QUESTION

17

Food Waste Behavior in an Urban Population Gail G. Harrison, William L. Rathje, and Wilson W. Hughes

In the previous profile, we gained some idea of the range of applications of garbology. This selection is an example of such study. Excess food waste is a serious problem, partic- ularly in the global context of widespread hunger, the pop- ulation explosion, and dwindling environmental resources.

Understanding patterns offood use and waste, how- ever, is not a simple task; researchers cannot get reliable data by just asking members of households. On the one hand, people simply do not know. They may answer a ques- tion about food consumption, but as we see in this article, they can befarfrom accurate. On the other hand, respon- dents to a questionnaire might give the answers they think the researcher wants to hear or the answers they think are expected or appropriate. For example, they may under- report their consumption of beer (orfatteningfoods) be- cause of our image of what is acceptable and appropriate.

Although archaeology developed as a means of study- ing the past—and it still is that—it is becoming an impor- tant tool for understanding the problems of the present, and as we will see in the next reading, developing new tools for the future.

As you read this selection, ask yourself the following questions:

El How might being told you are part of a study offood waste affect your behavior?

El Do you think respondents to a survey questionnaire can accurately estimate the amount offood wasted in their household?

LI What is meant by the term nonreactive measure?

El What proportion (percent) of household foods were wasted in Tucson in 1973 and 1974?

El Would you expect less waste of beef during a beef shortage than during times of plenty? Why?

The following terms discussed in this selection are in- cluded in the Glossary at the back of the book: demography epidemiological methods garbology nonreactive measure of behavior sample waste behavior

Reprinted with permission, Journal of Nutrition Education, vol. 7 (1):13-16,1975, Society for Nutrition Education.

09

Qrowing awareness of the finite limits of natural re- sources under the pressure of an exploding population has made it necessary to look at human utilization of food resources in a new light. The concept of effi- ciency—ecological and economic—has assumed a new priority in nutrition policy and planning. At the household level, economic inflation has made efficient use of food resources more obviously important to more consumers than it has been in the past.

Recent analyses of the U.S. food production sys- tem',' have made it clear that food production in this country is extremely energy-intensive, and that the U.S. food system is reaching the point at which further investments of energy-intensive technology may pro- duce only marginal increments in output. Notably ab- sent from such analyses, however, is an evaluation of the extent, nature, and effects of food waste. No doubt some waste of food is inevitable in any system of pro- duction, distribution, and consumption, but little is known about how much waste of food takes place, why, or how much might be avoided.

Food waste in the field and in storage and trans- portation has been recognized as a significant factor in affecting the availability of food supplies.' It has been estimated that up to 40% of the total grain crop in some areas of the developing world may be lost through spoilage or other damage in the field, in storage, and in handling and processing. Opinion varies as to the po- tential for reducing such losses.4

Food waste at the household or consumer level has been studied even less. The fact that household food waste in industrialized countries is substantial has been often remarked upon but seldom documented. The U.S. Department of Agriculture, which conducts household food consumption surveys in the United States, has long recognized the need for reliable data on food waste. In the late 1950s, USDA undertook some studies of household food waste using records of weighed food waste kept by volunteer respondents.-' These studies utilized small, nonrepresentative sam- ples, and the authors noted that the behavior of the re- spondents was changed by participation in the study. Even so, caloric loss from waste of household food supplies in these studies ranged from 7 to 10% of total calories.'

A problem in studying food waste is that the con- cept of waste is fraught with moral implications in our culture. Few Americans like to admit that they unnec- essarily waste food, and mere participation in a study of waste behavior is sure to bias results. What is needed, then, is a nonreactive measure—a means of estimating food waste which does not affect the be- havior of the subjects! We propose that the methods of archaeology may be useful in this context.

HOUSEHOLD REFUSE AS A NONREACTIVE MEASURE OF BEHAVIOR

The Garbage Project of the University of Arizona has been studying household refuse in Tucson, Ariz., for two years. The project is archaeological in background, theory, and method. Archaeologists have traditionally studied refuse and the remains of material culture in order to make inferences about ancient civilizations— their ways of life, social structures, and utilization of the environment. The Garbage Project is based on the assumption that the methods and theory of archaeol- ogy may offer useful perspectives for dealing with con- temporary problems of resource utilization.'

The project is accumulating data on a wide variety of resource management behaviors including recycling behavior and purchase of food, drugs, household and personal sanitation items, and other consumables. As a method for studying food utilization patterns and waste behavior, the study of household refuse offers two significant advantages.

First, it is a nonreactive measure of behavior. What goes into the trash can is evidence of behavior which has already occurred. It is the evidence of what people did, not what they think they did, what they think they should have done, or what they think the interviewer thinks they should have done. In this way, the study of household refuse differs from accepted methods of col- lecting data on household-level food consumption pat- terns,9"0 all of which suffer from problems of reactiv- ity—distortion of the behavior itself or the recall of the behavior.

The study of household refuse has its own, but dif- ferent, limitations as a measure of food utilization pat- terns. In no way can the evidence of food input to the household, as reflected by packaging or other items in the garbage, be used as a measure of nutritional ade- quacy or of quantitative consumption of food by the in- dividual household. Garbage disposals, meals eaten away from home, feeding of leftover food to household pets, fireplaces, compost piles, and recycling of con- tainers all introduce biases into the data acquired from the trash can. However, these biases all operate in one direction—they decrease the amount of refuse. Thus garbage data can confidently be interpreted as repre- senting minimum levels of household food utilization and waste. On this basis, population segments can be compared and changes over time observed.

A second major advantage to the study of house- hold refuse is that it is inexpensive, relatively easy to do, and requires no time or active cooperation on the part of the subjects. The logistics of a study of house- hold refuse should not be minimized (The Garbage Project requires the efforts of a full-time field supervi-

110

FOOD WASTE BEHAVIOR IN AN URBAN POPULATION 111

sor, even at present sample size), but compared to other methods of monitoring food consumption and nutritional behavior, to which the study of refuse may offer a supplement, the study of household refuse is relatively simple. Data collection can be accomplished by workers with relatively little previous training; and there is little need for special equipment or facilities. This is a major departure from traditional epidemio- logical methods which usually demand a high level of subject input)' As a result, household refuse may be studied in a community on an ongoing basis or at fre- quent intervals in order to detect short-range changes in food utilization behavior.

METHODOLOGY

The Sample

The city of Tucson is an urban community of slightly under 450,000 inhabitants located in southern Arizona. It is characterized by rapid growth in population. The two major ethnic groups are Anglos (whites) and Mexican-Americans, with the latter comprising 27.1 percent of the population in 1973; the proportion of el- derly individuals is relatively high, with 12% of the population aged 65 or over. 12

The sampling unit for The Garbage Project was the census tract. Tucson's 66 urban census tracts were grouped into seven clusters derived from 1970 federal census demographic and housing characteristics. Fac- tor analysis was used to derive groups of significantly associated census variables, and cluster analysis was then used to order census tracts into clusters based on their association with these derived factors of census variables." Data from 13 census tracts in 1973 and 19 in 1974, drawn to be representative of the seven census tract clusters identified by statistical analysis of the data, form the basis for this report.

Data Collection

Refuse was collected for the project by Tucson Sanita- tion Department personnel from two randomly se- lected households within each sample census tract, bi- weekly in 1973 and weekly in 1974. Refuse was collected for a four-month period (February through May) in 1973 again for the same period in 1974. Ad- dresses were not recorded, in order to protect the pri- vacy and anonymity of sampled households. Specific households were not followed over time; that is, a new random selection of households was done each time refuse was collected. Data from all collections in a

given census tract were pooled; thus data analysis is based on the census tract as the unit sampled. Total refuse studied includes the equivalent of that from 222 households in 1973 and 350 in 1974. Households were not informed that their garbage was being studied, although there was local newspaper, radio, and tele- vision publicity on the project at frequent intervals with emphasis on procedures taken to protect the anonymity of sampled households. Thus far commu- nity reaction to the project has been overwhelmingly supportive.

Fifty student volunteers sorted, coded, and re- corded the items in the refuse working at tables pro- vided in the Sanitation Department maintenance yard. After sorting and recording, all items in the refuse were returned to the Sanitation Department for de- posit in the sanitary landfill. While the students were not paid for their participation in the project, they had the option of receiving academic credit for archaeolog- ical field experience, since they gained experience with the methods and theory of field archaeology while working on the project. Student workers were pro- vided with lab coats, surgical masks, and gloves, and were given appropriate immunizations. In almost three years of the project's operation, there have been no illnesses attributable to garbage work.

Items found in the refuse were sorted into 133 cat- egories of food, drugs, personal and household sanita- tion products, amusement and entertainment items, communications, and pet-related materials. For each item, the following information was recorded onto precoded forms: Item code; type (e.g., "ground chuck" as a type of "beef"); weight, as derived from labeling; cost; material composition of the container; brand; and weight of any waste. Fifty-two of the category codes referred to food items.

Waste was defined as any once-edible food item except for chunks of meat fat. Bone was not included, nor were eggshells, banana or citrus peel, or other plant parts not usually deemed edible. Food waste was further classified into two categories: straight waste of a significant quantity of an item (for example, a whole uncooked steak, half a loaf of bread, several tortillas), and plate scrapings, which represent edible food but which occur in quantities of less than one ounce or are the unidentifiable remains of cooked dishes. Potato peels were classified separately, and are not included in "straight waste" for purposes of this paper. It is our guess (yet to be investigated) that "straight waste" may be more susceptible to directed change than is the type of waste we have classified as "plate scrapings."

For purposes of this report, the total weight of a given food item coming into sampled households, as de- rived from labeling on associated packaging materials

112 ARCHAEOLOGY

which are discarded into the trash can, is termed "input" of that food item. It must be kept in mind that these "input" figures are minimal, and their deviation from actual household food utilization of a type of food item is variable depending on the characteristics of the given households sampled.

RESULTS AND DISCUSSION

The following data summarize the evidence of food utilization and waste patterns for the entire sample for the time period specified. (Analysis of the data accord- ing to the socioeconomic characteristics of the individ- ual census tracts is presented elsewhere. 14)

1. The refuse analyzed showed that sampled households waste a significant proportion of their food resources. In 1973, 9.7% of the total food input, by weight, was wasted; in 1974, 8.9% was wasted. (The downward trend was not statistically significant.) Actual waste, of course, was higher since 21.3% of the house- holds in sampled census tracts have garbage disposals in good working order" and probably grind up a great deal of their food waste. We are currently undertaking a study which will allow us to estimate the effect of differential use of garbage disposals on the food waste found in garbage cans. These data on waste do not in- clude milk or other beverages, since beverage waste usually goes down the drain; thus, weight of beverages including milk was elimi- nated from the input figures for calculation of the above percentages.

2. In 1973, straight waste accounted for 55.3% of the food waste and in 1974 it totaled 60.6%. (The change is statistically significant at p <.001 using the difference-of-proportions test described by Blalock.15) Thus although the percentage of total food wasted remained stable from 1973 to 1974, the percentage of straight waste versus "plate scrapings" in- creased significantly.

3. There were some changes between 1973 and 1974 in evidence of utilization and waste of spe- cific food groups (see Tables 1 and 2). The total input of meat, poultry, and fish was signifi- cantly smaller in 1974 than in 1973 (normalized to the same sample size). The percentage of these animal protein foods which was wasted (total waste/ total evidence of input, by weight) showed a sharp and statistically significant drop from 12% in 1973 to less than 4% in 1974, mainly due to a decline in the rate of waste of

beef from 9% in 1973 to 3% in 1974. We find this interesting for two reasons. One is that the high 9% waste of beef occurred during the beef shortage in the spring of 1973. It is possible that during the shortage consumers were overbuy- ing or purchasing unfamiliar cuts or quantities which could not be used efficiently. The change in beef waste is also interesting since there was front-page local newspaper coverage of The Garbage Project, reporting the high level of beef waste (and only beef was mentioned) just at the start of the 1974 data collection period. We don't know whether the publicity had any ef- fect on waste behavior but believe that con- trolled investigations should be carried out to determine whether heightened awareness of waste behavior could have any effect on actual behavior.

Vegetable input decreased between 1973 and 1974 (again, normalized to the same sam- ple size), but vegetable and fruit waste in- creased. Waste of fresh vegetables accounted for most of the increase. In both years, vegetable and fruit waste made up a larger percentage of straight waste than of the evidence of house- hold input of food. Input of grain products in- creased from 1973 to 1974, but proportional waste of grain products decreased. In both years, grain products made up a larger percent- age of straight waste than of evidence of house- hold input, the waste being for the most part due to waste of bread.

Sweets and packaged foods in both years made up a smaller percentage of straight waste

Table 1 ITEM PERCENTAGE OF TOTAL HOUSEHOLD INPUT EVIDENCE AND WASTE

Item

1973 1974 %of total input

evidence

%of waste-

excluding leftovers

%of total input

evidence

%of waste-

excluding leftovers

Selected protein foods* 19.56 21.74 18.50 11.84

Vegetables 24.40 34.77 19.85 38.62 Fruits 13.64 14.25 15.26 17.26 Grain

products 11.23 14.68 14.8 15.8 Packaged

goods 4.53 4.28 7.41 5.89 Sugar and

sweets 10.10 5.74 9.72 6.55 Other 16.54 4.64 14.46 14.04

* Meat, fish, poultry, eggs, cheese, and nuts.

Table 2 PERCENT OF FOOD ITEMS WASTED'

Percent of item wasted

Item 1973 1974

Selected protein foods (meat, fish, poultry, cheese, and nuts) 12.09

344**

Vegetables Fruits

7.65 5.61

10.47** 6.09**

Grain products (excluding pies, cakes, and other sweet pastries) 7.02 5.73

Packaged foods (TV dinners, take-out meals, packaged soups, stews, and sauces) 4.96 4.28

Baby foods 3.01 2.42

Fats and oils 1.39 1.08 Dairy (excluding liquid milk) .92 .73

Spices .77 4.49 Dips, whips 4.07 1.54 Sugar and sweets (including sweet pastries) 3.04 3.63

* Waste (weight) as percent of total input (weight). ** Significantly different from 1973 value at p <.05.

FOOD WASTE BEHAVIOR IN AN URBAN POPULATION 113

than of household input. Perhaps the most re- markable change in input occurred in packaged and convenience foods: TV dinners, take-out meals, canned stews, soups, and sauces. Evi- dence of household input of these items in- creased by over 30% between 1973 and 1974. The only explanation we can offer is to point out that the percentage of households in Ari- zona in which two persons held jobs increased sharply in the same period from 14% in No- vember 1973 to 21% in March 1974. With more households with two adults in the labor force, the consumption of convenience foods might be expected to rise.

4. The cost of the food waste we observed is high. Extrapolating average household waste (total food waste, divided by the number of house- hold equivalents in the sample) over a full year and figuring at June 1974 prices, Tucson's an- nual food waste bill may run between 9 and 11 million dollars. For an average household over a year, the cost of waste was between $80 and $100 of edible food (see Table 3). The biggest contributors to the cost of waste were beef and other meats (in spite of the decline in waste, beef waste is expensive), cheese, fresh vegetables and fruits, take-out meals, bread, and pastry.

Extrapolating from our data to the esti- mated 110,000 households in Tucson, we esti- mate that Tucson was likely to throw out 9,538 tons of edible food in 1974. It may be easier to grasp the significance of this waste if we focus

on one item. The average sample household threw away 1.5 ounces of meat, fish or poultry (straight waste) in each garbage collection. That comes to 5.1 tons each time the garbage is col- lected in Tucson, which is twice a week. Using 1965 USDA data, we can estimate that a two- person urban household may consume about 9.4 pounds of meat, poultry, and fish each week. 16 Tucson's waste in one week would pro- vide a week's worth of meat, poultry or fish for over 2000 such households or a year's worth for 42 two-person households.

5. The quantitative estimates of food input to households derived from packaging materials in the garbage are similar to the quantitative es- timates of food consumption for similar house- holds achieved by the USDA household food consumption surveys. 16 If we extrapolate for a year from the evidence of food input by weight in the average Garbage Project sample house- hold, we estimate that the food input in our sample averaged 1.069 tons of food per house- hold in 1973 and .9763 ton in 1974. The median household size in the census tracts in our sam- ple is two persons. 12 If we add together the quantitative estimates for all food categories for the two-person urban household in the Spring,

Table 3 AN EXTRAPOLATION OF THE COST OF WASTE/HOUSEHOLD/YEAR'

1973 1974

Beef $20.80 $ 5.20 Other meat 4.58 5.10 Poultry 1.98 1.45 Cheese 3.11 3.86 Fresh vegetables 11.32 12.06 Canned vegetables 1.80 1.25 Frozen vegetables 1.29 .95 Fresh fruit 6.18 7.34 TV dinners .82 1.01 Take-out meals 4.68 7.90 Soups. stews, etc. .39 .31 Bread 5.12 4.21 Noodles .24 1.58 Chips, crackers 1.54 1.28 Candy 1.36 .81 Pastry 5.93 6.83 Baby food .50 .27 Potato peels 2.18 .92

Total $73.82 $62.33 Total with plate scrapings: 99.14 82.91

Plate scrapings at 340/lb. 25.31 20.58

Calculated by multiplying average quantities wasted per garbage pickup times the number of pickups a year (104) times current (7 June 1974) averaged Tucson prices.

114

ARCHAEOLOGY

Table 4 PERCENTAGE OF TOTAL HOUSEHOLD FOOD INPUT

Food groups as percent of

household food consumption

by weight, USDA, urban households

Spring 1965

Food groups as percent of total evidence for food input, by weight, Garbage Project

1973 1974

Selected protein foods

26.1

19.6 18.5

Vegetables

25.3

24.4 19.8

Fruits

18.3

13.6 15.3

Grain

11.6

11.2 14.8

Sugar and sweets

6.0 10.1 9.7

* Meat, fish, poultry, eggs, cheese, and nuts.

1965 USDA household food consumption sur- vey," we get a total of .9752 ton of food-ex- tremely close to the estimates obtained in our sample by observation of household refuse.

Although the categories of food are not strictly comparable mall details, it is interesting to compare Garbage Project data for the two years with the percentage of total household food consumption obtained in the 1965 USDA survey for urban households" (see Table 4). To the extent that the comparison can be made, it appears that people in Tucson in 1973 and 1974 were consuming somewhat less of some animal protein foods, less fruit, and more gram prod- ucts, sweets, and fats and oils than the USDA sample was in 1965. The overall similarity of the food input pattern shown in Table 4 with the independent USDA household food con- sumption data is an encouraging indication of the validity of refuse data as an index .of food utilization patterns on the community level.

CONCLUSIONS

These preliminary data show that the study of house- hold refuse offers a simple, inexpensive, and nonreac- five means of monitoring food utilization and waste behavior on the community level. The data accumu-

lated to date clearly indicate that food waste is a sig- nificant factor in food resource utilization and should be seriously considered by nutrition planners and educators.

REFERENCES

1. Piinentel, D., Hurd, L. E., Billotti, A. C., Forster, M. J., Oka, I. N., Sholes, 0. D., and Whitman, R. J. 1973. Food production and the energy crisis, Science, 182:433.

2. Steinhart, J. S., and Steinhart, C. E. 1974. Energy use in the U.S. food system, Science, 184:307.

3. Woodham, A. A. 1971. The world protein shortage: pre- vention and cure, World Rev. Nutr. & Dietet., 13:1.

4. Berg, A. 1973. The Nutrition Factor: Its Role in National De- velopment. Washington, D.C.: The Brookings Institution.

5. Adelson, S. F., Asp, E., and Noble, I. 1961. Household records of foods used and discarded, J. Am. Dietet. Assn., 39:578.

6. Adelson, S. F., Delaney,!., Miller, C., and Noble, I.T. 1963. Discard of edible food in households, J. Home Econ., 55:633.

7. Webb, E. J., Campbell, D. T., Schwartz, R. D., and Sechrist, L. 1966. Unobtrusive Measures: Nonreactive Re- search in the Social Sciences. Chicago: Rand McNally.

8. Rathje, W. L. 1974. The Garbage Project: A new way of looking at the problems of archaeology, Archaeology, 27:236.

9. Young, C. M., and Trulson, M. F. 1960. Methodology for dietary studies in epidemiological surveys II. Strength and weaknesses of existing methods, Am. I. PubI. Health, 50:83.

10. Pekkarinen, M. 1970. Methodology in the collection of food consumption data, World Rev. Nutr. & Dietet., 12:145.

11. Marr, J. W. 1971. Individual dietary surveys: purposes and methods, World Rev. Nutr. & Dietet., 13:105.

12. Bal, D. G., O'Hora, J. H., and Porter, B. W. 1974. Pima County ECHO Report, Tucson, Arizona, Pima County Health Department.

13. Tyron, R. C., and Bailey, D. 1970. Cluster Analysis. New York: McGraw-Hill.

14. Harrison, G. G., Rathje, W. L., and Hughes, W. W. 1974. Socioeconomic correlates of food consumption and waste behavior: the Garbage Project. Paper pre- sented at the annual meeting of the American Public Health Association, New Orleans, La., Oct. 21, 1974 (unpublished).

15. Blalock, H. M. 1960. Social Statistics. New York: McGraw-Hill.

16. Dietary Levels of Households in the United States, Spring, 1965: Household Food Consumption Survey, 1965-1966, Re- port No. 6, USDA/ARS, Washington, D.C.: U.S. Depart- ment of Agriculture.