Академический Документы
Профессиональный Документы
Культура Документы
Key words: attitudes, fat intake, food choice, theory of reasoned action.
387
388 R. Shepherd and G. Towler
accounted for in these studies is only about self-reported consumption of foods contri-
3 and I%, respectively. buting highly to fat in the diet (Shepherd &
Where the behaviour of interest is the Stockley, 1985, 1987). Subsequent work
consumption of particular foods or the has extended this to include beliefs and
intake of particular nutrients, the relation- evaluations for four food types (meat, meat
ships are even less clear, with studies products, dairy products and fried foods)
reporting only a small number of statisti- (Towler h Shepherd, 1992) and shown that
cally significant relationships (Eppright et the overall model predicts the intention to
al., 1970; Grotkowski & Sims, 1978) or none consume the foods. Females, higher social-
at all (Werblow et al., 1978; Perron & class subjects and younger subjects had
Endres, 1985). In research where the be- more negative beliefs, attitudes and inten-
haviour of interest is measured as some tions towards consuming the foods. Indivi-
score for general ‘nutritional practices’ (e.g. dual belief-evaluation scores related to
eating three meals a day) rather than food health and taste were the most important
or nutrient intake, there does tend to be a in predicting attitudes.
significant relationship between attitudes Shepherd b Stockley (1987) found that
and behaviour (Jalso et al., 1965; Schwartz, nutrition knowledge did not correlate with
1975; Carruth et ol., 1977; Foley et al., 1983; attitude, intention or behaviour. This was
Douglas & Douglas, 1984). possibly because the nutrition knowledge
In social psychology, similar problems of questionnaire was too short and had not
a lack of a clear attitude-behaviour link been validated. A nutrition knowledge
led to the development of structured atti- questionnaire was subsequently developed
tude and belief models. One such approach and tested for its ability to differentiate
is the theory of reasoned action developed between groups likely to be high and low
by Fishbein & Ajzen (1975; Ajzen & Fish- in nutrition knowledge (Towler & Shep-
bein, 1980), which offers a coherent frame- herd, 1990). This questionnaire was used
work within which to measure and relate in the present study. It includes four
beliefs, attitudes and behaviour. This sections on nutrient density, with the
model has attracted a good deal of atten- section on fat of particular interest here,
tion and experimental validation and has along with a set of multiple-choice ques-
shown good predictive power for a variety tions (see Towler & Shepherd, 1990, for the
of different behaviours (Ajzen & Fishbein, full questionnaire).
1980). In the particular case of food choice, The aims of the present study were to
this model has shown clear relationships test the Fishbein & Ajzen theory of rea-
between food choice and attitudes (Shep- soned action in relation to consumption of
herd & Stockley, 1985, 1987; Shepherd & foods contributing highly to fat in the diet
Farleigh, 1986; Tuorila, 1987). and to determine whether there is an
Within the theory of reasoned action, association with nutrition knowledge us-
behaviour is predicted by intention, which ing this type of approach. In addition to
in turn is predicted by attitudes and per- having a large group of subjects, a secon-
ceived social pressure, This perceived so- dary aim was to include a substantial
cial pressure has been found not to be number of men in the age range 34-54
important in the food-choice context years, as these are at greatest risk from
(Shepherd & Stockley, 1985, 1987; Shep- future coronary heart disease (COMA,
herd, 1989). Attitudes are predicted by the 1984) and therefore their attitudes towards
sum of beliefs about outcomes of the fat in foods are of particular interest.
behaviour multiplied by evahations of
these outcomes:
attitude - 2 belief x evaluation. Methods
We have previously used this approach to Subjects were recruited from the work
investigate the intake of dietary fat and force of a large insurance company. Five
showed good prediction of intention and hundred and thirty-eight subjects com-
Nutrition knowledge, attitudes and fat intake 389
pleted questionnaires; however, a number highly to dietary fat, using part of the
had missing responses on some items and Fishbein & Ajzen theory of reasoned ac-
the analyses therefore vary in the number tion. Meat, meat products, dairy products
of subjects included. The age and sex and fried foods were the four food groups
profile of the subjects is shown in Table 1. chosen for the study, based on previous
This sample was relatively homogeneous work (Shepherd & Stockley, 1985, 1987).
in social class. The meat products and dairy products
categories included a short list of
examples, for example, ‘sausages, meat
Table 1. Numbers of subjects in each age and pies, burgers, etc.’ for the meat products.
sex category The four groups chosen contribute approx-
imately 70% of the total dietary fat in the
Age Male Female Total
average UK diet. This section of the ques-
15-24 77 81 158 tionnaire contained 55 questions with
25-34 116 64 180 seven-point category scales with, apart
3544 79 32 111 from the behaviour questions, only the end
45-54 49 11 60 categories labelled.
55-65 23 6 29 There was one behaviour question for
Total 344 194 538 each of the four food groups. Subjects
responded using a seven-point frequency
scale from ‘less than once per month’ to
‘more than 2 or 3 times a day’. Subjects
The subjects completed the question- were asked if they intended to eat the food
naire in three sections. The first contained type during the next week, with the re-
demographic questions, including age, sex, sponse scale end points labelled ‘likely’
weight, height and position within the and ‘unlikely’.
company. The second part was the previ- There were three attitude questions for
ously validated nutrition knowledge ques- each food group. Each had wording of the
tionnaire (Towler & Shepherd, 1990). The type ‘My eating meat is good (or beneficial
third part was an attitudes questionnaire or pleasant)’, and the end categories of the
which was developed from previous work seven-point category response scales were
(Shepherd & Stockley. 1985, 1987) and labelled ‘strongly agree’ and ‘strongly dis-
tested on 240 quota sampled subjects agree’.
(Towler & Shepherd, 1992). The belief items were derived from short
The nutrition knowledge questionnaire structured interviews on salient beliefs
included nutrient density sections on pro- about the four different food types, with a
tein, carbohydrate, fat and dietary fibre, separate group of 34 subjects following the
and a section of nine multiple-choice ques- procedure described by Ajzen & Fishbein
tions. The nutrient density sections each (1980). Six salient beliefs were included for
included a list of 20 foods and subjects each food group but the beliefs were not
were asked to tick the ten foods that were the same for all food types. Three belief
highest in, for example, protein content. items were common to all food types,
The comparisons were to be made on an relating to the food being seen as healthy,
equal-weight basis. The selection of the high in fat and tasting good. The three
foods to include in these sections followed remaining items for each food type de-
extensive testing and validation (Towler & pended upon which beliefs were found
Shepherd, 1990) and were foods which to be salient in the pre-interviews (see
were found to differentiate between Table 6). The response categories were
subjects with high or low nutrition know- labelled ‘strongly agree’ and ‘strongly dis-
ledge. agree’ at the extremes. Each belief had
The third section of the questionnaire a corresponding outcome evaluation
was designed to measure attitudes towards statement, for example, ‘Food which is
the consumption of foods that contribute high in fat is desirable’, again with re-
390 A. Shepherd and G. Towler
sponses from ‘strongly agree’ to ‘strongly The degree of association between com-
disagree’. ponents was assessed using correlational
analysis and multiple regression, and the
differences between groups of subjects
Intake validation were assessed using analysis of variance
There was a particular interest in this (ANOVA).
study in males between the ages of 35 and
-
54 years. From this sex and age subgroup, a
validation group (n 30) was randomly
selected to measure dietary intake, with Results
weighed intakes completed over 3 days,
including 1 weekend day. The correlations between the components
Each subject was supplied with a set of of the attitude model for the whole group
digital scales (Soehnle), which weighed of subjects are shown in Table 2. These
amounts up to 1 kg to an accuracy of 2 g, a results are very comparable with those
record book, a plastic plate, full instruc- found previously using this same question-
tions and a written example. There was an naire (Towler & Shepherd, 1992). In the
oral explanation of the weighing and re- previous study the relationships between
cording procedure prior to the study. Sub- the components of the model were less
jects were to weigh and note down the clear for dairy products.
weight and a description of each food The correlations were also calculated
eaten; composite dishes were to be entered separately for the male subjects aged 35-54
as ingredients if possible and the fullest years (n -128). and the results are also
details of foods were requested. The foods shown in Table 2. The pattern of the
eaten were coded by experienced dietary correlations is very similar to that for the
research staff using McCance and Widdow- group as a whole, showing this group to
son’s The Composition of Foods (Paul & have a similar structure of the relationship
Southgate, 1978). Analysis was done using between beliefs, attitudes and behaviour.
the Institute of Food Research nutrient Similar correlations were found also for
data base. Mean daily intakes were calcu- the female subjects.
lated for each subject and used in subse- The mean scores and ranges for nutrition
quent analyses. knowledge are shown in Table 3; these are
shown for the nutrient density section for
protein, fat, carbohydrate and fibre, the
Data analysis multiple-choice section, and the overall
In the nutrient density section of the total. These are comparable with scores
nutrition knowledge questionnaire, items previously found for subjects not trained in
were scored as +I for a correct response nutrition (Towler & Shepherd, 1990). The
and -1 for an incorrect response. The correlations between the scores for total
possible scores ranged from - 2 0 to +2O nutrition knowledge (and the knowledge
for each of the four sections. The multiple- components) against the sum of belief-
choice maximum score was 14, although evaluations, attitude, intention and behav-
again negative scores were possible. Thus, iour scores for each of the food types are
the maximum-possible score for the whole shown in Table 4. The correlations were
questionnaire was 94,with negative scores generally negative, with those subjects
also possible. having higher knowledge also having a
All of the belief, evaluation, attitude and more negative predisposition toward con-
intention items were scored -3 to +3, suming these foods. The correlations were
with +3 representing a positive predispo- generally small although a number were
sition to the behaviour. The data were statistically, significantly different from
combined using the formulae from Ajzen & zero, given the large number of subjects
Fishbein (1980). The attitude responses included in the analyses. The significant
were summed to give a total attitude score. correlations tended to be for the meat and
Nutrition knowledge, attitudes and fat intake 391
Table 2. Correlation coefficients, for four food groups, between components of the
Eishbein & Ajzen theory of reasoned action for all subjects (n= 538), male subjects
aged between 35 and 54 years (n = 128). All significant at P c 0.001
Sum of
belief-evaluations Attitude Intention
V. V. V.
Food groups d.f. attitude intention behaviour
All subjects
Meat 502 0.70 0.64 0.73
Meat products 501 0.63 0.59 0.78
Dairy products 515 0.58 0.40 0.57
Fried foods 509 0.59 0.61 0.77
Males 35-54 years
Meat 116 0.65 0.51 0.55
Meat products 120 0.51 0.51 0.74
Dairy products 121 0.61 0.41 0.49
Fried foods 121 0.63 0.60 0.80
meat products sections of the attitudes significantly, higher than for males. It is
questionnaire. interesting to note that this difference was
Two-way ANOVAs, with factors of sex reversed for intention and behaviour. Ana-
(two levels) and age (five levels), were lysis based upon age showed fewer differ-
calculated for each component of the atti- ences, with the tendency for more negative
tude model for each food type. The factors attitudes, etc. in the older subjects. This
of sex and age were confounded in the was most marked for the dairy products.
present study, with unequal numbers of Correlations were calculated between
males and females in the different age the individual belief-evaluation items and
groups (Table I). Therefore the analyses do attitude scores for each of the four food
not allow testing of interactions between types (Table 6).The highest correlations in
these factors. There were significant differ- all cases were for the items relating to taste
ences between males and females on a and health, with the food being seen as
number of the variables, with females high in fat having a lower correlation.
generally having more negative attitudes Other beliefs, such as expense and conve-
(Table 5). The exception was with dairy nience, were less correlated with attitudes.
products, where the sum of belief-evalua- A multiple regression was also calculated
tions was non-significantly, and attitude for each food type between the individual
392 R. Shepherd and G. Towler -
Table 4. Correlation coefficients, for four food groups, between components of the Fishbein &
Ajzen theory of reasoned action and the scores on the nutrition knowledge questionnaire for the
four nutrient density sections (protein, carbohydrate, fat and fibre) and the total nutrition
knowledge score
Sum of
belief-
Knowledge score evaluations Attitude Intention Behaviour
Intake Discussion
Mean intakes of energy, protein, fat, car- The attitude model showed clear relation-
bohydrate and fibre for the 30 male sub- ships between the sum of belief-evalua-
jects taking part in the validation study are tions, attitude, intention and self-reported
shown in Table 7. The correlations behaviour, both for the whole group and
between fat intake and the question- for the subgroup of males. Nutrition
Nutrition knowledge, attitudes and fat intake 393
Table 5. Mean scores and significance levels from analyses of
variance for sex. The possible ranges for the means are:
belief-evaluations -54 to +54, attitude - 9 to +9, intention 1 to 7
and behaviour 1 to 7
Mean
Meat
Belief-evaluations 6.9 5.3
Attitude 3.2 1.3 ***
Intention 6.4 5.4 ***
Behaviour 4.6 4.0 ***
Meat products
Belief-evaluations -2.3 -4.7
Attitude 0.7 -1.0 ***
Intention 5.3 4.0 ***
Behaviour 3.4 2.7 ***
Dairy products
Belief-evaluations 4.5 5.1
Attitude 3.3 4.1 *
Intention 6.6 6.3 **
Behaviour 5.1 5.0 t
Fried foods
Belief-evaluations -7.6 -14.2 ***
***
Attitude -3.4 -6.0
Intention 4.2 2.9 ***
Behaviour 3.0 2.5 ***
Nutrition knowledge
Total 15.7 18.6 **
Protein 3.8 4.0
Carbohydrate 3.2 4.2 *
Fat 0.9 2.2 ***
Fibre 4.5 4.8
knowledge was less clearly related to these edge on fat to be most negatively related to
variables. the sum of belief-evaluations, attitude,
The total scores for nutrition knowledge intention and behaviour scores. There was
correlated with attitude and sum of belief also no tendency for the knowledge scores
-evaluations for the meat and meat pro- to be most closely related to the sum of
ducts sections of the questionnaire, but not belief-evaluations scores, even though
for dairy products or fried foods. In one they are conceptually most closely related
case for dairy products there was a signifi- to this part of the Fishbein & Ajzen theory
cant positive correlation with the score of reasoned action.
from the protein nutrient density section The overall results from this study con-
but this was significant only at P(O.05 firm those of our previous study using this
and, given the large number of correlations attitude questionnaire (Towler & Shep-
calculated, should not be viewed as impor- herd, 1992), with good correspondence
tant. There was no tendency for the knowl- between the components of the attitude
394 R. Shepherd and G. Towler
Table 6. Correlations between individual belief-evaluation items and
attitudes, along with the standardized (beta) coefficients and t values from
multiple regressions of the belief-evaluation items against attitudes
Meat
Healthy
(d.f. = 502)
0.54. * * 0.25
(d.f. -
497)
7.7***
Fat 0.28*** 0.12 4.l***
Taste 0.68*** 0.53 16.9***
Expense 0.20*** 0.04 1.2
Protein 0.36*** 0.12 4.l***
Vitamins 0.32*** 0.07 2.2*
Meat products (d.f. = 503) (d.f. = 498)
Healthy 0.52*** 0.27 7.8***
Fat 0.34*** 0.12 3.7***
Taste 0.64*** 0.51 15.6***
Expense 0.13** -0.01 -0.5
Convenience 0.11* -0.01 -0.3
Additives 0.35*** 0.16 4.8***
Dairy products (d.f. = 517) (d.f. = 512)
Healthy 0.56*** 0.42 12.2***
Fat 0.13** 0.18 5.2***
Taste 0.43*** 0.31 9.2***
Protein 0.37*** 0.16 4.8***
Vitamins 0.30"' 0.09 2.8***
Calories 0.03 -0.03 -0.9
Fried foods (d.f. = 512) (d.f. = 507)
Healthy 0.32*** 0.16 4.6***
Fat 0.32*** 0.13 3.7***
Taste 0.61*** 0.51 14.0***
Convenience 0.12** 0.01 0.3
Greasy 0.36** 0.15 4.2***
Smell 0.10' -0.03 -0.9
knowledge, attitudes, and fat consumption. J. U.S.Department of Health and Human Services
Am. Diet. Assoc. 87, 615-619. Public Health Service (1988) Surgeon Gener-
Towler, G. & Shepherd, R. (1990) Development al’s Report on Nutrition and Health. Washing-
of a nutritional knowledge questionnaire. J. ton, DC: U.S.Government Printing Office.
Hum. Nutr. Diet. 3,255-264. Weathersbee, P.S., Lodge, J.R. & Olsen, L.K.
TDwler, G. & Shepherd, R. (1992) Application of (1978) Milk intake of university students.
Fishbein and Ajzen’s expectancy-value model Illinois Res. 20, 19.
to understanding fat intake. Appetite 18, Werblow, J.A., Fox, H.M.& Henneman, A. (1978)
15-27. Nutrition knowledge, attitudes, and food pat-
Tuorila. H. (1987) Selection of milks with vary- terns of women athletes. I. Am. Diet. Assoc. 73,
ing fat contents and related overall liking, 242-245.
attitudes, norms and intentions. Appetite 8,
1-14.