The role of attitudes, intentions and habit in predicting

European Journal of Clinical Nutrition (2000) 54, 540±545
ß 2000 Macmillan Publishers Ltd All rights reserved 0954±3007/00 $15.00
www.nature.com/ejcn
The role of attitudes, intentions and habit in predicting actual
consumption of fat containing foods in Italy
A Saba1*, M Vassallo1 and A Turrini1
1
Istituto Nazionale di Ricerca per gli Alimenti e la Nutrizione, via Ardeatina 546, 00178 Rome, Italy
Objective: To explore which were the most important predictors of the actual consumption of foods containing
fat and to test if there were different pro®les of beliefs and attitudes.
Design: A total of 1200 households participated in a food consumption survey for 7 days. A questionnaire,
containing questions on belief, attitude and intention items based on Ajzen and Fishbein's model, was completed
on the ®rst day of the survey.
Setting: The study was conducted at the beginning of 1994 in Italy.
Subjects: In all, 860 subjects, randomly selected for each household participating in the food consumption
survey, were included in this study.
Results: A cluster analysis based on average daily intake per capita of nine types of food containing fats was
carried out on 860 subjects. The three clusters represented people with different criteria of preference for milk.
The relationship between the variables was tested using the forward selection method in multiple regression. The
attitude entered sequentially into the model assessing the intention of consuming, after the habit component.
Habit was found to be the most the most important predictor of the actual consumption of each type of food. The
intention entered into the equation as predictor of the consumption only for butter and olive oil. Habit has been
found more important than intention in in¯uencing the consumption of all types of food, whereas the intention
did not enter into the model as determinant of the consumption with exception of butter and olive oil in which it
remains as a signi®cant predictor.
Conclusions: The results of the cluster analysis suggested that subjects of cluster 1 might be more concerned
about fat intake. The ®ndings of the regressions showed that the subjective measure of habit entered in the model
as the most important predictor of the actual consumption of each type of food. The low prediction of the actual
consumption indicated that there could be other important variables in¯uencing the consumption not included in
this study.
Descriptors: attitudes; habit; actual consumption
European Journal of Clinical Nutrition (2000) 54, 540±545
Introduction
There have been suggestions that cardiovascular diseases
may be related to a high fat intake (Keys, 1970; WHO,
1990). Approximately, 37% of the total energy intakes of
the Italians is derived from fat (FAO, 1996). Although
public health nutrition education efforts have highlighted
dietary recommendations to decrease fat intake (INN, 1986,
1996), the consumption patterns have shown little change
in Italy (FAO, 1996).
One way to change people's food habits is by a change
in attitude (Ajzen & Fishbein, 1980). Information on
attitudes and beliefs held by people should help to understand the relative importance of beliefs related to different
factors in in¯uencing food choice (Shepherd, 1990). The
theory of reasoned action (TRA) developed by Ajzen and
Fishbein (1980) has been widely used in the food choice
area, demonstrating strong relationships between attitudes,
*Correspondence: A Saba, Istituto Nazionale di Ricerca per gli Alimenti e
la Nutrizione, via Ardeatina 546, 00178 Rome, Italy.
E-mail: [email protected]
Guarantor: A Saba.
Contributors: A Saba was responsible for the design of the study and
writing of the manuscript. M Vassallo was responsible for the statistical
analysis. A Turrini was responsible for the food consumption database. All
authors took part in the discussion of all results.
Received 16 November 1999; revised 11 February 2000; accepted
21 February 2000
beliefs and food choice (Tuorila & Pangborn, 1988a;
Shepherd, 1990; Shepherd et al, 1991=1992; Towler &
Shepherd, 1992). Within the TRA, behaviour is predicted
by the conscious decision to perform behaviour, ie the
behavioural intention, which in turn is predicted by a
combination of people's attitude towards performing the
behaviour and their perceptions of the social pressure put
upon them to perform that behaviour, ie the subjective
norms. Recent empirical works suggested modi®cations of
Fishbein and Ajzen's theory of reasoned action, including
other components as determinants of food consumption in
order to increase the predictive power of the model.
Particularly, habit was found to play an important role in
the context of food choice when consumption of food is
frequently performed (Tuorila & Pangborn, 1988b; Towler
& Shepherd, 1991=1992). Habit can be considered as
frequently repeated past behaviour (Triandis, 1977) or as
behaviour that is in some way automatic or out of the
awareness of the subject (Ronis et al, 1989). In the food
context, habit measured using `out of habit' type questions,
has been shown to be related to consumption of sweet, salty
and fatty foods (Tuorila & Pangborn, 1988b). A meÂasure of
habit incorporating the idea of actions performed `out of
habit' and `out of force of habit' was found to be an
important predictor of self-reported behaviour, although it
was not as important as intention (Towler & Shepherd,
1991=1992).
Attitudes, habit and actual consumption
A Saba et al
Ajzen and Fishbein (1980) pointed out that the link
between intention and behaviour can be reduced by lack of
temporal correspondence. However, the Fishbein and Ajzen
model has been applied frequently to situations in which
the intention measure was asked at the same time of the
current behaviour (see the review of Sta¯eu et al,
1991=1992). Also, the behaviour was generally measured
as self-reported consumption (Sapp, 1991; Sta¯eu et al,
1991=1992; Towler & Shepherd, 1992; Brug et al, 1994;
Raats et al, 1995), whereas there is little research examining the relationships between the behaviour intention and
the behaviour measured by observation of actual consumption of food (Shepherd & Farleigh, 1986; Saba & Di Natale,
1999).
The objective of the present study was twofold: (a) to
explore which were the most important predictors of the
actual consumption of foods containing fat within the
Fishbein and Ajzen framework including a measure of
the habit component; (b) to test if subjects of a large
sample had different pro®les of beliefs and attitude towards
the consumption of foods containing fats.
Since previous research in the food area has found that
the subjective norm component was less important than
attitude in in¯uencing the intention of consuming (Sapp,
1991; Sta¯eu et al, 1991=1992; Towler & Shepherd,
1991=1992; Thompson et al, 1994), it was not examined
in the present study.
Methods
Subjects
One thousand and two hundred households participated in a
food consumption survey for 7 days. They were selected at
random from the electoral register of medium and largesized towns of Italy at the beginning of the 1994. During
the weekly survey all members of the families recruited
were asked to record their individual food consumption
according to a methodology described below. A sub-sample
of 1968 individuals was selected in order to assess the
reference diet for an average person. For this purpose,
outliers and suspected underreporting subjects have been
excluded. The selection was performed by calculating for
each individual daily energy requirement for a sedentary
life style (SINU, 1996) and daily energy intake (kcal) by
the application of the Italian food composition table to the
consumption quantities (Carnovale & Marletta, 1997). The
comparison of the two estimates provided the group of
consumers assuming a habitual diet. The validity of the
selection has been tested by calculating the ratio between
the daily energy intake and the Basal Metabolic Rate
(FAO=WHO=UNU, 1985).
Finally, only one subject over 14 y of age, who participated over 4 days of food consumption, was randomly
selected for each household in order to remove possible
correlated responses between the members of the same
family. The ®nal sample was composed of 860 subjects.
Measurement of food consumption
The methodology adopted was a mixed technique including
the weighing method for household food consumption and
a record of daily food consumption for each component of
the household. The person who took care of feeding
weighed the amounts of foodstuffs used for the preparation
of the meals as well as food preparation wastes. Individuals
were asked to weigh their food portions or give the amount
consumed estimated in household measures, such as
spoons, glasses, etc. Meals eaten outside the home during
the survey period were taken into account as well. When
the exact composition of different dishes was not known,
the components included were calculated by the aid of
standard recipes. Well-trained dietitians during the survey
supervised the individuals. The validity of those survey
methods has been reviewed by other authors (Pekkarinen,
1970; Bingham, 1987).
In the present study the behaviour was measured as
mean daily quantity of consumption of the nine foods
containing fats and explaining the 80% of energy intake,
according to the results of a study on food consumption
carried out some years previously in Italy (Saba et al,
1990). The foods included red, white and preserved meat,
semi-skimmed and whole fat milk, cheese, olive oil, seed
oil and butter. Red meat included beef, veal and pork meat;
chicken and turkey meat were included in white meat,
whereas preserved meat included ham, salami and other
preserved meat.
541
Measurement of intention, beliefs, attitudes and habit
The questionnaire, completed on the ®rst day of the survey,
contained questions on belief, attitude and intention items
based on Ajzen and Fishbein's model (1980) along with
two questions on habit. All questions were related to the
consumption of each of the types of food. Responses to
questions on attitudes and intention of consuming were on
seven-point category scales, with the exception of those for
habit, which were on ®ve-point category scales.
Behavioural intention was measured, for each of the
nine foods, by asking subjects if they intended to eat the
food during the following days of the week. Responses
were on seven-point category scales, ranging from `extremely unlikely' to `extremely likely'.
Two questions on attitude, for each of the nine types of
food, eg `My eating cheese is (good or bene®cial)', were
rated on two sets of responses ranging from `strongly
disagree' to `strongly agree'.
The behaviour beliefs were elicited from a pilot sample
of 30 subjects, following the procedure suggested by Ajzen
and Fishbein (1980). For each of the nine types of foods
there were salient belief items relating to nutritional
aspects, weight concern and functional properties. For
whole milk and semi-skimmed milk nine salient beliefs,
for the other foods eight beliefs were elicited. The belief
items were worded according to the following format: `By
drinking whole fat milk I get a food that is high in calcium'.
The response categories were on a seven-point scale
labeled `strongly agree' and `strongly disagree' at the
extremes.
Each behaviour belief had a corresponding outcome
evaluation, as `Milk that is high in fats is desirable' and
the responses were on a seven-point category scale labelled
from `strongly disagree' to `strongly agree'.
Two questions assessed a subjective measure of habit of
consumption for each the eight types of food. The ®rst
question, taken from Tuorila and Pangborn (1988a, 1988b),
had the wording `I consume butter out of habit'. The
responses were rated from `not at all' to `very much'.
The second question, taken from Towler and Shepherd
(1991=1992), read `How often do you consume butter out
of habit?'. The responses were rated from `never consuming' to `consuming once a day'.
European Journal of Clinical Nutrition
Attitudes, habit and actual consumption
A Saba et al
542
Questions concerning socio-demographic information
were included in a second questionnaire.
Table 1
Data analysis
Semi-skimmed milk
Whole milk
Cheese
Preserved meat
Red meat
White meat
Olive oil
Seed oil
Butter
In order to estimate the association between the variables,
the attitude and habit responses were summed to give,
respectively, a total attitude and habit score (Ajzen &
Fishbein, 1980). Cronbach's alpha coef®cients of internal
reliability were assessed for attitude and habit components
(Cronbach, 1951).
Scores on belief responses multiplied by the corresponding outcome evaluations were calculated for each of the
nine types of food. The behaviour was measured as mean
daily quantity of actual consumption. The association
between variables was assessed using Pearson correlation
coef®cients.
The extended model of Fishbein ± Ajzen was tested by
regression analysis, by using the method of forward selection (Brook & Arnold, 1985). The forward selection
method adds each predictor variable sequentially only if
the increase in the proportion of variation explained by the
regression is highest and signi®cant.
Cluster analysis was applied in order to group respondents on the basis of the similarity of their consumption of
each of the nine types of food. The clustering was performed by applying a K-means procedure (MacQueen,
1967). Means of the components attitudes and beliefs for
each of the resulting clustered segments were calculated.
Differences between groups of subjects on attitudes and
beliefs scores were assessed using the non-parametric
analyses of variance (Kruskal ± Wallis).
The analyses were carried out using the statistical
package SPSS (1991).
Results
The sample of 860 subjects was composed of 481 females
and 379 males. The mean age was 41 y ( + 16.3 y). The
educational level was high (40% had a high school education and 17% had a degree).
The association between the components of the Fishbein
and Ajzen model was calculated (data not shown). All
correlations were signi®cant. The correlations were low for
each of the nine types of food, with the exception of the
correlations between attitude and intention, attitude and
habit, habit and intention, with values ranging from 0.4 to
0.8. The component behaviour showed higher correlations
with habit than with intention.
The internal reliabilities measured by Cronbach's alpha
coef®cients were quite high with the exception of the
attitude component for preserved meat, which low value
of 0.21 indicated a lack of homogeneous components for
the attitude item (Table 1). The sum of the beliefs responses
multiplied by the corresponding evaluations (Sb6e) for
olive and seeds oils showed low values of internal
reliabilities (0.35 and 0.43, respectively).
There were no signi®cant differences between attitudes
and beliefs by sex and age in the total sample (data not
shown). In order to explore pro®les of beliefs and attitudes
of clusters of subjects with different average food intake, a
cluster analysis was performed on the average daily intake
per capita of each the nine types of food. A classi®cation
with three clusters was considered. The ®rst cluster, cluster
1, had a total of 196 individuals; the second group, cluster
European Journal of Clinical Nutrition
Cronbach's alpha coef®cient for multi-item components
Food product
Attitudes
S b6e
Habit
0.79
0.74
0.54
0.21
0.66
0.76
0.77
0.84
0.54
0.57
0.58
0.61
0.58
0.55
0.44
0.35
0.43
0.67
0.90
0.91
0.77
0.82
0.82
0.74
0.72
0.86
0.84
2, was composed of 476 individuals; and the third group,
cluster 3, was composed of 188 subjects. Cluster 1 was
composed of 39% males and 61% females; cluster 2 had
44% males and 56% females; cluster 3 was composed of
51% males and 49% females. There were no signi®cant
differences between clusters by classes of age and educational level (data not shown). In Figure 1 the pro®les of
consumption of each cluster are shown. It is clear that the
three clusters represent people with different criteria of
preference for milk. The semi-skimmed milk was the most
consumed type of milk (206.4 g=per capita=day) for the
subjects of cluster 1, whereas the whole milk was the most
consumed type of milk for the subjects of cluster 3
(227.7 g=per capita=day). Also, the subjects of cluster 3
showed, on average, higher consumption of red meat,
cheese and butter. The attitudes and beliefs that were
signi®cantly different among the three clusters are shown
in Figure 2. The subjects of cluster 3 had higher scores of
attitudes and beliefs items for whole fat milk, with the
exception of the belief `Drinking whole milk I gain
weight', whereas the subjects of cluster 1 had higher
scores of attitude and belief components for semi-skimmed
milk. Also, there were more differences between the
clusters 1 and 3 on beliefs regarding fat intake for red
meat, cheese and butter. The subjects of cluster 1 had
higher scores of beliefs `By eating red meat I get a food that
is high in fat', and `By eating red meat (or cheese or butter),
I get a food containing harmful fats'.
The extension of the Fishbein and Ajzen model was
tested using the forward selection method in multiple
regression. Initially the attitude was regressed on the beliefs
multiplied by the evaluations for each of the nine types of
food. The intention was regressed on the attitude and habit
components for each type of food. Multiple regressions of
intention and habit on behaviour were calculated. There
were no differences in the magnitude of the relationships
between the variables for each of the nine types of food
among the three clusters of subjects and the total sample of
subjects. Thus, to minimise complexity of presentation,
only the results of the multivariate regressions applied on
the total sample are shown in Table 2. The attitude entered
sequentially into the model, assessing the intention of
consuming, after the habit component. Habit has been
found more important than intention in in¯uencing the
consumption of butter and olive oil, whereas the intention
did not enter into the model as determinant of the consumption of the other types of food.
Discussion
The classi®cation in three clusters represented people with
different criteria of preference for milk. The semi-skimmed
Attitudes, habit and actual consumption
A Saba et al
543
Figure 1 A representation of food pro®le by cluster and total sample of subjects. The lines correspond with the daily average per-capita consumption of
each of the nine types of food. The average values are reported between brackets.
milk was the most consumed type of milk by the subjects of
cluster 1, which was composed of more females than the
other two clusters. The whole milk was most consumed by
the subjects of cluster 3, which included a higher percentage of males than the other two clusters. It is interesting to
note that there were signi®cant differences mainly on
beliefs and attitudes towards the consumption of the two
types of milk between the clusters 1 and 3. The cluster 3
had higher scores of attitudes and beliefs items for whole
milk among the three clusters, whereas the subjects of
cluster 1 had higher scores of attitudes and beliefs items for
semi-skimmed milk. Additionally, there were signi®cant
differences on the scores of the beliefs concerning fat
content of red meat, cheese, and butter between clusters 1
and 3. In particular, the subjects of cluster 1 had higher
scores of beliefs and showed lower consumption of red
meat, butter, and cheese. This result might suggest that the
subjects of cluster 1 paid more attention to fat intake.
The high correlations between habit and intention for
olive and seed oil and butter suggest that habit may predict
intention of consuming food better than the attitude component assessed by the technique of Fishbein and Ajzen
(1980). This implies that habit may have an important role
in changing behaviour.
The extension of the model of Fishbein and Ajzen
provided limited prediction of actual consumption from
the components included. The ®ndings of the regressions
showed that the subjective measure of habit, incorporating
the idea that the behaviour is not only a frequently repeated
behaviour but also is partly out of conscious control,
outweighed attitude in the impact of intention on consuming. Even though the attitude component had a minor role,
it maintained a signi®cant weight in in¯uencing the intention. Habit entered into the model as the most important
predictor of the actual consumption of each type of food.
The intention entered into the equation as a predictor of the
consumption only for butter and olive oil. This ®nding
contrasts with previous studies on the consumption of milk
(Saba et al, 1998) in which intention entered into the model
as a predictor of consumption, whereas habit did not, on
consumption of sweet, salty and fatty foods (Tuorila &
Pangborn, 1988b) and on chip consumption (Towler &
Shepherd, 1991=1992), in which habit was shown to be
signi®cantly related to consumption of foods, although less
than intention was. As food choice is a repeated behaviour
(Triandis, 1977) and habit is likely to play an important role
in the maintenance of a frequently repeated behaviour
(Ronis et al, 1989), it is necessary to pay more attention
to the conceptualisation of habit in food choice. Also, it is
possible that, by adopting a different de®nition, the in¯uence of habit could change its importance in the prediction
models (Tuorila & Pangborn, 1988b). However, what the
European Journal of Clinical Nutrition
Attitudes, habit and actual consumption
A Saba et al
544
Figure 2 Means of ratings of beliefs and attitudes for subjects divided into three clusters. Only the beliefs and attitudes signi®cantly different among the
three clusters are shown (P < 0.001). The beliefs and attitudes are scored from 7 3 (strongly disagree) to ‡ 3 (strongly agree).
Table 2 Results of regressions on behaviour, intention and attitude. The determination coef®cients (r2) for each successive variable are added in, and the
®nal regression coef®cients (b) are showna
Dependent variable
Behaviour
Intention
Attitude
Behaviour
Intention
Attitude
Behaviour
Intention
Attitude
Variables entered
b
Semi-skimmed milk
Habit
18.289
Habit
0.622
Attitude
0.145
S b6e
0.084
Red meat
Habit
8.148
Habit
0.426
Attitude
0.095
S b6e
0.054
Olive Oil
Habit
1.034
Intention
1.130
Habit
0.354
Attitude
0.169
S b6e
0.047
r2
Variables entered
0.314
0.650
0.666
0.227
0.085
0.355
0.372
0.139
0.038
0.044
0.349
0.386
0.185
b
r2
Whole milk
21.482
0.704
0.072
0.070
White meat
Habit
6.676
Habit
0.349
Attitude
0.131
S b6e
0.034
Seed oil
Habit
0.909
0.028
Habit
Attitude
S b6e
0.509
0.522
0.215
Habit
Habit
Attitude
S b6e
0.679
0.088
0.108
0.315
0.650
0.654
0.205
0.064
0.281
0.313
0.068
Variables entered
b
Cheese
Habit
5.109
Habit
0.367
Attitude
0.121
S b6e
0.032
Preserved meat
Habit
3.653
Habit
0.516
Attitude
0.104
S b6e
0.062
Butter
Habit
0.830
Intention
0.298
Habit
0.714
Attitude
0.129
S b6e
0.071
r2
0.080
0.336
0.363
0.073
0.074
0.413
0.433
0.202
0.108
0.114
0.452
0.476
0.246
a
S b6e indicates the sum of the beliefs b responses multiplied by corresponding evaluations e.
subjects perceive as habitual behaviour could be the result
of the interaction of other factors in¯uencing the food
selection (Mela, 1999). Thus, the de®nition of a measure
of habit is so complex that it requires further attention in
future research.
The coef®cients of determination for actual consumption
were very low for all the nine types of food, indicating that
there could be other important variables in predicting the
European Journal of Clinical Nutrition
consumption not included in this study. It needs to be borne
in mind that human selection of food is derived from the
prevailing food, agro-economic and cultural environment,
cognitive and biological characteristics of individuals, and
the intrinsic and extrinsic attributes of foods themselves
(Shepherd, 1990; Mela, 1999). Factors related to the food,
such as food appearance, and to external environmental
in¯uences, such as cultural and religious factors, act
Attitudes, habit and actual consumption
A Saba et al
through food preferences (Khan, 1981). Physical characteristics of the product linked up with brand name and price
may in¯uence the choice of food (Grunert et al, 1996).
Also, it is possible that potential biases in the responses
on the consumption or intention to consume each of the
nine types of food may have in¯uenced the degree of
prediction in this study. It can be seen, however, that the
magnitude of the determination coef®cient for whole fat
and semi-skimmed milk were the highest, indicating that
habit could represent an important determinant of the
consumption of milk.
In conclusion, the Fishbein and Ajzen model was con®rmed as a useful basic framework within which to
measure and relate the variable of interests to understand
the complexity of causative factors in food consumption.
The modi®cations investigated in this paper, such as the
inclusion of habit, con®rmed the need to offer a framework
for empirical research rather than exploring merely the
in¯uences of factors likely to be important.
Acknowledgements ÐThe research reported in this article was carried out
as a part of the project `Food consumption and behaviour in Italy' funded
by the Italian Ministry of Agricultural Food and Forest Resources.
References
Ajzen I & Fishbein M (1980): Understanding Attitudes and Predicting
Social Behaviour. Englewood Cliffs, NJ: Prentice Hall.
Bingham SA (1987): The dietary assessment of individuals; methods,
accuracy, new techniques and recommendations. Nutr. Abstr. Rev. (Ser.
A) 57, 705 ± 742.
Brook RJ & Arnold GC (1985): Applied Regression Analysis and Experimental Design. New York: Marcel Dekker.
Brug J, Van Assema P, Kok G, Lenderink T & Glanz K (1994): Self-rated
dietary fat intake: association with objective assessment of fat, psychological factors, and intention to change. J. Nutr. Educ. 26, 218 ± 223
Carnovale E & Marletta L (1997): Tabelle di composizione degli Alimenti.
Rome: Istituto Nazionale della Nutrizione.
Cronbach LJ (1951): Coef®cient alpha and the internal structure of tests.
Psychometrika 16, 297 ± 334.
FAO (1996). Food Balance Sheets. Food and Agricultural Organization of
the United Nations, 1991 ± 1994.
FAO=WHO=UNU (1985): Energy and Protein Requirements, Technical
Reports Series. Geneva: WHO.
Grunert KG, Hartvig Larsen H, Madsen TK & Baadsgaard A (1996):
Market Orientation in Food and Agriculture. Boston, MA: Kluwer
Academic.
INN (1986): Linee Guida per una sana alimentazione italiana. Roma:
Istituto Nazionale della Nutrizione.
INN (1996): Linee Guida per una sana alimentazione italiana. Roma:
Istituto Nazionale della Nutrizione.
Keys A (1970): Coronary Heart Disease in Seven Countries. Circulation
41, 1 ± 23.
Khan MA (1981): Evaluation of food selection patterns and preferences.
CRC Crit. Rev. Food Sci. Technol. 15, 129 ± 153.
MacQueen JB (1967): Some methods for classi®cation and analysis of
multivariate observations. Proceedings of the Fifth Berkeley Symposium
on Mathematical Statistics and Probability, Vol 1, pp 281 ± 297.
Mela DJ (1999): Food choice and intake: the human factor. Proce. Nutr.
Soc. 58, 513 ± 521.
Pekkarinen M (1970): Methodology in the collection of food composition
data. In: World Review of Nutrition and Diet pp 145 ± 171. Basel:
Karger.
Raats MM, Shepherd R & Sparks P (1995): Including moral dimensions of
choice within the structure of the theory of planned behaviour. J. Appl.
Soc. Psychol. 25, 484 ± 494.
Ronis DL, Yates JF & Kirscht JP (1989): Attitudes, decisions, habits as
determinants of repeated behaviour. In: Attitude Structure and Function, ed. A Pratkanis, S Breckler & A Greenwald, pp 213 ± 239.
Hillsdale, NJ: Erlbaum.
Saba A & Di Natale R (1999): A study on the mediating role of the
intention in the impact of habit and attitude on meat consumption. Food
Qual. Pref. 10, 69 ± 77.
Saba A, Turrini A, Mistura G, Cialfa E & Vichi M (1990): Indagine
Nazionale sui consumi alimentari delle famiglie 1980 ± 84: alcuni
principali risultati. L Riv. Ital. Sci. Alimentazione 19, 453 ± 465.
Saba A, Moneta E, Nardo N & Sinesio F (1998): Attitudes, habit, sensory
and liking expectation as determinants of the consumption of milk.
Food Qual. Pref. 9, 31 ± 41.
Sapp SG (1991): Impact of nutritional knowledge within an expanded
rational expectations model of beef consumption. J. Nutr. Educ. 23,
214 ± 222.
Shepherd R (1990): Overview of factors in¯uencing food choice. Br. Nutr.
Found. Bull. 15(Suppl 1), 12 ± 30.
Shepherd R & Farleigh CA (1986): Preferences, attitudes and personality
as determinants of salt intake. Human Nutr. Appl. Nutr. 40A, 195 ± 208.
Shepherd R, Sparks P, Bellier S & Raats MM (1991=1992): Attitudes and
choice of ¯avoured milks: extensions of Fishbein and Ajzen's theory of
reasoned action. Food Qual. Pref. 3, 157 ± 154.
SINU (1996): Livelli di assunzione raccomanda ti di energia e nutrienti per
la popolqzione italiana. LARN Istituo Nazionale della Nutrizione:
Roma.
SPSS (1991): Base System User's Guide, SPSS Inc.
Sta¯eu A, De Graaf C, Van Staveren WA & Schroots JJF (1991=1992): A
review of selected studies assessing social-psychological determinants
of fat and cholesterol intake. Food Qual. Pref. 3, 183 ± 200.
Thompson KE, Haziris N & Alekos PJ (1994): Attitudes and food choice
behaviour. Br. Food J. 96, 9 ± 13.
Towler G & Shepherd R (1991=1992): Modi®cation of Fishbein and
Ajzen's theory of reasoned action to predict chip consumption. Food
Qual. Pref. 3, 37 ± 45.
Towler G & Shepherd R (1992): Application of Fishbein and Ajzen's
expectancy-value model to understand fat intake. Appetite 18, 15 ± 27.
Triandis HC (1977). Interpersonal Behavior. Monterey, CA: Brooks=Cole.
Tuorila H & Pangborn RM (1988a): Prediction of reported consumption of
selected fat-containing foods. Appetite 11, 81 ± 95.
Tuorila H & Pangborn RM (1988b): Behavioural models in the prediction
of consumption of selected sweet, salty and fatty foods. In: Food
Acceptability, ed. HDM Thomson, pp 267 ± 279. London: Elsevier
Applied Science.
WHO (1990): Diet, Nutrition and the Prevention of Chronic Diseases.
Geneva: WHO.
545
European Journal of Clinical Nutrition