The Roles of Beliefs, Information, and Convenience. in the American Diet

Transcription

1 The Roles of Beliefs, Inforation, and Convenience in the Aerican Diet Selected Paper Presented at the AAEA Annual Meeting 2002 Long Beach, July 28 th -31 st Lisa Mancino PhD Candidate University of Minnesota Departent of Applied Econoics Copyright 2002 by Lisa Mancino. All right reserved. Readers y ake verbati copies of this docuent for non-coercial purposes by any eans, provided that this copyright notice appears on all such copies.

2 Introduction Over the past two decades there has been a proliferation in the aount of scientific inforation substantiating a link between eating a good diet and aintaining good health. Fro this has sprung an increasing nuber of national capaigns intended to educate Aericans on the iportance of a ore healthful diet. The success of these capaigns have likely contributed to the findings of a 2001 Food Marketing Institute (FMI) study, where nearly 60 percent of sapled shoppers reported their grocery purchases were strongly affected by soe health concern and 76 percent felt eating well was a better way to anage their health than edication (FMI 2001a, p.7). Not surprisingly, then, the last decade has seen an increase in sales of organic, light, low-fat, and low-salt foods. In the natural food industry 1 alone, there has been an average growth rate of percent per year. In coparison, the conventional food arket has reported average growth rates of three to five percent per year (Richan, 1999). Yet while Aericans are ore aware of the links between diet and health, the rising nubers of dual incoe and single parent failies have ade the average faily ore pressed for tie and ore willing to pay for greater convenience. Since 1960, the percentage of woen working full tie has increased over 60 percent. Meanwhile, there has also been an escalation in the aount of dining out, fast food eals, and hoe eal replaceent (FMI 2001b; Senauer et al. 1991, p. 5). As of 1995, Aericans were eating nearly 30 percent of their eals away fro hoe, an over 80 percent increase fro 1977 (Lin, Guthrie, and Frazao, 1999). A potential proble with this escalation is that overall Aericans deonstrate they know very little about the nutrient content of food prepared away fro hoe and regularly underestiate the fat and caloric content of such eals (Kennedy et al, 1999). 1 Natural foods are defined as being produced with inial processing, free of artificial ingredients, preservatives, and cheicals, 2

3 Another potential health risk fro the increased deand for convenient foods and foods prepared away fro hoe is that they tend to be denser in calories. This is not to say that food ade fro scratch cannot be high in fat, but it is often difficult to find healthful foods that are convenient and flavorful. The lack of healthful alternatives at fast food and take out restaurants was cited as the strongest ipedient to having a ore healthful diet (FMI 2001a, p. 19). Not surprisingly, while Aericans have reduced their overall fat consuption, they have ade less progress reducing the aount of fat in food consued away fro hoe. In 1995, Aericans derived an average of 31.5 percent of their calories fro fat when eating food prepared at hoe. However, they received nearly 41 percent of their calories fro fat when eating food prepared away fro hoe (Lin, Guthrie, and Frazao). The increased consuption of ore convenient foods ay be a significant contributor to the draatic increases in both the incidence and level of obesity aong Aericans. As of 2000, it was estiated that 56.4% of all Aericans were overweight and one in five US adults were obese (Associated Press, 2001). Consequently, there has been a parallel rise in the incidence of diseases highly correlated with poor nutrition and over consuption: cancer, strokes, heart disease and diabetes. With total econoic costs of these four diseases conservatively estiated at $70.9 billion per year, this ay prove to be a costly trend (Frazao 1999, p. 23). The surgeon general has even predicted that obesity ay soon overtake tobacco as the priary cause of preventable deaths (News Service, 2001) Such conflicting trends highlight a striking paradox: While Aericans clai to be eating better and iproving their understanding of diet and health, they are getting heavier and increasing their risk of suffering fro diet related illnesses. The cause of this paradox is unclear. It ay be because Aericans just eat too uch of everything; there ay be a clear division between the people who eat poorly and 3

4 the people who eat healthfully; or alternatively, it ay be that individuals usually try to incorporate their beliefs about healthy eating into their food choices, but due to tie constraints, stress, and the need for convenience, ust soeties forego good intentions for ore iediate gratification. An inconsistency between what an individual chooses to do and what she believes she should do is an exaple of a tie-inconsistent choice. More forally, a tie-inconsistent choice is one where an individual akes a choice that would not have been ade under a ore detached, objective perspective. For exaple, one ay clai that for toorrow s dinner, she would prefer a healthy salad at 6:30 to a haburger at 5:35. However, when evaluating this sae decision for tonight s dinner while feeling hungry and pressed for tie, the sae individual ay decide that a haburger in five inutes is preferred to a salad in one hour. Although the delay between alternatives does not change, reducing the tie to either reward equally by 24 hours ay switch the ranking of alternatives. Such inconsistencies are well docuented in behavioral studies, where both huan and anial subjects will switch how they choose between alternatives when the tie delay is decreased equally for each alternative. Thus as advocated by Hoch and Lowenstein (1991) a ore coplete understanding of consuer behavior ust recognize that people are influenced both by long-ter rational concerns and by ore short-ter eotional factors (p.492). The ai of this study is to develop a odel that explicitly incorporates both long-ter health objectives and short-ter satisfaction objectives in individual food choices. This will help to identify factors that lead to ore tie-inconsistent choices. To that end, this study odels how one s deand for convenience changes with tie pressures and hunger, easured as the interval an individual has gone without food. It is hypothesized that as people becoe hungrier and busier, their short-ter satisfaction objective ay take priority over their long-ter health objectives. Thus, they ay be apt to 4

5 prefer ore convenient foods. Since ost convenient foods are higher in calories, fat, cholesterol and sodiu, there is an ancillary increase in the consuption of these nutrients. Therefore, not accounting for one s level of hunger and deand for convenience ay ispecify the roles of prices, incoe, and inforation on nutrient deand. Theoretical Model The ajority of studies of nutrient deand are based on the traditional, neoclassical econoic odel where the consuer solves: Max (1) U(F, NF) F, NF subject to the budget constraint p F F + p NF Y where F is a vector of food goods and NF is a NF vector of non-food goods consued by the consuer with prices p F and p NF. Iposing separability of food and non-food goods results in a syste of deand functions X n = f ( pf, Y) for each of the n different nutrients. Many studies have expanded on this approach to include health status, health inforation, and other deand shifters but rarely provide a theoretical fraework that argues for their inclusion. Moreover, ost of the studies aggregate individual eals and food consued throughout the day, week or year. This aggregation over observations ignores the individuals level of hunger when they ake their food choice and where they procured the food. This ay then result in isspecifying the relationship between inforation and nutrient consuption. The theoretical odel in this study begins with the Becker household production odel, where individuals are assued to axiize utility, subject to their production functions, budget constraint and tie constraint. To develop a odel that ore accurately depicts how individuals ake their food choices, this odel also assues that individuals; 5

6 i. use household tie to create food, health and relaxation, ii. iii. iv. ake their food and nutrient consuption choices on a per-eal 2 basis, are affected by the prospect of iediate gratification, convenience and tie delay, and are ore affected by these factors as their hunger increases. Specifically, it is assues that an individual axiizes his or her utility: M (2) Max U = U ( F, NF, H, T F NF H Γ T h = 1,,,, " " h ; Γ, ε o, ε u ) F is a K diensional vector of food consued at eal. In this odel, individuals not only receive satisfaction directly fro food, but also consuption of non-food ites (NF ), healthy leisure tie ( T" h ), and the individual s health status (H ). It is assued that individuals ake their consuption decisions on a per-eal basis () over soe finite planning period that ends at M. For exaple, if the planning period were one day, then M would be the last eal of the day. It is also assued that individuals get utility fro the quality of their leisure tie. In this way, leisure tie when one is sick yields less satisfaction then when she is healthy. Healthy leisure tie is the product of tie spent in leisure activities ( T" ) and the quality of leisure activities Ω H ), which indicates the flow of health services per unit of leisure tie (Grossan 1970). In this fraework, a person with perfect health would get one full hour of quality leisure tie for every leisure hour. If this person s health were to ( decline, then she would receive less than one hour of quality leisure tie. h (3) T" = T" Ω(H ) The level of satisfaction received fro these variables is indirectly affected by the following factors: an individual s observable exogenous factors, such as socio-deographic characteristics (ε o ); her 2 In this study, the word eal is used for siplicity, but also includes snacks and all other eating occasions. 6

7 exogenous unobservable characteristics, such as taste (ε u ); and her endogenous level of hunger at eal (Γ ), which decreases with the aount of food she consued at the previous eal ( F 1), and increases with the aount of tie between eals ( I ) such that : (4) Γ = Γ I, F ). ( 1 It is assued that the indirect affects of hunger on the arginal utilities fro each arguent are as follows: Increasing hunger increases the utility received fro food and leisure tie, but also leads to teporary discofort. The rationale behind this is that, as one becoes hungrier, she ay receive ore enjoyent fro food, but ay also experience health probles, such as low blood sugar, fatigue, and irritability. These increasing ill health effects will increase one s sensitivity to tie delay, and eventually, deand for convenience. In this fraework, greater convenience is odeled as a reduction in tie spent producing food. Since the total aount of tie is finite, leisure tie is negatively related to tie spent preparing food. Thus, increasing hunger leads to a decreased deand for preparation tie in favor of ore leisure tie. The first constraint, the individuals health production function (H ), defines the links between health inputs: nutrients (n ); tie spent in health pursuits, such as exercising and becoing ore infored about good nutrition (Th ), and hunger ( Γ ). The health production function is also affected by exogenous observable characteristics, such as gender, age, and education(ε oh ) and exogenous unobservable factors (ε uh ) such as genetic endowent. This function is assued to be continuous, strictly concave, twice continuously differentiable and exhibit positive and diinishing arginal utilities with respect to each arguent except hunger, and the nutrients that should be consued in oderation 3. In hunger, the production of health is increasing (or neutral) up to soe hunger level Γ*. 3 According to USDA recoendations, individuals should liit intake of fats, sugar, cholesterol and sodiu. 7

8 Once an individual surpasses this critical level, the ill health effects of denying the body nutrients begin to have a negative effect on the production of health until soe axiu level of hunger Γ, after which the individual will die of hunger. As such, the health production function is defined as follows: oh uh (5) H = h(n,th, Γ ; ε, ε ) The purpose of including the hunger variable is to odel not only the iportance of the aount and nuber of nutrients consued, but also the tiing of which they are consued. Typical health production functions only analyze the aount and nuber of nutrients consued in a given observation period, yet excluding the tiing of consuption is tantaount to assuing that it is irrelevant. If tiing is irrelevant, then an individual who eats nothing for six days, and gorges hiself on 7 breakfasts, 7 lunches and 7 dinners on the seventh day would have the sae estiated health production as an individual who spaces these eals out over 7 days. This does not correspond with current beliefs about diet and nutrition. For that reason, this study explicitly assues that the tiing of nutrient consuption is relevant in the health production function. The nutrient intake of individuals is decided by a perceived apping function ( τˆ ), which translates the foods consued, F, into nutrients, such as gras of fat, protein, carbohydrates, and cholesterol. The accuracy of this apping function depends on the individual s knowledge ( η ), her stock of huan capital (E ) and where the individual procured the food (FS ). This apping function uses a Lancaster fraework, where ( η ) dictates the individual s perception of how uch a specific characteristic flows fro each food ite. For exaple, an individual who is well aware of the health risks linked with consuing too uch fat will be better able to deterine the aount of fat in a food ite than soeone who is unaware of these links. Also, it is assued that an individual will be better able to 8

9 assess the nutrient content of food prepared at hoe than food purchased away fro hoe. Thus, where she purchased her food affects her perception of nutrient content. (6) n = F τˆ (7) τ ˆ = τ( ;E, FS ) η The second constraint the individual faces is the food production function, which is assued to be continuous, strictly concave, twice continuously differentiable and exhibit positive and diinishing arginal utilities with respect to each arguent. This odel assues that kth food ite produced by the individual at the th eal is a function of j purchased inputs ({ x } J kj ) and tie spent preparing j= 1 food (TF k ), given the individuals stock of huan capital (E ). This yields the individuals per eal food production function: J Fk F xkj Tf j = 1 k (8) = ({ }, ; ) E With the additional constraints of the individual s tie and budget, the First Order conditions are solved to yield the following per-eal reduced for deand functions for the jth food inputs at the th eal: * X j o u (9) X j = D ( PX, PNF, w, A, E, η, ε, ε, I, F 1, FS ) Where w is the individual s labor wage, A is assets, while o ε and u ε are observable and unobservable exogenous characteristics. Substituting the optial aounts of food inputs and prep-tie into (6) and (7) yields a derived nutrient deand function: * n o u (10) n = D ( PX, PNF, w, A, E, η, ε, ε, I, F 1, FS ) 9

10 Data and Assuptions The data to be used coes fro the USDA s Continuing Survey of Food Intake by Individuals (CSFII ) and the copanion Diet and Health Knowledge Survey (DHKS). The purpose of the CSFII is to onitor food use and consuption patterns in the U.S. and provide data on food and nutrient intake. This data set contains detailed inforation on an individual s food intake, and her personal and household characteristics, such as age, level of education, sex, weight, height, race, incoe, education level of faily ebers, and faily size. The DHKS is the first national survey of individuals dietary attitudes and nutrition knowledge that can be linked to the nutrient intakes. It provides inforation on peoples attitudes and knowledge about dietary guidelines and their ability to put this knowledge into practice. It also provides inforation on the perceived adequacy of one s own food and nutrient intake. The CSFII data were collected via in-person interviews, where survey respondents were asked to recall their food intake over the last 24-hours This was done on 2 non-consecutive days spaced 3-10 days apart. In each CSFII household, the DHKS was adinistered to 1 adult over 20 years old who reported at least one day of food intake. This survey was adinistered over the telephone 2 to 3 weeks after the collection of CSFII data. For the purposes of this study, only individuals who also answered the DHKS will be included in the econoetric analysis to aintain a clear linkage between one s inforation and beliefs and her nutrient intake. The survey was a stratified, ultistage area probability saple that over-sapled low-incoe failies. The use of sapling weights is iportant because it copensates for variable probabilities of selection, different response rates and potential deficiencies in the sapling frae and are designed to yield estiates of the actual population which allows inferences drawn fro saple estiates to be applied to the population. 10

11 Econoetric Issues Fro the reduced deand equation (8), nutrient deand is a function of arket prices, wages and full incoe, given household and individual socio-deographic characteristics, individual health related characteristics, an individual s level of health inforation, her sensitivity to tie delay at a specific eal, and whether the eal was prepared at hoe or away fro hoe. However, given the data set and nature of the proble, this study will need to address soe of the anticipated econoetric issues. The first is that nutrient intake is analyzed on a per-eal basis and will therefore include several observations on one individual. The resulting estiated error ters will likely be correlated across observations for a given individual. Because of this, OLS estiates would be inefficient and the significance of paraeter estiates would be inaccurate. Using a rando effects odel will help to circuvent this issue. Another issue that can lead to probles with the econoetric estiation is that, as is coon with cross-sectional data, there is no inforation on expenditures or prices. However, since the individuals intake choices were ade at a single point in tie, it is not unusual to assue there is little variation in prices across households and that these differences can be captured by the geographic location and urbanization of the household (Variya et al, 1995, 1996). A third econoetric proble is that the hunger variable is a function of both the interval between eals and the aount of food energy consued at the previous eal. Thus, one of the independent variables is a lagged variable fro the previous eating occasion, resulting in estiated error ters that are likely correlated across eals for a given individual. Again, OLS estiates would be inefficient. Assuing a first order autoregressive process will help to skirt this issue. 11

12 A fourth econoetric issue is that several of the right hand side variables, naely health inforation, hunger, and food source, are arguably endogenous and are ost likely easured with error. The standard econoetric ethod of correcting probles of endogeneity and easureent error is to use soe type of instruental variables (IV) estiators. For the IV approach to yield consistent estiates, the instruents used ust eet the conditions of exogeneity and relevance. Yet, as suarized by Park and Davis (2001), there are three properties of cross sectional data and nutrient deand analysis that lead to relevancy condition not being satisfied: (i) While in the short run, the conditional deand for nutrients is a function of prices, full incoe, tie constraints, health inforation and individual endowents, the long run, or unconditional, deand for nutrients is a function of prices, incoe, tie constraints, and individual endowents; (ii) Most cross sectional data sets do not have inforation on arket prices, tie constraints, and full incoe; (iii) The correlation between any variables in cross-sectional data is very low. Thus condition (i) suggests that theoretically, variables in the unconditional deand equations can provide instruents for the conditional deand equation. Due to property (ii) the only available instruents are the individual s personal and household characteristics. However, there is no theoretical reason for partitioning these characteristics into either the conditional or the unconditional deand equations. Finally, the low correlation aong variables eans that IV estiators ay still be biased and inefficient. Based on these probles and the results of a coparison between OLS and IV estiates using cross-sectional data, Park and Davis concluded that OLS estiates were preferred and suggest eploying ultiple odel estiators and specifications. For this study, instruental variables were not used for the econoetric analysis of nutrient intake. 12

13 Econoetric Analysis The deand odel used for estiation is a linear equation relating individual i s nutrient consuption to household and individual socio-deographic (ε i1 ) characteristics, individual health related characteristics (ε i2 ), an individual s level of health intentions ( η i ), her sensitivity to tie delay at a specific eal ( Γ i ), and the source of the eal ( FS i ): (11) N i β o + β ' ε i + β ' ε + β ηi + β Γi + β FSi + ei + ui = 1 2 i 2 ' In equation (11) β 0 is an intercept ter, β 1 through β 5 are structural coefficients, e i is the rando disturbance for the ith individual at the th eal, u i is the rando disturbance for the ith individual and is constant for all eals. A list of variables, excluding health intentions, is found in Table 1. This paper groups health inforation into five general categories: Knowledge, Beliefs, Perceptions, Iportance, and Intentions. Justification for this is based on the Theory of Reasoned Action, a general theory of huan behavior that links beliefs, attitudes, and intentions to observed behavior. It initially used priarily in arketing but has been extended to econoic analysis of resource anageent (Fishbein and Manfredo, 1992). Doing this will help illuinate how different aspects of inforation are used when aking food choices. For exaple, although one ay be fully aware of the links between being overweight and health probles, if she does not think it is iportant, she will be less likely to act on this inforation. Answers to questions that for the knowledge index have definitive right and wrong answers, where as answers to belief questions are ore subjective. Perception variables were ade by coparing how respondent ranked their own diet quality copared to how their diet actually scored using coponents of the Healthy Eating Index (Variya, Shi and Blaylock, 2001). Past studies have created inforation proxies based on the nuber of correct answers to the DHKS survey. However, as the arket axio suggests, perception is reality. Thus, it ay be that what 13

14 soeone perceives to be true is a better predictor of behavior than siply whether or not soeone believes what is deeed to be true. The list of variables that have been used to proxy one s health intentions are found in Table 2. Results The coefficient estiates of the per-eal fat consuption fro equation (11) are reported in Table 3. Not surprisingly, we find that several of the ore traditional explanatory variables have a significant effect on per-eal fat consuption. Older individuals, individuals that coe fro larger households, live in the southern United States or live in rural areas consue significantly ore fat at each eal. However, we also find that situational factors, such as the interval between eals, the aount and kind of food consued at the previous eal, and the where the eal was procured also have significant explanatory power. These results show that when an individual has a longer interval between eals, consues ore siple carbohydrates, and eats at a bar or restaurants she will consue food that is significantly denser in fat. Conversely, an individual who has consued a larger aount of calories at the previous eal or obtained the food fro a grocery store is ore likely to consue lower fat food at the current eating occasion. Finally, these results show that individuals who have ore accurate perceptions of their own diet quality or put ore iportance is placed on diet quality also consuer foods that are significantly lower in fat. Conclusion The purpose of this paper is to present a new approach to exaining the deand for nutrients that incorporates any of the advances in behavioral econoics and the analysis of nutrition intake. By including variables such as hunger and tie pressures, this study illuinates how sensitivity to tie delay contributes to ore tie-inconsistent behavior. Naely, as individuals becoe hungrier and busier, their use of inforation and long-run health objectives play a lesser role to ore iediate 14

15 gratification. A soewhat obvious recoendation fro these findings would be to increase the convenience of relatively ore healthful foods or iprove the nutrient content of convenient foods. These findings also suggest that it could be beneficial to provide nutritional inforation at the point where an individual purchases ore convenient food and food prepared away fro hoe. Although labeling laws has iproved our understanding of the nutrient content of food purchased at a grocery store, there is still roo to iprove our understanding of foods prepared away fro hoe. Another noteworthy finding is that not all aspects of inforation have the sae affect in influencing observed behavior. These results show that perceptions and iportance have ore of an effect than straight-up inforation and beliefs. Two siple recoendation fro these findings would be to ake Aericans better aware of their own diet quality and to find ways to ake the believe that aintaining a healthy diet is indeed iportant. The ain liitation of this odel is epirical applicability due to liited availability of surveys that collect all the data of interest. For exaple, the epirical analysis cannot precisely identify how and why eating foods prepared away fro hoe leads to a higher consuption of fat. It ay be that tieconstraints lead to an increased deand for convenience and convenient foods tend to be denser in fat. It ay be that individuals are less infored about the nutrient content of food prepared away fro hoe. It ay be a cobination of these two factors. However, as alternative odels appear ore often in the econoics literature, data collection activities ay eventually better eet the needs of these econoic odels. 15

16 Table 1:Independent Variables and Definitions Category Variable Definition Mean Std. Dev Dependent Dfat Percent of calories at eal fro fat Variable Household Characteristics Incoe Progra Size Midwest South West Northeast Urban Suburban Rural Total household incoe in $1,000 1 if participate in food assistance progra Nuber of ebers in household 1 if Midwest 1 if South 1 if West 1 if Northeast 1 if central city 1 if suburb 1 if rural Socioeconoic and Deographic Characteristics of Main Meal Planner Health Related Factors of Main Meal Planner Sensitivity to tie delay Food Source: Age Feale College White Black Hispanic Vegetarian Allergy BMI Soker TV BFPL Involved Interval Calories Glyceic Hours worked Free Captive Cheap Social Planned Age of eal-planner in years 1 if eal planner is feale; 0 otherwise 1 if attended school beyond 12 th grade 1 if White 1 if Black 1 if Hispanic 1 if vegetarian 1 if individual has any food allergies Body weight (kgs)/ height 2 (eters) 1 if soker Average hour of t.v. watching per day 1 if pregnant, breast feeding or lactating 1 if plans, prepares or shops for eals Tie elapsed between eals Calories consued at previous eal, as a percent of their RDA Ratio of carbohydrates to fiber consued at previous eal Nuber of hours worked previous week 1 if eal cae fro soeone else 1 if eal cae fro cafeteria, dining center 1 if food cae fro fast food restaurant, pizza place, vending achine 1 if eal cae fro a sit down restaurant or tavern 1 if eal was prepared with food purchased at a grocery store

17 Knowledge (0.816) Nuber of correct answers on health questions Beliefs (1.177) Agreeent with 'soe people are born to be fat..not uch you can do..' Table 2: Intention Variables* Perceptions Iportance (0.751) (1.297) Perception of protein, cholesterol, and fat intake Source of high fiber or low fat diet, if any Intentions (1.759) High Fiber diet Nuber of correct servings of grains, fruits, vegetables, etc. the individual is able to identify Agreeent with '.. so any recoendations.. it's hard to know what to believe' Perception of own body weight Any existing edical condition Diabetic diet Agreeent with 'What you eat can ake a difference in chance of getting a disease' Perception of own diet quality Iportance of aintaining a healthy weight Exercise frequency Nuber of diseases associated with eating too uch cholesterol Iportance of liiting saturated fat Whether or not respondent uses labels Nuber of diseases associated with being overweight Iportance of liiting fat and cholesterol intake Whether or not respondent is involved with soe aspect of eal planning Nuber of diseases associated with eating too uch fat Iportance of eating lots of fiber * The nuber below the variable s nae is the ean. The nubers in parentheses are the standard deviation. 17

18 Table 3: Effect of explanatory variables on per-eal fat consuption Explanatory Variable Estiate t INTERCEPT INCOME** PROGRAM INVOLVED HHSIZE** MIDWEST SOUTH** WEST SUBURB** URBAN** BLACK HISP COLLEGE AGE** BMI_SP VEGGIE** ALLERG SMOKE TV BFPL HOURS WORKED** INTENTIONS INFORMATION BELIEFS PERCEPTIONS** IMPORTANCE** INTERVAL** GLYCEMIC** CALORIES** FREE** CAPTIVE SOCIAL** PLANNED** ** Variable is significant at α=

19 References Associated Press, (2001) Governent Urges Change as Obesity, Diabetes Rates Clib. Star Tribune. October, A15. Becker, Gary, (1965) A Theory of the Allocation of Tie, The Econoics Journal (75), Septeber: Fishbein, M. and M.J. Manfredo (1992) A Theory of Behavior Change. In Influencing Huan Behavior: Theory and Applications in Recreation, Touris and Natural Resources Manageent. ed M.J. Manfredo. Sagaore, Chapaign, IL. Food Marketing Institute, (2001a) Reaching Out to the Whole Health Consuer. in Prevention Magazine. Rodale Incorporated. Food Marketing Institute, (2001b) What Working Woen Want: Based on Trends in the United States: Consuer Attitudes and the Superarket, Frazao, Elizabeth, (1999) High Costs of Poor Eating Patterns in the United States. in Aericas Eating Habits Changes and Consequences, ed E. Frazao. Agriculture Inforation Bulletin Nuber 750, Econoic Research Service, United States Departent of Agriculture. Hoch, S.J and G.F. Loewestein, (1991) Tie-Inconsistent Preferences and Consuer Self Control. The Journal of Consuer Research. (17:4) March: Kennedy, E., S.A. Boan, M. Lino, S.A. Gerrior and P. Basiotis, (1999). Diet Quality of Aericans: Healthy Eating Index, in Aericas Eating Habits Changes and Consequences, ed E. Frazao. Agriculture Inforation Bulletin Nuber 750, Econoic Research Service, United States Departent of Agriculture. Lancaster, Kelvin J., (1966) A New Approach to Consuer Theory. Journal of Political Econoy. (74): Lin, Biin-Hwan, J. Guthrie, and E Frazao, (1999) Nutrient Contribution of Food Away Fro Hoe in Aericas Eating Habits Changes and Consequences, ed E. Frazao. Agriculture Inforation Bulletin Nuber 750, Econoic Research Service, United States Departent of Agriculture. Lowenstein, George and D. Prelec, (1992) Anoalies in Interteporal Choice: Evidence and Interpretation, in Choice Over Tie, ed George Lowenstein and John Elster. New York: Russell Sage Foundation. News Service, (2001) Obesity May Soon Kill More Aericans than Cigarettes. Star Tribune. Deceber 14. Park, J and G.C. Davis, (2001) The Theory and Econoetrics of Health Inforation in Cross-Sectional Nutrient Deand Analysis. Aerican Journal of Applied Econoics. (83:4) Noveber: Richan, Nessa J., (1999) The Natural Foods Market: A National Survey of Strategies for Growth. Policy Studies Report Nuber 12. The Henry A. Wallace Institute for Alternative Agriculture. Greenbelt, Maryland. Senauer, Ben, Elaine Asp, and Jean Kinsey, (1991) Food Trends and the Changing Consuer. Eagan Press, Minnesota. Variya, J.N., Blaylock, J. and D. Sallwood (1996) A Probit Latent Variable Model of Nutrition Inforation and Dietary Fiber Intake. Aerican Journal of Agricultural Econoics. (78) August: Variya, J.N., Blaylock, J. and D. Sallwood (1995) Modeling Nutrient Intake: The Role of Dietary Inforation. Food and Consuer Econoics Division, Econoic Research Service, U.S. Departent of Agriculture. Technical Bulletin No Variya, J.N., Y. Shi, and J. Blaylock (2001) Consuer Misperceptions of Diet Quality. Journal of Nutrition Education. (33:6) Noveber/Deceber: