Shopping For Variety in Red Meat, Poultry, and Fish

ABSTRACT - Two definitions are used in this paper to measure "variety" in meat purchases. The first is the number of types of meat, poultry, of fish items purchased and used by a household. The second measure is the amount of dispersion in the quantity of various meat types purchased and used. These measures become the dependent variables in regression analyses of the effects of household composition and income on the consumption of variety. Results indicate that age-sex composition and income of households affect these measures of the amount of variety in the weekly meat, poultry, and fish bundle.

Citation:

Christine J. Hager (1988) ,"Shopping For Variety in Red Meat, Poultry, and Fish", in NA - Advances in Consumer Research Volume 15, eds. Micheal J. Houston, Provo, UT : Association for Consumer Research, Pages: 19-21.

Advances in Consumer Research Volume 15, 1988      Pages 19-21

SHOPPING FOR VARIETY IN RED MEAT, POULTRY, AND FISH

Christine J. Hager

ABSTRACT -

Two definitions are used in this paper to measure "variety" in meat purchases. The first is the number of types of meat, poultry, of fish items purchased and used by a household. The second measure is the amount of dispersion in the quantity of various meat types purchased and used. These measures become the dependent variables in regression analyses of the effects of household composition and income on the consumption of variety. Results indicate that age-sex composition and income of households affect these measures of the amount of variety in the weekly meat, poultry, and fish bundle.

INTRODUCTION

The importance of variety as a characteristic households seek when they shop for food products has received scant attention (McAlister 1979; McAlister and Pessemeir 1982; Shonkweiler, Lee, and Taylor 1985; Theil and Finke 1983). Demand and consumption equations for meat have usually estimated responses to price and income changes (Chavas 1982; Purcell and Raunikar 1971), but have not considered variety. Most studies of food characteristics have focused on nutrients to the exclusion of no-nutrient characteristics, including taste, processing, and variety (Adrian and Daniel 1976; Eastwood, Brooker, and Terry 1985; Ladd and Suvannunt 1976; Wohlgenant 1985). However, certain foods may be purchased because shoppers believe a varied diet is a healthy diet or because their families derive benefits from variety in addition to the nutritional benefits of the foods. Few attempts have been made to consider variety as a characteristic of the food bundle and to analyze household factors that affect the amount of variety purchased in the food bundle.

The present study attempts to operationalize the concept of "variety" and to analyze the consumption of variety in the meats used by households. Two measures of variety are defined and used as dependent variables in regression analyses -- a measure of the number of meat types in the food bundle and a measure of the dispersion in the amount of each type of items included in the bundle. This study analyzes effects of household income and composition on these two measures of "variety" as contained in a bundle of red meat, poultry, and fish items purchased and used by a household in a week.

DATA

The data are from the Spring 1977 household phase of the 1977-78 Nationwide Food Consumption Survey. The survey collected information about foods used by households during a 7 day period in Spring 1977. A 15 digit code was assigned to each item. The code provides information about the type of food item, the specific cut of the item, and further detail about the item such as whether it contained a bone and the type of processing (fresh, frozen, smoded, canned, etc.).

The data contain specific information about each item the household used during the week. For this study red meat, poultry, and fish items are grouped by type of meat as follows:

BACON = bacon, sausage, luncheon meats, franks

PORK = all pork items not included in BACON

POULTRY = all poultry items not included in BACON or mixed with other meats

BLOIN = beef from the loin or rib portions

RLIVER = liver and other organ meats from red meats = all fish except shellfish

FISH = shellfish

MISC = ground beef, chuck, stew beef, meat mixtures, and all other cuts not included above

The individual red meat, poultry, and fish items purchased and used by a household during a week are aggregated into a bundle of such items. When households purchase a bundle of individual items, they are also purchasing the characteristics of this bundle, including variety, nutrients, and non-nutritional services (deboning, processing, butchering into smaller cuts, etc.).

A total of 1,850 households having complete income and demographic information on their record reported purchasing and using 17,660 items during the survey week.

Two definitions of variety are used for the study. Variety is first defined as the number of different groups of red meat, poultry, and fish items chosen by the households during the survey week. Households may use items from as many as eight groups, but the average number of groups from which items were selected was 3.81, ranging from 1 to 7 groups (standard deviation = 1.29). Because variety is an elusive characteristic of meat purchases, a second measure of variety is used to reflect the variation in the quantity of meat eaten from each group. This second measure is the standard deviation in the number of pounds of each of the categories of items used by a household (DISPERSION).

The households in this study used an average of 6.66 different items for an average quantity of 12 pounds of meat products in the week, mainly from the bacon, poultry, pork, and miscellaneous categories (table 1). Households used amounts of each type of item within two pounds of the mean for the group.

Table 1 indicates the average pounds of each type of item, the standard deviation in the pounds across households, and the standard deviation (dispersion) in the pounds across the eight items. The average dispersion in the pounds of each type of item used is 2.06, with one standard deviation equalling about 1.56 pounds.

MODEL

The approach used is a simple regression analysis of a single equation consumption model that indicates the effects of household characteristics and income on the amount of variety in the bundle, as measured by number of pounds purchased and used of each type of item:

Y = f(A,I)

where

Y = dependent variable (number of types of items or dispersion among the pounds of items)

A = set of household characteristics

I = set of income variables

TABLE 1

MEANS AND STANDARD DEVIATIONS FOR POUNDS OF ITEMS OF DIFFERENT TYPES

TABLE 2

DEFINITIONS OF HOUSEHOLD SIZE VARIABLES

Household size if defined in terms of the number of members of given ages and sexes (table 2). Specifying household composition in terms of numbers of individuals in given age and sex groups is more consistent with the reporting of nutrient levels and other consumption information than is the life cycle approach. The parameters from the regression can then be interpreted as the effect on the quantity of the characteristic contained in the bundle, ceteris paribus, of adding one more individual in the specific age and sex group (Hager 1985).

Another variable represents an adjustment for the number of meals eaten away from home by members of the household that are not offset by additional meals eaten by guests (MADJ). If each household member consumes three meals per day at home, the total number of meals served to that member out of household supplies would be 21. The meal adjustment is defined as the difference between the total number of meals actually served out of household supplies minus 21 times the size of the household. No adjustment is made for the number of meals served. Household members may desire less variety at home if they are eating more meals away from home or they may want more variety at home if they are eating more meals away from home and these meals consist of fast food hamburgers or chicken.

Other socioeconomic variables are included because they are expected to serve as proxy variables for tastes and the value of time of the food shopper. Employed shoppers (FSEMP = 1) may have higher opportunity costs of time for food preparation, indicating less time to plan variety into their meals. Education (ED1 = 1 if the food shopper has at least a high school diploma) may also reflect opportunity costs of time. Ethnic differences between whites and nonwhites (RACE = 1 if the respondent is nonwhite) may indicate different preferences for a variety of meats.

Income indicates the ability to purchase variety, but a variety of meats could be purchased because different cuts are on sale. Income is measured as monthly after tax income divided by 1,000 (YAT1), this measure squared (YAT1SQ), and the bonus value of food stamps. The quadratic measure is included because it is likely that households do not respond to increases in income in a linear manner. That is, at some income level, they will no longer be willing to spend an additional dollar for more variety.

RESULTS

Table 3 indicates the results for the single equation models for each of the two measures of variety. Household size and composition variables influence the amount of each measure of variety, as expected; but as the number of persons in each age-sex group increases, the amount of variety in the bundle also increases. These effects differ by age and sex group.

Additional males ages 34 to 64 increase the number of groups of items selected by .38 and additional females in this age group increase the number of groups by about the same amount. However, a household with more males ages 34 to 64 has more dispersion in the number of pounds of items in each category, indicating that more meat items may be purchased from fewer categories rather than evenly divided among categories.

The meal adjustment variable has a small, significant, and positive effect on the variety in the bundle. Additional guest meals lead to an increase in the number of types of items served and to a slightly wider difference in the number of pounds of items purchased in each category. An alternative explanation is that as more meals are eaten away from home, less variety is sought in meals at home.

The variable for race has a significant, positive effect on variety suggesting that nonwhite households purchase a greater number of items from different groups than whites. This result may be due to an increased use of red meat, poultry, and fish items. Nonwhite food shoppers may be more responsive to sales or have a greater preference for a variety of meats used in a week.

Households with an employed food shopper may have less dispersion in their food supplies because their members frequently eat away from home or eat more roasts and leftovers from meats prepared in large quantities. Education of the food shopper does not affect the number of types of items. Neither employment nor education of the food shopper affects the dispersion in the number of pounds purchased in the selected groups. Income and the food stamp bonus have small, significant, and positive effects on the number of types of items used, but do not significantly affect the dispersin in the number of pounds of items. Therefore, it seems that as incomes increase, diversity in types of items increase even though households may not significantly change the relative quantities they consume among types of meat items.

TABLE 3

REGRESSION PARAMETERS FOR MEASURES OF VARIETY

CONCLUSIONS

These results are not conclusive. Variety is an elusive notion. One family may consider their diets to be varied if they have beef prepared five different ways during a week, or have beef every day one week and chicken every day the next week. Other families may use products from each group each week -- chicken on Monday, ground beef on Tuesday, etc.. Another family might have luncheon meats readily available, but use a wide variety of other meats for dinner meals. Still other families do not consider their diets varied unless they include eggs, cheese, beans, and various vegetables but eat mainly chicken, pork, or beef as meats. The measures of variety used in this paper are arbitrary definitions and rely on an arbitrary classification of foods into groups.

The model presented is a very simple one and does not include a measure of "price" as a model of demand would. It is not the price of any one type of item that would be of interest, but rather the relative expense of choosing items from more than one category. However, how to measure this "price" is not clear.

The amount of variation explained by the equations is small but significant. The model does add to the limited amount of information available about variety by illustrating responses of two measures of variety to the same socioeconomic variables. The results imply that the family characteristics and income only explain a small amount of the variation in the number of types of items and the dispersion in pounds of items of different types. Preferences for variety may not be easily captured by the specified socioeconomic variables. In addition, these results emphasize the difficulties in operationalizing the concept of "variety" in diets, defining the implicit value of variety in the food bundle, and estimating the influence of socioeconomic characteristics on the choice of frequently purchased items.

REFERENCES

Adrian, John and Raymond Daniel (1976), "Impact of Socioeconomic Factors on the Consumption of Selected Food Nutrients in the United States," Amer. J. Agric. Econ., 50, 31-38.

Chavas, lean-Paul (1983), "Structural Changes in the Demand for Meat," Amer. J. Agric. Econ., 65, 148-53.

Eastwood, David B., John R. Brooker, and Danny E. Terry ( 1986), "Household Nutrient Demand: Use of Characteristics Theory and a Common Attribute Model," South. J. of Agric. Econ., 235-45.

Hager, Christine J. (1985), "Demand for Nutrient and Non-nutrient Components in Household Purchases of Red Meat, Poultry, and Fish Products Using a Hedonic Approach," unpublished dissertation, Department of Economics and Business, N.C. State Univ., Raleigh, N.C..

Ladd, George and Veraphol Suvannunt (1976), " A Model of Consumer Characteristics," Amer. J. of Agric. Econ., 58, 504-10.

McAlister, Leigh (1979), " Choosing Multiple Items from a Product Class," J. of Cons. Res., 6 (December), 213-24.

McAlister, Leigh and Edgar Pessemeir (1982), "Variety-Seeking Behavior An Interdisciplinary Review." J. of Cons. Res., 9 (December), 311-22.

Purcell, Joseph C. and Roben Raunikar (1971), " Price Elasticities from Panel Data: Meat, Poultry, and Fish," Amer. J. of Agric. Econ., 51, 216-21.

Shonkweiler, J.S., J. Lee, and T.G. Taylor (1985), "The Significance of a Varied Diet: A Household Production Function Approach," paper presented at the annual meeting of the American Agricultural Economics Association, Ames, Iowa.

Theil, Henri and Renate Finke (1983), " The Consumer's Demand for Diversity," Eur. Econ. Rev., 23, 395-400.

Wohlgenant, Michael (1982), " Structural Changes in Demand for Meats: Tastes and Quality Changes," paper presented at the annual meeting of the American Agricultural Economics Association, Utah State University, Logan.

----------------------------------------