Family Economics and Nutrition Review [Volume 8, Number 4]

              Feature Articles
2 Relationship of Knowledge of Food Group Servings Recommendations
to Food Group Consumption
Joanne F. Guthrie and Lois H. Fulton
Health Status Transitions of the Elderly, By Residential Location:
1984 to 1990
Carolyn C. Rogers
Trends in Food and Alcohol Consumption Away From Home
Nancy E. Schwenk
Research Summaries
41 Relationship Between Cigarette Smoking and Other High-Risk
Behaviors Among Our Nation's Youth
44 How Does Living Alone Affect Dietary Quality?
47 Household Debt
Regular Items
50 Charts From Federal Data Sources
52 Recent Legislation Affecting Families
53 Research and Evaluation Activities in USDA
56 Estimated Annual Expenditures on Children by Families, 1994
63 Data Sources
64
65
66
67
68
69
70
Journal Abstracts
Cost of Food at Home
Consumer Prices
Guidelines for Authors
Index of Authors in 1995 Issues
Index of Articles in 1995 Issues
Reviewers for 1995
U.S DEPOSITORY
.P..ROPERTY OF TH LIBRAR
A 22199
The University f or h Carolln
t reensboro
UNITED STATES DEPARTMENT OF AGRICULTURE
Volume 8, Number 4
1995
Dan Glickman, Secretary
U.S. Depa1tment of Agriculture
Ellen Haas, Under Secretary
Food, utrition, and Consumer Services
Eileen Kennedy, Executive Director
Center for utrition Policy and Promotion
Jay Hirschman, Director
Nutrition Policy and Analysis Staff
Editorial Board
Mohamed Abdei-Ghany
University of Alabama
Rhona Applebaum
Nati onal Food Processors Associati on
Johanna Dwyer
ew England Medical Center
Jean Mayer USDA Human utri tion Research Center
on Aging at Tufts Uni versity
Helen Jensen
Iowa State University
Janet C. King
Western Human Nutrition Research Center
U.S. Department of Agriculture
C. J. Lee
Kentucky State University
Rebecca Mullis
Georgia State Uni versity
Suzanne Murphy
Uni versity of Californi a-Berkeley
Donald Rose
Economic Research Service
U.S. Department of Agriculture
Ben Senauer
University of Minnesota
Laura Sims
Uni versity of Maryland
Retia Walker
Uni versit y of Kentucky
Editor
Joan C. Courtless
Editorial Assistant
Jane W. Fleming
Family Economics and Nutrition Review is
written and published each quarter by the
Center for Nutrition Policy and Promotion,
U.S. Department of Agriculture, Washington, DC.
The Secretary of Agriculture has determined
that publication of this periodical is neces­sary
in the transaction of the public business
required by law of the Department.
This publication is not copyrighted. Contents
may be reprinted without permission, but
credit to Family Economics and Nutrition
Review would be appreciated. Use of
commercial or trade names does not imply
approval or constiMe endorsement by USDA.
Family Economics and Nutrition Review is
indexed in the following databases: AGRICOLA,
Ageline, Economic Literature Index, ERIC,
Family Resources, PAIS, and Sociological
Abstracts.
Family Economics and Nutrition Review is
for sale by the Superintendent of Documents.
Subscription price is $8.00 per year ($1 0.00
for foreign addresses). Send subscription
orders and change of address to Superinten­dent
of Documents, P.O. Box 371954,
Pittsburgh, PA 15250-7954. (See subscrip­tion
form on p. 71.)
Suggestions or comments concerning this
publication should be addressed to: Joan C.
Courtless, Editor, Family Economics and
Nutrition Review, Center for Nutrition Policy
and Promotion, USDA, 1120 20th St., NW,
Sutte 200 North Lobby, Washington, DC
20036. Phone(202)60€"4816.
USDA prohibtts discrimination in tts programs
on the basis of race, color, national origin,
sex, religion, age, disability, political beliefs,
and marital or familial status. (Not all prohibited
bases apply to all programs.) Persons with
disabilities who require alternative means
for communication of program information
(Braille, large print, audiotape, etc.) should
contact the USDA Office of Communications
at (202) 720-2791 . To file a complaint, write
the Secretary of Agricu~ure, U.S. Department
of Agriculture, Washington, DC 20250, or call
(202) 720-7327 (voice) or (202) 720-1127
(TOO). USDA is an equal employment
opportunity employer.
Center for Nutrition Policy and Promotion
Feature Articles
2
18
30
Relationship of Knowledge of Food Group Servings
Recommendations to Food Group Consumption
Joanne F. Guthrie and Lois H. Fulton
Health Status Transitions of the Elderly,
By Residential Location: 1984 to 1990
Carolyn C. Rogers
Trends in Food and Alcohol Consumption Away From Home
Nancy E. Schwenk
Research Summaries
41 Relationship Between Cigarette Smoking and Other High-Risk
Behaviors Among Our Nation's Youth
44 How Does Living Alone Affect Dietary Quality?
47 Household Debt
Regular Items
50
52
53
56
63
64
65
66
67
68
69
70
Charts From Federal Data Sources
Recent Legislation Affecting Families
Research and Evaluation Activities in USDA
Estimated Annual Expenditures on Children by Families, 1994
Data Sources
Journal Abstracts
Cost of Food at Home
Consumer Prices
Guidelines for Authors
Index of Authors in 1995 Issues
Index of Articles in 1995 Issues
Reviewers for 1995
Volume 8, Number 4
1995
2
Feature Articles
Relationship of Knowledge of Food
Group Servings Recommendations
to Food Group Consumption
By Joanne F. Guthrie
Nutritionist
Center for Nutrition Policy and Promotion
Lois H. Fulton
Supervisory Home Economist
Agricultural Research Service (retired)
The USDA Food Guide provides recommended numbers of servings of five
major food groups: (1) bread, cereal, rice, and pasta; (2) vegetables; (3) fruit;
(4) milk, yogurt, and cheese; and (5) meat, poultry, fish, dry beans, eggs,
and nuts. The objective of this study was to examine the relationship of
knowledge of recommended servings of the five major food groups to
reported food group consumption among female adult meal planners using
data from the 1990 and 1991 Continuing Survey of Food Intakes by Individuals
and Diet and Health Knowledge Survey conducted by the U.S. Department
of Agriculture. Because about 99 percent of Diet and Health Knowledge
Survey respondents gave incorrect responses for grain products, the effects
of correct information on consumption of this group could not be analyzed.
For the remaining four food groups studied, knowledge of serving recom­mendations
was significantly associated with food group consumption after
controlling for the effects of a number of other factors that may influence
food consumption behavior. Results provide support for the use of a food
guide-based approach to dietary guidance.
ITJ he science of nutrition
attempts to answer the
question, "What should
we eat to be healthy?"
That question can be addressed on
two levels: the scientific level and the
consumer level. At the scientific level,
recommended amounts of essential
nutrients and other ctietary components
have been established by expert groups,
such as the National Academy of Sciences
(19,20). Consumers, however, choose
foods rather than nutrients. To aid
consumers in selecting a healthful diet,
nutritionists have traditionally provided
guidance in terms of food choices. A
food guide translates recommendations
on intakes of nutrients and other dietary
components into recommendations for
food consumption, with the goal of
making nutrition advice understandable
and usable to consumers.
The U.S. Department of Agriculture
(USDA) published its first food guide
for the general public in 1916. As
Family Economics and Nutrition Review
scientific knowledge increased, USDA
developed new food guides to accom­modate
new information on nutrition
needs (34). In 1985, USDA published
its current food guide (30). To develop
this guide, USDA used criteria based on
the Recommended Dietary Allowances
established by the National Academy
of Sciences (21) and the Dietary Guide­lines
for Americans-the official state­ment
of Federal dietary guidance policy
published by the USDA and the U.S.
Department of Health and Human
Services (32). The overall goal of the
USDA Food Guide is to offer consumers
guidance on planning a total diet that
would be both adequate in essential
nutrients and moderate in food compo­nents
for which excess intakes are
associated with health risk (e.g., fat,
saturated fatty acids, cholesterol,
sodium, and sugars).
The USDA Food Guide provides
recommended numbers of servings to
be consumed each day from each of the
five major food groups. For the bread,
cereal, rice, and pasta group (called the
"grain group" in this article, for brevity),
6 to 11 servings per day are recommended.
Three to 5 servings from the vegetable
group and 2 to 4 servings from the fruit
group are recommended. For the milk,
yogurt, and cheese group ("milk group"),
2 to 3 servings are recommended. The
recommendation for the meat, poultry,
fish, dry beans, eggs, and nuts group
("meat and beans group") is 2 to 3
servings per day or the amounts of these
foods that would be equivalent to a total
of 5 to 7 ounces of cooked lean meat,
poultry, or fish daily. In addition, the
USDA Food Guide recommends mod­eration
in consumption of fats, oils,
sweets, and sodium.
1995 Vol. 8 No. 4
Figure 1. The Food Guide Pyramid: A guide to daily choices
Fats, Oils, & Sweets
USE SPARINGLY
Milk, Yogurt,
&Cheese
Group
2·3 SERVINGS
Vegetable
Group
3-5 SERVINGS
Meat, Poultry, Fish,
Dry Beans, Eggs,
& Nuts Group
2·3 SERVINGS
Fruit
Group
2-4 SERVINGS
Bread, Cereal,
Rice, & Pasta
Group
6-11
SERVINGS
Source: U.S. Department of Agriculture and U.S. Department of Health and Human Services.
The USDA Food Guide was used in
several USDA educational publications
during the 1980's and wa included in
the 1990 edition of the Dietary Guide­lines
for Americans. It gained further
prominence in 1992 with the release
of the Food Guide Pyramid (fig. 1 ), a
new graphic representation of the Food
Guide (31). After consumer testing, thi
graphic was selected because it was
found to be an effective visual image
for communicating the information
contained in the Food Guide (6,35).
Since its publication, it has been widely
disseminated not only as part of USDA
publications but also as part of many
educational and promotional materials
developed by other public and private
sector groups (35).
Given that so much emphasis has been
placed on the use of food guides for
nutrition education, it would be useful
to know to what extent a knowledge of
food guide recommendations is associ­ated
with actual food choices. Results
of some studies provide evidence that
knowledge of food group recommenda­tions
is associated with eating a healthier
diet. Butler and Raymond (7) found that
low-income consumers who were able
to name the four food groups defined
by the "Basic 4" food guide developed
by USDA in 1956 (34) had better diets
than those who could not. Several
evaluation studies have shown the
Expanded Food and Nutrition Education
Program (EFNEP), a nutrition education
program for low-income consumers that
u es a food-group-oriented approach to
3
4
Given that so much
emphasis has been
placed on the use
of food guides for
nutrition education,
it would be useful
to know to what
extent a knowledge
of food guide
recommendations
is associated with
actual food choices.
teaching nutrition, to be effective in im­proving
the diets of participants (2,28 ).
EFNEP, however, also teaches other
knowledge and skills, such as food
shopping and preparation, that may also
contribute to its success. Most recently,
in a longitudinal study of Australian
consumers, Smith et al. (25) found
knowledge of the Australian food guide
to be a significant predictor of the con­sumers'
likelihood of making positive
dietary changes following a nutrition
education program.
The objective of this study was to
examine the relationship of knowledge
of recommended servings of major food
groups to their reported consumption
among female meal planners using data
from the USDA's 1990-91 Continuing
Survey of Food Intakes by Individuals
(CSFII) and Diet and Health Knowledge
Survey (DHKS). These surveys are
unique in that, together, they provide
the only federally collected data set
capable of relating knowledge and
attitudes concerning diet and health
to actual dietary intake.
Methods
Data and Sample
The CSFII was designed to obtain a
nationally representative sample of
households in the 48 conterminous
United States and consists of an all­income
and a low-income sample. For
the all-income sample, all households,
including low-income households,
were eligible to be interviewed. For the
low-income sample, participation was
limited to individuals in households
with gross income for the previous
month at or below 130 percent of the
Federal poverty thresholds (29).
For the 1990-91 CSFII, trained inter­viewers
visited each household and
obtained socioeconomic and demo­graphic
data on households and their
members. Health-related information,
such as heights and weights of house­hold
members, was also collected.
Heights and weights were self-reported;
self-reported weights may be slightly
underestimated, especially among
overweight individuals (22). In
addition, the interviewers obtained
1 day of dietary intake data, using the
24-hour recall method, and household
members were asked to complete a
record of foods consumed on the 2 days
following the 24-bour recall. Thus, up
to 3 consecutive days of food consump­tion
information was obtained from
household members.
For the DHKS, one member of each
CSFII household was contacted about 6
weeks after dietary data were collected.
Ideally, the individual contacted was
the person who had identified himself
or herself as the household's main
meal planner/preparer. In some cases,
interviewers were unable to contact the
main meal planner/preparer, and about
6 percent of DHKS respondents were
not the main meal planner/preparer.
Most interviews were conducted by
telephone; in-person interviews were
conducted when this was not feasible.
DHKS respondents were asked a series
of questions on their knowledge, atti­tudes,
and practices related to diet and
health. One series of questions assessed
knowledge of food group recommenda­tions.
DHKS respondents were asked to
state how many servings of (1) fruit; (2)
vegetables; (3) dairy products; (4) grain
products; and (5) meat, poultry, or fish
Family Economics and Nutrition Review
a person should eat daily (fig. 2).1
Interviewers provided information on
sample serving amounts. For this analy­sis,
answers were coded as "correct,"
"above correct answer," or "below
correct answer," based on USDA Food
Guide recommendations. Any answer
within the recommended range was
acceptable. For example, for the fruit
group, 2 to 4 servings are recommended;
therefore responses of "2", "3", and "4"
were all considered correct.
In 1990 and 1991, DHKS and 3-day
food intake data were obtained from
2,960 respondents. From these, female
meal planners 18 years of age and over
were selected as the sample for this
analysis. The male DHKS respondents
were excluded because of the consider­able
difference in male and female
energy intakes. The small number of
female DHKS respondents who were
not meal planners were excluded because
nonmeal planners might have less
control over their food choices than
meal planners. Pregnant and lactating
women were excluded because their
physiological state might be expected
to create short-term changes in dietary
requirements and food consumption.
Women who consumed food products
not typical of a mixed diet of a healthy
adult, for example, medical nutritional
products and baby foods, also were
excluded. The final analysis data set
consisted of 2,17 4 women.
To adjust for oversampling of low­income
households and for differing
response rates among population sub­groups,
DHKS sample weights were
developed by USDA in cooperation
with Iowa State University (29). Use
1
Although the question does not specify all foods
that are included in the meat, poultry, fish, dry
beans, eggs, and nuts group, the analysis examines
consumption of all foods in this group.
1995 Vol. 8 No.4
Figure 2. Survey question assessing knowledge of food group
recommendations
Let's begin by talking about your opinion of the amount of food, such as fruits,
vegetables and meats that people should eat each day for good health. How many
servings of (READ ITEM) should a person eat each day if one serving equals
(READ AMOUNT)?
NUMBER OF
ITEM AMOUNT SERVINGS
a. Fruit One piece of whole fruit?
b. Vegetables A half cup of cooked vegetables?
c. Dairy products One cup of milk or a slice of
cheese?
d. Grain products One slice of bread or a half cup of
cooked cereal, rice, or pasta?
e. Meat, poultry A piece the size of a medium
or fish hamburger?
of these weights for descriptive statistics
is recommended, so that the weighted
sample will resemble more closely the
actual U.S. population ( 15); weighted
data were used in this study to calculate
all descriptive statistics.
Food Group
Consumption Measures
item or equivalent; and for milk products,
1 cup of milk or equivalent. For meat,
poultry, fish, dry beans, eggs, and nuts,
serving units were calculated in terms
of 1 ounce of lean meat or equivalent.2
In the case of mixed foods (for example,
a cheese and tomato sandwich), the
food was disaggregated and the contri­bution
of each food ingredient to a
major food group was estimated. That
is, the contribution of a serving of a
cheese and tomato sandwich to the
20ne ounce or equivalent was elected as the
unit of measurement for the meat and beans
group because the USDA advises consumers
The CSFII reports food intakes in grams.
Intakes were converted to serving
amounts as defined by the USDA Food
Guide using a data base previously
developed (14) . The servings desig­nated
for the USDA Food Guide were
used to determine the number of servings
or part of a serving represented in 100
grams of each food reported as eaten
in the 1989-1991 CSFII. For grains, a
serving size was one slice of bread or
equivalent; for vegetables and fruit, 112
cup of a chopped, cooked, or canned
to monitor daily intake of this group by estimating
consumption of each food in the group in terms
of equivalence to 1 ounce of cooked lean meat
(for example, 1/2 cup of cooked dry beans is
considered equivalent to I ounce of cooked lean
meat). The number of ounces or equivalent
consumed in a day should be totaled and com­pared
to the recommendation of 5-7 ounces of
cooked lean meat or equivalent per day (32).
5
grain, vegetable, and milk groups would
be estimated. All contributions to the
five major food groups were counted,
including contributions from condiments
and incidental ingredients (for example,
the rai ins in raisin bread would be
counted toward fruit intake, even
though the food is primarily a grain).
Thus, "number of servings consumed"
as defined in this paper, refers to the
total amount of a given food group
consumed, expressed in terms of USDA
Food Guide serving sizes.
Analysis of Association of
Knowledge of .Food Group
Recommendations with Food
Group Consumption
Mean food group intakes and total
energy intakes by meal planners who
reported the correct number of food
group servings were compared to food
group and energy intakes of other meal
planners. For this analysis, weighted
data were used and statistical tests were
conducted using the SUDAAN software
package, which accounts for the effects
of the complex design of the CSFII­DHKS
surveys (23). T-tests were used
for comparing means of two groups,
and multiple contrasts were used for
simultaneously comparing means of
three groups.
Besides knowledge of serving recom­mendations,
many factors can affect
food intake. Therefore, multivariate
analysis techniques were used to examine
the independent association of knowl­edge
of serving recommendations with
consumption of foods from the fruit,
vegetables, meat and beans, and milk
groups. (Because about 99 percent of
respondents gave incorrect responses to
the question on recommended servings
of grain products, analysis of the effects
of correct information was not under­taken
for this group.)
6
For the other four groups, we examined
to what extent food group serving con­sumption
was explained by knowledge
of serving recommendations while con­trolling
for the effects of other factors
that previous research indicated may
influence food intake. The following
factors were included in each model as
control variables: Age, race, the height
and body mass index3 of the individual,
household income4 as a percent of the
Federal poverty level, education, whether
the individual was on a weight-loss
diet, region of residence, urbanization,
season in which dietary intake was
reported, whether weekend intake
was included in the 3-day dietary data,
whether there were any days of reported
intake in which the meal planner
reported her food consumption to be
unusually low, and whether there were
any days of reported intake in which
the meal planner reported her food
consumption to be unusually high.
Age, race, household income, and
education were included because pre­vious
research indicated that they are
associated with differences in consump­tion
of particular food groups, such as
fruits, vegetables, and milk and milk
products (3,4,9,10). Region of residence
was included because it may influence
availability and price of some food
items, as well as local food preferences.
3Body mass index was calculated as the ratio of
self-reported weight in kilograms to the square of
elf-reported height in meters. These values were
calculated by the U.S. Department of Agriculture,
Agricultural Research Service and are available
on the data tape (29 ).
4Household income before taxes; includes house­hold
income from wages or salary, Social Security
or Supplemental Security, pension or retirement,
unemployment or workmen's compensation,
alimony, child support, public assistance not
including food tamps or WIC benefits, and any
other sources of income (29).
Urbanization may be associated with
availability; for example, central city
areas may have fewer and smaller
supermarkets, with less food selection
(33). Season in which dietary intake
was reported may influence price and
availability, particularly for fruits and
vegetables. Inclusion of a variable
assessing weekend food intake should
control for day-to-day variation associ­ated
with weekend versus weekday
eating patterns.
Finally, differences in total energy
intake can have an important influence
on food group intake. Unfortunately,
the use of energy intake as an inde­pendent
variable in a multivariate
equation is problematic because within­individual
variability in energy intake
introduce error that will produce
biased coefficients (use of an intake
variable such as food group consump­tion
as a dependent variable does not
create bias because within-individual
variability is subsumed into the error
term) (5). Therefore, several variables
that proxy differences in energy need
were used as control variables. These
include self-reported height and body
mass index, as calculated from self­reported
height and weight, since larger
individuals are likely to consume more
energy; being on a weight-loss diet; and
whether individuals reported any days
with either lower-than-usual or higher­than-
usual dietary intakes. In addition,
age influences energy needs and con­trols
for energy differences to some
extent.
For the vegetable, milk, and meat and
beans groups, ordinary least squares
regression was used. For the vegetable
and the meat and bean group , knowl­edge
of correct serving recommenda­tions
was entered into the equation a
Family Economics and Nutrition Review
a dichotomous variable with correct an­swers
compared with incorrect answers
below the recommendation. Because so
few meal planners gave responses that
were above the correct USDA Food
Guide recommendation for either of
these two groups, these individuals
were excluded from the analysis. For
the milk group, correct answers and
answers above the correct recommenda­tion
were compared with answers below
the correct recommendation. In accord­ance
with guidelines for the use of
USDA food consumption survey data
(15 ), unweighted data were used for
these multivariate analyses, and ordi­nary
least squares regression analyses
were conducted using the SPSS-X
statistical software package (26).
For the fruit gtoup, ordinary ieast squares
regression analysis was not appropriate
because of the large number of individuals
(n=354 or 16 percent of the sample)
who did not consume any fruit at all
over the 3-day period. In statistical
terms, this means that the dependent
variable (servings of fruit consumed)
cannot be considered a continuous vari­able
throughout its range, but is instead
limited at the zero point, and techniques
appropriate for limited dependent vari­able
analysis must be used (16). One
technique that has been proposed as
particularly suitable for analysis of
food group consumption is the two-step
analysis developed by Cragg (8,13).
The first step of this analysis (probit)
identifies factors associated with the
decision to consume fruit; the second
step (truncated regression) identifies
factors associated with quantity of fruit
consumed, conditional on fruit being
consumed. As with the vegetable group
and the meat and beans group, the small
number of meal planners who gave
answers above the correct response
were dropped from the analysis. Thus,
the analysis compared those who gave
"too low" responses and those who gave
correct responses. The two-step analysis
was conducted using the LIMDEP
statistical package ( 11 ). Unweighted
data were used in the analysis.
Interpreting Regression Coefficients
For the ordinary least squares regression analyses used to exanline consumption
of the vegetable, meat and beans, and milk groups, estimated coefficients can be
interpreted in terms of their independent effects on consumption. For example, if
the estimated coefficient associated with knowledge of vegetable recommenda­tions
is 0.26, that can be interpreted as meaning that given knowledge of serving
recommendations, an individual would consume 0.26 servings more of vegetables
than an individual with equivalent personal characteristics who believed that a
smaller-than-recommended number of vegetable servings should be consumed.
The estimated coefficients produced by the two-step analysis used to examine
fruit consumption cannot be interpreted in this manner. For this analysis, the
reader should interpret a significant coefficient as indicating that a relationship
exists between a given independent variable and fruit consumption, but the
estimated coefficient cannot be directly used to detetmine the magnitude of
that relationship.
1995 Vol. 8 No.4
Very few of the
meal planners-only
1 percent-reported
the correct number
of servings from the
grain group, while
99 percent gave
responses that
were below the
recommended
6 to 11 servings.
7
Results
Description of Study Population
The average age of the meal planners
was 49 (table 1). Average before-tax
household income was $35,218. Because
meal planners came from households of
varying size, income was also assessed
as a percentage of the Federal poverty
level, which accounts for household
size. Average household income as a
percentage of the Federal poverty level
was 361 percent. Eighty-four percent of
meal planners were white.
Most meal planners had at least a high
school education. Twenty percent had
not completed high school, whereas
37 percent were high school graduates,
and 43 percent had at least some college
education. Five percent were on a weight­loss
diet. All regions of the country
were represented in the study population,
with 22 percent of meal planners corning
from the Northeast, 24 percent from the
Midwest, 35 percent from the South,
and 19 percent from the West. A range
of urbanization levels was also repre­sented,
with 30 percent of the study
population from the central city, 4 7
percent from a suburban area, and 23
percent from a nonmetropolitan area.
To control for seasonal and day-of-week
variation in intake, dietary data were
collected from survey participants at
all seasons of the year and on all days
of the week. Approximately one-quarter
of meal planners provided dietary data
during each of the four seasons. For 57
percent of meal planners, the 3 days of
dietary intake data included at least 1
weekend day; for the remainder, week­end
dietary intake was not assessed.
Dietary intake data can also be affected
by fluctuations in day-to-day intake.
8
Table 1. Description of sample1
Variable
Age (years)
Before-tax annual household income
Annual household income as percent of
Federal poverty level
Body mass index (BMI)
Height (inches)
Race
White
Non-White
Education
Less than high school
High school
At least some college
On weight loss diet
Yes
No
Region of residence
Northeast
Midwest
South
West
Urbanization
Central city
Suburban
Nonmetropolitan
Season intake reported
Spring
Summer
Fall
Winter
Weekend day included in 3-day dietary data
Yes
No
At least 1 day of lower than usual intake
Yes
No
At least 1 day of higher than usual intake
Yes
No
1 n=2, 174, weighted data.
2Valid percent for each variable.
Mean Range
49 18-97
$35,218 $500 - $250,000
361 6- 3007
25 14-63
64 48 -73
Frequency
(Percent)
84
16
20
37
43
5
95
22
24
35
19
30
47
23
25
25
24
26
57
43
28
72
13
87
Family Economics and Nutrition Review
Twenty-eight percent of meal planners
stated that on at least 1 of the 3 days of
dietary intake reported, they consumed
less than they usually ate. Thirteen
percent indicated that on at least 1 of
the 3 days of dietary intake reported,
they consumed more than they usually
ate. Finally, amounts consumed can be
affected by an individual's size. The
meal planners averaged 64 inches
(1.6 m) in height and a body mass
index of 25.
Knowledge of Food Group
Servings Recommendations
The meal planners' knowledge of food
group servings recommendations varied
considerably by food group (table 2).
Very few of the meal planners-only
1 percent-reported the correct number
of servings from the grain group, while
99 percent gave responses that were
below the recommended 6 to 11 servings.
The majority-73 percent-provided
correct responses for fruits, and 34
percent provided correct responses for
vegetables. For the meat and beans
group, 52 percent of meal planners
provided correct responses.
Almost all of the incorrect responses for
the fruit, vegetable, and meat and beans
groups were below the correct recom­mendation.
For these three food groups,
only 1 to 2 percent of meal planners
named amounts that were above recom­mendations.
For the milk group, however,
12 percent of meal planners reported
serving recommendations that were
above the 2 to 3 servings recommended
by the USDA Food Guide. Sixty percent
of meal planners reported the correct
recommendation, and 28 percent
reported amounts that fell below
recommendations.
1995 Vol. 8 No.4
Table 2. Knowledge of food group servings recommendations of adult
meal planners, CSFIIIDHKS 1990-91, 3-day data set1•2
Answer correct
Answer below according to Answer above
correct USDA correct
Food group recommendation food guide recommendation
Bread, cereal, rice, and
pasta group 99
Fruit group 26
Vegetable group 64
Milk, cheese, and yogurt group 28
Meat, poultry, fish,
dried beans, eggs, 46
and nuts group
1n=2,174, weighted data.
2Valid percent for each food group.
Based on their 3-day diet records, meal
planners generally consumed smaller
amounts of the five major food groups
than are recommended by the USDA
Food Guide. Food consumption data
based on self-reports may be under­reported
(17). Examination of total
reported caloric intakes revealed mean
intakes of 1,479 kilocalories for the meal
planners, which is below the average
energy allowance for adult women
with light-to-moderate activity levels,
as established by the National Academy
of Sciences (20). It is possible, therefore,
that these results may underestimate
food group consumption.
Percent
1 0
73 <1
34 2
60 12
52 2
Food Group Intakes and Energy
Intakes of Meal Planners
Table 3, p. 10, presents mean food group
intakes and energy intakes of meal
planners by knowledge of food group
servings recommendations. Meal planners
who reported the correct number of
recommended servings of vegetables
consumed significantly more servings
of vegetables per day-2.9 servings,
on average, compared with 2.5 servings
consumed by those who gave answers
below the correct recommendation. An
average of 1.4 servings of fruit per day
was consumed by those who reported
the correct number of recommended
servings of fruit, significantly more
than the 1.0 servings averaged by those
who gave answers below the correct
recommendation.
9
Table 3. Mean food group intakes and total caloric intakes by female meal planners, CSFIIIDHKS 1990-91,
3-day data set1
Reported number of
recommended food
group servings
Below correct
recommendation
(Total caloric intake)
Correct
(Total caloric intake)
Above correct
recommendation
(Total caloric intake)
1n=2,174, weighted data.
2Does not include legumes.
Grain
group
4.9
(1479)
-Indicates too few respondents for reliable estimates.
Mean number of servings consumed per day
Meat, poultry, fish,
Vegetable Fruit Milk dried beans, eggs, and
group2 group group nuts group
2.5L l.OL l.lL,H 4.4 oz. or equivalent
(1456) (1446) (1414) (1462)
2.9 1.4 1.4H 4.5 oz. or equivalent
(1506) (1491) (1467) (1490)
1.8
(1699)E
L=Food group intake significantly lower than that of women with correct answers.
H=Food group intake significantly lower than that of women who gave answers above correct recommendation.
E=Energy significantly higher than that of women who gave correct answers and answers below correct recommendation.
For the meat and beans group, those
who provided a correct response to
the question on number of servings
consumed an average of 4.5 ounces
of cooked lean meat, poultry, fish or
the equivalent, compared with 4.4
ounces consumed by meal planners
who gave answers below the correct
recommendation; this difference was
not significant. Meal planners who
provided correct answers to the
questions on recommended servings
for the vegetable, fruit, and meat and
beans groups averaged higher caloric
intakes than those who provided answers
below the correct recommendation, but
the difference was not significant.
10
So few meal planners provided correct
responses to the question on recommended
servings of grains that average intakes
for the "correct answer" group cannot
be reliably estimated. The meal planners
who provided answers below the correct
recommendation, however, averaged
4.9 servings of foods from the grain
group.
For the milk group, meal planners who
reported the correct number of recom­mended
servings consumed 1.4 servings,
whereas those who gave answers below
the correct recommendation consumed
1.1 servings. The meal planners who
gave answers that were higher than the
USDA Food Guide recommendation
consumed 1.8 servings per day. All
three of the groups differed significantly
from each other in terms of milk group
servings consumed. Meal planners who
gave answers below the correct recom­mendation
had caloric intakes that were
53 calories lower, on average, than the
intakes of those who gave the correct
answer, but this difference was not
significant. Meal planners who gave
answers that were higher than the USDA
Food Guide recommendation had
average caloric intakes that were
significantly higher than the caloric
intakes of meal planners who gave
answers that were either correct or
below recommendations.
Family Economics and Nutrition Review
Association of Knowledge of
Food Group Recommendations
With Food Group Consumption
For all four food groups studied,
knowledge of food group serving
recommendations was found to be
positively associated with consumption
of the corresponding food group after
controlling for the other factors included
in the multivariate analyses. Several
of these other factors also influenced
consumption of particular food groups,
although none influenced food group
consumption as consistently as did
knowledge of serving recommendations.
Regression models used to examine
consumption of the vegetable, meat and
beans, and milk groups explained 8 to
10 percent of variance, similar to results
of other analyses using personal charac­teristics
to explain differences in con­sumption
as assessed by national food
consumption survey data (18) .
For the vegetable group (table 4),
knowledge of the correct number of
recommended servings was signifi­cantly
associated with increased
consumption of servings of vegetables.
In addition, vegetable consumption
was positively associated with age,
household income, having at least
some college education, and living in
the Northeast or Western regions as
compared with living in the South. It
was negatively associated with living
in a central city area, as compared with
living in a suburban area, and having at
least 1 day of lower than usual reported
dietary intake.
1995 Vol. 8 No.4
Table 4. OLS regression coefficients for factors related to number of
vegetable servings per day, CSFIIIDHKS 1990-91, 3-day data set1
Independent variable
Knowledge of correct number of vegetable servings
(base= Answer below correct recommendation)
Age
Height
Body mass index
White (base= Non-White)
Household income as percent of Federal poverty level
Education (base = No high school)
High school
At least some college
On weight loss diet (base = Not on weight loss diet)
Region of residence (base = South)
Northeast
Midwest
West
Urbanization (base= Suburban)
City
Nonmetropolitan
Season intake reported (base = Winter)
Spring
Summer
Fall
Weekend day included in 3-day report
(base= No weekend day)
At least 1 day of lower than usual intake
At least 1 day of higher than usual intake
Constant
R2
*p < .05.
12,023 observations included in analysis.
Estimated
coefficients
0.26*
0.008*
0.01
0.002
-0.007
0.0008*
0.06
0.25*
-0.08
0.56*
0.13
0.22*
-0.21 *
0.08
0.11
-0.04
-0.12
0.06
-0.52*
0.02
0.80
0.10
11
For the fruit group (table 5), two-step
analysis revealed that knowledge of the
serving recommendation was not associ­ated
with the decision to consume fruit
but was associated with the amount of
fruit consumed. Variables that had a
positive significant association with the
decision to consume fruit were being
older, being taller, having a higher
household income, having at least some
college education, and living in the
Northeast or West compared with the
South.
Besides knowledge of fruit servings
recommendations, other variables that
were positively associated with the
amount of fruit consumed, conditional
on the decision to consume fruit, were
being older, having at least a high
school education, and living in the
Northeast compared with the South.
Amount of fruit consumed was nega­tively
associated with having a higher
body mass index and having reported
dietary intake during a period that
included at least I weekend day.
Consumption of servings of foods from
the meat and beans group was positively
associated with knowledge of the serving
recommendations for this group (table 6).
For this food group, being older, being
white rather than non-white, being on
a weight-loss diet, and eating less than
usual on at least 1 of the 3 days of re­ported
dietary intake were all associated
with lower numbers of servings con­sumed.
Having a higher household
income, a higher body mass index, a
reported dietary intake during a period
that included at least 1 weekend day,
and living in the Northeast region were
factors that were positively associated
with intake of this food group.
12
Table 5. Estimated coefficients for factors related to number of fruit
group servings per day using two-step analysis, CSFIIIDHKS 1990-91,
3-day data set1
Independent variable
Knowledge of correct number of fruit servings
(base =Answer below correct recommendation)
Age
Height
Body mass index
White (base= Non-White)
Household income as percent of Federal poverty level
Education (base= No high school)
High school
At least some college
On weight loss diet
(base= Not on weight loss diet)
Region of residence (base = South)
Northeast
Midwest
West
Urbanization (base = Suburban)
City
N onmetropolitan
Season intake reported (base = Winter)
Spring
Summer
Fall
Weekend day included in 3-day report
(base= No weekend day)
At least 1 day of lower than usual intake
At least 1 day of higher than usual intake
Constant
Log-likelihood
Chi-square statistic
*p < .05.
12,032 cases included in analysis.
Estimated coefficients
Truncated
Pro bit regression
0.11 2.26*
0.02* 0.10*
0.03* -0.005
-0.007 -0.08*
0.02 0.76
0.0005* 0.001
0.14 1.15*
0.43* 2.39*
-0.08 1.09
0.25* 1.42*
0.10 0.19
0.30* 0.71
-0.16 0.70
-0.08 0.32
-0.03 -0.09
0.07 0.65
-0.15 -0.74
0.13 -0.93*
-0.13 -0.81
0.06 0.44
-2.53* -12.99*
-858 -2211
192*
Family Economics and Nutrition Review
Table 6. OLS regression coefficients for factors related to consumption
of meat, poultry, fish, dried beans, eggs, and nuts1 per day, CSFIIIDHKS
1990-91, 3-day data set2
Independent variable
Reported number of recommended meat, poultry,
fish servings
(base = Answer below correct recommendation)
Age
Height
Body mass index
White (base = Non-White)
Household income as percent of Federal poverty level
Education (base= No high school)
High school
At least some college
On weight loss diet (base= Not on weight loss diet)
Region of residence (base = South)
Northeast
Midwest
West
Urbanization (base= Suburban)
City
Nonmetropolitan
Season intake reported (base = Winter)
Spring
Summer
Fall
Weekend day included in 3-day report
(base= No weekend day)
At least 1 day of lower than usual intake
At least 1 day of higher than usual intake
Constant
R2
*p <.OS.
1 Expressed as ounces of cooked lean meat or equivalent.
22,0 18 observations included in analysis.
1995 Vol. 8 No.4
Estimated
coefficients
0.32*
-0.01*
0.01
0.02*
-0.70*
0.001 *
-0.05
-0.16
-0.76*
0.29*
0.0003
-0.21
0.17
0.16
0.22
0.11
-0.14
0.19*
-0.70*
0.26
3.80*
0.09
For the milk group (table 7, p. 14), the
relationship of knowledge to intake is
somewhat more complex. Both knowl­edge
of correct serving recommendations
and belief that even higher numbers of
servings from the milk group should
be consumed were associated with con­sumption
of significantly more servings
from the milk group, compared with
belief that fewer than two servings are
recommended. However, the values
of the estimated coefficients differ.
For the correct answer, the value of the
estimated coefficient is 0.31, indicating
that given knowledge of the correct
recommendations, the female meal
planner will consume 0.31 more servings
from the milk group than if he believed
the recommendation to be lower, all
other factors being equal.
For the answer above the correct
recommendation, the estimated coeffi­cient
is 0.49, indicating that this belief
is likely to lead to an increase in milk
group consumption of 0.49 servings.
Thus, belief that even more servings
from the milk group should be consumed
than are recommended by the USDA
Food Guide leads to an even greater
increase in milk group consumption
than knowledge of the correct
recommendations.
Several other factors also influenced
milk group consumption. Being white
rather than non-white was positively
associated with consumption of this
food group, as was having at least orne
college education, living in the Midwest
region as compared with the South, and
being taller. Having a higher body mass
index, reporting food intake during the
summer or fall months as compared
with winter, and reporting at least 1 day
of lower than usual intake were nega­tively
associated with number of servings
of milk and milk products consumed.
13
Table 7. OLS regression coefficients for factors related to number of
milk servings per day, CSFII/DHKS 1990-91, 3-day data set1
Independent variable
Reported number of recommended milk group servings
(base = Answer below correct recommendation)
Correct answer
Answer above correct recommendation
Age
Height
Body mass index
White (base= Non-White)
Household income as percent of Federal poverty level
Education (base= No high school)
High school
At least some college
On weight loss diet (base = Not on weight loss diet)
Region of residence (base = South)
Northeast
Midwest
West
Urbanization (base= Suburban)
City
Nonmetropolitan
Season intake reported (base = Winter)
Spring
Summer
Fall
Weekend day included in 3-day report
(base= No weekend day)
At least 1 day of lower than usual intake
At least 1 day of higher than usual intake
Constant
R2
*p < .05.
12,051 observations included in analysis.
14
Estimated
coefficients
0.31 *
0.49*
-0.0004
0.03*
-0.009*
0.34*
-0.00003
0.08
0.17*
-0.02
0.12
0.12*
0.01
0.03
-0.07
-0.08
-0.19*
-0.14*
-0.04
-0.15*
0.11
-0.59
0.08
Conclusions
For all four major food groups analyzed,
it was found that knowledge of USDA
Food Guide servings recommendations
was independently associated with food
group consumption after controlling
for a number of characteristics. For
the fruit group, knowledge of serving
recommendations was not associated
with the decision to consume fruit but
was associated with amount consumed.
It may be that over a 3-day period indi­viduals
consume at least small amounts
of fruit for a variety of reasons, for
example as a small part of another
food. Consumption of larger amounts,
however, seems to be more likely
when individuals know serving
recommendations.
For the vegetable, fruit, and meat and
bean groups, knowledge of correct
recommendations was compared only
with incorrect answers that were lower
than the correct recommendation, since
fewer than 3 percent of meal planners
gave answers that were above recom­mendations
for these food groups. Only
for the milk group was there an appre­ciable
number of meal planners who
believed that even more servings than
are recommended in the USDA Food
Guide should be consumed.
Believing one should consume higher­than-
recommended amounts of a given
food group may encourage its intake,
but it may also increase caloric intake
or displace other food groups from the
diet. The meal planners who believed
they should consume higher-than­recommended
amounts from the milk
group did consume more milk and milk
products than either those with correct
or "too low" answers. They also con­sumed
significantly more calories and
did not consume smaller-than-average
Family Economics and Nutrition Review
amounts of other food groups (data not
shown). Their total caloric intake was
232 to 285 calories higher than that of
women in other groups, much more than
would typically be provided by 0.4
servings from the milk group. It appears
that their increased consumption of
milk and milk products was associated
with a pattern of higher-than-average
overall caloric intake rather than with
displaced consumption of other foods.
The only one of the five major food
groups for which an association be­tween
knowledge of USDA Food Guide
recommendations and food group con­sumption
could not be established was
the grain group. Too few meal planners
knew the recommended number of
grain servings for it to be possible to
analyze the effects of correct informa­tion
on consumption of these foods.
These data were collected before the
publication of the Food Guide Pyramid
graphic, which has given more publicity
to USDA Food Guide recommendations.
It may be that consumers are now
more aware of recommendations for
consumption of grains.
Data on knowledge of food group
serving recommendations and food
consumption are currently being
collected as a part of USDA's 1994-96
Continuing Survey of Food Intakes by
Individuals/Diet and Health Knowledge
Survey. Comparison of the results ob­tained
after release of the Food Guide
Pyramid with the results obtained by
this study will provide some informa­tion
on the effectiveness of the Food
Guide Pyramid in transmitting knowl­edge
of recommended grain intake.
1995 Vol. 8 No.4
A major concern when examining the
relationship of knowledge of recom­mendations
to food group intake is the
possible effect of confounding variables.
One major variable of interest is energy
intake. Individuals differ in energy in­take,
due to differences in size, activity
level, etc. Individuals who consume
more total energy (kilocalories) might
be expected to consume larger amounts
of food groups, regardless of knowledge.
Therefore, several variables reflecting
differences in energy intake were in­cluded
in analyses as control variables.
The significance of knowledge, after
controlling for factors that would influ­ence
energy intake, lends more support
to the conclusion that differences found
are attributable to quality of food choices,
not just quantity of food consumed.
Underreporting of energy intake might
also affect identified relationships.
Unfortunately, dietary data based on 3
days of reported intake cannot be used
to categorize specific individuals as
underreporters, since some reported low
intakes may simply reflect day-to-day
variation in intake. Overweight has been
identified as being associated with
underreporting (24). Body mass index
was included in the analysis and was
found to be significantly associated
with decreased consumption of foods
from the fruit group and the milk group
and increased consumption of foods
from the meat and beans group. These
relationships may be associated with
underreporting or with distinctive
patterns of food consumption related
to having a higher body mass index.
Since it would be surprising for higher­weight
individuals to underreport fruit
and milk products but not foods from
the meat and beans group, the latter
explanation may be more probable.
Although many other factors were
found to have significant impacts on
individual food groups, knowledge
of food group recommendations had
the most consistent, significant, positive
association with food group consump­tion
of any of the factors included in
the multivariate analyses. This supports
the general usefulness of a food-group­based
approach to nutrition education
as a means of encouraging an overall
healthful diet. At the same time, nutrition
educators may want to consider the
role other variables, such as household
income, education, and place of residence,
may play in consumption of particular
food groups. For example, results of
these analyses indicate that, when factors
such as race and household income are
controlled, living in a central city area
is associated with decreased consump­tion
of vegetables. Nutrition educators
working with urban populations may
wish to consider this finding and
investigate any particular problems,
such as access to stores stocking a
variety of vegetables, that may need to
be considered in developing a nutrition
promotion program that would be
effective in an urban population.
Despite the positive effects of knowl­edge,
meal planners who knew USDA
Food Guide serving recommendations
did not, on average, consume the mini­mum
number of servings in the recom­mended
range. Although this result
could reflect underreporting, it may
also indicate further educational and
motivational needs of consumers. Lack
of knowledge of recommended serving
sizes for each food group has been cited
as a source of consumer confusion ( 1 ).
Educational efforts that provide this
information may assist consumers.
15
Other types of education that build on
the basic message of the Food Guide
may also increase its effectiveness ( 1 ).
These results provide some support for
the importance of nutrition education
efforts to teach food group servings
recommendations to consumers. It may
be that this simple information by itself
can encourage consumption of major
food groups. However, it is generally
agreed that many other factors, such as
attitudes toward food and health and
lifestyle factors, also play important
roles in influencing food-related
behavior ( 12). It may be that knowledge
of recommended servings, as used in
this analysis, is a proxy for other variables,
such as more positive attitudes toward
following dietary guidance or more
detailed nutrition knowledge. Further
investigation of the relationship of
nutrition knowledge to food group
consumption is indicated.
16
References
1. Achterberg, C., McDonnell, E., and Bagby, R. 1994. How to put the Food Guide Pyramid
into practice. Journal of the American Dietetic Association 94(9): 1030-1035.
2. Amstutz, M.K. and Dixon, D.L. 1986. Dietary changes resulting from the expanded food
and nutrition education program. Journal of Nutrition Education 18(2):55-60.
3. Axelson, M.L. 1986. The impact of culture on food-related behavior. Annual Reviews of
Nutrition 6:345-363.
4. Axelson, M.L. and Brinberg, D. 1989. A Social-Psychological Perspective on Food­Related
Behavior. Springer-Verlag, New York.
5. Beaton, G.H. 1994. Approaches to analysis of dietary data: Relationship between planned
analyses and choice of methodology. The American Journal of Clinical Nutrition
59(Supp1.):253S-261 S.
6. Bell Associates, Inc. 1992. An Evaluation of Dietary Guidance Graphic Alternatives.
Prepared for U.S. Department of Agriculture, Food and Consumer Services, Alexandria, VA.
7. Butler, J.S. and Raymond, J. 1987. Knowledge Is Better Than Money: The Effect of the
Food Stamp Program on Nutrient Intake. Institute for Research on Poverty Discussion Paper:
No. 828-87. University of Wisconsin-Madison, Madison, WI.
8. Cragg, J.G. 1971. Some statistical models for limited dependent variables with application
to the demand for durable goods. Econometrics 39:829-844.
9. Cronin, F.J., Krebs-Smith, S.M., Wyse, B.W., and Light, L. 1982. Characterizing food
usage by demographic variables. Journal of the American Dietetic Association 81:661-673.
10. Enns, C.W., Tippett, K.S., Basiotis, P., and Goldman, J. 1994. Diets of Americans. In:
Nutrition: Eating for Good Health, AlB 685, pp. 2-11 . U.S. Department of Agriculture,
Washington, DC.
11. Greene, W.H. 1992. LIMDEP Version 6.0: User's Manual and Reference Guide.
Econometric Software, Inc., Bellport, NY.
12. Guthrie, J.F. 1994. Quantitative nutrition education research: Approaches, findings,
outlook. Journal of Nutrition 124:1813S-1819S.
13. Haines, P.S., Guilkey, D.K., and Popkin, B.M. 1988. Modeling food consumption
decisions as a two-step process. American Journal of Agricultural Economics 70:543-552.
14. Kennedy, E.T., Ohls, J., Carlson, S., Fleming, K. 1995. The Healthy Eating Index: Design
and applications. Journal of the American Dietetic Association 95( 10): 1103-1108.
15. Kott, P.S., Gray, B.C., and Guenther, P.M. 1989. Guidelines for the Use of Weights When
Analyzing and Reporting HNIS Survey Data: October, 1989. U.S. Department of Agriculture,
Human Nutrition Information Service.
16. Maddala, G.S. 1983. Limited Dependent and Qualitative Variables in Econometrics.
Cambridge University Press, Cambridge.
17. Mertz, W., Tsui, J.C., Judd, J.T., Reiser, S., Hallfrisch, J., Morris, E.R., Steele, P.D., and
Lashley, E. 1991. What are people really eating? The relation between energy intake derived
from estimated diet records and intake determined to maintain body weight. American
Journal of Clinical Nutrition 54: 291-295.
Family Economics and Nutrition Review
18. Murphy, S.P., Rose, D., Hudes, M., and Viteri, F.E. 1992. Demographic and economic
factors associated with dietary quality for adults in the 1987-88 Nationwide Food Consumption
Survey. Journal of the American Dietetic Association 92:1352-1357.
19. National Academy of Sciences, National Research Council, Food and Nutrition Board.
1989. Diet and Health: Implications for Reducing Chronic Disease Risk. National Academy
Press, Washington, DC.
20. National Academy of Sciences, National Research Council, Food and Nutrition Board.
1989. Recommended Dietary Allowances (lOth ed.). National Academy Press, Washington, DC.
21 . National Academy of Sciences, National Research Council, Food and Nutrition Board.
1980. Recommended Dietary Allowances (9th ed.). National Academy Press, Washington, DC.
22. Rowland, M.L. 1990. Self-reported weight and height. American Journal of Clinical
Nutrition 52:1125-1133.
23. Shah, B.V., Barnwell, B.G., Hunt, P.N., and LaVange, L.M. 1991. SUDAAN User 's
Manual: Professional Software for Survey Data Analysis for Multi-Stage Sample Designs.
Research Triangle Institute, Research Triangle Park, NC.
24. Schoeller, D.A. 1990. How accurate is self-reported dietary energy intake? Nutrition
Reviews 48(10):373-379.
25. Smith, A.M., Baghurst, K., and Owen, N. 1995. Socioeconomic status and personal
characteristics as predictors of dietary change. Journal of Nutrition Education 27(4): 173-181.
26. SPSS, Inc. 1988. SPSS-X User 's Guide (3rd ed.). SPSS, Inc., Chicago.
27. Tippett, K.S., Mickle, S.J., Goldman, J.D., Sykes, K.E., Cook, D.A., Sebastian, R.S.,
Wilson, J.W., and Smith, J. 1995. Food and Nutrient Intakes by Individuals in the United
States, I Day, 1989-91. U.S. Department of Agriculture, Agricultural Research Service.
NFS Report No. 91-2.
28. Torisky, D.M., Hertzler, A.A., Johnson, J.M., Keller, J.F., Hodges, P.A.M., and Mifflin,
B.S. 1989. Virginia EFNEP homemakers' dietary improvement and relation to selected
family factors. Journal of Nutrition Education 21(6):249-258.
29. U.S. Department of Agriculture. 1993. Data tapes and documentation for 1990 and 1991
CSFll/DHKS. National Technical Information Service Accession No. PB93-504843 (1990
tape) and PB94-500063 (1991 tape). Computer tapes.
30. U.S. Department of Agriculture, Human Nutrition Information Service. 1985. Developing
the Food Guidance System for "Better Eating for Better Health," A Nutrition Course for
Adults. Administrative Report No. 377.
31. U.S. Department of Agriculture, Human Nutrition Information Service. 1992. The Food
Guide Pyramid. Home and Garden Bulletin No. 252.
32. U.S. Department of Agriculture and U.S. Department of Health and Human Services.
1980. Nutrition and Your Health: Dietary Guidelines for Americans. Home and Garden
Bulletin No. 232.
33. Weinberg, Z. 1995. No Place to Shop: The Lack of Supermarkets in Low-Income
Neighborhoods. Public Voice for Food and Health Policy, Washington, DC.
34. Welsh, S., Davis, C., and Shaw, A. 1992. A brief history of food guides in the United
States. Nutrition Today, November/December, pp. 6-11.
35. Welsh, S., Davis, C., and Shaw, A. 1992. Development of the Food Guide Pyramid.
Nutrition Today, November/December, pp. 12-23.
1995 Vol. 8 No.4 17
18
Health Status Transitions of the
Elderly, by Residential Location:
1984 to 1990
By Carolyn C. Rogers
Demographer
Economic Research Service
Transitions in the health status and living arrangements of community­resident
elderly persons are examined to determine whether declining
health and changes in social support networks are likely to result in changes
in living arrangements. The Longitudinal Study of Aging (LSOA) is used to
follow a sample of elderly people 70 years and older over a 6-year interval.
Results show that the level of disability at the baseline date (1984) affects
health outcomes over time, with fewer of the initially nondisabled entering
nursing homes or dying by 1990. The nonmetro elderly experienced a some­what
greater decline in health status over the 6-year interval than did the
metro elderly. A smaller proportion of elderly persons initially living with their
spouse end up in nursing homes or die 6 years later, compared with those
initially living either alone or with others. The longitudinal data show that the
risk of institutionalization is roughly the same for metro and nonmetro elderly.
Elderly people adjust their living arrangements in response to changes in
their health and social support networks.
[!] he growing number of older
people in the United States,
their greater risk of disability,
and their higher use of
health care services have increased the
need for a more complete understanding
of the nature of changes in health status
later in life. This study uses the Longitu­dinal
Study of Aging (LSOA) to follow
a sample of elderly people 70 years and
older living in the community in 1984
over a 6-year interval. Changes in health
and disability are examined in relation
to transitions in living arrangements
and by metropolitan-nonmetropolitan
(metro-nonmetro) residence. Under­standing
the relationship between changes
in health status, living arrangements,
and residential location is essential
to the allocation of health care and
community resources and the future
planning of appropriate health care
services in local communities.
This study examines individual transi­tions
both into and out of functionally
impaired states. In order to have a better
basis on which to plan interventions in
functional loss in the elderly population,
nationally representative estimates are
needed of functional status transition
rates specific by age, prior functional
status, and residence. Longitudinal data
make it possible to examine the 6-year
incidence of functional limitations, the
rates of improvement or loss of function
FamiJy Economics and Nutrition Review
for currently disabled persons, and
the risks of institutionalization and
mortality on the community-resident
elderly population.
Previous Research
Several studies have examined changes
over time in the functional status of the
elderly and the risk of institutionaliza­tion
and death using the LSOA and
other surveys (2,4,6,10,13). Manton
found that the majority of elderly
persons who were not initially disabled
(about 82 percent) remained nondisabled
over a 2-year period, and there was a
significant probability of long-term
improvement in functional status even
at very high levels of impairment.
Examining individual transitions both
into and out of functionally impaired
states, the level of disability was found
to strongly predict differentials in
mortality and risk of institutionaliza­tion;
the higher the level of disability,
the greater the risk of becoming institu­tionalized
or dying.
Crimmins and Saito (4) found that the
vast majority of older persons initially
free of functioning difficulties remained
that way over a 2-year period. Looking
at both older people with difficulties
in the initial period and tho e without
difficulties, they found that less than
10 percent of those without functioning
deficiencies at the first interview de­velop
a deficiency. Older persons with
a greater number of functioning diffi­culties
at the baseline are less likely to
improve in the interval. A much lower
likelihood of improvement occurs in
activities of daily living (ADL's) and
instrumental activities of daily living
(IADL's) with an onset duration greater
than 1 year. Crimmins and Saito (4)
1995 Vol. 8 No.4
conclude that a return to functioning
is most likely to occur where overall
functional status is higher, loss is
recent, and impairment is not severe,
and that decline in each individual
function is more likely to occur when
general health and functioning levels
are lower.
Health status was the best predictor of
the living arrangements of older people
2 years later; elderly persons who had
no difficulty performing personal care
and home management tasks were more
likely to be alive and living in the com­munity
2 years later (6,7). Changes in
marital status and household composi­tion,
such as death of a spouse, and
declining health are typical life cycle
events related to the aging process
that may encourage elderly persons
to change residence (3). Elderly non­movers
are more likely to live with
their spouse in independent households,
compared with movers who are more
likely to be widowed and to move in
with their children or other relatives (1).
Speare, Avery, and Lawton (13) found
that both the initial level of functional
ability and amount of change in func­tioning
(disability) from 1984 to 1986
predicted changes in living arrangements
and residential mobility for many
elderly people. Because most persons
kept the same living arrangements
between surveys, living arrangements
in 1984 were a strong predictor of
living arrangements in 1986. Disability
(difficulties in ADL's and IADL's)
predicted a change to more dependent
living arrangements; level of disability
had a significant and positive effect on
living with others. Changes in functional
limitations were even more strongly
related to changes in living arrangements
than the initial levels of these measures.
Health status and level of disability
in 1984 were major predictors of both ,
entering an institution in the 2-year
interval and death by 1986. Reasons
for moving indicated that health and
disability were important considerations
for many elderly movers.
Elderly moving decisions are based
on complex and interrelated health and
social motives (5). A life course typo­logy
by Litwak and Longino (8) consists
of three stages of elderly migration:
(1) amenity-related mobility in early
retirement, (2) mobility motivated by
moderate forms of disability, making it
difficult to perform activities of daily
living (ADL's), and (3) institutional
moves in late old age due to chronic
disability. Younger, recently retired
migrants may move because their health
is good-poorer health would deter
them from moving; alternatively, the
poorer health of the older elderly may
prompt them to move to long-term care
institutions (1 1). The elderly who move
because of dependency or assistance
needs have higher use of chronic health
care facilities, rely more on family and
friend support networks, and have lower
overall levels of well-being.
These previous studies show that changes
in health and functional status of the
elderly over time include both improve­ment
and deterioration. Elderly people
adjust their living arrangements or
relocate in response to changes in health
and family support networks. Changes
in health and disability measures should
help predict risks of institutionalization
and mortality. Differences by metro­nonmetro
residence will impact on
planning interventions in functional
loss and appropriate health care services
in local areas.
19
Purpose and Objectives
The purpose of this research is to examine
changes in functional status of the
elderly over time, to assess the relation­ship
between health and adjustments in
living arrangements, and to determine
the impact of residence on these changes.
The expected path of transitions is from
living independently to dependency and
then to institutional care. The basic
tenet of this research was that changes
in health and/or social support would
result in changes in living arrangements
or residential mobility, especially to­ward
places with better access to health
and social services. One would expect
elderly persons initially free of disability
to have better health outcomes over
time and lower risks of entering nursing
homes or dying. One would also expect
both those in declining health and the
recently widowed to experience adverse
changes in living arrangements and
greater risks of institutional care or
death.
This article focuses on transitions over
the 1984 to 1990 interval. The following
questions were addressed: (1) What
changes can be expected in the health
of a cohort of people 70 years and older
over an interval of 6 years and how are
transitions in health affected by initial
health status? Do these patterns differ
by metro-nonmetro residential location?
and (2) What transitions can be expected
in the living arrangements of a cohort
of people 70 years and older over an
interval of 6 years and what effect do
changes in health status and marital
status (widowhood) have on these living
arrangements? Do these patterns differ
by metro-nonmetro residential
location?
20
Data and Methods
The LSOA is designed to measure
changes in functional ability and in
living arrangements (including move­ment
into and out of nursing homes)
of a cohort of older people. The LSOA
describes the continuum from function­ally
independent living in the community
through dependence, possible institu­tionalization,
and finally to death. The
1990 data file provides baseline and
6 years of follow-up information 1 on
7,527 noninstitutionalized persons 70
years or older when they participated in
the 1984 Supplement on Aging to the
National Health Interview Survey. By
1990, 55 percent of the initial sample
completed the interview, 15 percent
were not interviewed for various
reasons, and 30 percent had died.
1Those interviewed in 1984 were reinterviewed
in 1986, 1988, and 1990.
Defmitions
Data include demographic and health
characteristics, including disability
measured by activities of daily living
(ADL's) and instrumental activities of
daily living (IADL's); changes in those
characteristics and reasons for change;
and doctor, hospital, and nursing home
use. The LSOA also contains informa­tion
about elderly persons who (1)
remained outside of institutions (with
unchanged living arrangements, or
living alone, or having moved to another
residence, or living with someone else
in their residence); (2) became institu­tionalized;
and (3) died. Data on physical
limitations provide information on
persons who remained the same or
changed in disability, difficulties with
physical movement (such as walking,
climbing stairs, and lifting), and the
provision of help with ADL's or IADL's.
Two measures are used to assess the degree of disability. Activities of daily
living, or ADL's, are the basic tasks of everyday life, including bathing or
showering, dressing, eating, transferring (getting in or out of a chair or bed),
walking, getting outside, and using/getting to a toilet. When people are unable
to perform these activities, they need help in order to cope, either from other
persons or via mechanical aids or devices. With advancing age, a higher propor­tion
of persons have difficulty performing personal care or home management
activities. ADL's, especially measures of mobility, such as walking and getting
outside, are key indicators of one's ability to live independently in the community
and are also significant predictors of admission to nursing homes, use of paid
home care, and use of both hospital and physician services.
ADL's do not measure the full range of activities necessary for independent
living in the community, and instrumental activities of daily living (IADL' s)
were developed to partially fill this gap. IADL's include meal preparation,
shopping for personal items, managing money, using the telephone, doing heavy
housework, and doing light housework. IADL disabilities capture those activities
that are more complex and less severe than ADL difficulties.
Family Economics and Nutrition Review
Use of medical care is obtained from
nursing home stays since 1984, hospital
stays in the year before the interview,
and contacts with doctors in the year
before the reinterview.
This study defines disability in terms of
a scale of impairment: (1) the healthiest,
who are free of disability, (2) those with
one or more IADL disabilities only, and
(3) the most impaired-those unable
to perform one or more ADL's, which
makes independent living difficult.
Transitions were examined over a
6-year interval, 1984 to 1990, so that
the widest range of change could be
observed. Intervals of 2 and 4 years
may reveal real transitions as well as
more temporary fluctuations.
Because the estimates in this article are
based on a sample rather than the entire
population of those 70 years and older,
the estimates are subject to sampling
error. Unless otherwise noted, all state­ments
of comparison in the text are
statistically significant at the 95-percent
level of confidence.
Results
Changes in Health Status
Among all persons 70 years and older
living in the community in 1984, 35
percent were not disabled by 1990,
13 percent had one or more IADL's
only, 6 percent had one or more ADL's,
25 percent had both ADL and IADL
difficulties, 6 percent were in nursing
Table 1. Health status transitions of the elderly 70 years and older
homes, and 15 percent had died by
1990 (table 1). Among those who were
not disabled in 1984 ( 68 percent of all
persons 70 years and older living in the
community), 48 percent remained free
of disability, 38 percent declined in
health, 4 percent entered nursing homes,
and the remainder had died by 1990.
Lower proportions of initially nondis­abled
persons ended up in nursing homes
or had died by 1990 than the elderly
who had some level of disability in
1984. Not surprisingly, the health of
those with ADL difficulties in 1984 (the
most disabled at baseline) deteriorated
more than their less disabled counter­parts,
and higher proportions ended
up in nursing homes or died. Some
improvement in health can also be seen
Health status 1990 Health status
Not Both ADL's Nursing
Measure of health In 1984 disabled 1 +IADL's1 1 +ADL'~ +IADL's home Deceased
Percentage
Self-reported health
Excellent-very good 40.4 48.2 12.6 5.7 18.8 3.7 11.0
Good 31.4 36.8 13.4 6.5 24.9 6.1 12.4
Fair or poor 28.2 15.7 12.1 6.0 34.9 8.5 22.6
Total 100.0 35.4 12.7 6.0 25.2 5.8 14.7
Functional status
Not disabled 67.8 47.6 13.6 5.9 18.4 3.8 10.8
1 + IADL's 10.0 13.5 20.0 4.5 35.2 8.3 18.5
1 +ADL's 21.7 7.5 6.9 7.3 42.4 10.9 24.9
Total 100.0 35.4 12.7 6.0 25.3 5.9 14.7
11 + IADL's: those with one or more IADL difficulties only. Instrumental activities of daily living (IADL's) include meal preparation, shopping for personal
items, managing money, u ing the telephone, doing heavy housework, and doing light housework.
21 + ADL's: tho e with one or more ADL difficulties, may also have IADL's. Activities of daily living (ADL's) include bathing or howering, dressing, eating,
transferring (getting in or out of a bed or chair), walking, getting outside, and using/getting to a toilet.
1995 Vol. 8 No. 4 21
22
Lower proportions of
initially nondisabled
persons ended up in
nursing homes or
had died by 1990
than the elderly who
had some level of
disability in 1984.
over the 6-year interval-about 15
percent of elderly persons with ADL
or IADL limitations in 1984 improved
in health by 1990.
Elderly persons whose self-assessed
health was very good to excellent at
baseline were more likely to be free
of disability by 1990 than were those
initially in poorer health. A higher
proportion of those initially in fair or
poor health had both ADL and IADL
limitations by 1990; they were also
more likely to have entered nursing
homes or died by 1990. The best health
outcomes over the 6-year interval
occurred to the elderly who rated their
health as excellent or very good at the
baseline, compared with those who
rated their health as fair or poor.
Metro-Nonmetro Residence
(see box table)
As measured by self-assessed health
and functional limitations, the elderly
in suburban areas are healthier than
their counterparts in nonmetro areas
and central cities (12). Nonmetro elders
have more functional limitations and
are also more likely to have certain
chronic conditions, such as arthritis,
that have a strong effect on their ability
to perform various activities of daily
living. Rogers (12) found that health
status differences by residence persist
even when other factors-age, race,
social support networks, income, and
education-are held constant. More­over,
residential location affects health
status indirectly in that nonmetro elders
are more likely to have those charac­teristics
associated with poorer health.
Nonmetro elders are likely to be less
educated and financially worse off than
their metro counterparts, and lower
socioeconomic status is strongly
associated with poor health.
Age and gender distribution of
the elderly, by metro-nonmetro
residence
Characteristic Metro Nonmetro
Age
70-74
75 -79
80-84
85+
Gender
Male
Female
Percentage
42.2
29.8
16.9
11.1
37.6
62.4
39.5
32.9
17.0
10.6
41.5
58.5
The overall pattern of health transitions
was similar by residential location
(fig. 1). No metro-nonmetro differences
were found in rates of institutional care
or death over the 6-year interval. The
major difference in health at the base­line
(1984) was that nonmetro elders
were more likely to have functional
disabilities than metro elders; 26 per­cent
of nonmetro elders had ADL
difficulties in 1984, compared with
20 percent of metro elders.
Among the nondisabled elderly at base­line,
a higher proportion of nonmetro
elders experienced both ADL and
IADL difficulties by 1990; health status
declined for 40 percent of nonmetro
elders, compared with 37 percent of
metro elders. Among the elderly who
had one or more ADL difficulties at
baseline (the most disabled), a lower
proportion of nonmetro elders improved
in health, and a slightly higher propor­tion
entered nursing homes or died,
compared with metro elders. The poorer
initial health and greater decline in
Family Economics and Nutrition Review
Figure 1. Health status transitions of initially nondisabled elderly, by residence, 1990
Not disabled 1 + IADL's
health of the nonmetro elderly may
reflect in part the "aging in place" of
many nonmetro communities, where
the older and more disabled elderly
persons remain in the community.
Transitions in Living Arrangements
Changes in the living arrangements of
the elderly, including entering a nursing
home, reflect adjustments to changes in
their social support networks. In 1984,
48 percent of the elderly age 70 and
older were living with their spouse; by
1990, only 31 percent were still living
with their spouse. By 1990, 6 percent of
the elderly had entered nursing homes,
and 15 percent had died. Among those
initially living with their spouse, 63
percent remained so by 1990. A higher
proportion of those living with others in
1984, compared with their counterparts
who initially lived with their spouse,
ended up in nursing homes or died by
1990.
1995 Vol. 8 No.4
1 + ADL's Both
ADL's + IADL's
Metro-Nonmetro Residence. The main
difference in living arrangements of the
elderly by metro-nonmetro residence is
that in both 1984 and 1990 nonmetro
elders were more likely to be living with
their spouse and less likely to be living
with others (table 2, p. 24). Similar pro­portions
of metro and nonmetro elderly
lived alone. Among those living with
their spouse in 1984, nonmetro elders
were more likely to remain living with
their spouse over the 6-year interval.
For those initially living alone, metro
elders were more likely than nonmetro
elders to remain living alone, but non­metro
elders were more likely to have
died by 1990 ( 18 percent) than their
metro counterparts (13 percent). Social
support is associated with better outcomes
over time, with fewer of those initially
with their spouse entering nursing
homes or dying 6 years later.
• Metro
B Nonmetro
Nursing home Deceased
Effect of Health on Transitions
in Living Arrangements
Initial Health Status. The initial health
status of elderly persons is expected to
affect subsequent living arrangements.
Elderly persons without disabilities
in 1984, regardless of initial living
arrangement, were more likely to be
living with their spouse or alone in 1990
and less likely to have entered nursing
homes or to have died than those who
had either ADL or IADL limitations at
baseline (table 3, p. 25).
The elderly who lived with their spouse
and were not disabled in 1984 were
more likely than other elderly to remain
with their spouse (67 percent) and less
likely to have entered a nursing home
(4 percent) or to have died (13 percent)
by 1990. Seven or 8 percent of those not
initially living with their spouse had
23
Table 2. Transitions in living arrangements of the elderly 70 years and older, by metro-nonmetro residence
Living arrangement
Metro-nonmetro residence In 1984
Metro
With spouse 46.9
Live alone 36.4
With others 16.7
Total 100.0
Nonmetro
With spouse 52.0
Live alone 36.3
With others 11.7
Total 100.0
entered nursing homes by 1990. Four­teen
percent of those initially living
alone and 19 percent of those living
with others had died by 1990. The
combination of being disabled and
living with others at baseline increases
the risk of institutional care and/or
death 6 years later.
Changes in Health. Since initial health
affects transitions in living arrangements,
one would expect changes in health to
operate in a similar way. Comparisons
were made between the elderly whose
health deteriorated and those whose
health either improved or remained the
same over the 6-year interval. By 1990,
46 percent of elders with unchanged
health and 35 percent of those in better
health were living with their spouse,
whereas only 22 percent of those in
24
1990 Living arrangement
With
spouse Alone
Percentage
61.5 15.1
1.1 67.7
1.8 16.1
29.6 34.4
64.5 13.7
1.8 61.8
3.6 20.4
34.6 32.0
declining health lived with their spouse
(table 4, p. 26). Higher proportions of
elderly persons in unchanged or better
health lived alone in 1990, compared
with those in poorer health. The elderly
whose health deteriorated over the
interval were less likely to live with
their spouse and more likely to shift
to nursing homes or to die.
Among the elderly initially living with
their spouse, 80 percent of those with
unchanged health remained with their
spouse over the 6-year interval, as did
70 percent of those in better health;
only 49 percent of those in worse health
remained with their spouse by 1990.
Those whose health improved in the
interval were more likely to be living
alone in 1990 than either those with
unchanged health or worse health.
With Nursing
others home Deceased
6.1 3.6 13.6
11.3 6.9 13.0
55.0 8.9 18.2
16.2 5.7 14.2
4.5 4.2 13.0
10.3 8.5 17.6
49.3 6.7 20.1
11.8 6.0 15.5
Among the elderly with declining
health, those who initially lived with
their spouse had better outcomes by
1990 than their counterparts who either
lived alone or with others at baseline.
Of elderly persons in declining health,
a lower proportion who initially lived
with their spouse had entered nursing
homes by 1990, compared with those
who either lived alone or with others
at baseline. The social support from
living with one's spouse appears to
have a beneficial effect on subsequent
transitions in living arrangements.
Furthermore, the most pronounced
change in living arrangements of the
elderly is the increased institutionaliza­tion
and death for elderly persons in
declining health.
Family Economics and Nutrition Review
Table 3. Transitions in living arrangements of the elderly, by initial health status and living arrangements
1984 Living arrangement 1990 Living arrangement
Health status With With Nursing
Health status in 1984 spouse Alone others home Deceased
Percentage
Lived with spouse in 1984
Not disabled 74.7 66.6 15.0 5.4 2.4 10.5
1 + IADL's1 7.5 56.5 12.1 4.9 7.0 19.5
1 + ADL's2 17.9 48.3 14.8 6.8 7.4 22.8
Total 100.0 62.6 14.8 5.6 3.7 13.3
Lived alone in 1984
Not disabled 65.7 1.6 71.7 10.5 5.7 10.4
1 + IADL's 11.1 1.5 61.9 12.0 7.0 17.6
1 + ADL's 23.2 0.5 52.0 12.1 12.4 23.1
Total 100.0 1.3 66.0 11.1 7.4 14.2
Lived with others in 1984
Not disabled 52.5 3.1 20.2 60.2 4.0 12.5
1 + IADL's 15.6 0.8 17.0 52.3 12.0 17.9
1 + ADL's 31.9 1.4 11.6 43.8 13.9 29.3
Total 100.0 2.2 17.0 53.7 8.4 18.7
11 + IADL's: those with one or more IADL difficulties only. Instrumental activities of daily living (IADL's) include meal preparation, shopping for personal
items, managing money, using the telephone, doing heavy housework, and doing light housework.
21 + ADL's: those with one or more ADL difficulties, may also have IADL's. Activities of daily living (ADL's) include bathing or showering, dressing, eating,
transferring (getting in or out of a bed or chair), walking, getting outside, and using/getting to a toilet.
Metro-Nonmetro Residence. The
effect of changes in the health of elderly
persons on transitions in living arrange­ments
is similar by metro-nonmetro
residence. For elders whose health
either improved or remained unchanged
over the interval, the main residential
difference is that by 1990, nonmetro
elders were more likely to live with
their spouse and less likely to live alone
or with others than metro elders (fig. 2,
p. 27).
1995 Vol. 8 No.4
Even nonmetro elders in deteriorating
health were more likely to live with
their spouse (25 percent) than metro
elders (20 percent). Otherwise, the
effect of declining health is the same
by residential location, with 10 percent
entering nursing homes and 25 percent
dying.
Marital Change
Changes in marital status, especially
widowhood, are good indicators of
the shifting social support networks
of elderly people. As previously seen,
strong social support tends to have a
beneficial effect on health. Also,
married elderly persons living with
their spouses are more likely to rate
their health as very good to excellent
25
Table 4. Transitions in living arrangements of the elderly, by change in health status
Health status 1984 Living
change arrangement
Better health
With spouse 49.5
Live alone 36.6
With others 13.9
Total 100.0
No change in health
With spouse 56.1
Live alone 32.3
With others 11.6
Total 100.0
Worse health
With spouse 44.1
Live alone 38.2
With others 17.7
Total 100.0
-Indicates no elderly were in this living arrangement.
and less likely to have difficulty
performing the various activities of
daily living than widowed, divorced, or
separated persons. A higher proportion
of elders with no change in marital
status remained free of disability by
1990 (42 percent), compared with those
who became widowed (38 percent) in
the interval. Among the nondisabled
elderly at baseline, fewer widows
remained free of disability by 1990
26
1990 Living arrangement
With
spouse Alone
Percentage
69.6 26.8
85.7
4.3 22.7
35.1 47.8
79.5 15.8
3.0 87.7
5.7 26.2
46.2 40.2
48.6 12.7
0.7 52.7
0.8 12.4
21.8 27.9
(48 percent) than those without a
change in marital status (54 percent).
At advanced ages, nearly all changes in
marital status result from the death of
one's spouse, which clearly affects the
living arrangements of elderly persons.
Widowhood involves a shift from living
with one's spouse to living alone or with
others (fig. 3). It also signals a shift to
living in nursing homes.
With Nursing
others home Deceased
3.6
14.3
73 .0
17.1
4.7
9.3
68.2
13.6
6.7 6.9 25.1
11.5 12.1 23.1
46.8 12.5 27.6
15.6 9.9 24.8
Metro-Nonmetro Residence. Among
the elderly without a change in marital
status, nonmetro elders initially living
with their spouse are slightly more
likely than metro elders to remain living
with their spouse (82 vs. 78 percent)
and less likely to live with others (4 vs.
6 percent) (fig. 4, p. 28). Nonmetro
widows are somewhat more likely to
enter nursing homes than their metro
counterparts (10 vs. 6 percent). In
general, transitions in living arrange­ments
by changes in marital status are
similar by metro-nonmetro residence.
Family Economics and Nutrition Review
Figure 2. Transitions in living arrangements of the elderly, by
residence and change in health status, 1990
In better health
Metro
Non metro
No change in health
Metro
Non metro
In worse health
Metro 205°o 285% ~~ 24.6%
Non metro
• With spouse • Alone • With others Nursing home Deceased
Figure 3. Transitions in living arrangements of the elderly
initially living with their spouse, by change in marital status, 1990
Widowed by 1990 No change in marital status
• With spouse • Alone • With others Nursing home
1995 Vol. 8 No.4
Change in Residence
Residential mobility among the elderly
is lower than among the general popula­tion.
Whereas 44 percent of all persons
age 5 and older moved between 1984
and 1990, only 9 percent of elderly persons
age 70 and older moved during this
period. This percentage was only slightly
higher in metro areas than in nonmetro
areas. The most frequently given reasons
for moving are associated with poor
health, social support networks
(remarriage, moving to be closer to
family), fmancial considerations, and
other/multiple reasons. Among the
nonmetro elderly, 28 percent moved
because of poor health, 18 percent moved
to be closer to social support networks,
15 percent moved for financial reasons,
and 18 percent for other or multiple
reasons. A similar pattern is found
among metro elderly persons .
Since poor health is frequently given as
a reason for moving, changes in health
are expected to influence residential
mobility. Among the metro elderly
who were initially nondisabled, a
lower proportion of movers had moved
between counties (20 percent) than
had moved locally. In contrast, among
nonmetro elders who were initially free
of disability, 31 percent of movers had
moved between counties. Perhaps the
metro elderly in better health moved
to nonmetro areas after retirement for
amenity-related reasons, whereas the
mobility of more disabled nonmetro
elders was motivated by the location
of health care services in metro areas.
Although type of residence at destina­tion
was not determined, some nonmetro
elders, particularly those at greater
distances from metro areas, may have
moved to be closer to relatives and/or
health care and social services in metro
areas .
27
28
The elderly whose
health deteriorated
over the interval
were less likely
to live with their
spouse and more
likely to shift to
nursing homes or
to die.
Figure 4. Transitions in living arrangements of the elderly
initially living with their spouse, by residence and change in
marital status, 1990
~----t 3.4%
4.0%
Metro Nonmetro Metro Nonmetro
Widowed by 1990 No change in marital status
• With spouse • Alone • With others D Nursing home
Summary and Conclusion
This study, based on 6-year rates of
functional change, supports the findings
of previous research showing that some
elderly exhibit long-term functional
improvement but more commonly, there
is a decline in their functional ability.
Both the initial level of functional ability
and amount of change in functioning
influenced changes in living arrange­ments
and residential mobility. Further­more,
this research documents the link
between functional decline and increased
risk of institutionalization, death, and
other changes in living arrangements
and residence.
The majority of elderly people living in
the community are in good health, and
about half of the elderly who were not
disabled in 1984 remained so 6 years
later. The level of initial disability
affected health outcomes over time,
with fewer of the initially nondisabled
entering nursing homes or dying by
1990. In general, the nonmetro elderly
had poorer initial health than their
metro counterparts and experienced
a somewhat greater decline in health
status than their metro counterparts over
a 6-year interval. Rates of institutional
care and dying were similar by metro­nonmetro
residence.
Family Economics and Nutrition Review
Elderly people make adjustments in
their living arrangements in response
to changes in their health and social
support networks. Elderly persons
initially living with their spouse were
less likely to enter nursing homes or
die within 6 years, compared with those
who initially lived either alone or with
others. Nonmetro elderly persons were
more likely to be living with their spouse
than the metro elderly, and such social
support may ameliorate their poorer
health to some extent. Having some­one
in the household, primarily one's
spouse, who could offer assistance is
beneficial to the health of the elderly and
acts as a buffer to institutionalization.
Regardless of metro-nonrnetro residence,
deteriorating health results in an in­creased
likelihood of entering a
nursing home or dying in the interval.
The elderly who were initially disabled
and also living with others had the
greatest risk of institutionalization and
death. Advanced age and widowhood
were also strong predictors of a person's
entering a nursing home or dying.
About 9 percent of the elderly moved
between 1984 and 1990. Decisions to
move are based on complex interrelated
health and social motives. The most
common reasons for moving are associ­ated
with poorer health, moving closer
to family, financial considerations, and
other/multiple reasons. Some nonmetro
elders may move to be closer to rela­tives
and to obtain health care services
in metro areas. This may be especially
true of the more disabled nonmetro
elderly who relocate for both health­care
resource and social support
considerations.
A substantial and growing number of
the elderly have, or are at risk of devel­oping,
chronic conditions that impair
their ability to function independently.
1995 Vol. 8 No.4
The ability or inability of the elderly to
obtain help with difficult personal care
activities is an important factor in deter­mining
which individuals are able tore­main
in the community and which must
enter nursing homes or other institutions
for needed care and assistance. The
incidence and duration of disability has
References
important consequences for long-term
care and federal spending as well as for
effective local planning for health care
and other services. Furthermore, residen­tial
moves that are strongly associated
with health factors can have potentially
large impacts on local public resources,
particularly health and social services.
1. Biggar, J.C. 1980. Who moved among the elderly, 1965 to 1970: A comparison of types
of older movers. Research on Aging 2(1):73-91.
2. Branch, L.G. and Ku, L. 1989. Transition probabilities to dependency, institutionalization,
and death among the elderly over a decade. Journal of Aging and Health 1(3):370-408.
3. Bryant, E.S. and El-Attar, M. 1984. Migration and redistribution of the elderly: A challenge to
community services. The Gerontologist 24:634-640.
4. Crimmins, E.M. and Saito, Y. 1990. Getting Better and Getting Worse: Transitions in
Functional Status Among Older Americans. Paper presented at the annual meeting of the
Population Association of America, Toronto, Canada.
5. Heaton, T.B ., Clifford, W.B., and Fuguitt, G.V. 1980. Changing patterns of retirement
migration: Movement between metropolitan and nonmetropolitan areas. Research on Aging
2(1):93-104.
6. Kovar, M.G. 1987. The Longitudinal Study of Aging: Some Estimates of Change Among
Older Americans. Proceedings of the 1987 Public Health Conference on Records and Statistics,
National Center for Health Statistics. DHHS Publication No. 88-1214.
7. Kovar, M.G. 1988. Aging in the eighties, people Hving alone-two years later. Advance
Data from Vital and Health Statistics No. 149. National Center for Health Statistics,
Hyattsville, MD.
8. Litwak, E. and Longino, Jr., C. 1987. Migration patterns among the elderly: A developmental
perspective. Gerontologist 27:266-272.
9. Longino, Jr., C., Wiseman, R. , Biggar, J., and Flynn, C. 1984. Aged metropoHtan­nonmetropolitan
migration streams over three census decades. Journal of Gerontology
39:721-729.
10. Manton, K.G. 1988. A longitudinal study of functional change and mortality in the
United States. Journal Of Gerontology: Social Sciences 43(5):153-161.
11. Patrick, C. H. 1980. Health and migration of the elderly. Research on Aging 2(2):233-241.
12. Rogers, C. C. 1993. Health Status and Use of Health Care Services by the Older Population.
Rural Development Research Report No. 86.
13. Speare, Jr., A., Avery, R., and Lawton, L. 1991. Disability, residential mobiHty, and
changes in living arrangements. Journal of Gerontology 46(3): 133-142.
29
30
Trends in Food and Alcohol
Consumption Away From Home
By Nancy E. Schwenk
Consumer Economist
Center for Nutrition Policy and Promotion
In the early 1970's, American households spent about one-fifth of their
food dollar on food away from home. From the mid-1980's to the present,
households have been spending about twice that proportion on food away
from home. According to the 1992 Consumer Expenditure Survey, U.S.
households allocated 38 percent of their food dollar to food away from home
and 46 percent of their alcohol dollar to alcohol consumed outside the home.
Consumers who spent the greatest share of their food dollar on food away
from home were in the highest income quintile, under age 25, or living
alone. Sales at eating and drinking places were up 134 percent between
1980 and 1993, with a corresponding 48-percent increase in the number of
employees at these establishments. Factors that influence the decision to
dine out include the increasing numbers of women in the labor force, the
trend toward more one-person households, price competition among
restaurants, and the interest in restaurants that offer some type of
entertainment. Consumption trends when dining out, demographic
influences, and other issues of concern to nutritionists, food policymakers,
and restaurateurs are presented.
e make food choices many
times each day-what to
eat, how much to eat, when
to eat, and where to eat.
Food and drinks can be purchased at
food retail stores and consumed either
at home (in someone's home) or pur­chased
away from home in restaurants
and other establishments. A number of
factors influence our choice of where to
eat, including time, convenience, cost,
and nutrition. Many people, after a full
day at work, Jack the energy or interest
needed to cook. Very often, people
dine in restaurants because they make
a last-minute decision and just feel like
going out or they want to socialize
with friends and family. This article
examines various trends related to food
and drinks purchased away from home,
including food prices; aggregate,
household, and government expendi­tures;
and restaurant trends-food
retailing and findings from surveys of
restaurant customers.
Prices
In 1994, prices for food, as measured
by the Consumer Price Index (CPI),
rose 2.4 percent over 1993 (table 1).
This annual increase was slightly less
than the 2.6-percent increase for all
items during the same period. The price
of food at eating places-food away
from home-was up 1.7 percent,
Family Economics and Nutrition Review
Table 1. Annual percent change in prices of food and alcohol, 1993-94
and average annual change, 1984-93. Consumer Price Index for all
urban consumers [1982-84=100]
Annual percent Average annual
change percent change
Group 1993-94 1984-93
All items 2.6 3.8
Food 2.4 3.5
At home 2.9 3.5
Away 1.7 3.7
Lunch 1.7 3.7
Dinner 1.8 3.5
Other/snacks 1.6 3.8
Alcohol 1.3 4.1
At home .2 3.6
Away 2.5 5.3
Source: U.S. Depamnent of Labor, Bureau of Labor Statistics, CPI Detailed Report, January issues.
whereas the price of food purchased at
supermarkets and other grocery stores­food
at home-was up 2.9 percent.
Between 1984 and 1993, the average
annual percent change in prices for all
goods and services was 3.8 percent,
compared with 3.5 percent for food at
home and 3.7 percent for food away
from home.
Compared with the other major com­ponents
of the CPI, prices for food
increased less in 1994 than housing
(2.5 percent), transportation (3 .0 per­cent),
medical care (4.8 percent), enter­tainment
(2.9 percent), or personal and
educational expenses (5.9 percent) (14).
Since 1989, the increa e in the overall
CPI has been greater than the increase
for food away from home (fig. 1, p. 32).
In 1994, prices for alcohol, as measured
by the CPI, rose 1.3 percent, less than
the average annual increase of 4.1
1995 Vol. 8 No.4
percent between 1984 and 1993
(table 1). The price of alcohol away
from home increased 2.5 percent in 1994,
compared with an increase of only 0.2
percent for alcohol at home (15).
Expenditures
Aggregate Expenditures
Americans spent $197.8 billion on
eating away from home in 1993, an in­crease
of 9 percent over 1992 (table 2,
p. 32). The expenditure for food at home
in 1993 was $329.5 billion, only 2 per­cent
more than was spent in 1992. As a
portion of disposable personal income,
food away from home increased from
3.7 percent in 1970 to 4.2 percent in
1993; food at home, however, decreased
from 10.3 percent of disposable per­sonal
income in 1970 to 7.0 percent in
1993. Although dollars spent on food
increased greatly over the years, the
gain in disposable income was greater.
Americans spent
$197.8 billion on
eating away from
home in 1993, an
increase of 9 percent
over 1992.
31
After adjustment for inflation, food
expenditures per capita increased 21
percent between 1970 and 1993, while
per capita income increased 45 percent.
As a result, the proportion allocated to
total food (home and away) dropped
19 percent between 1970 and 1993. As
income rises, the proportion allocated to
food goes down, as there is more money
to spend on other discretionary items
(12).
Americans spent $85.5 billion on
alcohol in 1993, of which $37.4 billion
was on alcoholic drinks consumed away
from home at eating and drinking
places; the remainder was spent on
packaged alcohol purchased at liquor
stores, food stores, and convenience
stores. Although there was an increase
from 1992 of $732 million, or 2 percent,
in spending on alcohol away from home,
this was offset by a 2-percent decrease
of about $1.1 billion spent on alcohol at
home (12).
Household Expenditures
The 1992 Consumer Expenditure
Survey (CE), conducted by the Bureau
of Labor Statistics, was used to obtain
data on households' food and beverage
purchases away from home. 1 Expendi­tures
on meals and snacks eaten away
from home in 1992 averaged $1,631 per
household (table 3). Whites and others2
spent $1,717, whereas Blacks spent
1The Consumer Expenditure Survey defines food
away from home as all meals or snacks purchased
in restaurants, cafeterias, cafes, drive-ins, carry­outs,
and vending machines, including trips, plus
meals as pay, school lunches, special catered
affairs, and meals away from home on trips.
Food at home is the total expenditures for food
and grocery stores or other food stores (excluding
nonfood items) and food prepared by the
consumer.
2Category includes people who are White,
American Indian, Aleut, Eskimo, Asian, and
Pacific Islander.
32
Figure 1. Changes in consumer prices of all items and food
away, 1984-94
Annual percent
change in CPI-U
6
5
4
3
2
I
\;
I
All items
Food away
0 1~9784~----~876------~88~----~90~----~92~-----9~4
Source: U.S. Department of Labor, Bureau of Labor Statistics, CPJ Detailed Report, January issues.
Table 2. Food expenditures by families and individuals as a share of
disposable personal income, selected years, 1970-93
Disposable Expenditures for food
Year personal income At home Away from home Total
Billion$ Billion$ Percent Billion$ Percent Billion$
1970 722.0 74.2 10.3 26.4 3.7 100.6
1975 1,150.9 115.2 10.0 45.9 4.0 161.1
1980 1,952.9 179.1 9.2 85.2 4.4 264.4
1985 2,943.0 230.7 7.8 129.4 4.4 360.1
1990 4,050.5 306.7 7.6 172.4 4.3 479.1
1991 4,236.6 320.6 7.6 174.9 4.1 495.5
1992 4,505.8 322.1 7.1 181.7 4.0 503.7
1993 4,688.7 329.5 7.0 197.8 4.2 527.4
Source: Putnam, J.J. and Allshouse, J.E., 1994, Food Consumption, Prices, and Expenditures, 1970-1993,
Statistical Bulletin No. 915, U.S. Department of Agriculture, Economic Research Service.
Family Economics and Nutrition Review
Table 3. Average annual expenditures of CE households on food away
from home, by demographic characteristics, 1992
Demographic characteristic
All
Income quintiles
Lowest
2nd
3rd
4th
Highest
Age (years)
Under 25
25-34
35-44
45-54
55-64
65 and over
Composition of household
Husband and wife only
Husband and wife with children
Single parent (at least one child under age 18)
One person
Race
White and other
Black
Region
Northeast
Midwest
South
West
Urbanicity
Urban
Rural
Mean dollars
$1,631
612
1,030
1,464
2,092
3,168
1,181
1,732
2,017
2,131
1,521
987
1,921
2,170
1,131
1,004
1,717
953
1,686
1,610
1,550
1,730
1,675
1,349
Source: U.S. Department of lAbor, Bureau of lAbor Statistics, 1992, Consumer Expenditure Survey,
unpublished data.
1995 Vol. 8 No. 4
$953. Households in the West spent
the most ($1,730), whereas households
in the South spent the least ($1,550).
Spending on food away from home
increased as income increased. Those
who had higher than average expendi­tures
included married-couple house­holds
(both with and without children),
homeowners, and those in urban areas,
whereas those who had lower than aver­age
expenditures included households
headed by someone either under age 25
or age 65 or older, single-parent house­holds,
single persons, and those in rural
areas (14).
Overall, households spent 38 percent
of their food dollar on food eaten away
from home in 1992 (fig. 2, p. 34). A
similar proportion was spent in 1991,
down from the 42-percent share spent
in both 1990 and 1989. By comparison,
about 20 percent of the food dollar in
the early 1970's was spent on food
away from home.
Households that allocated a greater than
average portion of their food dollar to
food away from home included those in
the highest two income quintiles, those
headed by a person age 54 and younger,
husband-wife only households, and
Whites and others. One-person house­holds
also allocate a very large portion
of their food dollar ( 44 percent) to
eating out (14). The number of one­person
households increased from
21.4 million in 1970 to 41.8 million in
1992 (13). Households that allocated a
smaller than average portion to food
away from home included those in the
lowest three income quintiles, those
headed by a person age 55 and older,
those families with children at home
(both husband-wife and single-parent),
and Blacks (14).
33
Figure 2. Portion of the food dollar allocated to food away from home, 1992
Demographic characteristic
All
Income quintiles
Lowest
2nd
3rd
4th
Highest
Age
Under25
25.34
35-44
45-54
55-64
65 and over
Race
White and other
Black
Percent
38 ...................
27
31 r-----------------~-.
36
41
45
45
41
39
41
35 r---------------------,----'
31
39 .................. .
30
• ~ All household average • < All household average
Source: U.S. Department of Labor, Bureau of Labor Statistics, 1992, Consumer Expenditure Survey, unpublished data.
In 1992, 46 percent of alcohol expendi­tures
were for alcohol consumed away
from home (fig. 3). This proportion was
highest in the highest income quintile,
in the two youngest age categories
(under 25 years old and 25 to 34 years),
and among those employed in technical/
sales or managerial/professional occupa­tions
or self-employed (14).
34
According to the 1992 Consumer
Expenditure Survey, dinner accounted
for nearly half of dining-out expendi­tures,
lunch for about one-third, with
the remainder consisting of snack,
breakfast, and brunch expenditures
(fig. 4, p. 36). Blacks spent equally on
lunch and dinner out, whereas Whites
and others spent 35 percent more on
dinner out than on lunch out. By occupa­tional
groups, managers and professionals
spent the most on breakfasts and lunches
out, whereas construction workers spent
the most on snacks. Retired people
spent the least on dining out (14).
The cost of food eaten on out-of-town
trips in 1992 averaged $167 for all
households. Households residing in the
Family Economics and Nutrition Review
Figure 3. Portion of the alcohol dollar allocated to alcohol away from home, 1992
Demographic characteristic
All
Income quintile
Lowest
2nd
3rd
4th
Highest
Age
Under25
25-34
35-44
45-54
55-64
65 and over
Occupation
Self-employed
Managerial/professional
Technical/sales
Service worker
Construction worker
Operator
Retired
Percent
46
44
39
45
45
47
53
57
43
46
34
31
49
50
52
44
39
39
33
• ;;:.. All household average • < All household average
Source: U.S. Department of Labor, Bureau of Labor Statistics, 1992, Consumer Expenditure SuNey, unpublished data.
West spent the most ($201), whereas
households in the South spent the least
($139). Another type of expenditure for
food away from home is school lunches,
which averaged $46 for all households.
Husband/wife families with an oldest
child between 6 and 17 years had the
highest average school-lunch expendi­ture
($184) (14).
1995 Vol. 8 No. 4
Governmental Expenditures
In addition to household expenditures
for school lunches, households make
further expenditures on food away
from home indirectly through their tax
dollars. The Federal Government, in
cooperation with State and local govern­ments,
operates five food assistance
programs to provide meals and snacks
to preschool and school-age children.3
Expenditures for these programs-the
National School Lunch, School
Breakfast, Special Milk, Child and
Adult Care, and Summer Food Service
Programs, totaled $7.1 billion in fiscal
1993, a 6.6-percent increase over fiscal
1992 (3).
3Food assistance programs, such as the Food
Stamp Program and WIC-the Special Supple­mental
Nutrition Program for Women, Infants,
and Children, lower food-at-home expenditures
if their value is not included.
35
The largest of these programs, the
National School Lunch Program, served
an average of 24.9 million children per
day in fiscal 1993 at a cost of $4.1
billion, up from $3.9 billion spent and
24.7 million children per day served
in fiscal 1992. The School Breakfast
Program served an average of 4.9 mil­lion
children per day in fiscal 1992 and
5.3 million children per day in fiscal
1993. Expenditures rose from $787
million in fiscal 1992 to $868 million
in fiscal 1993. Many of these meals
are available free or at reduced prices to
economically qualified households (3).
The Child and Adult Care Food Program
serves meals to children in nonresiden­tial
child-care centers and family day­care
homes and to chronically impaired
adults and persons over age 60 enrolled
in adult day-care centers. The program
served 1.3 billion meals in fiscall993
with an average daily participation of
2.06 million people. This was up from
1.2 billion meals served and an average
daily participation of 1.93 million in
fiscal 1992 (3).
Restaurant Trends
Food Retailing­Sales
and Employment
Between 1980 and 1990, retail sales
at eating places rose 120 percent and
continued to climb through 1993. Retail
sales at drinking places increased 22
percent between 1980 and 1990, peaked
in 1992, then fell 2 percent in 1993
(1 3). Declining alcohol consumption,
with price stability, was responsible for
declining alcohol sales. Consumption of
alcohol by the adult population age 21
years and over, which peaked in 1981
at 43.1 gallons per person, declined 8
percent between 1990 and 1993-from
36
Figure 4. Allocation of dining-out expenditures, 1992
47%
Dinner
Snacks,
drinks
Breakfast,
brunch
Source: Calculated from U.S. Department of Labor, Bureau of Labor Statistics, 1992,
Consumer Expenditure Survey, unpublished data.
40.0 to 36.8 gallons. The consumption
of distilled spirits declined 14 percent,
and wine and beer consumption de­clined
7 percent during this time (12).
Millions of Americans are dependent
on the food retailing industry for their
livelihood-about 3 in 10 employees in
service occupations work in food prepa­ration
and service (13). In 1993, 6.9 mil­lion
workers were employed at the over
400,000 eating and drinking places in
the United States, up from 6.5 million
workers in 1990 and 4.6 million in 1980
(table 4). The average hourly earnings
of production workers in this industry in
1993 was $5.35, up from $3.69 in 1980.
The number of employees at eating and
drinking places grew at an annual rate
of 4.5 percent during the 1980's but is
projected to slow to a growth rate of
1.9 percent between 1990 and 2005,
based on assumptions of moderate
growth. Specific job categories that are
projected to grow faster than this rate
are: Restaurant cooks (2.8 percent annual
increase), food counter and fountain
workers (2.3 percent), dining room and
cafeteria attendants and bar helpers
(2.3 percent), short-order and fast-food
cooks (2.2 percent), food service managers
(2.2 percent), and food preparation
workers (2.1 percent). The growth rate
for waiters is projected to slow to I. 7
percent (1 3).
Family Economics and Nutrition Review
T~ble. 4. Number of employees and average earnings at eating and
drmkmg places, selected years, 1980-93
Production workers 1
Total employees average earnings
Year (in thousands) (dollars per hour)
1980 4,626 $3.69
1990 6,509 4.97
1991 6,571 5.18
1992 6,485 5.29
1993 6,863 5.35
I Over 90 percent of employees at eating and drinking places are classified as production workers.
Source: U.S. Department of Commerce, Bureau of the Census, 1994, Statistical Abstract of the United
States, 1994, [I 14th ed.]
Findings from Surveys of
Restaurant Customers
Impulse Meals. A recent Roper Starch
Worldwide survey found that 71 percent
of Americans would eat most dinners
at home, even if money were no object.
However, 56 percent had eaten dinner
at a restaurant or fast-food place during
the week preceding the survey. Only 16
percent of survey respondents had not
eaten out during the previous month.
Most restaurant meals appear to be
bought on impulse, serving as a time­and/
or labor-saving device for many.
According to Waldrop (16), the most
recent decision to eat out for 51 percent
of Americans was made at the last
minute. The most likely age group to
decide to eat out on impulse was young
adults under age 30 (64 percent).
Seventy-five percent of trips to fast­food
restaurants and 42 percent of
dinners at full-service restaurants were
last-minute decisions. The most frequent
reason given by Roper respondents for
eating dinner at a restaurant was that
the respondent "just felt like going out."
The next most frequent reason was
"socializing with friends" (16).
1995 Vol. 8 No.4
Generation X. American consumers
born between 1965 and 1976, often
referred to as "Generation X," allocate
nearly 25 percent of their discretionary
income to eating out, and they go out to
dinner more than any other age group,
according to the National Restaurant
Association's Consumer Reports on
Eating Share Trends. At all types of
restaurants, this age group prefers
Mexican food, hamburgers, and sand­wiches,
with fast food accounting for
80 percent of their restaurant visits.
These young adults tend to be less
concerned about health and nutrition
and are less likely than middle-age
adults to consciously restrict their sugar
and cholesterol intake. Generation X
is interested in restaurants that offer
all-you-can-eat specials, as well as
restaurants with a lively, entertaining
atmosphere, such as display cooking
or live music (5).
Men. Men, particularly those age 55 or
older, are likely to be frequent patrons
of lower check (average check under
$10 per person) table-service restaurants.
According to Nutrition and Restaurants:
A Consumer Perspective, one-third of
this sex/age group ate at this type of
restaurant more than once a week. In
contrast, only 13 percent of women
reported doing so (11).
Fast-food restaurants were also more
popular among men, particularly those
under age 35. Twenty-five percent of
men, and nearly 33 percent of men ages
18 to 24, frequented fast-food places
more than once a week, compared with
13 percent of women. Those age 55
and older were the least likely group
to patronize fast-food restaurants. Men
ages 35 to 54 were the group most likely
to frequent self-service cafeterias or
buffets, with 14 percent patronizing
these establishments more than once
a week (11).
Carryout food from all types of restau­rants
was popular among men, with 20
percent reporting having purchased a
carryout meal more than once a week.
This figure rose to 30 percent for men
ages 18 to 34. Also popular among
these young men was purchasing a
freshly prepared meal from a super­market,
convenience store, or deli-17
percent reported purchasing this type of
meal more than once a week, compared
with 12 percent of all men and 6 percent
of all women. Delivery of meals from
fast-food or table-service restaurants
was used by only 2 percent of consumers
more than once a week but by 39 per­cent
at least once a month. Home
delivery was reported most often by
families with children and men ages
18 to 34 (11).
37
38
When dining out for
a special occasion,
55 percent of adults
were not concerned
with nutrition . ..
Menu Choices. According to the
National Restaurant Association's 1993
survey, Nutrition and Restaurants: A
Consumer Perspective, most Americans
would like to see restaurants offer a
wider array of healthy menu selections,
including food offerings for people on
restrictive diets. However, restaurant
patrons' behavior does not always
mirror their concerns. Only 55 percent
of adults reported that they pay atten­tion
to the nutritional content of the
food they eat. In addition, half of adults
said they eat whatever they want when­ever
they feel like it. When dining out
for a special occasion, 55 percent of
adults were not concerned with nutrition,
a proportion that had not changed
substantially since 1986 when the
study was first undertaken (10).
Over three-fourths of respondents
indicated that restaurants should offer
different sized portions for different
sized appetites, lessening the tendency
to overeat. The National Restaurant
Association's 1993 Menu Analysis
compared 66 representative restaurant
menus from 1988 with 1993 menus from
the same restaurants: menus offering
entrees with more than one portion size,
such as "queen-size" and "king-size"
steak, increased 12 percent during this
period (10).
Whereas, the 1986 survey reported
consumers' concerns about sodium in
menu items, the 1992 survey revealed
consumers' concerns about dietary fat
( 10 ). However, customers are often
unwilling to sacrifice the satisfying taste
of fat when dining out, regardless of
what they say to market researchers.
For example, recent attempts by the fast­food
industry to offer lowfat hamburgers,
skinless chicken, and light Mexican fare
were not received well by consumers.
The targeted audience of health­conscious
consumers who do not
ordinarily frequent fast-food restaurants
did not materialize. These marketing
failures point out that consumers are
very inconsistent when it comes to fat
in foods-they tell researchers they
are more interested in lowfat foods
than they really are (1).
Another way in which restaurants are
addressing customers' concerns about
fat is by offering a wider selection of
meatless entrees, particularly pasta.
The number of meatless main dishes
on menus was up 23 percent between
1988 and 1993. In addition, 71 percent
of surveyed customers reported that
restaurants are usually responsive to
special requests, such as serving salad
dressing on the side. Consumers may
have had their fill of nutrition advice
from the news media-nearly half of
respondents said they were tired of
hearing about which foods are good
or bad for them (10).
Ethnic Entrees. The availability of
ethnic-inspired entrees on restaurant
menus rose from 37 percent of menu
offerings in 1988 to 47 percent in 1993,
according to the 1993 Menu Analysis.
Most popular among ethnic offerings
were Mexican and Italian, although
Chinese, Thai, and Japanese were each
up about 10 percentage points during
this period. Ethnic food may address
two possible consumer concerns:
nutrition and cost. Unlike meat-centered
American dishes, many ethnic dishes
use meat only as a minor ingredient, if
at all. Meatless dishes tend to be among
the lowest priced menu entrees (8).
Family Economics and Nutrition Review
Pasta. Pasta orders in restaurants
increased 38 percent between 1989
and 1993. During this time, the number
of pasta entrees on menus rose nearly
60 percent. In addition, 90 percent of
consumers surveyed in the 1993 Menu
Analysis believed that pasta is a good
value for the money and a healthy
choice. Pasta experienced its greatest
growth in casual-dining restaurants,
where orders increased 69 percent from
1989 to 1993. Pasta also gained in popu­larity
in fine dining, or higher check
(average check over $1 0) restaurants,
as well as in fast-food restaurants.
Establishments offering pasta dishes
are responding to customer demand for
vegetarian alternatives-over 40 per­cent
offered meatless pasta dishes in
1993, up from about 25 percent in 1988.
Restaurant patrons with household in­comes
of $60,000 or more and those
with professional or managerial occupa­tions
were the most likely groups to
order pasta, whereas blue-collar
customers were the least likely (6).
Fast Food and Pizza. Several fast-food
hamburger chains have added "value
meals," also called "combo meals," to
their menus. According to the National
Restaurant Association, these value
meals, which usually consist of a large
sandwich, French fries, and a soft drink,
are especially popular at dinner with a
bargain-hungry public. Customers may
have shifted some of their allegiance
from pizza places to quick-service
hamburger places in 1992. According to
Edmondson (2), the market for pizza is
now mature, having reached a low but
stable growth rate. Customer counts at
pizza places were up only 1 percent in
1992, compared with a 5-percent gain
in 1989 and 3-percent gains in both
1990 and 1991. Customer counts at
quick-service hamburger places were
1995 Vol. 8 No.4
up 3 percent in 1992. In an effort to
compete with value meals, the major
pizza chains added bigger pizzas to
their menus (9).
Quick-service pizza restaurants are
heavily concentrated in the Northeast
and the Midwest, whereas many areas
of the South have none. After rapid
growth over several decades, the num­ber
of stores (about 58,000) has not
changed since 1992. The most frequent
customers are young, affluent, college­educated
adults (2).
Coffee. In 1970, Americans drank 33.4
gallons of coffee per person, but by
1988, consumption had fallen to 25.7
gallons (12). However, the popularity
of specialty coffees and the emergence
of coffee bars helped per capita con­sumption
reach 27.8 gallons in 1992.
Coffee bars may serve as alternatives to
traditional alcohol bars for socializing.
Operators of table-service restaurants
reported in the National Restaurant
Association's 1994 Tableservice Operator
Survey that customers were ordering
more specialty and premium coffees
in 1993, particularly in restaurants with
higher average checks. Coffee consump­tion
at fast-food places increased by 36
percent between 1980 and 1993. Fast­food
places accounted for 42 percent of
coffee consumed away from home in
1993. Coffee is increasing in popularity
among young adults and even teenagers.
There is a growing perception among
younger consumers that coffee can be
enjoyed throughout the day, not only at
mealtime. Coffee ordered as a between­meal
snack has risen steadily in recent
years, whereas coffee ordered with
lunch or dinner has declined. In 1993,
39 percent of coffee orders occurred
with breakfast, 24 percent with lunch,
22 percent with dinner, and 15 percent
as snacks (4).
Premium Beer. During the 1990's,
restaurant sales for all types of beer
have been increasing, according to the
National Restaurant Association's
Tableservice Trends 1994. Although
consumption per person dropped 7 per­cent
between 1990 and 1993, this was
offset by a price increase of 16 percent.
In particular, consumption of premium
beers--craft-brewed, made by a micro­brewery
or regional brewer in small
batches-rose. The number of craft
breweries operating in the United States
grew from 30 in 1985 to 382 in 1993.
The number of barrels of craft-brewed
beer increased 40 percent between
1992 and 1993, whereas there was little
change in the number of barrels of total
beer sold domestically. Consumers
seem to be more willing to pay pre­mium
prices for beer than for other
products-over 40 percent of surveyed
adults believed that some blends of beer
are different and worth paying more for.
According to the Consumer Reports on
Eating Share Trends, between 1989 and
1993, the proportion of consumers with
household incomes of $60,000 or more
who ordered beer in restaurants increased
from 15 to 25 percent. At the same
time, the proportion of consumers
with household incomes of less than
$40,000 who ordered beer at restaurants
declined from 58 to 45 percent. During
the 1980's, 43 percent of beer orders in
restaurants were from customers with
professional or managerial occupations;
22 percent were from blue-collar workers;
20 percent were from agricultural workers,
retirees, or unemployed people; and 15
percent were from those in clerical or
sales jobs (7).
39
Summary
For many years, prices of food away
from home have risen more slowly than
prices for most other major commodities,
adding to the appeal of dining out.
Demographic characteristics that influ­ence
household expenditure for food
away from home include: Income
(spending increases as income in­creases),
race (higher among non-Black
households), age of household head
(highest among those ages 25 to 64),
household composition (higher in
married-couple families both with and
without children), and region (higher in
the West and in urban areas). Alcohol
consumption among adults has declined
over the past decade.
Restaurant trends include: a deeper
understanding of consumers' requests
for wider menu selections and choices
of p