An evaluation of the pilot food certificate program in Chicago, Illinois and Bibb County, Georgia

              \
An Evaluation of the Pilot Food Certificate Program
in Chicago, Illinois and Bibb County, Georgia
A Research Report Prepared by:
Dr. Robert E. Wunderle
Dr. David L. Call
The Graduate School of Nutrition
Cornell University
Ithaca, New York
On Contract From:
The Food and Nutrition Service
United St ates Department of Agriculture
LI BRARY
MAY 1 7 1971
April, 1971
ASER UNIVI:.n::.IIY Ul" 1'-l uRTH CAKOLINA
AT GREENSBORO
0447A
l \
! I
1 l
l
i\ ' {
' (
\
\ ~
\
j r l
TABLE OF CONTENTS
I. An Evaluation of the Food Certificate Program
II.
A Summary of the Findings
Methods . . . . • . •
Sample Description
Dietary Intakes . . •
Interviews in Health Clinics and Food Stores
Program Acceptance
Conclusions and Recommendations
Population Characteristics
The Pilot Food Certificate Program
Recommendations • ." . .
Study Design and Methodology
Introduction
The Economic Context of the Program
Scope of the Study
Sample Design • . .
Sample Selection
Estimation of Sample Size
Procedure for Determining Nutrient Intakes and Dietary
Patterns • • • • • •
Computation of Nutrient Intakes •
The Questionnaire • . .
Field Survey Procedure • • . .
Method of Analysis . • • .
Interviews with Clinic Personnel
Interviews with Retail Food Store Managers
III. Results of the Field Survey .
Part I - Sample description
Significant Differences Between Groups
Family Income and Food Expenditures •.
Infant Birth Weight Comparisons . . . • •
Summary of the Sample Description Data
Part II - Infant nutrient intakes . • • • • • •
Nutrient Intakes as Related to Per Capita Income
Infants Receiving Vitamin and Mineral Supplements •
Comparison of Results with Other Surveys •...
Distributions of the Nutrient Intakes • • • • • .
Significant Between-Group Differences in Nutrient Intakes •
Food Sources of Nutrients . . • • . • • • • . .
Part III - Infant consumption of instant dry infant cereal,
l
l
2
2
3
4
5
6
6
"7
8
10
10
ll
14
16
17
18
19
20
21
21
22
23
23
25
25
26
27
28
29
30
31
32
32
32
34
35
commercial formulas, and milk • • • • . . • • • • . • . 35
Part IV - Family member milk intakes . . • • • . • • • •
Pregnant and Not Pregnant Women's Milk Consumption •
Incidence of Milk Intolerance • • • .
Maternal Vitamin and Mineral Supplements
Household Milk Consumption . . . . • . •
Preschool Children's Milk Consumption
Summary of Program Effects (Parts II, III and IV)
37
37
37
38
38
38
38
IV. Interviews With Health Clinic Personnel and Food Store Managers 94
Interviews with Health Clinic Personnel • • • . • 94
Complications of Pregnancy and the Continuity of Pre- and
Post-natal Care . • . . • . . • • • . . • . • . 94
Infant Morbidity and Mortality . • • . . • • . • . . • . • • 95
Dietary and Vitamin and Mineral Supplement Recommendations • 95
Nutrition and Its Role in Determining Maternal and Infant
Health . • • . • · . • • . . . • • . • • • • . . . . • . • 95
Problems in Program Administration and Recommended Program
Changes . . • • . • . . • • • • . . • • • • .
Health Clinic Data Availability for Program Monitoring
Summary of Health Clinic Interviews
Retail Food Store Interviews
Store Manager Interviews . . • . • • •
Summary of Interviews in Retail Food Stores
V. Participant Reactions to the Pilot Food Certificate Program and
96
96
97
98
98
99
the Use of Related Family Programs 104
Family Food and Education Programs •
Nutrition Education •.•.
Other Food Programs in the Pilot Areas •
Summary
VI. Extent of Participation and Reasons for Non-Participation in the
• 105
• 105
105
106
Pilot Food Certificate Program . • • • • • • 115
Extent of Participation - Chicago • • • •
Extent of Participation - Bibb County, Georgia •
Reasons for Non-Participation
APPENDIX Statistical Methodology
Linear Discriminant Analysis •
BIBLIOGRAPHY • .
115
116
• 117
. 119
119
124
<I
LIST OF TABLES
Table
3.1 Size of Subpopulations, Sample Groups, and the Determination
of Questionnaires Fielded - Chicago • • • • . . . • • • • . 40
3. 2 Size of Subpopula tions, Sample Groups, and the Determination
of Questionnaires Fielded - Georgia • . . • • 41
3.3 Subpopulation and Sample Group Sizes - Chicago 42
3.4 Subpopulation and Sample Group Sizes - Georgia 43
3.5 Number of 24 Hour Recall Records of Mothers and Infants Used
in the Analysis - Chicago . . . . . . . . . . 44
3.6 Number of 24 Hour Recall Records of Mothers and Infants Used
in the Analysis - Georgia . . . . . . . . . . 45
3.7 Family Composition by Sample Group - Chicago and Georgia 46
3.8 Per Capita Weekly Food Expenditures by Sample Groups -
Chicago and Georgia . . • . . . . . . • • . . • . . . 47
3.9 Per Capita Monthly Income and Food Program Maintenance by
Sample Group - Chicago and Georgia . • . • • • • . . . 48
3.10 Mother's Average Age and Education by Sample Group - Chicago
and Georgia . • • . • • • . . • . • . . . . 49
3.11 Significant Differences in: Sample Group Socioeconomic
Characterist ics - Chicago . . . . . . . . . . 50
3.12 Significant Differences in: Sample Group Socioeconomic
Characteristics - Georgia . . . . . . . 51
3.13 Average Birth Weights by Age of Sample Group - Chicago and
Georgia · · · • · · · · · · • • . . . . • • . . . . 52
3.14 Incidence of Birth Weights 2500 Grams or Less by Sample
Groups - Chicago and Georgia • • . . . . • 53
3.15 Neonatal and Infant Mortality Rates in the Sampled
Population - Chicago and Georgia . • . . . . . . . 54
3.16 Respondents Attending Pre-Natal or Well Baby Clinics by
Maternal Status by Sample Groups - Chicago and Georgia 55
3.17 Average Daily Calorie, Protein, Calcium and Iron Intakes for
Infants 1-5 Months of Age by Sample Group (24 hour recall)
- Chicago . . . . • • . . . . . . . . . . . . . . . . . . . 56
Table
3.19
3.20
3.21
Average Daily Vitamin Intakes for Infants l-5 Months of Age
by Sample Group (24 hour recall) - Chicago ••
Average Daily Calorie, Protein, Calcium and Iron Intakes for
Infants l-5 Months of Age by Sample Group (24 hour recall)
- Georgia . . . • . . . . . . . • . . . . • . . . • . • • •
Average Daily Vitamin Intakes for Infants l-5 Months of Age
by Sample Group ( 24 hour recall) - Georgia . . . .
Average Daily Calorie, Protein, Calcium and Iron Intakes for
Infants 6-12 Months of Age by Sample Group (24 hour recall)
- Chicago
3.22 Average Daily Vitamin Intakes for Infants 6-12 Months of Age
by Sample Group (24 hour recall) - Chicago . • ••.
3.23 Average Daily Calorie, Protein, Calcium and Iron Intakes for
Infants 6-12 Months of Age by Sample Group (24 hour recall)
3.24
3.25
3.26
3.27
3.28
3.29
3.30
3.31
- Georgia
Average Daily Vitamin Intakes for Infants 6-12 Months of Age
by Sample Group (24 hour recall) - Georgia •....•..
Percent of Infants Taking Vitamin and Mineral Supplements by
Sample Groups - Chicago ~~d Georgia • • . .
Comparison of Chicago and Georgia Nutrient Intakes for
Infants with U.S.D.A. 1965 for Calories, Protein, Calcium
and Iron
Percent of Infants l-5 and 6-12 Months Receiving Less Than
7o% of Their RDA in 3 or More Nutrients by Sample Groups
Chicago and Georgia . . . . . . . . . . . . . • •
Proportions of Infants l-5 Months Old Receiving Various
Levels of the Calorie, Protein, Calcium and Iron RDA's by
Samply Groups (24 hour recall) - Chicago . . . . •••
Proportions of Infants l-5 Months Old Receiving Various
Levels of the Calorie, Protein, Calcium and Iron RDA's by
Sample Groups (24 hour recall) - Georgia • • • . . . •
Proportions of Infants 6-12 Months Old Receiving Various
Levels of the Calorie, Protein, Calcium and Iron RDA's by
Sample Groups ( 24 hour recall) - Chicago . • . . • . .
Proportions of Infants 6-12 Months Old Receiving Various
Levels of the Calorie, Protein, Calcium and Iron RDA's by
Sample Groups (24 hour recall) - Georgia ••.....
57
58
59
60
61
62
63
64
64
65
66
67
68
69
,,
r
I)
Table
3.32 A Comparison of the Chicago and Georgia Percent of RDA
Distributions With Upper Middle Class Infant Intakes 70
3.33 Percent of Infants Receiving Vitamin and Mineral Supplements
by Levels of Iron Intake . • . • • . • . • • . . . • • 70
3.34 Significantly Different Nutrient Intakes for Infants 1-5
and 6-12 Months of Age - Chicago • • • . . • • • • . • 71
3.35 Significantly Different Nutrient Intakes for Infants 1-5
and 6-12 Months of Age - Georgia • • • • • • • • • 72
Percent Contribution of Prepared Infant Formula, Dairy
Products and Infant Cereals to Nutritive Value of One Day's
Food for Infants - Chicago and Georgia . • • . 73
3.37 Number of Mothers Breast Feeding by Sample Groups - Chicago
and Georgia . . 74
3.38 Incidence of Formula and Milk Intake by Sample Groups -
Chic.ago • • • . • • • • • . . • • • . . • • . . • . • • . . 75
3.39 Incidence of Formula and Milk Intake by Sample Groups -
Georgia 76
3.40 Average Infant Consumption of Milk, Evaporated Milk, Commer-cial
Formula, and Infant Cereal by Sample Groups (24 hour
recall) - Chicago and Georgia . . . • • • . . • . . • • • • 77
3.41 Significantly Different Commercial Formula, Milk and Instant
Infant Cereal Intakes for Infants 1-5 and 6-12 Months of
Age - Chicago and Georgia • . . . . . • • . • • . 78
3.42 Average Daily Fluid Milk and Total Fluid Intake by Maternal
Status by Sample Groups - Chicago . • • • • . . • • • • . 79
3.43 Average Daily Fluid Milk and Total Fluid Intake by Maternal
Status by Sample Groups - Georgia . • • • . • . • • • • . • 80
3.44 Significant Differences in the Milk Consumption of Pregnant
and Not Pregnant Women, Preschool Children, and the Total
Household - Chicago and Georgia . • • • • • . • . • • • • • 81
3.45 Reported Incidence of Milk Intolerance - Chicago and
Georgia . . . . . . . . . . . . . . . . . . . . . . 82
3.46 Percent of Women Taking Vitamin and Mineral Supplements by
Sample Groups - Chicago and Georgia • • • • • • . • • 83
3.47 Weekly Fluid Milk Consumption by Household by Sample Group -
Chicago and Georgia • . • • • . • • • • • • . • • • . . • 84
Table
3.48 Preschool Children's Daily Per Capita at Home Consumption of
Fluid Milk by Sample Group (24 hour recall) - Chicago and
Georgia . • • • . • • • • • •
4.1 Routine Maternal Visit Schedule and Clinic Determinations
85
Made - Chicago Board of Health Clinics . . • . . • • . . . 100
4.2 Routine Maternal Visit Schedule and Clinic Determinations
Made - Macon-Bibb County Health Clinic 101
4.3 Prices Observed for Food Certificate Eligible Foods - Macon,
Georgia - October 1970 . . . . . . . . . . . . . . . 102
4.4 Prices Observed for Food Certificate Eligible }foods - Chicago,
Illinois - August 1970 . . . . . . . . . . . 102
4.5 Proportion of Shelf Facings Allocated to Fortified Infant
Cereals and Concentrated Commercial Infant Formulas -
Chicago, Illinois - August 1970 . . . . . . . . . . . . . 103
4.6 Proportion of Shelf Facings Allocated to Fortified Infant
Cereals and Concentrated Commercial Infant Formulas -
Macon, Georgia - October 1970 . . . . . . . . . . . . . 103
5.1 Reported Difficulty in Using Food Certificates by Sample
Groups - Chicago and Georgia • • . . • . . . . . • • • • . 108
5.2 Reported Adequacy of the Food Certificate Stipend by Sample
Group - Chicago • . . . . • . • • • • . • • • • . . • • • . 109
5.3 Reported Adequacy of the Food Certificate Stipend by Sample
Group - Georgia . . • • • . • . • • . . • . . • • • . 110
5.4 Requests of Children's Foods to be Added as Eligible for
Purchase with Food Cert ificates by Sample Groups - Chicago
and Georgia • • • • • lll
5.5 Expenditure Preferences for a Hypothetical $15-$25 Increase
in Monthly Income by Sample Groups - Chicago and Georgia 112
5.6 Number of Respondents Reporting Receiving Information on Meal
Planning and Family Feeding During the Previous Two Months
- Chicago and Georgia . . . . . . • . . . • • . . • • . . 113
5.7 Number of Children Receiving Free School Lunches by Sample
Groups - Chicago . . . • . . . • . • . • . . • 114
5.8 Children Receiving Free School Lunches and Breakfasts by
Sample Group - Georgia . . . • . . . • • • • . . . . . 114
LIST OF FIGURES
Figure
land 2 Percentage of Distribution of Diets by Caloric and Nutrient
Intake - Chicago • • • • . . • • • • . • . • • • • . • . • 86-87
3 and 4 Percentage of Distribution of Diets by Caloric and Nutrient
Intake - Georgia • • . . • . • • . . • . • . • • . • 88-89
5 and 6 Percentage of Distribution of Diets by Percentage of
Individual RDAs - Chicago .•••..•..••••.•. 90-9l
7 and 8 Percentage of Distribution of Diets by Percentage of
Individual RDAs - Georgia ••••.•.•.••••••• 92-93 )'
"
'•
I. An Evaluation of the Food Certificate Program
A Summary of the Findings
This study evaluates the Pilot Food Certificate Program in Chicago,
Illinois and in Bibb County, Georgia. It examines the impact of the pro­gram
on infants' dietary nutrient intakes and food consumption patterns,
the milk intakes of women, and the operations of health clinics and retail
food stores. Supportive information was collected about participants'
reactions to the program, and the per cent of the target population parti­cipating
in the program was estimated. Data were collected for the analysis
of the programs six months after the program was initiated in Chicago and
after five months of operation in Georgia.
As set forth in the federal Register Title 7 - Agriculture, Chapter
II, Part 265, the purpose of the program is to provide "highl y nutri tious
foods through normal trade channels to certain low income persons who are
vulnerable to malnutrition, to wit: women during and for a period after
pregnancy and infants."
Although the objectives of the program have never been specified more
definitively, the program apparently was designed for the following reasons:
-To provide a replacement for the commodity distribution supplemental
foods program which becomes uneconomical when areas are switched
from the Commodity Distribution Program to the Food Stamp Program.
-To encourage the consumption of commercially prepared iron fortified
infant formulas and of iron fortified infant cereals.
-To increase the milk consumption of women during and for a period
after pregnancy.
The pilot program provides five dollars worth of certificates each
month to women during pregnancy and for the twelve month period following
birth. It also provides certificates valued at ten dollars per month for
each infant through 12 months of age. The Food Certificates may be used
only to purchase whole or skim milk, commercially prepared concentrated or
powdered infant formulas, instant precooked infant cereal, and nonfat dry
milk and evaporated milk. (Nonfat dry milk and evaporated milk were not
eligible for purchase with Food Certificates in Chicago.) Women living
within the program areas could be certified to receive Food Certificates
for the period of their eligibility if:
-They were receiving public assistance or,
-They were participating in the Food Stamp Program or,
-They had been referred to the local certificate issuance agency
(local welfare departments) by prenatal or well baby clinics or,
-Their income and resources did not exceed Food Stamp eligibility
requirements.
From the standpoint of acceptance by the target popul ation, the Food
Certificate Program was successful based upon the relatively high parti­cipation
rates and field reports. However, was the program successful in
terms of supplementing quantities of food intake, substituting higher
-1-
-2-
nutritional quality intakes, or were the Food Certificate expenditures
just substituted for previous cash expenditures, i.e., a type of income
maintenance?
Methods
To determine the impact of the Food Certificate Program on dietary
patterns and the nutritional value of the diets of the recipients, the
24-hour recall method was used to obtain information about the entire
daily food intake of infants, and about the milk consumption of pregnant
women and/or mothers, and preschool children. The nutritional value of
the food intakes was then calculated using food composition tables. This
approach was preferred over the biochemical and clinical tests of nutri­tional
status, primarily because of the time and financial constraints of
the study, but also because the program was designed to affect nutritional
status through its influence on the nutritional value of foods during
current consumption. Tpis study was designed to evaluate a specific food
intervention program on the basis of its ability to influence the nutri­tional
value of dietary intakes. It was not designed, nor was proper data
collected, to provide the basis for commentary on the presence, absence,
or extent of malnutrition or on the nutritional status of the target
population.
Because a1ongitudinal study to measure program effects was not possible,
a cross-sectional study was designed in which infants, their mothers, and
pregnant women were divided into five sample groups on the basis of their
family's participation or lack of participation in the Food Stampand Food
Certificate Programs. A fifth sample group (clinic referrals) was also
created for non-Food Stamp, non-public assistance, Food Certificate Program
participants that were eligible for certificates on the basis of attending
a health clinic. The five sample groups analyzed were:
Food Program Participation
Group I - Food Stamps + Food Certificates
Group II - Food Stamps
Group III - Food Certificates
Group IV - Food Certificates - Clinic Refer~als
Group V - No Food Program participation but received
public assistance
The sample was drawn by randomly selecting the names of pregnant women and
women with children less than one year old from Food Certificate records,
Food Stamp records, and Public Assistance rolls. Home interviews were con­ducted
by experienced professional enumerators who were compatible with
the social setting.
Sample Description
All of the respondents in Chicago were black, as were 89 per cent in
Georgia. Altogether, 1,031 interviews were completed in Chicago and 564 in
Georgia. Six hundred and sixty-six usable infant 24-hour recall records were
obtained in Chicago and 371 in Georgia. Complete fluid intake records were
obtained for 1,028 women in Chicago and 556 women in Georgia. The average
birth weights of infants in both the Georgia and Chicago samples was 3,090
grams, 10 grams more than the 1963 United States average birth weight for
infants born to non-white families with incomes less than $3,000. The
incidence of birth weights less than 2,500 grams (5 lbs. 8 oz.) was 15
per cent in Chicago and 14 per cent in Georgia as compared to the 1963 U.S.
average of 14 per cent for non-white families with incomes less than $3,000.
More than 60 per cent of the pregnant women and mothers of infants in both
localities reported attending a well baby or prenatal clinic.
Dietary Intakes
The calculated mean nutrient intakes of infants one through five
months of age in Chicago and Georgia for all sample groups were in excess
of the Recommended Dietary Allowances (RDA's) with the exception of iron,
which ranged from 64 per cent to 120 per cent of the RDA's. The sample
gro1lp mean nutrient intakes of infants 6 through 12 months of age in Chicago
and in Georgia were all equal to or in excess of the RDA's with the excep­tion
of one sample group's mean calorie intake (98 per cent) and again iron
intakes which ranged from 36 per cent to 58 per cent of their RDA's. The
nutrient intakes analyzed did not include vitamin and mineral supplements
which slightly more than half of the infants in both areas were reported
to be receiving.
Nutrient intakes as a per cent of individual RDA's based on the age
and weight were calculated for all infants. The distributions of the
nutrient intake as a per cent of the RDA for all nutrients except iron
indicated that a minimum of 5 per cent and a maximum of 26 per cent of
all infants age one through five months were receiving less than 70 per cent
of their individual RDA's in at least one nutrient. Forty-eight per cent
of the Chicago infants age one through five months were receiving less
than 70 per cent of their iron RDA's as were 45 per cent of the Georgia
infants in the same age group. The proportions of all infants 6 through 12
months of age receiving less than 70 per cent of their RDA's in at least
one nutrient, except iron, ranged from l per cent to 42 per cent. The
distributions of iron intakes for this age group indicated that 71 per cent
of the Chicago infants and 84 per cent of the Georgia infants were receiv­ing
less than 70 per cent of their iron RDA's. Because of variability be­tween
an individual's daily food intakes and the variability of an indi­vidual's
daily intake over time, one would expect to find some proportion
of individual nutrient intakes to be below 70 per cent of the RDA and
certainly below 100 per cent of the RDA in 24-hour recall data. To conclude
that the mere existence of nutrient intakes below the RDA's is indicative
of dietary deficiencies in the sampled populations is erroneous.
A comparison of the proportion of the sampled population's infants
receiving less tr~n 2/3 of their RDA's from one day's food intake with a
1954 study of infants from upper middle class families indicated that the
proportion of intakes of infants l through 5 months below 2/3 of the RDA's
were approximately the same in both studies. The same comparisons made for
infants 6 through 12 months of age indicated that larger proportions of
infants in this study were receiving less than 2/3 of their RDA's in iron,
vitamin A, and calcium.
-4-
Although all of the respondent families had incomes near public
assistance levels, a number of significant differences in socioeconomic
characteristics were found between the sample groups both in Chicago and
in Georgia. To account for between group . differences in socioeconomic
variables which could influence 'nutrient intakes and dietary patterns, a
multivariate discriminant analysis method was used to test for program
effects. To compare each of the sample groups in each age classification
for the Chicago and Georgia data, 360 individual tests for significant
differences in nutrient intakes were performed. Only 20 significant dif­ferences
were found, and no consistent Food Certificate and/or Food Stamp
Program effects were suggested by these differences. The effect of the
Food Certificate Program on increasing the infant's daily consumption of
cereal, milk, and formula is not supported by the evidence of tests for
significant differences. The purpose of the Food Stamp Program is not the
same as the purpose of the Food Certificate Program. The Food Certificate
Program is a "rifle or singular approach" to family feeding, since it was
designed for specific family members, while the Food Stamp Program is more
of a "shotgun or multi-approach" since it was designed to increase total
family food purchases and/or upgrade the nutritional quality of the family
diet. The Food Stamp Program per se was not evaluated in this study but ·
information collected indicated that it was quite successful in increasing
household per capita food expenditures.
Interviews in Health Clinics and Food Stores
Interviews were conducted with health clinic personnel and retail
food store managers to determine if the Pilot Food Certificate Program
was compatible with their normal operations and to solicit opinions about
the program and recommendations for change. The program was generally well
received by both clinic personnel and retail food store managers. Health
clinic personnel unanimously refused to isolate nutrition from environmental
stresses as a determinant of health status. However, those interviewed did
feel that nutritional status plays a synergistic role. The most substantial
improvements in the health status of pregnant women and infants less than
six months of age resulted probably from programs which emphasized prenatal
diet and care. Sufficient anthropometric data and clinical determinations
are recorded routinely at the health clinics to provide a basis for program
monitoring. The Food Certificate Program caused no problems from the stand­point
of administration but did tend to be time consuming at the expense of
dispensing medical care. The lack of coordination between local hospital
and Food Certificate Program formula provisions and recommendations has been
detrimental to the ease of acceptance and use of Food Certificates in Chicago.
Retail food store managers reported virtually no problems with the
Food Certificate Program. A shelf audit conducted in ten stores in Chicago
and nine stores in Georgia indicated that in all stores except one in each
area iron fortified formula was in stock. The prices for certificate
eligible items were found to be at the same levels as prices in higher in­come
neighborhoods. Store managers reported increases in the sales of infant
formulas and fluid milk since the initiation of the program, but could not
supply any exact product movement data. The only consistent problem report­ed
by store managers in handling certificates was that many certificate
users appeared to be uninformed about the products that were certificate
-5-
eligible. Although to measure accurately the extent to which certificates
were actually used for program foods was not possible, store observation
would lead one to a conclusion that there is some slippage.
Program Acceptance
Field survey data regarding certificate recipients' reactions to
the program indicated that the vast majority of participants have had no
problems using certificates. When problems were encountered, most often
they involved the retailer's refusal to accept the certificates during
particularly busy hours. The Food Certificate stipend was reported as
ade~uate by more than three-fourths of one book recipients, but inade~uate
by an average of 27 per cent of three or more book recipients. More than
half of the recipients in both areas re~uested that infant foods in jars
be made eligible for purchase with Food Certificates. In response to a
~uestion regarding expenditure preferences for a hypothetical $15-$25
monthly increase in income, Food Stamp recipients mentioned food expendi­tures
less fre~uently as the first expenditure priority than did non-Food
Stamp recipients. The consistency of the pattern of responses across
sample groups in both Chicago and Georgia suggests that Food Stamps suc­cessfully
lessened economic restrictions to food purchasing.
Education programs about nutrition in the study areas did not reach
the population of interest in this study. The free school lunch program
in Chicago is making a substantial contribution to feeding the target
population's school age children, but in Georgia it did not serve even
half of the potential numbers at the time of the survey.
As of November 30, 1970 in Chicago and December 31 in Georgia, the
Pilot Food Certificate Program was estimated to be serving 62 per cent of
the potential participants in Chicago and 42 per cent in Georgia. The
most prevalent reasons found for program non-participation were personal
pride or an aversion to "hand-outs", and lack of awareness of the existence
of the program.
-6-
Conclusions and Recommendations
The Pilot Food Certificate Program was designed to effect the quantity
and quality of the dietary intak~of a specific target population: mothers
during and for a period after pregnancy and their infants.
As characteristic of many economic aid programs which increase pur­chasing
power, the program can have a variety of combinations of the
following effects:
-It can increase the quantity of food consumed by the recipients
and/or household members, i.e., a supplemental dietary effect;
-It can increase the quality of the diets of the recipients by en­couraging
and enabling the purchase of more nutritious foods, i.e.,
a substitution of quality effect;
-It may not increase the quantity or quality of the food consumed and
simply replace cash expenditures with certificate expenditures, i.e.,
a substitution of purchasing power or an income maintenance effect.
The mere receipt of certificates obviously has an income effect on
the household. Whether the increase in income will increase the consump­tion
of foods provided by the program is determined by the household's
income elasticity. The preceding analysis in essence has tested the hypo­thesis
that the foods provided by the program and higher quality infant
foods have a zero income elasticity for the target families. Given the
determinants of income elasticity, the success of the program in increasing
incomes (program acceptance alone) and influencing the quantity and quality
of food intakes is a function both of program characteristics and the
characteristics of the target population. The following conclusions derived
from the study data refer to population characteristics as well as the Pilot
Food Certificate Program.
Population Characteristics
1. Based on the mean nutrient intakes and the distributions of intakes
of infants age 1 through 5 months, the diets of this age group appear
adequate. The distribution of their nutrient intak~is approximately
the same as found in a study of infants from upper-middle class families.
2. Infants age 1 through 5 months in all sample groups had an average
total consumption of milk plus formula approximately equal to infant
consumption levels found in populations where food intake was not
restricted by economic limitations.
3. Based on the mean nutrient intakes and the distributions of intakes of
infants 6 through 12 months of age, their diets possibly are inadequate,
especially with respect to iron.
4. Infants age 6 through 12 months in all sample groups had an average
total consumption of milk plus formula which was greater than those
levels found in higher income populations. This tends to support the
reported incidence of high milk content diets for 6 to 12 month old
infants in low income populations.
-7-
5. Based on the incidence of low birth weight infants (2500 grams or
less) there is substantial reason to believe that the health status
and perhaps the nutritional status of the target population's preg­nant
women is deficient. The 1963 National health statistics indi­cate
that the incidence of low birth weight infants for both whites
and non-whites decreases in the same proportion as average incomes
rise. However, the incidence of low birth weight infants for non­whites
is twice that found for whites regardless of income. This
suggests that providing income alone, whether in kind or cash, will
not significantly decrease the incidence of low birth weight infants.
6. The potentially vulnerable low income population of lactating mothers
is very small in the population sampled; 95 per cent of the mothers
were not breast feeding.
The Pilot Food Certificate Program
1. The rates of participation in the two pilot areas indicate that the
program successfully reached the target population.
2. The program successfully reached segments of the low income popula­tion
which did not receive Food Stamps and/or public assistance.
3. The pilot program was well accepted by the recipients, with few prob­lems
reported about the use of the certificates.
4. The amount of the food certificate stipend was reported to be adequate
by a majority of recipients.
5. The conduct of the Pilot Food Certificate Program was compatible with
the normal operations of both health clinics and retail food stores.
6. The pilot program did not significantly increase either the quantity
of milk and/or formula intakffiof infants agel through 5 months, nor
did it increase their nutrient intakes.
7. The pilot program did not significantly increase the quantity of milk
and formula intakes of infants 6 through 12 months of age, nor did it
increase their nutrient intakes. Since the Certificate Program did
not successfully encourage substitution of infant formula for whole
milk in this age group, the problem continues. Caloric intakes in this
age group are actually excessive in Georgia and the program could be
contributing to problems of overnutrition.
8. The program did not successfully increase the milk intakes of either
pregnant women or mothers of infants in a consistent fashion.
9. By implication, the income elasticity for program foods of the target
families is very small, i.e., near zero.
The Food Certificate Progvam did not effectively increase the quantity
or quality of the diets of the participants relative to the control groups.
Its failure to enhance the diets of infants l through 5 months of age can
-8-
be explained largely as a function of the relatively adequate diets of
infants in the control groups. The program's failure to increase mother's
milk consumption and the substitution of commercially prepared formulas
for milk in the diets of infants 6 through 12 months cannot be explained
in terms of consumption characteristics of the control groups. Since the
Pilot Program was widely accepted by the target population and because it
did not significantly influence either the qualitative or quantitative
aspects of the diets, the program performs the function of an income
maintenance pr ogram.
One can question the nutritional adequacy of the diets of infants 6
through 12 months of age in the sampled populations. To the extent that
diet influences pregnancy outcome, some pregnant women in the sampled
populations may have dietary inadequacies. Medical personnel serving the
target population have stated that poor maternal and infant dietary habits
are a function of lack of knowledge as well as economic restrictions.
Given the results of this study, one must conclude that either the provis­ions
of the Food Certificate and/or Food Stamp Programs do not sufficiently
overcome the economic limitations to diet or that lack of knowledge is a
substantial problem that must be considered if maternal and infant diets
are to be improved. The fact that infants from families receiving Food
Stamps did not have higher nutrient intakes, although the families had
higher per capita food expenditures, gives further support to the inference
that the limitations to improving infant feeding are not purely in terms of
food purchasing power.
Recommendations
Currently a substantial amount of research is examlnlng child feeding.
When available, the results should be integrated with the findings of this
study to provide a more comprehensive and complete source of information
about dietary practices. The present lack of current information about
infant and maternal dietary intakes and practices complicates a rigorous
interpretation of these findings. To facilitate data interpretation,
future program design, and evaluations, we recommend that additional re­search
and/or integration of current research findings be undertaken to
provide a better understanding and definition of the role of food intakes
and food programs in determining the nutritional and health status of
children and pregnant women in low income families, and to provide a basis
for the formulation of food, health, income and education programs and
policy for low income families. Given this data the areas in which further
inquiries and/or integration of findings should be done are:
1. Infant Dietary Intakes - Infant dietary intake data should be examined
for infants in families in low, middle, and high income brackets in­cluding
low income families participating in various family food pro­grams.
The nutrient data should be tabulated with and without vitamin
and mineral supplement contributions to determine their role in meet­ing
dietary inadequacies. This data would provide current information
which could be used to examine income and food program effects on
infant intakes and would serve to provide a basis for confirmation
and more rigorous interpretation and understanding of the data col­lected
in this study. Data already collected in the National Nutrition
Survey and other surveys will help fulfill this need.
-9-
2. Preschool Children's Diets - Data collected in this study suggest a
trend in decreasing dietary adequacy as infants grow older. Dietary
intake data should be examined for children age 1 through 5 years in
low income families including those participating in the various food
programs. The contributions of the Food Stamp Program and the food
provided at Headstart and Day Care Centers should be considered explici­tly.
3. Low Income Pregnant Women's Diets - Data should be examined to define
the dietary patterns and nutrient intakes of low income pregnant women
and the empirical relationships between diet and pregnancy outcome.
Food Stamp and non-Food Stamp recipients should be considered separately
by per capita income classifications and the associated diets of pre­school
children should be measured to determine if a competitive rela­tionship
exists between intakes of mothers and preschool children.
4. The Role of Food and Supportive Programs - To suggest that conclusive
research could be done to define the role of diet in reproductive
efficiency is unrealistic. (Maternal Nutrition and the Course of
Pregnancy, published by the National Academy of Sciences, attests to
the complexity of the problem.) However, the effects of the Food Stamp
and/or Food Certificate Programs or an income program in combination
with supportive health and educational programs could be measured, at
least empirically, by monitoring health clinic data for pregnant women
and infants participating in various experimental programs. Pilot pro­grams
of this nature are suggested purely as experimental programs to
be modified, expanded or dropped as implications from studies outlined
in 1, 2, and 3 above are understood.
5. Cost-Effectiveness and Potential Benefits - Additional effort should be
devoted to develop measures to define more clearly the effectiveness
of nutrition, education, and health care programs. Based on this study,
interdisciplinary and most likely interdepartmental programs are needed
to deal effectively with the health problems of the target population.
Future program design should realize the necessity of constructing pro­grams
with a combination of the most effective food, health, and educa­tion
elements provided by different governmental agencies. Better
measures should also be developed to determine the societal benefits
of improved infant and maternal health care. The societal benefits to
be derived from the elimination of these problems should be defined to
better reflect their significance in the framework of national priorities.
\
J II. Study Design and Methodology
Introduction
In February 1970, the Pilot Food Certificate Program was established
in a selected area of Chicago, Illinois, specifically, the four welfare
districts on Chicago's south s ide: Kenwood, Oakland, Robert Taylor
Homes and Woodlawn. In April 1970, the same program with minor modifica­tions
was established in Bibb County , Georgia. As of January 1971, pilot
programs had also been established in Brazos County, Texas; Yakima County,
Washington; and two counties in Vermont.
As set forth in the Federal Register Title 7 - Agriculture, Chapter
II, Part 265, the purpose of the program is to provide "highly nutritious
foods through normal trade channels to certain low income persons who
are vulnerable to malnutrition, to wit: women during and for a period
after pregnancy and infants."
The program provides free Food Certificates with a retail value of
$5 per month to women during and for 12 months after pregnancy, and $10
of Certificates for each of their infants through 12 months of age.
Thus, from the time of program certification until delivery , pregnant
women receive $5 worth of certificates each month and in the case of a
single birth delivery, for a 12 month period following the baby's birth
receive $15 worth of certificates monthly.
All pregnant women and/or women with an infant less than 12 months
of age living within the pilot program areas may become certified to
receive Food Certificates for the period of their eligibility if:
-They are receiving public assistance, or
-They are participating in the Food Stamp Program, or
-They have been referred to the state issuance agency (local
welfare departments in Chicago and Georgia) by prenatal or
well baby clinics and services, or
-Their income and resources, as determined by the State Issuance
Agency, do not exceed Food Stamp eligibility standards.
Foor Certificates may be used only for the purchase of whole or
skim milk; commercially prepared concentrated or powdered infant for­mulas,
preferably enriched with iron; instant precooked infant cereal,
preferably enriched with iron; and nopfat dry milk and evaporated milk
(nonfat dry milk and evaporated milk are not eligible for purchase
with Food Certificates in the Chicago Program area).
Although the objectives of the program have never been specified
more definitively than as stated in the quotation from the Federal
Register, one assumes that the program was designed for the following
reasons.
l. To provide a replacement for the commodity distribution supplemental
foods program which becomes uneconomical when areas are switched
from the Commodity Distribution Program to the Food Stamp Program.
-10-
•,
-ll-
2. To encourage the consumption of commercially prepared iron fortified
infant formulas and of iron fortified infant cereals.
3. To increase the milk consumption of women during and for a period
after pregnancy.
From the type of products eligible for purchase with Food Certifi­cates,
one tends to conclude that the program was designed primarily to
influence infant feeding and only secondarily to influence the diets of
pregnant women. Point two above also suggests that the program was
based, at least in part, on the assumption that infants from low income
families have insufficient dietary intakes, particularly with respect
to iron.
Given the construct of the Pilot Food Certificate Program, this
study evaluates the impact of the pilot programs in Chicago and in Bibb
County, Georgia on the nutrient intakes and dietary patterns of the
infants of certificate recipients, the milk consumption of pregnant
and not pregnant certificate recipients, and the operations of health
clinics and retail food stores in the program areas. Data were collected
for the analysis of the programs six months after the pilot program was
initiated in Chicago and after five months of operation in Georgia.
The Economic Context of the Program
The possible effects of the program on the consumption patterns of
the recipients can be developed from basic economic theory. Let us
assume first that the amounts of formula, cereal, and milk provided by
Food Certificates are the same as free goods which cannot be resold,
and second that these products provided by Food Certificates are perfect
substitutes for the formula, cereal and milk purchased by these same
households before their receipt of certificates. Infants would likely
consume milk and/or commercial formula and mothers would likely consume
some milk even if their families did not receive certificates.
For simplicity of presentation, household consumption of milk,
commercial formula, and cereal is called Program Foods (X) measured on
the X axis, and the consumption of All Other Goods (Y) is measured on
the Y axis. Before the receipt of Food Certificates the household
faces the budget line AEB and purchases OC of Program Foods and CE of
All other Goods (equal to CE amount of money). Upon receipt of the
Food Cf~tificates the household in effect receives OD = BB' of Program
Foods.:J Since the household cannot sell the Program Foods, it is now
faced with the budget line A'B' which is parallel with the original
budget line but does not intersect the Y axis because it cannot have
zero Program Foods if it spends all of its resources. If the household
!J Although the value of certificates provided each month was established
to fulfill all of the mother's milk needs and the baby's milk, cereal,
and formula needs, OD is not equated with OC in this example because
of the likelihood that household milk purchases are consumed by house­hold
_members other than the infant and the mother or pregnant woman.
\
I
rJ.l
'"!:)
0 g
-12-
B'
Program Foods (X)
totally eliminates cash expenditures for Program Foods, it will have
OD = BB' of Program Foods and its entire income would be devoted to
non-program goods and services. This amount of money is DA' = OA, and
the household's new equilibrium is at G or K or at some point between
G and K on the new budget line (assuming Program Foods are not inferior
goods). Assuming the new equilibrium is atE', then the household buys
only DF of Program Foods with non-certificate income (less than OC
which it bought before), and DF plus the free Program Foods OD yields
the total, OF of Program Foods. The household then has FE' of "All
Other Goods" (money) which is greater than the CE amount it had prior
to participating in the program.
The selection of E' as the new equilibrium point is done only to
illustrate the income and consumption effects of any Program Foods -
All other Goods (money) split between G and K. The actual equilibrium
reached by households receiving Food Certificates will depend on the
shape of their indifference maps (not shown) for All Other Goods
(Y=money) versus Program Foods (X) and the Px/Py price ratio (i.e. the
slope of the budget line). If households consider Program Foods either
"normal goods" or "luxury goods," then the new equilibrium will be at
-13-
K or to its left, but not at G on A' GE ' KB ' .~ If Program Foods are
considered as "necessity goods , " then the equilibrium will be at G or
marginally to its right. Thus, in the present example, the action of
receipt and use of certificates in and of itself causes a parallel
shift in the budget line (an increase in income), but the determination
of the equilibrium X,Y split on the new budget line is a function of
the marginal rate of substitution of X for Y at a higher income, which
is related to income elasticities.
Therefore, the ability of the Food Certificate Program to increase
the consumption of Program Foods is a function of the certificate re­cipient's
income elasticity for these foods. Similarly the ability of
the program to have the secondary effects of the substitution of higher
quality (presumably higher cost) non-program foods int o the daily diet
is a function of the income elasticity of those foods (e. g., jar baby
foods). If the income elasticity of Program Food expenditures (X) is
perfectly inelastic (=0), the new equilibrium will be at G, but if it
is elastic (~1), the new equilibrium will be closer to K. The income
elasticity of Program Foods would be expected to be determined by the
role of these foods in the household consumption patterns. If Program
Foods are necessities, they are by definition highly income inelastic;
their need biologically determined and limited; and the program effec­tively
increase consumption only if incomes are near or below subsis­tence.
If the near subs i stence situation exists, one would expect the
program participant equilibrium to be at or near K on A'GE'KB'. If
certificate recipient household incomes are not near subsistence, one
would expect the equilibrium to be closer to G assuming as before that
Program Foods are considered necessities. However, if the certificate
recipient's equilibrium is at or near G, the recipient household incomes
may still be at subsistence levels. Infant feeding may have such a
high priority that Program Foods are purchased at the sacrifice of
other consumption. Therefore, household incomes may be at subsistence
levels and the consumption of Program Foods unchanged by the receipt
and use of Food Certificates because of the prior expenditure priority
attached to infant feeding.
In many practical contexts whether or not the quantity and quality
of goods consumed are envisioned as necessities (have high expenditure
priorities) is a function of the consumer's subjective valuations (e.g.,
television sets in contemporary' society). The possible effects of the
~The terms "necessity," "normal," and "luxury" goods are used in the
classical income-consumption curve context, i.e., "necessity goods"
are those whose consumption is biologically determined and limited,
and therefore perfectly income inelastic once consumption reaches the
physical saturation point; "normal goods" are those whose consumption
increases with successive increases in income throughout the entire
income range; and "luxury goods" are those which are initially con­sumed
only at relatively high income levels and whose consumption
increases as income increases.
I
-14-
Food Certificate Program on the recipient's consumption as discussed
above considers only the program participants' definitions of the role
and importance of Program Foods in her expenditure pattern. That is,
the recipient is believed to be economically rational within the confines
of her values which one would expect to be related to knowledge. Because
she personally defines both the quantitative and qualitative needs of
her family's consumption of Program Foods and fulfills them with cash or
certificate expenditures, the family's dietary needs still may not be
met from a nutritional standpoint. The service provided by the Pilot
Food Certificate Program is that of enabling the purchase of food. How ~
the enabling economic power is used will be determined by her economic
status as well as by her own judgments as to the family's needs and
possible benefits to be gained by increased consumption of Program Foods
versus other household goods.
The Program, before the initiation of this study, successfully
supplemented incomes because it was well accepted by the target popula­tion.
Whether or not the Program successfully ~mproves the quantitative
and qualitative aspects of the dietary intakes will be determined by the
levels of Program Food consumption prior to the receipt of certificates
and her knowledge and valuation of higher quality foods.
Economic theory, as discussed above, suggests that Food Certificates
can affect dietary intakes only if Program Foods are not perfectly or
near perfectly income inelastic. The following analysis tests statis­tically
the hypothesis that Program Foods and higher nutritional quality
baby foods are perfectly income inelastic, i.e., Ho: certificate
recipients' consumption of Program Foods and higher nutritional quality
baby foods is not significantly different from the consumption of those
foods by non-certificate recipients. Rejection of the null hypothesis
will indicate that the program, by supplying purchasing power, is effec­tive
in increasing nutrient intakes. If the hypothesis is not rejected,
then factors other than purchasing power are of primary importance in
affecting nutrient intakes.
Scope of the Study
The primary emphasis of this inquiry deals with nutritional con­siderations,
while the program's inherent involvement with food retailing
and health institutions is examined from the standpoint of their compat­ibility
with the conduct of the pilot program.
This study is designed only to evaluate a specific food intervention
program and not to identify the presence, absence, extent or cause of
malnutrition in the target population. The determination of malnutrition
necessitates dietary, biochemical, and clinical tests, while identifying
its causes is an extremely complex problem involving behavioral and
environmental considerations. Although the 24 hour recall is currently
the most efficient method available for measuring nutrient intakes,
the following points must be recognized in relation to the validity
of the estimates of the nutritional value of the food intakes.
-15-
1. The various amounts and composition of the foods reported as
consumed are based on the respondent's estimates; the various
quantities were not weighed.
2. The nutritional value of the foods reported to be consumed
were estimated from food composition tables and not deter~
mined for the specific foods by individual chemical analyses.
3. The interpretation of the calculated nutrient intakes utilizes
the Recommended Dietary Allowances (R.D.A.) which are also
estimates.
Therefore, this study is clearly inadequate on the basis of design and
data collection to provide the basis for any valid commentary on the
nutritional status of the target population.
While recognizing the limitations enumerated above, this study
usefully estimates whether the nutritional value of the food consumed
in a particular population is adequate to meet recommended levels of
intake. However, these data were collected in only two geographic
regions, both urban; thus, the sample does not permit the drawing of
inferences about the entire U.S. low income population. Any attempt
to use or interpret the following data without recognizing the above
qualifications must be considered inappropriate.
The development of the Pilot Food Certifi cate Program apparentl y
was based on the assumption that dietary deficiencies exist in the
target population. By virtue of the amounts and types of foods eligible
for purchase with Food Certificates, the dietary deficiencies presumably
were hypothesized to be a function of insufficient quantities of food
and/or the consumption of l ow nutrit i onal value foods.
As previousl y developed from economic theory, the Pil ot Food
Certificate Program can have a variety of combinations of the following
effects:
-It can increase the quantity of food consumed by the recipients
and/or household members, i.e., a supplemental effect;
-It can increase the quality of the diets of the recipients
and/or household members by encouraging and enabling the
purchase of more nutritious food i tems, i.e., a substitution
of quality effect;
-It can affect neither the quantity nor the quality of the food
consumed by the recipients and simpl y repl ace cash food expen­ditures
with certificate expenditures, i.e., a substitution
of purchasing power or a pure i ncome maintenance effect.
The dietary analysis sect i on of this study is designed to determine
if there have been program rel ated suppl emental effects on the quan­tities
of intake of mothers, infants, and preschool children, and if
there has been a substitution of program r elated better quality foods
into the diets of infants and mothers. If the program does not prove
to increase either the quantity or quality of the partici pants' food
-16-
intakes, one must assume that its primary effect is income maintenance,
in the amount of the cash commitment it releases from the food budget.
Since the Pilot Program was widely accepted by the target population
prior to this study, it is clear that the program's ability to have an
impact on the quantity and quality of the diets does not depend upon pro­gram
acceptance but on both the amount of food intake and the quality of
the diets of the program participants prior to their receipt of Food Cer­tificates.
Because it was not possible to do a before and after study
(longitudinal) of the dietary intakes of the program participants, a cross­sectional
study was done in which the quantity and quality of pre-program
diets was measured with two control groups of non-Food Certificate Program
participants.
Recognizing the important role that inter-agency cooperation and the
food distribution delivery system plays in the success of any publicly
sponsored food program, interviews were conducted with health clinic
personnel and food retailers to find out if they were experiencing any
problems with the program as currently administered and how it could be
made more effective.
Sample Design
Since it was not possible to do a longitudinal study, determining
the impact of the program on participants as well as assessing the nature
and extent of its overlap with the Food Stamp Program required a cross­sectional
analysis of five subpopulations within the target population.
The classification of mothers into each of the subpopulations was done
on the basis of their participation in the Food Stamp and/or Food Cer­tificate
Program and on the basis of their eligibility for Food Certifi­cates.
Defining the target population as the universe of mothers and
infants that are socioeconomically eligible for Food Certificates, the
subpopulations were defined as:
1. Those mothers recelvlng Food Certificates and Food Stamps
certified on the basis of Food Stamps.
2. Those mothers receiving Food Stamps, but not Food Certificates .
3. Those mothers recelVlng Food Certificates but not Food Stamps,
and certified on the basis of being public assistance recipients.
4. Those mothers receiving Food Certificates but not Food Stamps
nor public assistance and certified on the basis of being referred by
health clinics.
5. Those mothers not recelvlng Food Certificates or Food Stamps,
but eligible for Food Certificates on the basis of receiving public
assistance.
The following diagrammatic representation of these subpopulations,
henceforth referred to as "sample groups " or "groups", may aid in con­ceptualizing
their differences.
Sample Group I
Sample Group II
Sample Group III
Sample Group IV
Sample Group V
-17-
Participation in Food
Assistance Programs
F.C.* + F.S.**
F.S.
F. C.
F.C.
Basis of Food
Certificate Eligibility
Food Stamps
Food Stamps
Public Assistance
Clinic Referral
Public Assistance
* Food Certificate Program
** Food Stamp Program
These groups are mutually exclusive and serve to categorize all
members of the potential target population except for those mothers cer­tified
on the sole basis of having incomes sufficiently low to meet
Food Stamp eligibility requirements but not low enough to meet Public
Ass istance requirements. These potential participants were excluded
since their numbers were not large enough to justify the cost of identi­fication
and selection of another sample group and its associated
additional number of intergroup comparisons.
Sample Selection
To draw the samples for Groups I, III and IV a list of all program
participants that had received certificates during the current month and
would be eligible for Food Certificates the following month was compiled
from the Food Certificate Issuance Office's records (Form FCP-423 in
Chicago and Form 769 in Georgia). Participants were then classified as
belonging to Group I, III or IV depending on their basis of certification
and therefore their participation or lack of participation in the Food
Stamp Program. The samples for the control groups (II and V) were drawn
from similar socioeconomic areas lying outside the pilot program area.
If the control group samples were drawn only from the nonparticipating
segment of the target population in the program areas, the data would
very likely be biased and by definition could not reflect the whole of
the target population since it would be strongly weighted in favor of
potential participants who had at least tacitly rejected the program.
As a consequence of drawing the control groups from outside the program
area, determining the reasons for nonparticipation had to be accomplished
by talking with health clinic and public assistance personnel as well as
having health clinic personnel question clinic patients.
In Chicago the list of names from which the control groups were
drawn was compiled from Food Stamp and Public Assistance records for the
two welfare districts (Park Manor and Englewood) adjoining the Pilot Pro­gram
area. In Georgia the control groups were drawn from Savannah (Chatham
County) because of its socioeconomic comparability with Macon (almost all
of the Bibb County participants live in the city of Macon) and the avail­ability
of Department of Family and Children 's Services personnel to
compile the lists from Public Assistance case records.
Because the records used to classify households into one of the
five groups were in some cases filed six months prior to the scheduled
time of interview, it was expected that after the interviews some
-18-
respondents would have to be classified in groups other than originally
anticipated. This procedure was mandated since the comparisons of
interest concerned households that were participating in the various
programs at the time the intake data was recorded. The extent to which
their switching between groups is shown in tables 3.1 and 3.2.
After all of the name lists had been compiled, the name and address,
telephone number (when available), maternal status (pregnant, not preg­nant),
length of time on the program (Groups I, III and IV) and a group
identification code were typed on self-adhesive labels. Within each
group a simple random sample was drawn and the labels re-ordered by
their order of sample selection. Relying on the previous experiences
of researchers sampling low income populations, it was assumed from
the outset that due to transiency and the reliability of records, it
would not be practically feasible or even possible to contact all of
the respondents selected from a sample. For this reason and because
of the sample size re~uirements of the tools used to estimate mean
nutrient intakes, the actual sampling procedure used was one of ~uota
sampling where the order of selection was determined randomly.
Estimation of Sample Size
Since the purpose of the analysis is primarily to determine if
significant differences in nutrient intakes exist between each of the
sample groups, the determination of sample sizes was computed on the
basis of multiple comparison procedures rather than on the basis of
estimating parameters for a single population. A valid test criterion
for planned comparisons uses the least significant difference (lsd)
procedure where
lsd(.05) = t. 05 s.d.
t.o5 is the tabular value of "t" for error degrees of freedom and s.d.
is ~2s2/r with s2 the error variance and r the number of observations
per parameter estimate.
One thousand interviews were budgeted for the Chicago study with
e~ual sample sizes for each group, giving the ability to detect signi­ficant
differences of greater than 8 per cent. This assumed that the
five groups could be treated as having infinite populations. Once the
actual population sizes of each group were determined, it was obvious
that the populations had to be treated as finite and that the total
number of observations budgeted had to be reallocated.
The total number of observations to be taken (1000 in Chicago and
500 in Georgia) were allocated between the sample groups so that the
expected standard errors were e~uated after taking the finite population
correction factor into account.
The desired sample sizes derived for each group were the following:
-19-
Chicago Georgia
Group I 225 130
Group II 221 104
Group III 198 58
Group IV 196 104
Group v 160 100
The analytical framework actually used in this study utilizes Hotelling's
multivariate t2 test as well as the property of the simple two-way linear
discriminant function which permits the identification of variables by
which two samples differ. This procedure and the rationale for its sub­stitution
for the least significant difference approach will be discussed
later.
Procedure for Determining Nutrient Intakes and Dietary Patterns
The specific method used to determine nutrient intakes was a 24 hour
dietary recall in which mothers were asked to report the exact amounts,
kind, brand and composition of all foods their infants actually consumed
during the preceding 24 hours. The mothers and/or pregnant women were
also asked to report the amounts and types of beverages they had consumed
during the same period. The total food intake of infants was recorded
to assess the nutritional intakes of their total diets, but only fluid
intake was recorded for the women since one could not logically expect
to find any significant differences in the mother's total diet as a
result of a program which provides only milk for her consumption and
interviewing time constraints also ruled out a complete dietary for
the mothers.
The arguments for and against the various dietary survey metq~4s
are voluminous and are topics of discussion in many publicationsl/~~.
If one were to draw any conclusion from the discussions, it would most
likely be the obvious deduction that the method should be selected to
meet objectives of the particular study.
The 24 hour recall technique was selected for use rather than alter­native
dietary survey'methods such as food record, weighed intake, or
diet history because of its relative simplicity, efficiency, and appro­priateness
for the objectives of this study. The specific criteria for
selecting this method were the following:
l. Objective of the Food Certificate Program - The Food Certificate Pro­gram
is designed to influence dietary patterns and enhance nutri­tional
status through influencing current consumption.
Ibid., pp. 469-474
Morgan, Agnes (ed.). Nutritional Status, U. S.A., an Interregional
Research Publication, Berkeley: University of California Agricultural
Experiment Station, Bulletin 769, October 1959, pp. 12-15.
Guthrie, H. A. Introductory Nutrition, Saint Louis: C. V. Mosby
Company, 1967, pp . 306-308.
-20-
2. Units selected for data analysis - The program was designed to have
an impact on a target population, therefore the study was designed
to evaluate the program's impact on the basis of group comparisons.
Previous research has indicated that 24 hour recalls give approx­imately
the same estimates of nutrient levelg
1
as a seven day
record when groups are the units of inquiry.~
3. Sample size constraints - The study design provided sufficient numbers
of infant intake records to estimate meaningful mean nutrient
intakes from 24 hour recall data.1/§/
4. Type of dietary data to be collected - Complete intakes were requested
only for infants, thus the possible variability of food items and
difficulties with estimating quantities presented a much less
significant problem in this study than in others concerned with
older family members.
5. Respondent cooperation - It was felt that respondent cooperation
could be gained and reporting bias controlled more easily with
this method rather than other more involved procedures.
6. Compatibility with other infant dietary intake information - The 24
hour recall method was also used in the 10 state National Nutri­tion
Survey and for the analysis of infanx
1
nutrient intakes in
the 1965 U,S.D.A. Food Consumption Study.2J
Computation of Nutrient Intakes
A computer ~~ggram was developed to compute the infant intakes of
eight nutrients ~ from the 24 hour recall data. Three professional
nutritionists were employed to interpret, transform and record in the
appropriate format the various kinds and amounts of all foods and liquids
consumed by each infant. The amount of each of the eight nutrients in
one ounce portions of the baby and adult foods was determined from
"Composition of Foods," Agricultural Handbook No. 8, U,S.D.A. (1963)
and from food manufacturers' data for commercially prepared baby foods
and formulas. This data compr.ised a "food list" which was stored
"Cooperative Nutritional Status Studies in the Northeast Region, III.
Dietary Methodology Studies," Northeast Regional Publication No. 10,
University of Massachusetts Agricultural Experiment Station, Amherst,
Mass., Bulletin No. 469, August 1952, p. 48.
Ibid., pp. 45-47.
Manual for Nutrition Surveys, Interdepartmental Committee on Nutrition
for National Defense, Second Edition, 1963, p. 174.
"Food Intake and Nutritive Value of Diets of Men, Women, and Children
in the United States, Spring 1965," A Preliminary Report, Agricultural
Research Service, U.S. Department of Agriculture.
Calories, protein, calcium, iron, vitamin A, thiamine, riboflavin,
and vitamin C.
\.
-2l-internally
in the computer to be referenced. The nutrient intake of
each infant was calculated by matching the individual foods reported
with their respective nutrient values and multiplying each value by the
multiple and/or fractional number of ounces consumed. In cases where
breast feeding was reported (4.8 per cent of the infants in Chicago
and 5 per cent in Georgia), the ~uantity of mother's milk consumed
was estimated with the method developed by Beal~ and the nutritive
value of the milk consumed computed. The total nutrient intake was
then determined by adding together the amounts of each nutrient con­tributed
by all the foods consumed.
The Questionnaire
The ~uestionnaire was designed to provide the re~uired information
on food consumption, socioeconomic, and health characteristics, and
food certificate program participants' attitudes towards the program.
Copies of the actual ~uestionnaire are av~ilable from the authors.
The ~uestionnaire was pretested by conducting 98 interviews with
professional enumerators in two Chicago Health Clinics as well as in
households. Revisions were made for ambiguous ~uestions, the format
for recording some responses, and the skip pattern was clarified. The
~uestionnaire used in Chicago was also used in Georgia with minor
changes.
Field Survey Procedure
The data in Chicago were collected during the three week period,
August l7 through September 4, l970, and in Georgia from October 5 through
l6. Professional interviewing services were used to collect the survey
data. Only experienced enumerators compatible with the social setting
were employed and with one exception were all women.
The enumerators were given a ~uestionnaire instruction booklet,
familiarized with the peculiarities of the ~uestionnaire, and reviewed
standard sizes of baby food containers and baby bottles. Each enumerator's
first day's work.was checked immediately after its completion and in all
cases unacceptable work was refielded with an enumerator who had done
acceptable work. Enumerators were supplied with a set of measuring
spoons and a number of 8 oz. graduated measuring cups. Upon making
contact with the respondent the measuring cup was presented as an incen­tive
for cooperation and throughout the interview both the cup and
spoons were used in estimating intake ~uantities.
If the respondent was not at home when the enumerator made her
initial call, one call back was made at a different time of day. If
no contact was made on the call back, the name label was returned to
the supervisor who again fielded the label with a different enumerator
for two further call backs. If still no contact was made, the potential
W Beal, V. A., "Breast and Formula Feeding of Infants," Journal of
the American Dietetic Association 55(l), 3l-37, l969.
-22-
respondent was dropped from consideration. Similarly, respondents that
had moved were dropped from consideration if a new address could not be
determined.
In Chicago the census tract of residence was also typed on the
labels and each enumerator was given labels from tracts as near to each
other as possible. Fielding the work in this manner served to reduce
enumerator's travel time and by rotating the census tracts assigned to
each enumerator assured that each interviewer recorded intakes in all
of the sample groups . This procedure also served to control for any
systematic group response bias that might be encountered by having
enumerators complete interviews in only one group. There was no reason
to believe that any of the groups of respondents were prone to or
subject to any greater or lesser degree of bias than any of the others.
Method of Analysis
The original intention for an experimental design framework was
to utilize a multiple comparison procedure, but it was necessary to
reject this as the tool of analysis for the following reasons:
-The one-sided LSD procedure is appropriate only when the specific
group comparisons of interest are selected before the data is
observed, i.e., this procedure is appropriate only for testing
a hypothesized transitive ranking of the means. It was obvious
that the calculated means did not follow a transitive order
consistent with that expected from participation in the various
programs and based on other information, one clearly could not
accept tests based on the ordering.
-The subpopulation samples drawn were not socioeconomically homo­geneous
to the degree that one could be confident that observed
differences in nutrient intakes and consumption characteristics
were purely food program effects.
-Nutrient intake group means based on 24 hour recall data are
reputed to have large variances due to the peculiarities of the
test instrument as well as the usual inter-individual variability.
The coefficients of variation of the group mean nutrient intakes
were so much larger than anticipated (75% were greater than
.60) that it was very unlikely that a multiple comparison pro­cedure
would detect a program effect if one did indeed exist.
To account for the possible intercorrelations of the between group
differences in socioeconomic characteristics and to deal with the varia­tion
in variables that theoretically could indicate food program effects,
a multivariate procedure was selected for use.
The statistical tool used to identify differences between the five
sample groups was linear discriminant analysis. It is most often used
for the classification of sample observations ~nto various populations
and for determining the extent to which populations overlap or diverge
from one another. A detailed discussion of the procedure is included as
an appendix to this report.
-23-
Interviews with Clinic Personnel
The interviews with health clinic personnel (pediatri~~~s, obste­tricians,
and nurses) were conducted in six health clinics~ serving
the Chicago Pilot Program area and at the Macon-Bibb County Health
Clinic. Interviews were conducted to:
-Help identify the nature and extent of the health problems of
the target population,
-Solicit opinions on the relationship of health status and
nutrition,
-Determine the routine procedures regarding prescribing and dis­pensing
vitamin and mineral supplements,
-Determine what problems clinic personnel were experiencing
with the Food Certificate Program and recommendations for
changes,
-Determine the availability of health clinic data for future
program monitoring, and
-Determine the reasons for non-participation in the Certificate
Program.
Since the purpose of the interviews was exploratory in nature, those
interviewed were asked to discuss the above topics and no formal
questionnaire was used. In addition to soliciting the opinions of
in-clinic personnel and visiting nurses on the reasons for non-parti­cipation
in the program, two clinic nurses in Chicago Board of Health
Station #23 and two in #30 were enlisted for two days to ask program
non-participants why they had not signed up for Food Certificates and
to record the specific responses given.
Interviews with Retail Food Store Managers
Interviews were conducted with the store managers in ten retail
food stores in the pilot program area in Chicago and in eight food
stores and one pharmacy i~ Bibb County, Georgia. The particular stores
selected in each locality were those that had had the highest monthly
average of Food Certificate redemptions of all stores participating in
the program.
The store managers interviewed were asked to comment on the following:
-The impact of the Food Certificate Program on the sales of eligible
products,
-The operational requirements of accepting certificates,
-The problems customers have in using certificates, and
-Recommendations for program changes.
As was the case with the clinic interviews, no formal questionnaire
was used due to the exploratory nature of the topics on which comments
were solicited.
~ Chicago Board of Health Stations 19, 23, 30, 119, Woodlawn Child
Health Center, and Michael Reese Hospital.
-24-
Shelf audits were also conducted in each of the stores to establish
the level and range of prices of the certificate eligible foods and to
determine the proportion of shelf space allocay~q to fortified formulas
and cereals. Although the program regulations!]/ state that the partic­ipant's
purchase of iron enriched instant precooked infant cereals is
preferred, all of the certificate eligible cereals in the baby cereals
section of the stores visited had the same level of iron fortification
and the only infant cereal that both contained higher nutrient levels
and was merchandised as such was "Hi-Protein"ill, a specific product
of one manufacturer. Because "Hi-Protein" was the only instant pre­cooked
infant cereal product with higher nutrient levels, data on its
shelf space allocation and retail price was recorded and is presented
in the same manner as the data for iron fortified formula.
Federal Register Document 70-1037; Title - 7 Agriculture, Chapter 11,
Part 265, Section 265.2.
Beechnut Hi-Protein Instant Infant Cereal.
'1
..
III. Results of the Field Survey
The results of the food consumption sections of the participant field
surveys in Chicago and in Georgia are divided into four major parts:
Part I Sample description
Part II Infant nutrient intakes
Part III Infant consumption of instant dry infant
cereal, commercial formula, and milk
Part IV Family member milk intakes
Part I - Sample description
All of the respondents in Chicago were black as were 89 per cent of
those in Georgia. Since so few of the respondents (61) were other than
black, the five sample groups were not analyzed by ethnic characteristics.
Tables 3.1 and 3.2 present the estimated subpopulation sizes, the
number of questionnaires fielded for each sample group, the number of
refusals, not at home and moved. Also shown under the columns "Net Inter­group
Transfers" is the net number of respondents added to or deducted
from the respective groups on the basis of their food program participa­tion
at the time of the interview. As stated earlier, the declaration of
participation or lack of participation in the Food Stamp Program was in
some cases made six months prior to the interview. Therefore, after the
interviews were completed the classification of some respondents in groups
other than originally anticipated was necessary so that the hypotheses
of program effects between groups could be tested. Estimating the actual
sizes of the subpopulations would require some rather heroic assumptions
as evidenced by the Georgia Group III's increase in size of 61 per cent
on the basis of this field enumeration alone.
Tables 3.3 and 3.4 breakdown the status of the respondents and the
number of infants in each group as compared to the estimated composition
of the subpopulations. The subpopulation estimates were made from Food
Certificate application forms for Groups I, III, and IV and from Welfare
and Food Stamp records for Groups II and V. Of course, these estimates
are subject to the same complications of intergroup transfers as evidenced
by the respective subpopulation and sample sizes for the Georgia Group III
classifications.
Tables 3.5 and 3.6 show the number of usable infant and mother's
intake records as compared to the total number of records taken for each
group. Unusable intake records were those with inadequate enumerator
recording, lack of respondent cooperation (some mothers were unwilling to
describe the food fed to their baby as other than "table food"), sick and
teething infants, sick mothers, and excessive nutrient intakes. The process
of eliminating excessive intake records was simply the usual editing proce­dure
of c1liling observations which distorted the calculated group means.
To illustrate the distortions created by including the excessive intakes,
one vitamin A observation in a group of 40 intake records increased the
-25-
-26-
group mean nutrient intake by 1500 international units. In another case,
one vitamin C observation in a group of 37 intake records increased the
group mean nutrient intake by 12 mg. To attempt to formalize the usual
arbitrary process of editing, the following procedure was used. Excessive
nutrient intakes w·ere determined by listing the calculated calorie, iron,
vitamin A, and vitamin C intllices of all infants 1-5 and 6-12 months of
age in ascending order and examining the largest values. Beginning with
the largest nutrient intake value, the differences between the ordered
values were calculated. When any difference between the values of the
largest 2 per cent of the observations was more than four times as large ~
as the greatest difference observed in the interquartile range, the intake
record indicated by the larger value and all records ranked above it were
excluded from the analysis. The only constraint observed in eliminating
the records was that no more than 2 per cent of the total number of records
was to be eliminated from either of the age groups. Using this procedure,
21 records were eliminated from the Chicago data and 17 from Georgia.
(In only one instance was an excluded extreme nutrient intake record
associated with one of the oldest infants in the particular group.)
Significant Differences Between Groups
The average household size, number of children in school, and number
of preschool children in each of the sample groups appear in table 3.7
for Chicago and Georgia. The significant differences between the groups
when the other family socioeconomic characteristics were taken into account
are presented in table 3.11 for the Chicago data and in table 3.12 for the
Georgia data. In testing them for socioeconomic variables, ten comparisons
were made for each set of five groups to test all possible two group
comparisons. Eight socioeconomic variables were tested in each two group
comparison. Thus, 80 individual tests of significant differences were
made for the Chicago data and 80 for the Georgia data. With alpha set at
.05, one would expect to commit eight Type I errors in the process of doing
all of the tests, i.e. reject the null hypothesis of no difference when
no difference exists. Consequently, the large number of tests virtually
assures obtaining "significant" differences, which are spurious. The
variables tested are net of the influence of other variables, but inter­correlations
may obscure the net relationships.
The reader must remember that the tests for significant differences
were performed using a multivariate discriminant analysis technique that
tests each variable after accounting for the effects of other variables
in the function with which it may be correlated. For example, in using
the discriminant analysis technique to test for a significant difference
between Group I and Group II incomes, the statistical procedure "accounts"
for Group I and Group II between group differences in family size, com­position
and the other socioeconomic variables included in the function,
and tests for a difference in income net of influence of these other
variables. Therefore when examining the group mean values given in the
tables, some mean values which appear to be quite different may not be
significantly different (statistically) while some mean values which appear
to be quite similar may be found to be significantly different when inter­group
differences in correlated variables are accounted for.
•'
-27-
The following example demonstrates how tests for between group
significant differences were performed for each of the two group com­parisons
of socioeconomic, nutrient and food intake variables. Consider
testing for a significant difference between Chicago Group I and Group
II incomes:
Ho: Monthly family income is not significant in discriminating
between Group I and Group II*
Ha: Monthly family income is significant in discriminating
between Group I and Group II
A discriminant function is estimated for the total number of observations
from Chicago Group I (n=322) and Group II (n=l99) including all of the
socioeconomic variables except the income variable. The residual sum of
squares is calculated for this function (S.S.E.h = .0070417) with the
associated d.f.h = 511. The income variable is then added to the variable
set used in the. previous function, and a new discrlininant function is
estimated calculating the residual sum of squares (S.S.E.m = .0070061)
with the associated d.f.m = 510. The appropriate hypothesis test statistic
for this procedure is the F-test. With the associated degrees of freedom
and alpha set at .05, the critical value is,
c.v.(511-510),510,.o5 = 3- 84
Thus with alpha = .05 the null hypothesis will be rejected if the cal­culated
F-statistic is greater than 3.84. Substituting the appropriate
residual sums of squares in the F-statistic calculation, the F-statistic
is found to be,
F(511-510),510
Since F1 ,510 = 2.59
monthly family income
Group I and Group II.
expected in 95 out of
.0070417 - .0070061
.0070061
510
511 - 510
2.59
<3.84 we fail to reject the null hypothesis that
is not significant in discriminating between Chicago
We also can state that this same result would be
100 trials.
As stated above, this same hypothesis testing procedure was used for
the Chicago and Georgia data to test for significant differences in all
possible two group comparisons of each socioeconomic, nutrient and food
intake variable.
Family Income and Food Expenditure~
Table 3.8 depicts the per capita food expenditure and the number of
* With appropriate algebra, the significance of a variable in discriminat­ing
is functionally related to x1 - x2 where X1 = mean family income
2s
in Group I, X2 = mean family income in Group II, and s = the standard
deviation which is assumed to be the same in both populations.
-28-
households reporting food expenditures for each group. Per capita income
and food program maintenance for each sample group is shown in table 3.9.
The per capita monthly value of bonus Food Stamps was calculated by divid­ing
the family size into the difference between the amount paid each month
for Food Stamps and the food purchasing value of the stamps received. The
difference in the per capita monthly value of bonus Food Stamps between
Chicago and Georgia is explained by the relationship of income levels.
Households with lower income levels receive more bonus Food Stamps; thus
the households in Georgia, having lower incomes than the households in
Chicago, received a larger number of bonus Food Stamps. The per capita
monthly value of Food Certificates was determined by dividing the total
retail value of Food Certificates received each month by the total house­hold
size. Although the income and food expenditure data are presented
as per capita figures, the test of significant differences of these
variables was computed for the total household income and food expendi­tures
and not for the per capita estimates. This was done to examine
independently the household size variable. By including it in the discrim­inant
function, the significant differences in total family income and
food expenditures could be estimated net of the influence of family size.
The significant between-group differences of these variables is also
presented in tables 3.ll and 3.12.
The average age and education (last year of schooling completed) of
mothers in each of the sample groups is shown in table 3.10. Again the
significant differences of these variables is presented in 3.ll and 3.12.
Infant Birth Weight Comparisons
Because birth weight and infant mortality statistics are frequently
used as indicators of the health status of infant populations, a comparison
of the sampled populations in Chicago and Georgia with national averages
on these bases is instructive. Since the Food Certificate Program had
been in effect only a short time in each area, one should not expect to
find any differences in these measures as a result of the program. These
data are presented only to describe the sampled populations relative to
national statistics. Table 3.13 shows the average birth weights of the
infants in the sample cross tabulated by mother's age and by sample group.
There is not any consistent relationship between mother's age and average
birth weights in either Chicago or Georgia nor are there any substantial
between-group differences in either area. Combining mothers of all ages
and all groups in each area, the average birth weight in each area is
6 lbs. l3 ounces (3090 grams), which is comparable with the 1963 u.s.
average of 3080 grams for infants born to nonwhite households with
incomes less than $3000.
Comparing the incidence of low birth w·eight babies Y (2500 grams or
1/ This term is used on the recommendation of the World Health Organiza­tion
and medical groups in the United States and refers only to the
weight of the child at birth with no implications as to length of
gestation or any other measure of maturity. "Variations in Birth
Weight - Legitimate Live Births - United States 1963, " National Center
for Health Statistics, u.s. Department of Health, Education and
Welfare, Series 22, No. 8, 1968.
-29-
less) in the sample groups (table 3.14) indicates a between-group consis­tency
in both areas except for the Georgia Clinic Referral Group which had
only a 2 per cent incidence as compared to a 20 per cent incidence in the
Food Certificate Only Group. Combining all of the groups and computing
the incidence for each area indicates that 15 per cent of the infants in
Chicago and 14 per cent in Georgia are classified as low birth weight
infants. As shown in the table, both of these statistics are comparable
with the 1963 U.S. average of 14 per cent low birth weight babies in
nonwhite households with annual incomes less than $3000. The 1963 sta­tistics
indicate that the incidence of low birth weight infants in white
families of all incomes is 2.3 percentage points lower than that found
for white families with incomes less than $3000 per year. The analogous
comparison for nonwhite low birth weight incidence shows a decrease of
2.2 percentage points, but the incidence for both nonwhite income classi­fications
is substantially greater than that found for whites. The
implication of this observation is that income alone does not appear to
account for the incidence of low birth weight infants.
The neonatal and infant mortality statistics for the samples pre­sented
in table 3.15 show considerable variability between groups, with
the lowest mortality rates being found in the Clinic Referral Group in
Chicago and in the Control Group in Georgia. Combining all groups in
each area shows that the neonatal mortality rates, 17.9 in Chicago and
14.8 in Georgia, are considerably below the 1968 U.S. average of 23.8
for nonwhites. Similarly, the infant mortality rates, 27.9 in Chicago
and 26.2 in Georgia, are also below the 1968 National nonwhite average
of 35.9. Considering the inhere~t social unacceptability of personal
interview questions regarding the mortality of infants born to the res­pondents,
whether the mortality rates reported in the interviews are a
function of a respondent under-reporting bias, or whether the respondents
did indeed have a lower infant mortality rate, is unclear. The high
incidence of low birth weight infants and the low average birth weights
of the infants in the samples would tend to support a hypothesis of bias
in reporting infant mortality because low birth weight/has been shown to
contribute heavily to perinatal and infant mortality.~
An alternative explanation may be that the sampled population's use
of health clinic facilities has had some impact on reducing mortality
rates. As shown in table 3.16, 81 per cent of the mothers of infants
in Chicago attended well baby clinics, as did 64 per cent in Georgia.
Evaluating the relative influences of reporting bias and clinic attendance
is not possible with this data nor are there any national statistics on
mortality rates of urban infants who have had clinic care to use as a
base of comparison.
Summary of the Sample Description Data
There are a substantial number of differences in the socioeconomic
characteristics of the sample groups in Chicago and in those from Georgia
Maternal Nutrition and the Course of Pregnancy, Committee on Maternal
Nutrition/Food and Nutrition Board, National Research Council, National
Academy of Sciences, Washington, D. C., 1970.
-30-
(tables 3.11 and 3.12). Utilizing the discriminant analysis framework,
which allows the testing of significant differences in individual socio­economic
characteristics while the variation in others is taken into
account, all possible two group comparisons (10 comparisons for each set
of 5 groups) were made for the Chicago data and for the Georgia data.
The single variable which was most often significantly different
between the groups in Georgia and in Chicago was Family Food Expenditures.
This result should have been anticipated since the criteria for grouping
the responses was participation in family food programs. In descending
order of occurrence, number of preschool children, family income, and
family size were found to be significantly different between the sample
groups in Chicago. Other socioeconomic variables which were different
between some of the Chicago sample groups, but occurred less fre~uently
than those already mentioned, were number of school children, mother's
age, and days since income receipt. Socioeconomic variables, in addi­tion
to food expenditures, that occurred as being significantly different
between some sample groups in Georgia were, in descending order of occur­rence,
income, family size, number of preschool children, number of school
children, and mother's education.
The practical implication of these findings is that these differences
are sources of between group variation which must be taken into account
if any analysis is to measure a net food program effect.
The sample data from Chicago and from Georgia on birth weights and
the incidence of low birth weight infants (tables 3.13 and 3.14) is
virtually identical with 1963 United\ States averages for nonwhite families
with incomes less than $3000 per year. The 1963 statistics indicate that
factors other than income explain the relatively high incidence of low
birth weight infants in nonwhite populations. In neither area was there
any consistent relationship between mother's age and birth weights or
substantial differences in birth weights between groups. Neonatal and
infant mortality rates (deaths per 1000 live births, excluding still
births) from the sample ·data in both Chicago and Georgia were substantially
below the 1968 United States averages for nonwhite births. This dis­crepancy
may be accounted for by the nature of the ~uestions in a personal
interview setting as well as the sampled population's high rate of well
baby clinic attendance. The possible presence and extent of this un~uanti­fiable
bias complicates the use of mortality rates as determinants of
whether or not the samples reflect a "high risk" infant population.
Part II - Infant nutrient intakes
Tables 3.17 and 3.18 present the sample group mean intakes of eight
nutrients for Chicago infants 5 months of age or less while tables 3.19
and 3.20 present the same data for the Georgia infants 5 months of age
or less. The mean nutrient intakes for infants 6 through 12 months old
by sample group are presented in tables 3.21 and 3.22 for the Chicago
data and in 3.23 and 3.24 for the Georgia infants. Stated with each of
the mean nutrient intake values is the Recommended Dietary Allowance (RDA)
weighted for age and weight. The weighted RDA's were derived by using
the average age in each group to estimate the associated average weight
-31-
with the Harvard Standards.J/ The average weight was used then to cal­culate
the RDA's using the recommended nutrient intake to weight
4
7elation­ships
shown in the 1968 Table of Recommended Dietary Allowances.21
The following discussion of the mean nutrient intakes relies on
examining the mean intakes expressed as a percentage of their respective
weighted RDA's. Discussing the mean values per se is of little value
because of the differences in the age composition-of each group .
As shown in tables 3.17 and 3.18, the mean nutrient intakes for
Chicago infants 5 months of age or less are all at least 50 per cent in
excess of the respective weighted RDA's with the exception of calories
and iron. The mean intakes of calories are marginally above the RDA's
and range from 103 to 114 per cent. However , mean iron intakes range
from 83 to 120 per cent of their respective RDA's. Using the same method
of analysis for infants less than 5 months of age in Georgia, one again
finds all mean nutrient intake/weighted RDA percentages in excess of
150 per cent with the exception of calories and iron. The mean calorie
intake/RDA percentages range from 112 to 135 while the iron percentages
range from 64 to 106.
The mean nutrient intake/RDA percentages for infants 6 through 12
months old in Chicago (tables 3.21 and 3.22) are all in excess of 160
per cent with the exceptions of calories, iron and vitamin C. The
calorie percentages range from 98 to lll per cent, the vitamin C per­centages
from lll to 168 per cent, while the iron percentages ranging
from 50 to 58 per cent are substantially lower than those found for
infants less than 5 months of ~ge. The Georgia data for infants 6
through 12 months old (tables 3.23 and 3.24) show even higher mean
nutrient intake/weighted RDA percentages, except for calories, iron and
vitamin C, in excess of 190 per cent. The calorie mean intake/RDA per­centages
range from 123 to 144 per cent, the vitamin C percentages from
lll to 160 per cent and the iron values, as was the case in Chicago,
are substantially below the values for the infants less than 6 months
old, ranging from 36 to 53 per cent.
Nutrient Intakes as Related to Per Capita Income
To explore the possibility that infant nutrient intakes may be
related to family per capita income, all of the nutrient intake records
regardless of sample group were segmented into per capita income groups
in each area for infants l through 5 months, and 6 through 12 months of
age. There did not appear to be any positive relationship between per
capita income and nutrient intakes for either age group in either
Chicago or Georgia.
Presented in Jelliffe, D. B., The Assessment of the Nutritional Status
of the Community, Annex I, World Health Organization, Geneva, 1966.
Food and Nutrition Board, National Academy of Sciences-National Research
Council, Recommended Daily Dietary Allowances, Revised 1968.
-32-
Infants Receiving Vitamin and Mineral Supplements
These values reflect only the nutrient intakes from foods and do not
include the consumption of vitamin and mineral supplements. Table 3.25
shows that more than half of all the infants in Chicago and Georgia were
receiving vitamin and mineral supplements. This gross average tends to
underestimate the actual incidence of infants receiving supplemental
vitamins and minerals because infants consuming commercially prepared
formulas (all are fortified) are usually not prescribed vitamin and
mineral supplements. These percentages, therefore, should be considered
together with the incidence of commercial formula consumption.
Comparison of Results With Other Surveys
As an overall indication of how the Georgi a and Chicago sample
data compare with other infant dietary surveys, the mean nutrient intakes
of all the Chicago infants and all the Georgia infants is compared in
table 3.26 with 1965 U.S.D.A. Survey Statistics. Excluding iron intakes,
the comparison of the data in general shows that for infants 5 months
of age or less, the mean nutrient intakes of infants in the U.S.D.A.
survey is somewhat higher than the Chicago and Georgia data.
Mean iron intakes for infants 5 months of age or less indicate
that both the Chicago and Georgia means are higher than the U.S.D.A.
figures.
Considering the data for infants 6 through 12 months of age, the
Chicago mean nutrient intakes are in general below those in the U.S.D. A.
survey, while the Georgia intakes are higher. The consistency in which
the Chicago and Georgia mean iron intakes are higher than the U.S.D.A.
averages may be explained in part due to the five year time difference
in which the data was collected and the growth in cow~ercial formula
sales during this same time period.
In general, the Chicago children had nutrient intakes slightly
below those of the U.S.D.A. national average except for iron. On the
other hand, the Geor gia infants were quite comparable with the U.S.D.A.
average except for iron intakes in the younger children.
Distributions of the Nutrient Intakes
Although to comparison and evaluation of the mean nutrient intakes
of groups is a standard procedure, it is a questionable procedure to use
when discussing the nutritive adequacy of the diets of a population
because of the distribution of the intakes. As shown in table 3.27,
29 .4 per cent of the infants 5 months of age or less and 43.2 per cent
of infants 6 through 12 mou~hs of age in Chicago were receiving less than
70 per cent of their RDA 's2/ in three or more nutrients. In Georgia
a basically analogous situation was found with 22 per cent of infants
2/ Calculated for each infant using the Jelliffe tables of the Harvard
Standards.
-33-
5 months of age or less and 37 per cent of infants 6 through 12 months
receiving less than 70 per cent of their RDA's in three or more nutrients.
The distributions of the proportions of infants receiving various
levels of their calorie, protein, calcium and iron RDA's is given in
tables 3.28 and 3.29 for infants 5 months of age or less in Chicago and
Georgia respectively. As the tables indicate, very substantial propor­tions
of the populations in both areas are receiving less than 70 per
cent of their RDA' s in iron while relatively smaller but still sizeable
propor t ions are receiving less than 70 per cent of their RDA's in calories.
Tbis same phenomenon is even more accentuated by examining the distribu­+
lons of the intakes of infants 6 through 12 months of age (table 3.30
f or Chicago and 3.31 for Georgia). For all sample groups, the proportions
of infants in this age group receiving less than 70 per cent of their iron
RDA's is larger in Georgia than in Chicago. Conversely, a relatively
small number of infants in Georgia as compared to Chicago are receiving
less than 70 per cent of their RDA's in calories.
Because of the unknown serial correlation in infants' daily nutrient
intakes, to assess the significance of the proportion of individual
dietary intakes which are g~low various levels of the RDA's is difficult.
Davis, Gershoff and Gamble~ have reported a study showing the distribu­tions
of infants, age l through 12 months, nutrient intakes relative to
1968 RDA's. The data was collected on infants from upper middle class
families and is presented in table 3.32.
As the comparisons i ndicate, a smaller proportion of Chicago and
Georgia infants age l through 5 months have intakes less than 2/3 RDA
for vitamins C, A, and iron, but a greater proportion have low calciuu1
intakes. The proportions of Chicago and Georgia infants 6 through 12
months receiving less than 2/3 the RDA is greater for all nutrients than
the proportions reported by Beal. Given this admittedly fragmentary
evidence, there is reason to doubt the nutritional adequacy of the
dietary intakes of the Chicago and Georgia infants 6 through 12 months
of age, but little reason to doubt the adequacy of the diets of infants
less than 6 months of age.
Table 3. 33 is presented to show the relationship between dietary
iron intake and the use of vitamin and mineral supplements. As shown in
the table, there is no significant direct or inverse relationship between
dietary iron intake and the use of vitamin and mineral supplements for
Chicago infants in either age group. In Georgia, however, the incidence
of vitamin and mineral supplement use does appear to be directly related
to dietary iron intakes, particularly for infants l through 5 months of
age. What a platitude!
Concurring with the opinions offered by Davis, Gershoff and Gamble
that due to the dispersion of intake data and the periodic revisions of
§/ Davis, T. R. A., Gershoff, S. N., Gamble, D. F. "Review of Studies of
Vitamin and Mineral Nutrition in the United States (1950-1968 ),"
Supplement 1, Journal of Nutrition Education, Fall 1969, pp. 44-52.
-34-
the RDA's, "the results of nutrition surveys should be presented by plot­ting
the cumulative percentages of ~~bjects plotted against the actual
units of the nutrient in question. "'il Figures 1, 2, 3, and 4 have been
constructed.
Figures 1 and 2 show the cumulative percentages of Chicago infants
1-5 and 6-12 months plotted against the actual computed units of intake
of calories, protein, calcium, iron, vitamin A, thiamine, riboflavin and
vitamin C. Figures 3 and 4 present the Georgia nutrient intakes in the
same manner. With the provided calibrations, these figures can be used •
in a straightforward manner to determine the proportions of infants in
either age group in either area receiving less than any selected level
of the various nutrients. Because of the age distributions of the infants
in the 1 through 5 and 6 through 12 month categories, Figures 1 through 4
tend to distort the actual relationship of the proportions of the infants
receiving various levels of their RDA's. To determine the actual propor­tions
of infants in each of the age groups receiving less than various
levels of their individual RDA's, Figures 5 and 6 have been constructed
for the Chicago infants and 7 and 8 for the Georgia data. These figures
show the cumulative percentages of the age groups receiving various pro­portions
of RDA's based on the ages of each individual infant. The data
for these tables was generated by dividing each infant's nutrient intakes
by the specific RDA for his age. (RDA's for infants 1 through 12 months
of age were determined using Jelliffe's tables of the Harvard Standards
and the weight to nutrient intake relationships specified in the revised
1968 Recommended Dietary Allowances.) The distortions in evaluating the
distributions on the basis of intakes per se, not accounting for individual
RDA's, can be seen by comparing any of the distributions in Figures 1
through 4 with those in 5 through 8. When either Figures l through 4 or
5 through 8 are examined, one realizes that mean nutrient intakes are
inadequate as the sole basis for describing the nutritional adequacy of
the diets of an entire population.
Significant Between-Group Differences in Nutrient Intakes
The significant differences in nutrient intakes between the sample
groups after accounting for the between-group variability in socioeconomic
variables (i.e., the tests based on discriminant analysis) are shown in
table 3.34 for the Chicago data and in 3.35 for Georgia.
For the Chicago data, of the 80 possible differences (10 group com­parisons
of 8 nutrients) in each of the age groups, only seven significant
differences in nutrient intakes were found for infants 1 through 5 months
of age and eight for infants 6 through 12 months. Vitamin A accounted
for five of the seven significant differences in the infants 1-5 group,
while the other two differences were in protein intakes. Of the eight
significant differences in the infants 6 through 12 months of age group,
three were calories, two vitamin C, two vitamin A, and one thiamine.
There is no consistent food program effect (Food Stamp or Food Certifi­cate)
in either age group for any of the nutrients.
7.) Ibid., p. 53
-35-
The tests performed for the Georgia data indicate that of the 80
possible differences in each of the age groups only three were found for
infants l through 5 months of age and two for infants 6 through 12 months.
Two of the differences in the l-5 group were in iron intake, while the
other was vitamin C. Calcium and vitamin C were the two nutrients that
were significantly different for the 6 through 12 months of age group of
infants . As was the situation with the Chicago data, no consistent food
program effect emerged from the analysis. Because so few differences
were found , an analysis was made in which the residual sum of squares
wa ~ calculated for the function with and without the entire set of nutrient
vnriables. This was undertaken simply to determine if the addition of
·.nowledge about the group nutrient intakes significantly reduced the un­explained
differences between groups (reduced the residual sum of squares).
Of the 40 differences evaluated, ten for each age group for each area,
only four significant differences were found.
Food Sources of Nutrients
When examining the nutrient intakes of groups consideration of the
food sources of the nutrients is also instructive . Table 3.36 shows the
per cent contribution of commercial infant formulas, dairy products and
infant cereals to the nutrient intakes of infants for Chicago and Georgia
respectively . As would be expected for infants l-5, commercial formula,
milk, and cereal are the primary sources of virtually all nutrients.
The only exception is vitamin A which is contributed by the fruits,
vegetabl es, and juices food group. Although commercial formula, all of
which has iron, was consumed by an average of 65 per cent of all of the
sample groups, the primary source of iron for the Chicago l-5 infants
was infant cereals. The Georgia data for the same age group shows the
same high dependence on formula, dairy products, and infant cereals for
the primary sources of nutrients. Infant cereal plays a smaller role
in providing iron for the Georgia infants than it did in Chicago. This
is primarily due to a lower level of cereal consumption in Georgia and
therefore a higher reliance on formula as a source of iron. The data
for the infants 6 through 12 months of age in Chicago and Geor gia show
a continued dependence on these three food groups as the primary source
of nutrients. As would be expected, the large contribution of formula
to nutrient intakes for infants l-5 is replaced by dairy products con­sumption
(primarily milk) for infants 6-12. The primary source of iron
continues to be cereal for the Chicago infants but becomes foods other
than the three listed for the Georgia infants. Although the nutritive
value of breast milk is included in computations of nutrient intakes,
it affects only a very small number of the calculations. As shown in
table 3.37, 4.8 per cent of the mothers in Chicago were breast feeding
as compared to 5 per cent in Georgia. Because the incidence was so
minor, its contribution to the nutrient intakes was not evaluated
separately.
Part III - Infant consumption of instant dry infant cereal,
commercial formulas, and milk
Tables 3.38 and 3.39 show the incidence of formula and milk consump­tion
by sample groups for Chicago and Georgi a respectively . The data for
-36-
infants 1-5 in Chicago indicates that the percentage of infants consuming
a commercially prepared formula ranges from 62 to 74 with no consistent
Food Certificate or Food Stamp Program effect.
The proportions of infants consuming an iron fortified formula
ranges from 8 per cent for the control group to 23 per cent for the Food
Stamp only group. The proportions for the infants receiving Food Cer­tificates
ranges from 12 to 17 per cent. The proportions of infants
6-12 months of age consuming a commercial formula ranges from 21 to 25
per cent with no apparent program effects and the proportions consuming
an iron fortified formula range from 4 to 8 per cent. The Georgia data
suggest both Food Stamp and Food Certificate Program effects for infants
1-5 months of age with commercial formula consumption ranging from 52
per cent for the control group and 85 per cent for the Food Stamps plus
Food Certificate group. The consumption of iron fortified formula
ranges from 14 per cent for the referral group to 44 per cent for the
Food Stamp plus Food Certificate group . The commercial formula consump­tion
of Georgia infants 6-12 months of age ranges from 9 per cent for
the Food Certificate only group to 31 per cent for the referral group.
As in Chicago, the proportions of infants 6-12 months of age consuming
an iron fortified formula are relatively small, ranging from 5 per cent
for the Food Stamp group to 8 per cent for the referral group .
The average consumption of formula as well as instant infant cereal
and milk is given in table 3.40 for Chicago and Georgia. The Chicago
data suggests few between-group differences for formula, milk, or
cereal for either age group. Using the co-variance analysis to account
for socioeconomic variables which could influence consumption, the only
significant differences due to program participation between groups was
found for milk consumption (table 3 .41). The differences were between
the control group and the referral group for both infants 1 through 5
months of age and 6 through 12, and also between the Food Stamp and
control group for the older infants.
The Georgia data (table 3.40) suggests program effects for formula,
milk and cereal consumption. As shown in table 3.41, there are three
significantly different intakes, formula, milk, and cereal occurring
once each, for infants 1-5 and two significant differences for infants
6-12 in the consumption of cereal. The order of the differences do not
indicate any consistent Food Certificate or Food Stamp program effects.
An infant feeding ang nutrient intake study of upper middle class
families reported in 1954~ stated that during the first two years of
life infant consumption of milk reaches its peak at age six months and
is approximately 27 ounces. Two national probability samples reported
Beal, V. A. "Nutritional Intake of Children, II Calcium, Phosphorous,
and Iron," ;:rournal of Nutrition 53: 1954, p. 504.
-37-
in 196321 and 1964~ gave evidence that the national average consumption
of milk by 6 to 8 month old infants is approximately 25-28 ounces. Con­sidering
that the control groups consumption of milk plus formula is in
this general range, one would not expect the Food Certificate Program to
have any substantial consumption supplement effects.
Part IV - Family member milk intakes
Pregnant and Not Pregnant Women's Milk Consumption
Table 3.42 presents the average daily milk and total fluid consump­tion
for pregnant and not pregnant women by sample group in Chicago. One
should consider the average daily total fluid intakes when analyzing
the data for program effects, since the intake of any specific beverage
is constrained as well as influenced by total fluid intake. Admittedly
this argument assumes that a substitution effect is more probable than
is an additive effect as the result of the Food Certificate Program.
Table 3.43 presents the same data in like manner for the Georgia women.
A possible seasonal effect in intakes (Chicago data was collected in
August; Georgia data in October) is indicated by the all group total
fluid intake average for not pregnant women in Chicago which is almost
10 fluid ounces greater than that for Georgia. When adjusted for average
daily water intake, both the Chicago and Georgia estimates of total fluid
consumption compare favorably with a 1969 estimate of the United States
average daily per capita consumption of commercial beverages. The
discriminant analysis indicates that for the milk intakes of pregnant
women there are no between-group differences in Chicago and only the
Food Certificate and Control group differ significantly in Georgia
(table 3.44). Considering the milk consumption of the not pregnant
women, the Food Stamp plus Food Certificates and the Food Certificate
Group differ significantly from the Control group in Chicago, while the
Food Stamp plus Food Certificate Group differs significantly from the
Food Stamp group in Georgia (table 3.44).
Incidence of Milk Intolerance
A factor that could influence the impact of the program on milk
consumption is the extent of allergies or intolerance to milk in the
target population. From the field s~rrvey relatively little milk intol­erance
among infants , preschool children, and the women interviewed
was discovered. As shown in table 3.45, the highest incidence of milk
intolerance was reported for Georgia women and accounted for 5.2 per
cent of the total sample. The same statistic for the Chicago women is
3.7 per cent,and for infants and preschool children in both areas was
less than l per cent.
21 Filer, L. J ., Martinez, G. A. "Caloric and Iron Intake by Infants in
the United States : An Evaluation of 4,000 Representative Six-Month­Olds,"
Clinical Pediatrics, Vol. 2, No . 9, 1963, p. 474.
!Qj Filer, L. J., Martinez, G. A. "Intake of Selected Nutrients by Infants
in the United States: An Evaluation of 4,000 Representative Six-Month­Olds,"
Clinical Pedi atrics, Vol. 3, 1964, p. 635 .
-38-
Maternal Vitamin and Mineral Supplements
Table 3.46 indicates that 42 per cent of the pregnant women in Chicago
and 74 per cent of those in Georgia were taking vitamin and mineral suppl e­ments
at the time of the interview. Substantiall y small er proportions,
15 per cent in Chicago and 21 per cent in Georgia, of the not pregnant
women were also taking supplements.
Household Milk Consumption
Table 3.47 shows the average mil k consumption per household and
average milk consumption per household member for Chicago and Georgia.
In general, the households participating in the Food Certificate Program
in Chicago consume more milk per week than do the households on Food
Stamps alone or households receiving neither stamps nor certificates.
However, when the gross consumption figures are adjusted for household
size, there are onl y minor between-group differences. The tests for
significant differences, which takes into account household size as well
as the other socioeconomic variables, indicate that the household con­sumption
of the Control group differs significantly from the Food Cer­tificate
and the Food Stamp plus Food Certificate groups (table 3.44).
The Georgia data, table 3.47, does not show any consistent difference
between certificate and.non-certificate recipient households but does
indicate that the average household consumption for all groups in Chicago
is more than 50 per cent above the all group average in Georgia. As was
the case with the Chicago data, adjusting the gross household purchase
figures by household size reduces the magnitude of the apparent between­group
differences. The only significant between-group difference was
for the Food Certificate and the Referral groups (table 3.44).
Preschool Children's Milk Consumption
Table 3.48 shows the average daily per capita milk consumption of
preschool children (ages 1 through 5 years) for the Chicago and Georgia
households. Only minor differences appear between the sample groups in
both areas but the lower average daily consumption in Georgia may be due
to seasonal effects. Significant differences in preschool children's
milk consumption were found only between the Food Certificate and Clinic
Referral groups in Georgia and between the Clinic Referral and Control
group in Chicago (table 3.44). Although the average daily per capita
consumption of nonfat dry milk is also presented for the Georgia children,
the differences were not tested for significance due to the low average
intakes.
Summary of Program Effects (Parts II, III and IV)
In anal yzing the nutrient intake data by comparing each sample group
with all others from the same locality and in the same age group, there
were 320 possible differences in nutrient intakes. Of the 160 possible
differences in Chicago only 15 statistically significant ones were found,
seven for infants 1 through 5 months and eight for infants 6 through 12
months of age. Five of the seven differences for infants 1 through 5
months were for vitamin A intakes, which is primarily contributed to the
-39-
diets by foods other than those provided by Food Certificates. The eight
significant differences for infants 6 through 12 months of age were com­prised
of three calorie differences, two vitamin A, two vitamin C and
one thiamine. No consistent Food Stamp or Food Certificate Program
effects are evident in either age group. Of the 160 possible differ­ences
in Georgia, three were found for infants l through 5 months and
two for infants 6 through 12 months of age. Two of the three differences
in intakes of infants l through 5 months were iron and the two differ­ences
for infants 6 through 12 months were calcium and vitamin C. As
in Chicago, no consistent Food Stamp or Food Certificate Program effects
emerged from the analysis.
With the level of significance (type I error) equal to .05, one
would expect to make 16 type I errors (i.e., reject the null hypothesis
of no difference when it is true) in 320 tests. Therefore, in appraising
the total of 20 significantly different intakes for all infants, the
reader must keep in mind that most or all of these differences likely
are the consequences of chance rather than real program effects.
Examining infant intakes of cereal, formula, and milk one finds
only seven significantly different intakes for all infants in both areas
(there are 120 possible differences) and the order of the differences
did not indicate any consistent Food Stamp or Certificate Program effects.
A similarly small number of significant differences in the milk con­sumption
of households, pregnant women, not pregnant women, and preschool
children were found. In both areas combined there were three significant
differences in the milk consumption of not pregnant women and one differ­ence
in the consumption of pregnant women. In three comparisons, the
total milk consumption of households was found to differ significantly
and in only one case was there any between-group difference found in the
milk consumption of preschool children.
Based on the small number of significant differences found in the
variables that could indicate a Food Certificate Program effect, one
must conclude that the Food Certificate Program has had little or no
consistent effect on the nutrient intakes of infants, milk, formula and
cereal consumption of infants or the milk consumption of women partici­pating
in the program. Considering the average quantities of milk plus
formula consumption in the control groups, this result is not surprising.
Similarly the distributions of the nutrient intakes in the control groups
are such that for the program groups to exhibit significantly higher
nutrient intakes would be difficult, except in the case of iron. Because
no consistent program effects were found influencing iron intakes, one
must conclude that the program did not increase successfully the con­sumption
of iron fortified formulas and cereal to the extent that it
significantly increased iron intakes.
Table 3.1 Size of Subpopulations, Sample Groups, and the Determination of Questionnaires Fielded
Chicago
Estimated* Number of Number of Net Completed
subpopulation questionnaires questionnaires fielded intergroup** question-size
fielded Refused Not at home Moved transfers naires
Group I
F. S. + F.C. 1070 389 12 52 51 +48 322
Group II
F. S. 973 240 4 23 32 +18 199
Group III
F.C. 648 331 17 23 43 -66 182
Group IV
F. C.- Referral 627 355 12 72 79 -ll-9 143
Group V
Control 362 257 12 50 50 +49 185
Total Sample 368o 1572 57 220 264 1031
(3 .6%) (14 .0%) (16 .7%)
*Groups I, III and IV estimated from Food Certificate Program participants' application forms;
Groups II and V estimated from Public Assistance and Food Stamp Records .
**Number of respondents added to or deducted from the respective groups on the basis of' their
food program participation at the time of' the interview.
I
+
}J
t ,.
Tabl.e 3.2 Size of Subpopulations, Sample Groups, and the Determination of Questionnaires Fielded - Georgia
Group I
F.S. + F.C.
Group II
F.S.
Group III
F.C.
Group IV
F .C. - Referral
Group V
Control
Total Sample
Estimated*
Subpopulation
Size
372
202
82
217
217
1090
Number of
Questionnaires
Fielded
254
155
So
l4o
155
684
Number of Questionnaires
Fielded
Refused Not at Home Moved
2
5
6
13
(L9%)
35
24
9
17
31
116
(17 .0%)
33
14
8
16
19
104
(15.2%)
Net Completed
Intergroup** Questi on-
Transfers naires
-41 143
0 112
+50 113
-16 91
+6 105
564
*Groups I, III and IV estimated from Food Certificate Program participants' application forms; Groups II
and V estimated from Public Assistance and Food Stamp Records.
**Number of respondents added to or deducted from the respective groups on the basis of their food program
participation at the time of the interview.
I
+o­f-'
I
Table 3.3 Subpopulation and Sample Group Sizes - Chicago
Sample NUMBER OF WOMEN NUMBER OF INFANTS
group Pregnant Not pregnant
size Subpopulation Sample Subpopulation Sample Subpopulation Sample
Group I
F.S. + F.C. 322 289 96 781 226 912 248
Group II
F.S. 199 350 76 623 123 681 129
Group III
F.C. 182 174 41 474 141 570 149
I
..r:-
Group IV ~
F. C. - Referral 143 275 46 352 97 427 103
Group V
Control 185 137 74 225 111 264 121
Total Sampl e 1031 1225 333 2455 698 2854 750
~
Table 3.4 Subpopulation and Sample Group Sizes - Georgia
Sample NUMBER OF WOMEN NUMBER OF INFANTS
Group Pregnant Not pregnant
Size Subpopulation Sample Subpopulation Sample Subpopulation Sample
Group I
F.S. + F.C. 143 106 47 266 96 270 llO
Group II
F. S. ll2 49 26 153 86 155 88
Group III
F. C. 113 17 29 65 84 7l 86
I
Group IV .::- w
F. C.-Referral 91 127 44 90 47 97 48 I
Group V
Control 105 27 25 190 So 201 83
Total Sample 564 326 171 764 393 794 415
Table 3.5 Number of 24 Hour Recall Records of Mothers and Infants Used in the Analysis - Chicago
Group I
F.S. + F.C.
Group II
F.S.
Group III
F.C.
Group IV
F .C.-Referral
Group V
Control
Total Sample
Infants
Number of 2
hour recalls
recorded
248
129
149
103
121
750
complete food intakes)
Number of Number of
unusable *24 . usable 24
hour recalls hour recalls
25 223
20 109
15 134
14 89
10 111
84 666
Mothers
Number of 2
hour recalls
recorded
322
199
182
143
185
1031
fluid intakes only)
Number of Number of
unusable*24 usable 24
hour recalls hour recalls
1 321
199
182
1 142
1 184
3 1028
*Unusable intake records include those with i nadequate enumerator recording, lack of respondent
cooperation, sick and teething infants, and atypic