Appetite, 1998, 31, 205–228
Article No. ap980180
Fruit and Vegetable Consumption, Nutritional Knowledge and
Beliefs in Mothers and Children
E. L. GIBSON, J. WARDLE and C. J. WATTS
Health Behaviour Unit, Department of Epidemiology and Public Health,
University College London
Increasing fruit and vegetable consumption is an important health behaviour.
Parental and other psychosocial inﬂuences on children’s fruit and vegetable
consumption are poorly understood. The contribution of a variety of psychosocial
and environmental factors to consumption of fruit and vegetables by children
aged 9–11 years was explored. Ninety-two mothers and children (48 girls and 44
boys) were recruited via urban primary health-care practices. Socio-economic and
educational level, nutritional knowledge and health- and diet-related beliefs
and attitudes were assessed in mothers and children by questionnaires and
semistructured interviews. Mothers’ diets were measured by a food frequency
questionnaire, while children’s diets were assessed by 3-day diaries (N=80). The
pattern of inﬂuence of the various measures on fruit and vegetable consumption
was compared with that on children’s confectionery intake. The children’s intakes
of macronutrients were typical for the U.K. (37% fat, 50% carbohydrate and
13% protein by energy; 12 g/day ﬁbre), while median fruit, fruit juice and vegetable
intake amounted to about 2·5 servings/day. Univariate correlations and subsequent
multiple regression analyses revealed quite diﬀerent inﬂuences on the three
food types. Independent predictors of children’s fruit intake included mothers’
nutritional knowledge (b=0·37), mothers’ frequency of fruit consumption (b=
0·30) and mothers’ attitudinal conviction that increasing fruit and vegetable
consumption by their children could reduce their risk of developing cancer
(b=0·27; multiple r
=0·37, p<0·0001). Children’s vegetable consumption was
independently explained by the child’s liking for commonly eaten vegetables (b=
0·36) and the mother’s belief in the importance of disease prevention when
choosing her child’s food (b=−0·27; r
=0·20, p<0·001). Children’s confectionery
consumption was predicted by the mother’s liking for confectionery (b=0·32)
and the children’s concern for health in choosing what to eat (b=–0·26; r
p<0·005). Children’s consumption of fruit and vegetables are related to diﬀerent
psychosocial and environmental factors. Promotion of this behaviour may require
attention to nutritional education and child feeding strategies of parents.
1998 Academic Press
Carolyn Watts is now at the Ministry of Health, Wellington, New Zealand.
The Health Behaviour Unit is funded by the Imperial Cancer Research Fund, U.K.
The authors gratefully acknowledge Kate Wrightson and Wendy Solomons for their considerable
contribution to interviewing the mothers and children. The authors are also grateful to all the families
that took part in this research.
Address correspondence to: Professor Jane Wardle, Health Behaviour Unit, Department of Epi-
demiology and Public Health, University College London, 2–16 Torrington Place, London WC1E 6BT,
0195–6663/98/020205+24 $30.00/0 1998 Academic Press
E. L. GIBSON ET AL.
There is considerable epidemiological evidence for the importance of fruits and
vegetables in reducing risk of cancer (Block et al., 1992; Negri et al., 1991) and
cardiovascular disease (Key et al., 1996). Reviews of the evidence have suggested
that ﬁve servings of fruits and vegetables per day would confer considerable beneﬁts
to health, while representing a feasible target which the public could strive to achieve,
as well as being a memorable and marketable number (Foerster et al., 1995). In the
U.S., the National Cancer Institute instigated the Five-a-Day Program, aimed at
encouraging the public to eat a minimum of ﬁve servings of fruit and vegetables per
day, as well as promoting knowledge of the health beneﬁts provided by fruit and
vegetables (Foerster et al., 1995). Similarly, an increase in consumption of fruit
and vegetables has been recommended by U.K. and other European authorities
(Department of Health, 1994; Scottish Oﬃce, 1993), and the World Health Or-
ganisation (WHO, 1990) has recommended a minimum daily intake for adults of
400 g of fruit and vegetables per day. In the U.K., the Department of Health (1994)
has recently recommended that intake of fruits and vegetables should be increased
by at least 50%, with the suggestion that this target could be met by a minimum
intake of two servings of fruit and four of vegetables per day, excluding potatoes.
In the U.S., to attain the Five-a-Day target, the public has been recommended to
double their daily consumption of fruit and vegetables (U.S. Department of Health
and Human Services, 1990).
The extent of the U.S. National Five-a-Day Program is impressive, and yet early
indications suggest that its impact may be disappointing: whilst there have been
signiﬁcant improvements in the public’s awareness of the link between fruit and
vegetables and disease prevention, and people’s recognition of the shortcomings of
their own diet, actual consumption of fruit and vegetables does not appear to have
increased signiﬁcantly, even in California where the Five-a-Day campaign originated
and has been in existence since 1988 (Foerster et al., 1995). This, together with other
studies of factors inﬂuencing fruit and vegetable consumption (Laforge et al., 1994),
emphasizes the resistance to dietary change. Such diﬃculties have led to calls for
greater co-operation between nutritional and behavioural scientists in the study of
nutritional behaviour (Booth, 1994), since no matter how accurate the expert
advice, dietetic exhortations are impotent without successful behavioural change.
Furthermore, despite evidence supporting a protective eﬀect of fruit and vegetables
against cancer and cardiovascular disease, no clear beneﬁcial role for particular
nutrients has been deﬁned (Block et al., 1992; The Alpha-Tocopherol Beta Carotene
Cancer Prevention Study Group, 1994): therefore, the emphasis here is on factors
inﬂuencing consumption of fruits and vegetables rather than intake of the nutrients
which these foods provide.
Studies of relationships between consumer characteristics and fruit and vegetable
intake have typically concentrated on sociodemographic qualities, especially socio-
economic status (Department of Health, 1989; Laitinen et al., 1995; Wolfe &
Campbell, 1993); but such ﬁndings by and large beg, rather than answer, questions
about psychosocial inﬂuences on healthy food choice. Studies of psychosocial
inﬂuences on fruit and vegetable consumption have tended to favour qualitative
methods, including focus group reports, one in Dutch adults (Brug et al., 1995), and
one among U.S. school children (Domel et al., 1995), or restricted applications of
FRUIT AND VEGETABLE INTAKE BY CHILDREN
theoretical sociocognitive models (Brug et al., 1995; Resnicow et al., 1997). It is clear
from the observations in these studies that beliefs and attitudes about diet-disease
relationships, and availability of and aﬀective reactions to fruits and vegetables, are
important inﬂuences on their consumption. Habitual use of, and exposure to, fruits
and vegetables from an early age are also likely to be critical (Domel et al., 1993;
Krebs-Smith et al., 1995; Sullivan & Birch, 1994). In addition, the extent of nutritional
knowledge has been found to predict dietary quality (Axelson et al., 1985), albeit
with varying degrees of success (Shepherd & Towler, 1992), but this has not been
investigated speciﬁcally as a determinant of fruit and vegetable consumption. Given
that nutritional education is a central component of intervention strategies, the
association between nutritional knowldge and fruit and vegetable consumption is a
critical area for investigation.
A greater understanding of the early development and interaction of these factors,
i.e. accounting for the reasons that children eat particular amounts of fruits and
vegetables, is of particular importance for three main reasons: (1) food preferences
and habits are often established in childhood and tend to be maintained into
adulthood (Kelder et al., 1994; Rozin, 1990); (2) nutritional inﬂuences on health can
be established in childhood, with long-term consequences (Berenson et al., 1989;
Kimm et al., 1990); (3) childhood may be a time of greater openness to modiﬁcation
of food choice (Birch, 1990; Wardle, 1995), while children’s consumption of fruits
and vegetables is especially poor (Basch et al., 1994; Department of Health, 1989;
Krebs-Smith et al., 1996). In adults, formal models have been used to determine
relative psychosocial inﬂuences on dietary quality with some success (Laforge et al.,
1994; Shepherd & Towler, 1992): however, a recent application of such a model to
fruit and vegetable intake in fourth- and ﬁfth-grade school children proved prob-
lematic (Domel et al., 1996), and it is likely that concepts employed in these models
are not entirely appropriate for explanations of food choice in school children of
this age. Therefore, the present study used a less structured approach, incorporating
environmental as well as psychosocial variables. The study was designed to investigate
the relative inﬂuences on children’s fruit and vegetable consumption of variables
describing mothers’ socio-economic and educational status, mothers’ and children’s
nutritional knowledge, dietary beliefs and preferences, diet–disease attitudes, and
mothers’ habitual consumption, as well as mother–child correlations in knowledge,
beliefs and behaviour.
Mothers were recruited who had at least one child aged between 9 and 11 years:
children of this age were chosen because they can be expected to provide reliable
responses to semi-structured interviews, may be acquiring a mature understanding
of health together with some knowledge of nutrition and are soon likely to have
considerably more independence in choosing foods; moreover, they are of an age at
which intervention may be critical for life-long health (Berenson et al., 1989).
As a comparison to ﬁndings on fruit and vegetable consumption, the inﬂuence
of the same group of factors on children’s confectionery consumption was studied.
This comparison is of interest because confectionery forms a substantial portion of
children’s diets (Department of Health, 1989; Kimm et al., 1990), and confectionery
snacks are likely to be in direct competition with attempts to increase fruit and
vegetable intake by encouraging their consumption between meals. Furthermore,
beliefs concerning the impact of confectionery on health might be expected to
contrast with such beliefs about fruits and vegetables (Resnicow & Reinhardt, 1991).
E. L. GIBSON ET AL.
Ninety-two mothers with 9–11 year-old children were recruited from primary
care patient registers within south London, England. Five local primary care practices
were identiﬁed as reﬂecting the range of neighbourhoods typical for this area of
London, and all these practices agreed access to their patient registers. All families
with children in the designated age group were identiﬁed and written to, then
subsequently telephoned. The letters sent to mothers invited them to participate in
a study of family health involving visits to the clinic (‘‘The Family Diet Study’’).
The study was approved by the Ethics Committee (Research) of the Institute of
Psychiatry, London, where the Unit was based at the time. Fifty-ﬁve percent of 167
suitable families contacted took part. Twenty percent agreed to take part but failed
to arrive for the interview; twenty-ﬁve percent refused to take part. Mothers and
their children were interviewed separately, when as well as completing semi-structured
interviews and a number of standardized instruments, physiological and an-
thropometric measures were taken. Comprehension of the questionnaires was es-
tablished by the interviewer. There were no exclusion criteria for ethnicity, provided
that mothers and children spoke and read English ﬂuently.
Weight and height were measured using standard procedures, to the nearest
0·1 kg and 0·5 cm, respectively. Waist, hip, blood pressure and serum cholesterol
were also measured while the subjects were at the clinic; details of these procedures,
together with results of mother–child correlations of anthropometric measures, will
be presented elsewhere.
Mothers completed questions on health behaviour, including their own and their
children’s health status over the last year, frequency of exercise, smoking and drinking
Socio-economic group. The socio-economic group of mothers was determined by
their partners’ occupations, or their own occupations if their partners were un-
employed or absent, using the standard occupational classiﬁcation (Oﬃce of Popu-
lation Censuses & Surveys, 1990). For analysis here, levels A, B and C1 were
categorized as ‘‘non-manual’’, while levels C2, D and E were labelled ‘‘manual’’.
Deprivation index. An index of economic deprivation was constructed by summing
presence of the following criteria: (1) neither owning nor having the use of a car;
(2) living in rented accommodation; (3) partner currently unemployed; (4) mother
not in full-time employment, in addition to partner being unemployed or not present.
Thus, the deprivation index score could range from 0 to 4 (low to severe deprivation).
FRUIT AND VEGETABLE INTAKE BY CHILDREN
Frequency of food consumption
Mothers completed a food frequency questionnaire, by recording their average
consumption, in the last year, of medium servings of 130 common foods and drinks.
Frequency measures were: never or less than once per month, 1–3 per month, 1 per
week, 2–4 per week, 5–6 per week, 1 per day, 2–3 per day, 4–5 per day, 6 or more
per day. This instrument was originally developed and validated for the European
Prospective Investigation of Cancer (Bingham et al., 1994). Total daily frequencies
of consumption of fruit (10 classes) and vegetables (21 classes) and confectionery,
were calculated from the questionnaire by summing consumption rate per month
for each item in a food group, and dividing by 30. Tinned fruit was excluded from
the fruit consumption calculation because: (1) no distinction had been made between
sweetened and unsweetened varieties, which could lead to conﬂation of attitudinal
eﬀects depending on perception of the sugar content of this item; (2) the potential
uniqueness of its use and availability amongst such families; (3) this non-speciﬁc
fruit item was listed after the other individual fruits, so that there was a risk that
duplication of estimated fruit use may have occurred. Frequencies for items labelled
‘‘melon’’, ‘‘peaches, plums, apricots’’ and ‘‘strawberries, raspberries, kiwi fruit’’ were
divided by three to adjust for seasonal availability: this was necessary to avoid
overrepresentation of annual use, because subjects had been asked to estimate their
use of these fruits speciﬁcally when in season. Five items were excluded from the
vegetable consumption calculation because of peculiarities of use or additional fat
content: these were ‘‘onions’’, ‘‘garlic’’, ‘‘coleslaw’’, ‘‘dried lentils, beans, peas’’, and
‘‘tofu, Soya meat, TVP, vegeburger’’. The confectionery variable was summed from
sweets, chocolate, chocolate snacks and sweet biscuits (both plain and chocolate).
Amongst these confectionery items, a ‘‘serving’’ was represented as either one biscuit
or snack or a ‘‘medium serving’’.
Three-day dietary diary
Children were asked to complete (with the help of their parents) a diary of
everything they ate and drank for 3 days: they were asked to choose two typical
weekdays and one weekend day. The diaries included extensive instructions for
completion of records for diﬀerent foods, as well as photographic guides to portion
sizes. Eighty children returned completed diaries. Nutrient intake was derived from
the diaries by a qualiﬁed dietitian using Microdiet (University of Salford, U.K.,
1993) nutrient analysis software. Food composition in this database derives from
the Royal Society of Chemistry Nutrient Databank (1992) and McCance and
Widdowson’s The Composition of Foods (Holland et al., 1994), supplemented by
manufacturers’ information. Total nutrient intakes derived from the diaries were
then averaged to give daily intakes for further analyses.
Fruit, vegetables and confectionery intakes were derived from lists of all foods
consumed by each child, as produced by Microdiet (a total of 500 food items were
eaten by 80 children over 3 days). These food lists were converted into an appropriate
data ﬁle—of average daily intake of each food item by each subject—essentially by
processing with WordPerfect (v. 5.1), Borland Paradox (v. 4.0) and SPSS/PC+(v.
5.0) or SPSS for Windows (v. 5.0). Four vegetable-based dishes were excluded
from vegetable intake data since they contained other ingredients, especially fat.
Confectionery intakes were summed from sweets and sweet biscuits, including
E. L. GIBSON ET AL.
Mothers. Three aspects of nutritional knowledge were assessed by questionnaire:
knowledge of nutrient content of common foods, knowledge of expert re-
commendations for changes in consumption of certain food groups and nutrients,
and a measure of the ability to apply both types of knowledge to practical dietary
choices (‘‘practical knowledge’’).
(1) Knowledge of nutrient content. A list of 19 common foods drawn from all the
major food groups was presented to subjects, alongside ﬁve columns headed ‘‘fat’’,
‘‘ﬁbre (roughage)’’, ‘‘protein’’, ‘‘sugar’’ and ‘‘starches (complex carbohydrates)’’.
Subjects were instructed to indicate which foods they thought were high in which
components. The 19 foods were: wholemeal bread/chapatti/roti; white ﬁsh (e.g. cod,
plaice)—grilled; oily ﬁsh (e.g. sardines, tuna tinned in brine); hard cheese (e.g.
cheddar); whole milk; skimmed milk; potatoes (boiled); red meat (e.g. lamb, beef,
bacon); poultry (e.g. chicken without skin); sunﬂower margarine; butter; pulses (peas,
beans, lentils); jam; chocolate; white rice (boiled); plain sweet biscuits (e.g. digestive,
rich tea); bananas; apples; bran ﬂakes (without milk).
On another page, subjects were asked in a similar manner to indicate whether
each of eight foods was high in iron, calcium, salt or vitamin C. The eight foods
were: oily ﬁsh (e.g. sardines, tuna tinned in brine); hard cheese (e.g. cheddar); whole
milk; skimmed milk; sunﬂower margarine; oranges; red meat (e.g. beef, lamb)—roast;
For scoring, lists of foods were selected that were high in each nutrient, together
with an equal number of foods that were not high in that nutrient. Foods were
deemed to be high in the nutrient if selected as such by at least 26 out of 33 qualiﬁed
dietitians (p<0·01 for v
), who had completed this questionnaire previously. For
instance, this resulted in selection of high-fat foods which were at least 40% energy
as fat, while the selected low-fat foods were less than 9% energy as fat. Subjects
correctly selecting those foods scored 1 for each food, while incorrect selection of
foods not high in the nutrient scored −1. This method prevented artiﬁcially high
scores due to blanket responding, such that random responding would tend to sum
to zero. Thus, summed scores were obtained for each nutrient, and these nine scores
in turn were summed to give a single score for total knowledge of nutrient content
(Cronbach’s alpha for internal consistency=0·69).
(2) Recommended changes in consumption. Subjects were asked to indicate whether
“health experts recommend that adults in this country should eat more, the same
amount, or less” of 16 foods, food groups or nutrients. The correct recommendations
were assessed by administering the same questionnaire to 33 qualiﬁed dietitians, and
validated as above. The 11 items included were: whole milk, wholemeal bread,
potatoes, fruit, green vegetables, ﬁsh, saturated fats, added sugar, starches, ﬁbre and
salt. The expert recommendation was to increase consumption of these items except
whole milk, salt, added sugar and saturated fats, for which the recommendation was
to eat less. Correct responses were scored as 1, incorrect as 0 and responses to the
above items were summed to provide a score for ‘‘knowledge of recommended
changes’’ (Cronbach’s alpha=0·61).
An additional item asked subjects to give the number of servings of fruit and
vegetables that health experts recommend we should eat per day (counting an average
piece of fruit as one serving).
FRUIT AND VEGETABLE INTAKE BY CHILDREN
(3) Practical nutritional knowledge. Subjects completed eight multiple-choice items
that addressed ways in which fat or ﬁbre content of the diet could be altered by
choosing particular foods or dishes. For example, one question asked ‘‘What kind
of sandwich do you think is healthier: two thick slices of bread with a thin slice of
cheddar cheese ﬁlling, or two thin slices of bread with a thick slice of cheddar cheese
ﬁlling?’’. Topics addressed by other questions included: healthiness of servings of
pasta vs. meat sauce; saturated fat in various meats; fat in diﬀerently cut chips
(french fries); saturated fat in cooking fat or oils; ﬁbre in cereals; suitable low-fat,
high-ﬁbre snacks or light meals. Correct responses were scored as 1, incorrect as 0
and responses to the eight items were summed to provide a score for ‘‘practical
knowledge’’ (Cronbach’s alpha=0·65).
Finally, the scores from the above three forms of nutritional knowledge were
summed to give an overall measure of nutritional knowledge that reﬂected knowledge
of nutrient content of foods, as well as knowledge of dietary recommendations and
their application. Thus, this summed nutritional knowledge score was used in
subsequent analyses as a potential predictor of dietary quality.
Children. Children’s knowledge of nutrient content of common foods was assessed
in a similar manner to that of their mothers, but using only foods likely to be
familiar to most children from this population. In appropriate columns alongside
15 common foods, children were asked to ‘‘tick which of the following describes
each food’’: has lots of sugar; has lots of fat; has lots of ﬁbre. The 15 foods listed
were: whole milk, skimmed milk, apples, chocolate bar, baked beans, cheese (e.g.
cheddar), broccoli, ice cream, baked potatoes, margarine, Branﬂakes
beefburgers, brown bread, chips, carrots. Responses were scored using the same
system as for the mothers’ knowledge of nutrient content, giving separate scores for
knowledge of sugar, fat and ﬁbre content. These scores were then summed to provide
an overall measure of knowledge of nutrient content for the children.
Beliefs and attitudes concerning diet
Factors inﬂuencing food choice. Mothers rated the extent to which the following
factors played a part in their decisions about what to eat themselves, and, separately,
about what to give their children to eat: prevention of disease, general health and
well-being, taste, cost, speed and convenience, and weight control. Each ﬁve-point
scale was anchored by ‘‘none at all’’ (=1) and ‘‘a great deal’’ (=5).
Children were asked to rate the importance of the following factors in deciding
what to eat: healthiness of the food, taste of the food, food that is quick and easy
to eat and how much they liked to eat the same as other people. They used the
following scale (scored from 1 to 5): not at all important/a bit >/quite >/very >/
Beliefs and attitudes about diet–disease relationships. Beliefs were assessed by mothers
rating, from ‘‘strongly agree’’ to ‘‘strongly disagree’’, separate disease categories, i.e.
heart disease, cancer, digestive problems, overweight, tooth decay and acne, in
response to the statement: ‘‘By eating more fruit and vegetables, I think that people
can reduce their chances of getting:’’.
Attitudes have been considered to be better predictors of behaviour, or behavioural
intention than are beliefs, because attitudinal measures take into account the perceived
E. L. GIBSON ET AL.
importance of a belief for a given individual (Shepherd & Towler, 1992). Furthermore,
the attitudes of interest were those likely to inﬂuence children’s diets, and so these
variables included items addressing mothers’ concerns about diet and their children’s
health. Thus, attitudinal variables were constructed from the product of the individual
diet–disease belief score (as above), the rated worry about their children getting
heart disease or cancer (1–5, ‘‘never’’ to ‘‘often’’), and the rated perceived personal
risk of their children developing heart disease or cancer (‘‘Compared with children
of his/her own age and sex, I think my child’s chances of developing heart disease
or cancer at some time in his/her life are:’’, 1–5, ‘‘much below average’’ to ‘‘much
above average’’). In this way, measures of attitudes to diet–disease relationships were
derived separately for attitudes to ‘‘more fruit and vegetables’’, and prevention of
‘‘heart disease’’ or ‘‘cancer’’.
Rated liking for, and healthiness of, fruit, vegetables and confectionery
Preferences for fruits, vegetables and confectionery were assessed by questionnaire.
Thus, both mothers and children were asked to rate how much they liked each of
30 commonly consumed foods, from −2=‘‘hate it’’ to +2=‘‘love it’’ (i.e. by circling
their choice on an appropriately labelled ﬁve-point numerical scale). From these
ratings, scores of overall preference were obtained by averaging for groups of foods
in the above categories, as follows: apples, oranges and bananas for ‘‘liking for
fruit’’, tomatoes, carrots and peas for ‘‘liking for vegetables’’; chocolate and sweet
biscuits for ‘‘liking for confectionery’’.
Similarly, mothers and children were asked to indicate how healthy they believed
the foods to be, by rating each one, again on a ﬁve-point scale, between −2=‘‘very
bad for you’’ and +2=‘‘very good for you’’. As above, average ratings were obtained
separately for these commonly eaten fruit, vegetables and confectionery.
Relationships of measures of nutritional knowledge, beliefs and attitudes to diet,
and between mothers and children, were assessed by bivariate Spearman’s rank
correlations. This non-parametric statistic was used because a large number of the
variables showed skewed distributions. Paired comparisons between variable means
were made by related or independent t-tests, with data transformed for normality
where appropriate (see below).
To determine the strength of individual predictors of children’s fruit, vegetable
and confectionery consumption in relation to the inﬂuence of other signiﬁcant
predictors, those variables having univariate correlations with intake of these food
types (Table 5) were entered into step-wise multiple regression analyses. To improve
the distribution of residuals, the dependent variable in each analysis (children’s fruit,
vegetable or confectionery intake) was transformed by natural logarithm after
addition of a constant to remove zero scores. Data were only included in these
univariate and multivariate analyses involving children’s dietary data from the 71
children who completed diaries but did not report being on restricted, i.e. ‘‘special’’
or vegetarian, diets.
Two-tailed levels of signiﬁcance (alpha=0·05) are given unless stated when
speciﬁed for unidirectional hypotheses.
FRUIT AND VEGETABLE INTAKE BY CHILDREN
Distribution of demographic characteristics of mothers (N=89
Characteristic Category N(%) Diary N(%)
Ethnicity White 63 (71) 59 (74)
Afrocaribbean 16 (18) 12 (15)
Others 10 (11) 9 (11)
Marital status Married or living as married 68 (76) 63 (79)
Was married 16 (18) 14 (17)
Never married 5 (6) 3 (4)
Number of children One 9 (10) 8 (10)
Two 39 (44) 34 (42)
Three 22 (25) 19 (24)
Four 12 (13) 12 (15)
Five–seven 7 (8) 7 (9)
Socio-economic group Non-manual (A, B, C1) 51 (57) 45 (56)
Manual (C2, D, E) 32 (36) 30 (38)
Unclassiﬁed 6 (7) 5 (6)
House owners 55 (62) 52 (65)
Use of a car 74 (83) 69 (86)
Deprivation index Low 45 (51) 43 (54)
Moderate 23 (26) 20 (25)
High 10 (11) 10 (12)
Very high 11 (12) 7 (9)
Education Secondary school, no qualiﬁcation 22 (25) 19 (24)
(achieved level) GCSE or equivalent 34 (38) 29 (36)
A Level, HND or equivalent 11 (12) 11 (14)
Degree or professional qualiﬁcation 22 (25) 21 (26)
Demographic data were not available for three mothers.
Of 167 suitable families contacted, 41 (25%) refused to participate, 34 (20%)
initially agreed but failed to attend the interview and the remaining 92 (55%)
completed the study. Letters were sent to a further 116 families initially identiﬁed as
appropriate, but these turned out to be unsuitable (child of wrong age, unable to
speak English, or no longer living with mother, N=30) or uncontactable due to
inaccurate addresses or telephone numbers (N=86).
Distributions for demographic characteristics are presented in Table 1. There
were 48 girls and 44 boys. All mothers and children were English speakers. Mean±SD
age of the mothers was 39·2±4·8 years. Mean±SD body mass index (BMI) (weight
in kg/height in m
) was 25·3±5·8; 37% of mothers had a BMI greater than 25,
compared to 57% of 25–54 year-old women in the national population (Oﬃce of
Population Censuses and Surveys, 1994). The mean age of the children was 10·7±0·9
years. Average age and weight of the children did not diﬀer between sexes, and mean
BMI (mean±SD=17·9±2·9) was the same as U.K. national data for this age group
(Department of Health, 1989).
E. L. GIBSON ET AL.
Children’s average daily nutrient intake from 3-day dietary
Nutrient Girls (N=42) Boys (N=38)
Total fat (%)
Saturated fat (%) 13·5±2·8 12·5±2·6
P:S ratio 0·48±0·21 0·55±0·29
Protein (%) 12·9±2·4 13·1±2·6
Carbohydrate (%) 50·0±4·6 50·1±6·2
Sugar (%) 20·9±6·2 21·8±7·7
Fibre (g/day) 11·3±3·7 12·6±4·5
Total energy (MJ/day) 8·5±1·0 9·6±2·0
(%)=Percentage contribution to total energy.
Sex diﬀerence: p<0·005.
Only nine mothers reported being in poor (or very poor) health over the last
year, and only four children were reported as being in poor health over the same
time. One in four mothers was a current regular smoker, which is similar to national
ﬁgures for this region of the U.K. (Department of Health, 1994), and 38% reported
never smoking, the remainder being occasional or ex-smokers. The percentages of
mothers reporting drinking no alcohol (17%) or greater than 14 units (3%) were
respectively higher and lower than U.K. national ﬁgures (12 and 11%; Department
of Health, 1994). Thirty-six percent of mothers reported taking exercise for 20 min
or more at least once a week.
Children’s dietary diaries
Children’s average daily macronutrient intakes calculated from the diary records
are summarized in Table 2. Macronutrient intakes were typical of the population:
total and saturated fat intakes were slightly greater than recommended by current
guidelines (Department of Health, 1994), while total carbohydrated intake ap-
proximated the dietary reference value (DRV) for this population. Dietary ﬁbre
intake (Englyst method; non-starch polysaccharides) was likewise typical of the
population, but about 15–25% lower than the minimum recommended, and intake of
sugars was approximately double the recommended maximum (overall mean±SD=
123·4±47·7 g/day) (Department of Health, 1991). Boys consumed signiﬁcantly more
total energy than girls [Table 2: F(1,78)=9·21, p<0·005], probably reﬂecting greater
energy demands, and average total energy intake was about 10% greater than the
current DRV (Department of Health, 1991). There were no sex diﬀerences in any
macronutrient intakes as a proportion of total energy, and so sex was not included
in any further analyses.
FRUIT AND VEGETABLE INTAKE BY CHILDREN
Children’s average daily fruit and
Food type Value g/day
Fruit 97·7±93·1 (66·7)
Fruit juice 136·9±178·3 (83·3)
Vegetables 97·6±63·8 (74·2)
Mean±SD (median); See Results text
for children’s intake in equivalent servings
per day, and mothers’ reported frequency of
fruit and vegetable servings per day.
These averages were not appreciably
altered by excluding children on restricted
Fruit and vegetable intakes of the children (Table 3) were widely distributed, and
for the most part well below the recommended level (Department of Health, 1994;
WHO, 1990). On the basis of one serving of fruit or vegetables being equivalent to
80 g (Cos et al., 1997), children were on average consuming little more than one
serving of fruit, and a similar amount of vegetables, each day. Assuming 200 g of
fruit juice to represent one serving, children consumed about two-thirds of a serving
of juice per day. Thus, it is clear that vegetable intake falls much further below the
suggestion to eat two servings of fruit and four of vegetables each day than does
fruit plus juice, albeit that this suggestion is intended only as one illustration of how
fruit and vegetable intake might be increased. Moreover, the median values represent
less than one serving of fruit or vegetables per day, and less than half a serving of
juice: in fact, only 5% of the children had total daily intakes of fruit and vegetables
in excess of 400 g, the WHO recommended minimum intake for adults (WHO, 1990).
Children’s reported daily average intake of confectionery items varied con-
siderably, ranging from zero to a maximum of 118·7 g/day, with an average of 39·3 g/
Mothers’ frequency of food consumption
On the food frequency questionnaire, mothers reported consuming 2·2 (1·6)
servings (mean±SD) of fruit and 3·5 (1·5) servings of vegetables per day (median=
1·8, 3·2, respectively). Mean summed fruit and vegetable servings eaten per day by
mothers was 5·6 (median=5·0), which approximates to the suggested minimum of
ﬁve servings per day: however, these averages may be considerable overestimates
due to the large number of relevant items in the food frequency questionnaire and
the use of a 1-year recall period (see Discussion). Only 17% of mothers believed the
recommended minimum to be more than four servings per day. Mothers reported
consuming 1·6 (1·9) servings of confectionery per day.
Mothers who reported eating fruit most frequently tended to have children
with high fruit consumption (N=80, r
=0·35, p<0·001, 1-tail), but there was no
relationship between mothers and children either for vegetable consumption (r
−0·02) or confectionery intake (r
E. L. GIBSON ET AL.
Relationships between mothers’ level of deprivation and children’s fruit, vegetable, ﬁbre
and confectionery intake
Moderate to Correlation of
Low deprivation very high intake with
Food (N=43) deprivation (N=37) deprivation
Fruit (g/day) 114·7±96·6 (90) 77·9
±86·1 (63) −0·20
Fruit juice (g/day) 160·1±195·9 (83) 109·9±153·5 (50) −0·18
Vegetables (g/day) 95·9±62·2 (75) 99·6±66·4 (72) 0·02
(g/1000 kcal/day) 5·9±1·7 (5·8) 5·1
±1·5 (4·9) −0·26
Confectionery (g/day) 40·3±27·4 (39) 38·0±26·4 (33) 0·04
Energy intake (MJ/day) 8·8±1·3 (8·7) 9·2±1·8 (8·9) 0·10
Deprivation index ranging from 0=Low to 3=Very high.
p<0·05, 1-tail, t(78)=1·79, for fruit intake, low vs. moderate to very high deprivation; p<0·05,
1-tail, for correlation with deprivation index.
p<0·05, 2-tail, t(78)=2·11, for ﬁbre intake, low vs. moderate to very high deprivation.
Socio-economic Status, Education and Diet
Children’s macronutrient intakes did not signiﬁcantly diﬀer between manual and
non-manual groups, nor were there any diﬀerences for children’s consumption of
fruit, vegetables or confectionery (data not shown). There was, however, a tendency
for children with mothers in the non-manual group to consume more fruit juice than
those in the manual group (mean±SD fruit juice per day: manual=95·6±177·6 g;
non-manual=175·2±178·3 g; t=1·90, p=0·03, 1-tail). In addition, mothers with
higher scores on the deprivation index had children who ate less fruit (Table 4: r
−0·20, p<0·05, 1-tail), and the correlation was larger when children on restricted
diets were excluded (Table 5). Children’s ﬁbre intake, as a proportion of total energy
consumed, was signiﬁcantly less for those in more deprived families (Table 4: r
−0·26, p<0·05). Children of less deprived mothers tended to eat a greater ratio of
polyunsaturated to saturated fat (N=80, r
=−0·19, p<0·05, 1-tail), but this cor-
relation was no longer signiﬁcant when children on restricted diets were excluded
=−0·18, p=0·13). Children’s intakes of fruit juice, vegetables and con-
fectionery were not signiﬁcantly related to their mothers’ levels of economic dep-
rivation (Table 4).
By contrast, both fruit and fruit juice consumption by children were positively
related to mothers’ achieved level of education (N=80, r
=0·31, p<0·01, r
p<0·05, respectively). Fibre intake, as a proportion of energy intake, was also greater
in more educated mothers (r
=0·19, p<0·05, 1-tail), but this relationship was not
signiﬁcant when children on restricted diets were excluded (r
=0·17, NS). Neither
vegetable nor confectionery consumption by children was related to their mothers’
=−0·04, −0·07, respectively, NS).
Mothers’ and children’s knowledge of speciﬁc nutrient content were positively
correlated: thus, scores for knowledge of fat and sugar content were signiﬁcantly
FRUIT AND VEGETABLE INTAKE BY CHILDREN
Univariate correlations (Spearman’s r
) for potential predictor variables and children’s
fruit, vegetable and confectionery consumption
Fruit Vegetable Confectionery
Variable intake intake intake
Mother’s consumption frequency for food 0·36
Mother’s liking for food type 0·04 0·14 0·28
Mother’s nutritional knowledge 0·40
Deprivation index −0·24
Mother’s educational level 0·28
Mother’s concern for disease prevention in 0·24
choosing child’s diet
Mother’s attitude to fruit, vegetables and 0·29
child’s cancer risk
Mother’s concern for child’s taste preferences −0·15 −0·17 0·17
Child’s concern for health in choosing what −0·05 0·06 −0·24
Child’s concern for taste in choosing what to 0·24
Child’s nutritional knowledge 0·09 0·01 0·11
Child’s liking for food type 0·03 0·36
Signiﬁcant correlations are highlighted in bold typeface.
=0·21, p<0·025; r
=0·26, p<0·01, 1-tail); the correlation for knowledge
of ﬁbre content was positive but weak (r
=0·16, p=0·065, 1-tail). Mother’s summed
nutritional knowledge (mean±SD=40·2±7·4) correlated signiﬁcantly with their
children’s summed knowledge of nutrient content (mean±SD=8·9±2·4; r
Nutritional Knowledge and Diet
Nutrient intake from diaries
Children’s summed knowledge of nutrient content of common foods was not
signiﬁcantly related to their intakes of any macronutrient. When diary data from all
children were included, mothers’ nutritional knowledge was negatively related to
children’s total energy intakes (r
=−0·25, p<0·05) and positively to their weight of
ﬁbre intake relative to energy consumed (r
=0·35, p<0·01). However, when the nine
children on restricted diets were excluded, only ﬁbre intake remained related to
nutritional knowledge (r
=0·29, p<0·05). Mothers’ nutritional knowledge was not
signiﬁcantly related to children’s intake of any other macronutrient.
Fruit, vegetable and confectionery consumption
Mothers’ nutritional knowledge was positively and quite strongly related to
children’s fruit consumption (N=80, r
=0·40, p<0·001), but by contrast, knowledge
showed no relationship to intake of fruit juice (r
=−0·01), nor to vegetable con-
=0·03). Children’s summed nutrient knowledge was not related to fruit
=0·06), fruit juice (r
=−0·17) or vegetable (r
E. L. GIBSON ET AL.
Neither mothers’ nor children’s nutritional knowledge scores were related to
children’s consumption of confectionery (r
=0·01, 0·06, respectively).
Beliefs and Attitudes Inﬂuencing Food Choice
Importance of health, taste, cost, convenience and weight loss
Mothers considered taste to be of greatest importance in choosing what to eat
for themselves (N=92, mean±SD=4·53±0·83), followed by health and well-being
(mean±SD=3·90±1·04), and then disease prevention (mean±SD=3·51±1·17)
and cost (mean±SD=3·38±1·25). In choosing food for their children, mothers
considered health to be most important (mean±SD=4·37±0·83), followed by taste
(mean±SD=4·22±0·85) and then prevention of disease (mean±SD=4·01±1·18).
The rank order of importance of all the rated factors (by mean) for mothers was
very similar when rating for themselves and their children (r
=0·96, p<0·001), except
that in the latter case health was more important than taste. All of the factors
mentioned above (taste, health and well-being, prevention of disease) were on average
rated as signiﬁcantly greater than the mid-point for the scale (i.e. 3·0; Wilcoxon’s
matched-pairs test, p<0·01, at least), while issues of convenience, weight loss and
cost in choosing children’s foods, were not (data not shown: cost was rated as greater
than 3·0 for mothers choosing their own foods).
As with mothers, children considered taste to be the most important factor
(mean±SD=3·64±1·03), with healthiness second (mean±SD=3·30±1·15), in
choosing food for themselves. These factors were rated signiﬁcantly greater than the
mid-point (i.e. 3·0; Wilcoxon’s test, p<0·0001, 0·02, respectively), whereas food that
is ‘‘quick and easy to eat’’ (mean±SD=2·26±1·14) and eating the same as others
(mean±SD=1·56±0·75) were both rated as less important than the mid-point
The following relationships between beliefs and children’s diet concern mothers’
ratings of importance of various factors in choosing food for their children, or
children’s own ratings of important factors in choosing what to eat. Only those
factors rated signiﬁcantly above the mid-point (see above) were considered further
in relation to the children’s diets. Mothers’ ratings of the importance of ‘‘general
health and well-being’’ and ‘‘prevention of disease’’ were signiﬁcantly correlated (N=
=0·57, p<0·001) but the latter measure provided greater variance and so is
used in subsequent analyses as a possible predictor of variation in the children’s diet.
Mothers with low ratings of importance of disease prevention in choosing foods for
their children had children who ate a higher percentage of fat (r
and less fruit or fruit juice (r
=0·24, p<0·05 and r
=0·38, p<0·001, respectively), but
surprisingly these mothers tended to have children who ate more vegetables (r
−0·26, p<0·05) (note that potatoes, fried or otherwise, are not included as vegetables
here). No other nutrient intake was related to this rating.
Children who considered health to be important for the foods they choose ate
less confectionery (r
=−0·24, p<0·05), but otherwise this rating was not related to
their consumption of any nutrient, or fruit and vegetables: no other rating by children
was related to their confectionery intake. However, children who rated taste (their
primary concern, on average) as important ate more fruit (r
=0·24, p<0·05) but less
FRUIT AND VEGETABLE INTAKE BY CHILDREN
–1 0 1
2–2 –1 0 1
F 1. Mean (+SEM) ratings of liking (a) and healthiness (b) by mothers and
children, for commonly eaten fruits, vegetables and confectionery (N=90–92).
Signiﬁcant diﬀerences between mothers and children within a food type:
p<0·001. Φ, mothers; ∆, children.
Ratings of liking and healthiness for fruit, vegetables and confectionery
Mothers’ and children’s average ratings of liking for commonly eaten fruits,
vegetables and confectionery are presented in Fig. 1(a). In general, mothers appeared
moderately to like all three food types, with fruit preferred the most, then vegetables,
then confectionery. Children’s liking for common fruit did not diﬀer from that of
their mothers [t(91)=1·06, NS]. However, children liked vegetables signiﬁcantly less
than did their mothers, t(90)=6·74, p<0·0001, whereas confectionery was liked more
by children than by their mothers, t(90)=3·00, p<0·005.
Children’s, but not mothers’, rated likings for vegetables were signiﬁcantly related
to their vegetable intake (r
=0·33, p<0·005; r
=0·17, NS, respectively). Conversely,
mothers’ but not children’s liking for confectionery was correlated to children’s
confectionery intake (r
=0·25, p<0·05; r
=0·13, NS, respectively). Children’s fruit
intake was not related to either their own or their mothers’ rated liking for fruit
=0·02 and 0·01, respectively). Excluding children on restricted diets did not
substantially alter these correlations (Table 5), except that children’s liking for
confectionery was then marginally signiﬁcantly related to their confectionery intake.
In contrast to ratings of liking, mothers and children did not diﬀer in their ratings
of healthiness of fruits, vegetables or confectionery [Fig. 1(b)]: on average, fruit was
rated as very healthy and slightly more so than vegetables, while confectionery was
rated as rather unhealthy. Mothers’ and children’s ratings of healthiness did not
correlate with children’s intakes of the respective food types, with the exception of
children’s ratings of healthiness of vegetables, which were negatively and signiﬁcantly
related to their intake of vegetables (r
=−0·27, p<0·025; excluding ‘‘restricted’’
=−0·24, p<0·05), i.e. children who gave vegetables the highest healthiness
ratings tended to eat the least vegetables.
Beliefs and attitudes about diet–disease relationships
There was very good agreement within this population that eating more fruits
and vegetables could reduce people’s chances of becoming overweight (Table 6):
since almost 90% of the sample agreed or strongly agreed with this notion, this belief
cannot sensitively discriminate dietary quality in their children. Almost three-quarters
E. L. GIBSON ET AL.
Distribution of mother’s beliefs in the inﬂuence of
fruit and vegetable consumption on the incidence of
Percentage agree or
Disease strongly agree
Heart disease 65·3
Digestive problems 73·7
Tooth decay 56·6
Percentage agreeing or strongly agreeing to the
statement ‘‘By eating more fruit and vegetables, I think that
people can reduce their chances of getting:’’, for various
Poor distribution, and so little predictive value.
of mothers believed that digestive diseases could be reduced by eating more fruits
and vegetables, and so this belief was also unlikely to predict variation in their
children’s diet. However, there was less consensus about a role for fruit and vegetables
in preventing cancer, heart disease, tooth decay and acne (Table 6), and so these
beliefs could have more predictive value. Even so, no signiﬁcant correlations were
found for these beliefs and children’s consumption of fruit, fruit juice, vegetables,
total fruit and vegetables, or confectionery.
By contrast to measure of beliefs per se, when more speciﬁc attitudinal variables
were used, i.e. those expressing mother’s concern for their own child’s risk of
developing cancer or heart disease, and the inﬂuence of fruit and vegetables on that
risk (see Methods), a signiﬁcant relationship was found: thus, mothers who were
concerned that more fruit and vegetables could reduce risk of cancer in their children
had children who tended to eat more fruit (r
=0·29, p<0·02). However, neither
vegetable nor confectionery consumption was related to the mothers’ attitudinal
Multivariate Inﬂuences on Children’s Diet
Twelve variables were selected for potential multivariate analysis that were
considered particularly plausible predictors of children’s dietary quality, representing
inﬂuences from mothers’ economic and educational status, mothers’ and children’s
nutritional knowledge, dietary beliefs and preferences, diet–disease attitudes and
mothers’ habitual consumption. Univariate correlations between each of these vari-
ables and children’s consumption of fruit, vegetables and confectionery are sum-
marized in Table 5. The relative contribution of each signiﬁcantly correlated variable
in the context of other predictor variables was assessed by step-wise multiple
regression, as summarized in Table 7.
FRUIT AND VEGETABLE INTAKE BY CHILDREN
Summary of stepwise multiple regression analysis of univariate correlates of children’s
fruit, vegetable or confectionery consumption
of fruit intake
BSE B Beta tp(t)
Mother’s nutritional 0·022 0·006 0·37 3·68 0·0005
Mother’s fruit consumption 0·077 0·026 0·30 2·94 0·0046
Mother’s attitude to fruit, veg 0·007 0·002 0·27 2·71 0·0085
and child’s cancer risk
=0·37 Adjusted r
=0·34 F(3,66)=13·10, p<0·0001
of vegetable intake
BSE B Beta tp(t)
Child’s liking for common 0·078 0·024 0·36 3·25 0·0018
Mother’s concern for disease −0·156 0·063 −0·27 −2·47 0·0160
prevention in choosing child’s
=0·20 Adjusted r
=0·17 F(2,66)=8·15, p<0·001
of confectionery intake
BSE B Beta tp(t)
Mother’s liking for 0·066 0·023 0·32 2·82 0·0063
Child’s concern for health in −0·040 0·017 −0·26 −2·28 0·0257
choosing what to eat
=0·16 Adjusted r
=0·14 F(2,67)=6·41, p<0·005
Only those variables for which p(t)<0·05 are included.
Dependent variables are Log
Variables not signiﬁcantly contributing to the regression: mother’s educational level, mother’s
concern for disease prevention in choosing child’s diet, child’s concern for taste in choosing what to
eat, deprivation index.
No variables excluded from regression equation.
Variables not signiﬁcantly contributing to the regression: child’s liking for confectionery.
Predictors of fruit and vegetable consumption
Univariate correlations between 12 possible inﬂuences on children’s diets revealed
six signiﬁcant relationships for fruit, and four for vegetable intake (Table 5). It can
be seen that consumption of fruit was, by and large, inﬂuenced by a diﬀerent set of
factors from intake of vegetables. This diﬀerence was conﬁrmed by multiple regression
analysis (Table 7): thus, of the variables entered, most variance in children’s fruit
intake was explained by mothers’ nutritional knowledge, followed by mothers’
reported frequency of fruit consumption. Nevertheless, an independent positive
inﬂuence was present for mothers’ attitude that more fruit and vegetables may reduce
their children’s risk of cancer. These three variables accounted for 37% of the variance
in children’s fruit intake, while, in their presence, mothers’ concern for disease
E. L. GIBSON ET AL.
prevention in choosing foods for their children, mothers’ educational and social
deprivation levels, and children’s concern for taste, did not signiﬁcantly contribute
to the variation in fruit intake.
By contrast, variation in children’s vegetable consumption was best explained by
their own rated liking for common vegetables, followed by mothers’ concern for
disease prevention when choosing foods for their children (Table 7), with this latter
variable having a negative inﬂuence on vegetable intake. These two variables were
the only ones from the 12 investigated in this analysis that were signiﬁcantly related
to children’s vegetable intake, and only 20% of the variance was explained.
Predictors of confectionery consumption
Two predictor variables univariately correlated to children’s confectionery con-
sumption, out of the same 12 potential predictor variables as used with fruit and
vegetable intakes (Table 5). Thus, mothers’ liking for common confectionery was
positively correlated to children’s intake of confectionery, while children’s concern
for health when choosing what to eat was negatively related to the amount of
confectionery they ate. Both of these univariate predictor variables independently
contributed to the variance in confectionery consumption, explaining 16% of the
variance (Table 7), while, in the presence of those variables, children’s liking for
confectionery made no signiﬁcant contribution. By comparison, none of these
variables signiﬁcantly explained any variation in children’s fruit or vegetable con-
sumption, emphasizing the unique inﬂuences on intake of these diﬀerent food groups.
The distribution of children’s macronutrient intakes determined from their dietary
diaries agrees well with recent nutritional surveys of children of similar age from
this region of the U.K. (Department of Health, 1989; Doyle et al., 1994), supporting
the validity of this dietary record, although average energy intake here was 5–10%
greater than in the other studies. As is typical for this population (Department of
Health, 1989), the majority of children exceeded the recommended intake of saturated
fat and sugar, but ate less than the recommended daily intake of ﬁbre (Department
of Health, 1991).
However, the main intention of this study has been to determine the relative
inﬂuences on fruit and vegetable consumption by children of a number of factors
believed to inﬂuence food choice and dietary quality in mothers and children. The
emphasis is thus on intake of fruit and vegetables rather than nutrients: for instance,
ﬁbre is strongly correlated to fruit and vegetable intake, but it is unlikely that
children choose to eat these foods as a source of ﬁbre per se. On average, the children
ate no more than half the suggested target intakes of fruit and vegetables (even less
by median), and consumption of vegetables in particular will have to be substantially
raised to meet the target, similar to the situation in the U.S. (Basch et al., 1994).
Although the food frequency questionnaire results indicated that half the mothers
achieved at least ﬁve servings of fruits and vegetables per day, this is likely to be an
overestimation (Cox et al., 1997): thus, when analyses of a similar questionnaire
were compared to 16-day weighed food records, fruit and vegetable consumption
was found to be overestimated by about 30%, with vegetable intake in particular
essentially doubled by the frequency questionnaire method (Bingham et al., 1994).
FRUIT AND VEGETABLE INTAKE BY CHILDREN
Therefore, a more accurate estimation would be an average of about three servings
of fruit and vegetables per day, i.e. well below the current recommended minimum.
Socio-economic status had little eﬀect on the distribution of fruit or vegetable
intake here, although fruit and ﬁbre intake was slightly lower in children from more
deprived families. It is possible that underrepresentation of more deprived families
here minimized the apparent impact of socio-economic status on the children’s fruit
and vegetable consumption; however, U.K. national ﬁgures for children of this age
group do not suggest more than a weak eﬀect of social class for intake of common
fruits and vegetables, with the exception of salad vegetables (including tomatoes)
and fruit juice (Department of Health, 1989). Similarly, a study of the diets of second
and ﬁfth grade school children (average ages, 8 and 11 years, respectively) in New
York State (Wolfe & Campbell, 1993) found that deprivation level was related to
lack of dietary diversity but not intake of fruits and vegetables. The lack of a strong
relationship to socio-economic status contrasts with that found for adults in the
U.K. (Department of Health, 1994; Scottish Oﬃce, 1993), and may reﬂect the
generally lower intake of fruit and vegetables by children of this age, irrespective of
family economic status. By comparison, mothers’ education was clearly related to
children’s fruit intake here, although not to their vegetable consumption. This
compares well with a recent study of 9- to 15-year-old Finnish children (Laitinen et
al., 1995), whose intake of fruit but not vegetables was greater for those with more
educated parents, while the fathers’ occupational strata had no signiﬁcant impact
on children’s consumption of either fruit or vegetables.
It is likely that greater education was functionally related to higher fruit con-
sumption through better nutritional knowledge, as well as positive attitudes towards
fruit and disease prevention. Thus, mothers’ nutritional knowledge was quite strongly
correlated to their children’s fruit intake, but not to intake of vegetables or con-
fectionery. Moreover, children’s level of nutritional knowledge was related to that
of their mothers, which is consistent with evidence of mother–child transmission of
nutritional awareness even in pre-school children (Anliker et al., 1990). However,
children’s nutritional knowledge did not correlate with fruit or vegetable intake. This
diﬀerence could be attributable to the diﬀering nature of the knowledge scores, i.e.
the children’s score was derived solely from knowledge of nutrient content of common
foods, whereas the mother’s score also incorporated assessment of the understanding
of recommendations and their application. However, it might also indicate a genuine
independence of children’s fruit and vegetable intake, and indeed confectionery
intake, from their nutritional knowledge.
Consideration of mothers’ and children’s beliefs about inﬂuences on their food
choice, and about fruit, vegetables and confectionery, provides a revealing pattern
of relationships. Clearly, issues of health and taste are conceived of as most inﬂuential
by mothers and children, as motives for food choice in general. However, while
mothers and children certainly both reported believing that fruit and vegetables are
healthy and confectionery is not, children reported liking confectionery more, and
vegetables less, than did mothers, with no diﬀerence in liking for fruit. Other studies
support a clear preference for fruit over vegetables in children of this age (Basch et
al., 1994; Domel et al., 1993).
However, it is apparent that fruit and vegetable intake was not simply motivated
in children by concern either for their healthiness or liking for their taste, as had
been inferred from categorization of motives to consume a range of foods (Michela
& Contento, 1986). The certainty with which both mothers and children rated fruit
E. L. GIBSON ET AL.
and vegetables as being ‘‘good for you’’ is no guarantee of their consumption by
children, and this contradiction of attitude and action has been observed elsewhere
in mothers and other adults (Fischler, 1986). Paradoxically here, both children’s
conviction of the healthiness of vegetables and mothers’ concern for disease prevention
when choosing their children’s food were associated with less consumption of
vegetables by the children. One possible explanation of these apparent paradoxes is
that children’s initial resistance to eating vegetables (perhaps because of unattractive
sensory properties, low energy density and neophobia) has been countered by mothers
(especially those who are more concerned about disease prevention) with threats of
dire, but distal, health consequences should the children continue to refuse to eat
the vegetables (Casey & Rozin, 1989; Wardle, 1995). There is evidence that such a
strategy will only serve to devalue the food for children (Birch, Marlin & Rotter,
1984), while they remain acutely aware of the declared healthiness of that food, i.e.
the rule of thumb might become ‘‘good (for you) but gross’’ rather than ‘‘naughty
It is clear from the results of the multiple regression analyses that quite separate
factors contributed to the variation in fruit, vegetable and confectionery consumption
by children. This is worth emphasizing, because both research studies (Brug et al.,
1995a, 1995b; Laforge et al., 1994) and interventions (Foerster et al., 1995) often
appear to treat fruit and vegetables as a unitary variable, perhaps because of
nutritional commonalities, and yet there are broad diﬀerences both in sensory
qualities (fruit is more often sweeter and not bitter) and in cultural use (fruit is more
often a snack or dessert, and eaten on its own), which might suggest control by
diﬀerenct factors. Furthermore, there is evidence that children of this age distinguish
between fruits and vegetables in terms of categories and preferences (Domel et al.,
It is encouraging that mothers’ nutritional knowledge is strongly and in-
dependently related to children’s fruit intake, because it suggests that nutritional
education might have a positive impact, irrespective of underlying educational level
or even diet–disease attitudes. A meta-analysis of earlier studies of the impact of
nutritional knowledge on dietary quality (Axelson et al., 1985) did support a positive,
if weak, correlation, at least in adults. However, it is likely that the particular
population and the type of food being studied, as well as the measure of knowledge,
will inﬂuence this relationship. For instance, knowledge of fat content might be
expected to be, and was, negatively related to consumption of fatty meat products
(Shepherd & Towler, 1992). Also, in studies where attitudes have been found to be
better predictors of food choice than was nutritional knowledge (Shepherd & Towler,
1992), the items assessing attitudes were explicitly linked to the food choice behaviour,
whereas those measuring knowledge were not. The only attitude predicting children’s
fruit consumption here was particularly explicitly linked to the beneﬁts of eating
more fruit and vegetables for cancer prevention, but its independent contribution at
least suggests that highlighting diet–disease issues may be beneﬁcial. When measures
of both knowledge and attitudes were similarly unfocused (Frederick & Hawkins,
1992), knowledge correlated more strongly with consumption of the target foods
than did attitudes.
The positive relationship between mothers’ and children’s fruit consumption
found here supports other ﬁndings that availability of and exposure to fruit (and
even vegetables) are important determinants of children’s intake (Birch, 1990; Domel
et al., 1993; Krebs-Smith et al., 1995; Rozin, 1990), although there may be an
FRUIT AND VEGETABLE INTAKE BY CHILDREN
intermediary factor not assessed in this study. Experimental studies in infants and
older children indicate that ‘‘mere’’ exposure to novel foods, including vegetables,
albeit perhaps in the presence of a respected other, can lead to increased acceptance
of those foods (Sullivan & Birch, 1990, 1994). It may be inferred from these ﬁndings
that eﬀective strategies to increase consumption would include innocuous exposure
to the food in a positive social context: just such a mechanism has been invoked to
explain children’s acquisition of acceptance of foods ﬂavoured with hot chilli (Rozin,
1990). However, there is a need for further research into the acquisition of likes and
dislikes for vegetables, especially, by children, which is highlighted here by the
dominance of children’s rated liking for vegetables in predicting their intake, in the
context of hedonic indiﬀerence, on average, even to commonly eaten vegetables.
This occurred despite the use of quite crude and general measures of liking here,
and underlines the need both to recognize the inﬂuence and explore the development
of these aﬀective, or hedonic, reactions to foods (Birch, 1990; Domel et al., 1993;
Rozin, 1990). Furthermore, the health consequences of a child’s food preferences
may extend beyond the particular child and into the entire household, since there is
evidence that parents may adapt their diets deleteriously to accommodate their
children’s likes and dislikes (Laforge et al., 1994; Wardle, 1995).
It was to be expected that children’s confectionery consumption would be related
to diﬀerent variables from intake of fruit or vegetables: confectionery intake will be
reinforced by sensory properties, energy density and intense marketing, none of
which applies to the same extent to fruits and vegetables. Indeed, there is evidence
that a primary barrier to fruit consumption in adults is the belief that ‘‘fruit
does not make a ﬁlling snack’’ (Scottish Oﬃce, 1993). Nevertheless, it is perhaps
encouraging that the two signiﬁcant independent predictors of children’s con-
fectionery intake found here were mothers’ liking for confectionery and children’s
concern for health in choosing what to eat. The former suggests at least that there
is considerable maternal inﬂuence, while the latter indicates one likely improvement
in children’s dietary quality through increasing awareness of diet and health issues.
Finding the means to curb confectionery consumption by children would likely not
only reduce fat and sugar content of their diet, but probably lessen the likely
displacement by confectionery of opportunities to eat more fruit and vegetables: for
instance, there is evidence that children’s intake of fruit juice substituted for sugary
drinks (Basch et al., 1994).
Shortcomings of this study include the dependence of interpretation on the
validity of the various measures employed. Moreover, the relationships between
predictor and dependent variables observed in the regression analyses describe the
situations as measured here (albeit with substantial variance remaining unexplained),
but deﬁnite conclusions about causality are not possible. Selection bias may have
produced a sample of mothers that were more conscious of both their own and their
children’s health, and more interested in nutritional issues, than a random sample
of the regional population. Such a bias may result in a sample having a healthier
diet and a greater understanding of its importance to health: this limits the gen-
eralizability of the ﬁndings, but does not invalidate use of statistics in this non-
experimental design. In addition, this study did not address the possible inﬂuence
of psychosocial measures from fathers (aside from socio-economic group) on their
children’s fruit and vegetable intake. However, prima facie, interventions to increase
fruit and vegetable consumption in children could well employ separate strategies
for these two food types. The results support attention to nutritional education of
E. L. GIBSON ET AL.
parents, including issues of disease, especially cancer prevention by diet, as well as
guidance to parents in strategies likely to encourage fruit and vegetable consumption
in their children.
Anliker, J. A., Laus, M. J., Samonds, K. W. & Beal, V. A. (1990). Parental messages and the
nutrition awareness of preschool children. Journal of Nutrition Education,22, 24–29.
Axelson, M. L., Federline, T. L. & Brinberg, D. (1985). A meta-analysis of food- and nutrition-
related research. Journal of Nutrition Education,17, 51–54.
Basch, C. E., Zybert, P. & Shea, S. (1994). 5-A-DAY: dietary behavior and the fruit and
vegetable intake of Latino children. American Journal of Public Health,84, 814–818.
Berenson, G. S., Srinivasan, S. R., Hunter, S., MacD., Nicklas T. A., Freedman, D. S., Shear,
C. L. & Webber, L. S. (1989). Risk factors in early life as predictors of adult heart disease:
the bogalusa heart study. American Journal of the Medical Sciences,298, 141–151.
Bingham, S. A., Gill, C., Welch, A., Day, K., Cassidy, A., Khaw, K. T., Sneyd, M. J., Key,
T. J., Roe, L. & Day, N. E. (1994). Comparison of dietary assessment methods in
nutritional epidemiology: weighed records v. 24 h recalls, food-frequency questionnaires
and estimated-diet records. British Journal of Nutrition,72, 619–643.
Birch, L. L. (1990). Development of food acceptance patterns. Developmental Psychology,26,
Birch, L. L., Marlin, D. W. & Rotter, J. (1984). Eating as the ‘‘means’’ activity in a contingency:
eﬀects on young children’s food preference. Child Development,55, 431–439.
Block, G., Patterson, B. & Subar, A. F. (1992). Fruit, vegetables, and cancer prevention: a
review of the epidemiological evidence. Nutrition and Cancer,18, 1–29.
Booth, D. A. (1994). Psychology of Nutrition. London: Taylor and Francis.
Brug, J., Debie, S., van Assema, P. & Weijts, W. (1995a). Psychosocial determinants of fruit
and vegetable consumption among adults: results of focus group interviews. Food Quality
and Preference,6, 99–107.
Brug, J., Lechner, L. & De Vries, H. (1995b). Psychosocial determinants of fruit and vegetable
consumption. Appetite,25, 285–296.
Casey, R. & Rozin, P. (1989). Changing children’s food preferences: parent opinions. Appetite,
Cox, D. N., Anderson, A. S., Reynolds, J., McKellar, S., Mela, D. J. & Lean, M. E. J. (1997).
Measuring fruit and vegetable intake: Is ﬁve-a-day enough? European Journal of Clinical
Department of Health. (1989). The Diets of British Schoolchildren. London: HMSO.
Department of Health. (1991). Dietary reference values for food energy and nutrients for the
United Kingdom. London: HMSO.
Department of Health. (1994). Nutritional aspects of cardiovascular disease. London: HMSO.
Domel, S. B., Baranowski, T., Davis, H., Leonard, S. B., Riley, P. & Baranowski, J. (1993).
Measuring fruit and vegetable preferences among 4th- and 5th-grade students. Preventive
Domel, S. B., Baranowski, T., Davis, H. C., Thompson, W. O., Leonard, S. B. & Baranowski,
J. (1995). A measure of outcome expectations for fruit and vegetable consumption among
fourth and ﬁfth grade children: reliability and validity. Health Education Research,10,
Domel, S. B., Baranowski, T., Davis, H. C., Thompson, W. O., Leonard, S. B. & Baranowski,
J. (1996). A measure of stages of change in fruit and vegetable consumption among
fourth- and ﬁfth-grade school children: reliability and validity. Journal of the American
College of Nutrition,15, 56–64.
Doyle, W., Jenkins, S., Crawford, M. A. & Puvandendran, K. (1994). Nutritional status of
schoolchildren in an inner city area. Archives of Disease in Childhood,70, 376–381.
Fischler, C. (1986). Learned versus ‘‘spontaneous’’ dietetics: French mothers’ views of what
children should eat. Social Science Information,25, 945–965.
Foerster, S. B., Kizer, K. W., Disogra, L. K., Bal, D. G., Krieg, B. F. & Binch, K. L. (1995).
California’s ‘‘5 a day—for better health!’’ campaign: an innovative population-based
FRUIT AND VEGETABLE INTAKE BY CHILDREN
eﬀort to eﬀect large-scale dietary change. American Journal of Preventive Medicine,11,
Frederick, L. & Hawkins, S. T. (1992). A comparison of nutrition knowledge and attitudes,
dietary practices, and bone densities of postmenopausal women, female college athletes,
and nonathletic college women. Journal of The American Dietetic Association,92, 299–305.
Holland, B., Welch, A. A., Unwin, I. D., Buss, D. H., Paul, A. A. & Southgate, D. A. T. (1994).
McCance and Widdowson’s The Composition of Foods. 5th ed. Cambridge: Royal Society
of Chemistry and MAFF.
Kelder, S. H., Perry, C. L., Klepp, K. I. & Lytle, L. L. (1994). Longitudinal tracking of
adolescent smoking, physical activity, and food choice behaviors. American Journal of
Public Health,84, 1121–1126.
Key, T. J. A., Thorogood, M., Appleby, P. & Burr, M. L. (1996). Dietary habits and mortality
in 11000 vegetarians and health conscious people: results of a 17 year follow up. British
Medical Journal,313, 775–778.
Kimm, S. Y. S., Gergen, P. J., Malloy, M., Dresser, C. & Carrol, M. (1990). Dietary patterns
of US children: implications for disease prevention. Preventive Medicine,19, 432–442.
Krebs-Smith, S. M., Heimendinger, J., Patterson, B. H., Subar, A. F., Kessler, R. & Pivonka,
E. (1995). Psychosocial factors associated with fruit and vegetable consumption. American
Journal of Health Promotion,10, 98–104.
Krebs-Smith, S. M., Cook, A., Subar, A. F., Cleveland, L., Friday, J. & Kahle, L. L. (1996).
Fruit and vegetable intakes of children and adolescents in the United States. Archives of
Pediatric and Adolescent Medicine,150, 81–86.
Laforge, R. G., Greene, G. W. & Prochaska, J. O. (1994). Psychosocial factors inﬂuencing
low fruit and vegetable consumption. Journal of Behavioural Medicine,17, 361–374.
Laitinen, S., Rasanen, L., Viikari, J. & Akerblom, H. K. (1995). Diet of Finnish children in
relation to the family’s socio-economic status. Scandinavian Journal of Social Medicine,
Michela, J. L. & Contento, I. R. (1986). Cognitive, motivational, social and environmental
inﬂuences on children’s food choices. Health Psychology,5, 209–230.
Negri, E., Vecchia, C., Franceschi, S., D’Avanzo, B. & Parazzini, F. (1991). Vegetable and
fruit consumption and cancer risk. International Journal of Cancer,48, 350–354.
Oﬃce of Population Censuses and Surveys. (1990). Standard occupational classiﬁcation, volume
2. London: HMSO.
Oﬃce of Population Censuses and Surveys. (1994). Health survey for England 1992. London:
Resnicow, K. & Reinhardt, J. (1991). What do children know about fat, ﬁbre, and cholesterol?
A survey of 5,116 primary and secondary school students. Journal of Nutrition Education,
Resnicow, K., DavisHearn, M., Smith, M., Baranowski, J., Doyle, C. & Wang, D. Q. T.
(1997). Social-cognitive predictors of fruit and vegetables intake in children. Health
Rozin, P. (1990). Development in the food domain. Developmental Psychology,26, 555–562.
Scottish Oﬃce (1993). The Scottish Diet. Edinburgh: HMSO.
Shepherd, R. & Towler, G. (1992). Nutrition knowledge, attitudes and fat intake: application
of the Theory of Reasoned Action. Journal of Human Nutrition and Dietetics,5, 387–397.
Sullivan, S. A. & Birch, L. L. (1990). Pass the sugar, pass the salt: experience dictates
preference. Developmental Psychology,26, 546–551.
Sullivan, S. A. & Birch, L. L. (1994). Infant dietary experience and acceptance of solid foods.
The Alpha-Tocopherol, Beta Carotene Cancer Prevention Study Group. (1994). The eﬀect of
vitamin E and beta carotene on the incidence of lung cancer and other cancers in male
smokers. New England Journal of Medicine,330, 1029–1035.
U.S. Department of Health and Human Services (1990). Healthy people 2000: National health
promotion and disease prevention objectives. Washington, DC: DHSS publication PHS 91-
Wardle, J. (1995). Parental inﬂuences on children’s diets. Proceedings of the Nutrition Society,
E. L. GIBSON ET AL.
Wolfe, W. S. & Campbell, C. C. (1993). Food pattern, diet quality, and related characteristics
of schoolchildren in New York State. Journal of the American Dietetic Association,93,
World Health Organisation. (1990). Diet, nutrition, and prevention of chronic diseases. Report
of a WHO study group. Geneva: WHO (Technical Report Series—797).
Received 17 March 1997, revision 28 May 1998