features were based on the variables which loaded on each
Mark Speece and Douglas L MacLuchlan
Half Gallons
MEAN LEVEL RATING
flavor 6.63 leak 6.54 break 6.29 pour 5.83 drip 5.63 recyc 5.29 open 5.21 dispos 4.54 weight 3.71
f laver= flavor protection recyc = recyclability leak = possibility of leakage open = ease of opening break = possibility of breakage dispos. disposability pour = ease of pouring ueight: light weight container drip = pourabilitg/dripping
Kote: The survey asked each respondent to list the two most and two least important attributes, so there are twice as many reeponses as respondents on these questions.
52 Journal of International Food & AgnblLFiness Marketing
TABLE 2. Factor Loadings of Attributes
' ROTATED FACTOR MATRIX: HALF GALLONS
FACTOR 1 FACTOR 2 FACTOR 3 FACTOR 4
flavor leak break drip pour open recyc dispos - weight -
FACTOR 5
ROTATED FACTOR MATRIX: CALLOSS
FACTOR 1 FACTOR 2 FACTOR 3 FACTOR 4 FACTOR 5
flavor 1 eak break drip pour open recyc dispos weight
Note: t indicates factor on which variable primarily loads.
factor in the factor analysis and on verbal descriptions by con- sumers in the pilot of what they thought those variables meant (Ta- ble 3).
A commercial software program Conjoint Designer (Bretton- Clark 1987) was used to determine an appropriate fractional facto- rial design for the study; out of a possible 96 different attribute bundles, sixteen were sufficient for a full orthogonal design. The main reason for presenting attribute bundles rather than two attri- butes at a time was to make the descriptions respondents evaluated somewhat more realistic. Also, some research has shown that this multiple-factor evaluation is slightly more reliable than two-factor
Mark Speece and Douglas L. MacLmchlan
TABLE 3. Package Features and Levels in Conjoint Study
Feature Level
Package design'. 1. easy to open, no dripping 2. easy to open, may drip a little 3. sometimes hard to open, no dripping 4. sometimes hard to open, may drip a little -
Package sturdyness 1. may shatter if dropped 2. may leak if dropped 3. unlikely to break or leak if dropped
Nutrition protection 1. protects nutrition 2. nutritional value of milk may decline over time
Convenience 1. no deposit, no return 2. returnable, requires deposit
Veight (gallon qucstionnaire) 1. weighs about 8 ounces 2. veighs about 3 pounds
evaluation (e.g., Segal1982), and used more often in the marketing research industry (e.g., Wittink and Cattin 1989). The attribute bundles were presented as brief descriptions in the questionnaire, and respondents were asked to rate how closely the description matched their idea of a good milk container for each respective size. In addition, they answered questions rating each container type on how closely it fit their idea of good performance on each feature. Finally, the questionnaire asked which type of container respon- dents preferred to buy milk in, and a few demographic questions.
Questionnaires were distributed in malls which contained grocery stores to people who were observed leaving the grocery stores. Re- spondents were further screened to determine that they actually pur- chased milk regularly. Sixty-five good questionnaires were re- turned. Nearly 55 percent of the respondents were male, just over 45 percent female. One third were in the age category 18 to 23 years, about a quarter in each of the two categories 24 to 30 and 31
54 Jownaf of International Food Br Agribwiness Marketing
to 40, with even distribution over the remaining three higher age categories. The conjoint analysis package Conjoint Analyzer (Bret- ton-Clark 1987) was used to analyze the data.
RESULTS
Table 4 summarizes the relative importances (i.e., pamvorth ranges) of each of the five factors/features. For half gallons, the
TABLE 4. Degree Which Package Has Good Feature and Relative Importance
features nutrition protection and package sturdiness were regarded as most important, at 28 and 25 percent, respectively, followed not very closely by package design. Convenience was least important, at slightly over eight percent. Weight was second least important, though about twice as important as convenience. Relative impor- tance was somewhat more evenly distributed across features for gal- lons, except for convenience, which ranked at zero percent to the digits reported in this program. Package design and package sturdi- ness ranked highest here, both around 27.5 and 25.5 percent, with the other two not far below. One important difference between the results for gallons and half gallons was expected. Weight was rela- tively more important for gallons, where glass bottles were much heavier than half gallon bottles.
Results on the questions asking about degree to which respon- dents felt different container materials met good performance on each feature are also summarized in Table 4. It is clear that glass containers are at a severe disadvantage on most features relative to the other two package types, although they are somewhat less far behind when people are thinking about half gallon containers. How- ever, for the important feature nutrition protection, glass does very well, coming out slightly (but not significantly) ahead of both paper and plastic. Plastic ranked best on all other features for gallon con- tainers, but paper came out slightly ahead on most features for half gallons. (Most of the differences among container types reported in Table 4 are significant at a .05 significance level if the difference is greater than about 1.2 for half gallons, and greater than about 1.6 for gallons.)
Stated preferences were similar to the results in the pilot (Ta- ble 5). Two-thirds of the respondents said they preferred plastic for gallon containers when given a choice. Most of the rest preferred paper. But for half gallon containers, nearly sixty percent preferred paper, while plastic and glass were nearly even among the rest. Stated preferences for half gallon glass containers were slightly more common in the survey than in the pilot, and slightly less com- mon for plastic. (Differences on gallons are not statistically signifi- cant.) In the survey, people had to think about package features and
56 Journal of International Food & Agribuiness Markeling
TABLE 5. Slated Preference for Container Type in Pilot and Survey
Pilot Survey
Package Half I Half Type Gallons Gallons 1 Gallons Gallons
Note: Figures are percent vho prefcr package type.
what they considered important in packages before they stated their preference. These results may indicate that even among those who do prefer plastic for half gallons, preference is somewhat weak if consumers can be encouraged to really evaluate their choice.
Relative utilities of the three packaging materials can be com- puted from these results. The summation to get the utility scores follows Stern, Ayers, and Shapanka (1975), and is written:
where u, is the utility of the ith packaging material, N is the number of attributes, w, is the importance weight of the nth attribute, and g. is the degree to which the ith packaging material contains the nth attribute. In form, this is just a very common attitude model which is widely used in marketing (ag., Wilkie and Pessemier 1973; Srinivasan and Shocker 1973; Parker and Srinivasan 1976). Stover (1978) operationalized a measure of perceived value of various gen- erations of technology with this form.
Since importance weights are typically measured on arbitrary scales, though, the common form of this model does not yield ratio- scaled values, so that adjusting prices by these utilities would not be appropriate. However,-the weights dktermined through conjoint analvsis here are ratio valued. Conioint analvsis allows estimation of iiterval-scaled utilities for differkt levelsf a particular feature
Mark Speece and Douglas L. Machchlan 57
(i.e., partworths of the levels of the feature, which are not the same as the utilities u, in the equation above). From these interval-scaled partworths, ratio-scaled relative importance of a particular feature can be computed by taking the spread in partworths on that feature and dividing by the sum of partworth spreads across all features. These are the relative importances reported above. Care was also taken to obtain ratio-scaled measures of a,. Thus, the summed product was ratio-scaled, and could legitimately be used to compute utility adjusted prices for comparison across packaging materials.
The order in which averages are done in the above equation causes slight changes in the resulting aggregate utility score, so three methods were used. The first takes individual utilities, nor- malizes them to paperboard's utility, then averages over individual utilities. The second method averages individual utilities first, then normalizes the aggregate to paperboard. The third takes the average relative importance and the mean degrees features are contained (from Table 4), so that the summation in the utility equation is done only once on aggregate values. The resulting values are reported in Table 6; the differences are slight. Plastic containers have most util- ity to consumers in gallon sizes, but paperboard hold the edge in half gallons. The table also shows that there is relatively less varia- tion on individual utilities for the container types which have higher utilities; i.e., the preferred type in most cases. There is most vari- ance in package types with the least utility. These results probably indicate that consumers are more sure about how they rate the pack- age types they use, and less sure about the ones they do not use very much.
For gallon containers, individual utility scores were usually con- sistent with the container type respondents said they preferred if given a choice. Seventy percent of respondents gave plastic the highest utility and all but one said they preferred plastic. The rest split fairly evenly, giving either paper or glass the highest utility. Most of them preferred the container which had the highest utility, but thirteen percent chose another one:Half gallons were not quite as consistent; about one third of the respondents said they preferred a different package type than the one to which they gave the highest
58 Joumal of International Food &Agnbuiness Madefing
TABLE 6. Relative Utilities and Variability (Normalized to Paper)
Rav utilities - individual utilities before normal- izing, i.e., sums of relative importance veights times degree feature contained.
utility. However, nearly all of these cases had utilities for the pre- ferred material which ranked second and within about ten percent of the highest. Only one person chose a container type with utility drastically below the highest one. There was simply less spread between utilities of the various packaging materials for half gallons, so that utility scores were not as accurate at predicting package preference.
For adjusting prices, we simply averaged (and slightly rounded to the nearest 0.05) the three utilities in Table 6. Thus, utilities used for gallon containers were 0.65, 1.00, and 1.30 for glass, paper-
Mark Speece and Douglas L. Machchlan 59
board, and plastic, respectively, and 0.70, 1.00, and 0.90, respec- tively, for the half gallon market. Pricing data for paperboard car- tons from 1964 to 1987 were obtained from a paperboard manufacturer. Unit prices in some years for glass and plastic (and paper) containers were available from industry publications (Chern- ical Week, July 9,1966, June 14,1972, March 11,1987, August 3, 1988; Food Processing, October 1974; Modern Packaging, July 1964, March 1965, February 1967, June 1967, September 1967, April 1978; Plastics World, February 1981, May 1981, November 1987, August 1988, November 1988; Paperboard Packaging, Octo- ber 1965; Pulp and Paper, April 1986, July 1986, October 1987), and in Stephens (1977), who covered 1965 through 1975. Price indices for glass beverage bottles and HDPE resin were also avail- able in USDL (various years), so that with the benchmark prices, complete price series could be computed. To get utility adjusted prices, the unit prices in these series for each container type were divided by the utilities for the container type.
Prices for glass half gallon bottles were nearly five times greater than for half gallon paperboard containers in 1964, and rose steadily to over seven times greater by 1987. Glass gallon bottles cost nearly four times as much as gallon paperboard cartons in 1964, and also rose steadily relative to paper. Price sensitivity alone could easily explain the rapid loss of market share to paperboard in the 1960s. However, it is clear that consumers see considerably more utility in paperboard than in glass milk containers, so utility adjusted price differentials probably more accurately reflect relative prices. After adjustment, the price differential was substantially greater.
It would be difficult to maintain that price sensitivity had any- thing to do with the rapid growth of plastic gallon containers, though, unless utility adjusted prices were used. Prices of plastic. gallon jugs did not drop below those of gallon paperboard until the early 1980s, long after plastic had captured most of the gallon mar- ket (Figure 3). But utility adjusted prices for gallon plastic jugs undercut paperboard almost as soon as plastic was introduced. It would also be difficult to see why plastic has not done well in half gallons, since prices relative to paperboard are slightly higher than for gallons, but closely parallel gallon prices. Figure 4, though,
60 Journal of International Food & Agribusiness Marketing
FIGURE 3. Prices of Plastic Relative to Paperboard
Year s Gallons + HaUGallons
shows that while gallon plastic utility adjusted prices have nearly always been lower than paperboard, half gallon plastic utility ad- justed prices have always been higher than paper.
CONCLUSION
We have illustrated a research approach for studying packaging preferences. Results obtained here must be applied with caution if they are to be used for management planning. The sample is rela- tively small, and the population itself may not be representative of other areas in the U.S. Further, the utilities are likely to change over time. Indeed, the utilities for gallon milk containers derived by Speece (1989) from Stevens (1977) and Stern, Ayers, and Sha- panka (1975) show a wider spread, with glass at relatively lower utility and plastic relatively higher. Those utilities were not derived in the same manner as the ones used in this study, but the pattern is
Mark Speece and Douglas L. Machchlan 61
FIGURE 4. Utility Adjusted Prices of Plastic Relative to Paperboard
nevertheless suggestive. Also interesting were slight differences in utilities observed among subgroups in the consumer survey. Older people, who were more likely to have had some first-hand experi- ence with glass gallons, assigned lower utilities to glass gallon bot- tles than did younger people.
Within the limitations mentioned, we believe we have a good evaluation of milk container preferences in Alaska, which are likely similar to other parts of the U.S. Certainly, the results give a more concrete idea about package characteristics consumers think are im- portant, and about how well they think paper or plastic perform on those characteristics. By combining importance with package char- acteristics, relative utilities can be computed, and it becomes clear why plastic has been so successful in gallon sizes but not in half gallons. Plastic is perceived as having more utility than paperboard in gallons, but less in half gallons. By computing utility adjusted prices, we can further see that plastic enjoys a price advantage for
62 Journal of Internarional Food & Agribusiness Marketing
gallons, but paperboard maintains the price advantage in half gal- lons.
This kind of information is essential to manufacturers of compet- ing packaging materials. They need to have a good idea about what consumers perceive to be strengths and weaknesses of alternative package types. Managers need to know which of the package fea- tures are relatively more or less important to consumers. Perhaps most significantly, information on consumer perceptions of overall utility is essential, since consumers are unlikely to adopt new pack- age types unless they perceive them to be an improvement on exist- ing packaging. Utilities can be further used to investigate relative prices, once prices have been adjusted for utility, to get an idea of how value-conscious consumers view prices. The approach pre- sented here obtains all of this information and is relatively easy to apply. It can serve as a guide for further research on consumer perceptions of packaging materials.
LITERATURE CITED
Bretton-Clark (1987), Conjoint Analyzer, Version 2.0. New York: Bretton-Clark. Bretton-Clark (1987), Conjoin! Designer, Version 2.0. New York: Bretton-Clark. Chemical Week (July 9, 1966), "PE Scores in Milk Bottle Battle," 99(2): 51-52. Chemical Week (June 14, 1972), "Milk Bottles: Healthy New Market for Polyeth-
ylene," llO(24): 27. Chemical Week (March 1 1 , 1987), "For HDPE: Rapid Growth, Promising New
Processes," pp. 8-10. Chemical Week (August 3, 1988), "Polethylene Shortage Shifts the Spotlight to
Traditional Materials," pp. 20-23. Food Processing (October 1974), "Returnable Plastics Reduce Container Costs
by 46%," 35(10): 85-86. Green, P.E., & V. Srinivasan (1978), "Conjoint Analysis in Consumer Research:
Issues and Outlook," Journal of Consumer Research, 5: 103-123. Harper, W.J., & C.W. Hall (1976), Dairy Technology and Engineering. West-
port, C T AVI Publishing Company, Inc. Henderson, J.L. (1956), "Market Milk Operations, 1906 versus 1956," Journal
of Dairy Science, 39(6): 812-818. Henderson, J.L. (1971), The Fluid Milk Industry. Westport, a: AUI Publishing
Co., Inc. MacLachlan, D.L., M.G. Mulhern, & A.D. Shocker (1988), "Attribute Selec-
tion and Representation in Conjoint Analysis: Reliability and Validi!y Issues," Proceedings of the Sawtooth Software Conference on Conjoint Analysis, Per-
Mark Speece and Douglas L. MacLuchLn 63
ceptual Mapping, and Computer Interviewing. (June), Ketchum, ID: Sawtooth Software.
Modem Packaging (July 1964), "Milk in Plastics Has Arrived," 37(11): 120- . -- ILL.
Modem Packaging (March 1965), "The Plastics Milk Bottle: Time of Decision," 39(7): 162-169, 293-294.
~ o d e m Packaging (June 1967). "More on Plastic Milk Bottles," 40(10): 178. Modem Packaging (September 1967), "New Pastures for Milk Cartons," 40(13):
136-139. Modem Packaging (April 1978), "Milk Packaging: Still in Transition," 51(8):
37-40. Paperboard Packaging (October 1965), "Paperboard vs. Plastics: A Positive
View." 50(10): 71-73. Parker; B.R., & V. Srinivasan (1976), "A Consumer Preference Approach to the
Planning of Rural Primary Health Care Facilities," Operafionr Research, 24(Sept.-Oct.): 991-1025.
Plastics World (February 1981), " ~ a l f Gallon Milk Bottles: Is there Any Hope for Plastic?" 39(2): 39-43.
Pfasfics World (May 1981), "Paperboard is Inviting Target," 39(5): 18, 24. Plasrics World (November 1987), "Resin Forecasts: Poly-ethylenes," 45(11):
46(8): 35-37. Plastics World (November 1988), "Resin Forecasts: Poly-ethylene," 46(11):
54-57. Pulp and Paper (April 1986), "Liquid Packaging Declines in Milk Markets but
Grows in New End Uses," 60(Apr.): 137-140. Pulp and Paper (July 1986), "Plastic Nightmare," 60(July): 7. Pulp and Paper (October 1987), "Plastic Inroads in Milk Canons Slow,"
61(0ct.): 33. Reibstein, D., J.E.G. Bateson, and W. Boulding (1988). "Conjoint Analysis
Reliability: Empirical Findings." Marketing Science, 7(Summer), 271-286. Roadhouse, C.L., & J.L. Henderson (1950), The Market Milk Industry, Second
Edition. New York: McGraw-Hill. Segal, M.N. (1982), "Reliability of Conjoint Analysis: Contrasting Data Collec-
logy, Multiattribute Technological Forecasting," Technological Forecasiing and Social Change, 20(3): 215-233.
Speece, M.W. (1989), "Pricing in a Multi-Generation New Product Diffusion Model." Presented at TIMS XXIX, Osaka, Japan (July).
Srinivasan, V., & A.D. Shocker (1973), "Estimating the Weights for Multiple
64 Journal of International Food & Agn'business Marketing
Attributes in a Composite Criterion Using Pailwise Judgements," Psychome- trika, 38(Dec.): 473r493.
Stern, M.O., R.U. Ayres, &A. Shapanka (1975), "A Model for Forecasting the Substitution of One Technoloev for Another." Technolom'cal Forecastina ond - ~ocia l Change, 7: 57-79.
- Ste~hens, K.L (1977). Technolo~ical Substitution: A Model for Marker Forecast-
jng. ~ n ~ u b l i s h e d ~ ~ ~ ~he$s, University of Washington. Stover, J.G. (1978), "Use of Decision Modeling for Substitution Analysis: Appli-
cation to Acceptance of New Electricity-Generating Technologies," Techno- logical Forecasting and Social Change, 12(4): 337-351.
U.S. Department of Agriculture, Dairy Division, Agricultural Marketing Service (1987), Packaged Fluid Milk Sales in Federal Milk Order Markets (Report 14, June 1987). Washington, DC: U.S. Government Printing Office.
U.S. Department of Labor, Bureau of Labor Statistics (Various dates, 1974- 1988), Producer Prices and Prices Indexes. (Prior to March 1978: Wholesale Prices and Price Inderes.) Washington, DC: U.S. Government Printing Of- fice.
U.S. Federal Trade Commission (1973). Economic Repon on the Dairy Industry, March 19, 1973. Washington, DC: U.S. Government Printing Office.
Wilkie, W.L., & E.A. Pessemier (1973), "lssues in Marketing's Use of Multi- Attribute Attitude Models," Journal of Marketing Research, 10(Nov.): 428- 441.
Wittink, D.R. & Cattin, P. (1989), "Commercial Use of Conjoint Analysis: An Update," Journal of Marketing, 53(July), 91-96.
