Revista de Psicología Social: International Journal of Social Psychology

Revista de Psicologa Social
International Journal of Social Psychology
ISSN: 0213-4748 (Print) 1579-3680 (Online) Journal homepage: http://www.tandfonline.com/loi/rrps20
I love chips and (h)ate them too: the role of

ambivalence and contextual cues on attitudinally
based eating behaviour / Me encantan las patatas
fritas, las como y (tambin) las odio: el papel de
la ambivalencia y de las claves contextuales en el
comportamiento alimentario basado en la actitud
Maria T. Batista, Maria-Lusa Lima, Ccero Pereira & Hlder Alves
To cite this article: Maria T. Batista, Maria-Lusa Lima, Ccero Pereira & Hlder Alves (2014) I
love chips and (h)ate them too: the role of ambivalence and contextual cues on attitudinally
based eating behaviour / Me encantan las patatas fritas, las como y (tambin) las odio: el papel
de la ambivalencia y de las claves contextuales en el comportamiento alimentario basado en la
actitud, Revista de Psicologa Social, 29:3, 430-461, DOI: 10.1080/02134748.2014.981387
To link to this article: http://dx.doi.org/10.1080/02134748.2014.981387
Published online: 15 Dec 2014.
Submit your article to this journal
Article views: 29
View related articles
View Crossmark data
Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=rrps20
Download by: [Universidad Nacional Andres Be]
Date: 23 November 2015, At: 21:32
Downloaded by [Universidad Nacional Andres Be] at 21:32 23 November 2015
Revista de Psicologa Social / International Journal of Social Psychology, 2014

Vol. 29, No. 3, 430461, http://dx.doi.org/10.1080/02134748.2014.981387
I love chips and (h)ate them too: the role of ambivalence and
contextual cues on attitudinally based eating behaviour / Me
encantan las patatas fritas, las como y (tambin) las odio: el papel
de la ambivalencia y de las claves contextuales en el
Maria T. Batistaa, Maria-Lusa Limaa, Ccero Pereirab, and Hlder Alvesa
a
Instituto Universitrio de Lisboa (ISCTE-IUL), CIS-IUL; bInstituto de Cincias Sociais

da Universidade de Lisboa
(Received 1 March 2013; accepted 10 July 2013)
Abstract: Although past work proposed ambivalence as moderating the consistency between attitudes and behaviours, empirical evidence is mixed. We address
this matter by taking into account social influence processes and observing actual
eating behaviour. We predicted that less ambivalent attitudes are more consistent
with behaviours independently of context cues. In two studies, we first evaluated
participants ambivalence towards the target food. A week later we observed their
eating behaviour (Study 1), and analysed the influence of context cues on it
(Study 2). Results showed that only less ambivalent attitudes are good predictors
of actual behaviour. Importantly, context cues only influenced behaviour among
individuals with ambivalent attitudes.
Keywords: attitudes; attitudinal ambivalence; eating behaviour; social
influence
Resumen: Aunque trabajos anteriores sugieren que la ambivalencia modera la
consistencia entre actitudes y comportamientos, las evidencias empricas no
son, sin embargo, convergentes. En este estudio abordamos esta cuestin
teniendo en cuenta los procesos de influencia social y observando comportamientos alimentarios reales. Predecimos que actitudes menos ambivalentes
sern ms consistentes con los comportamientos, independientemente de las
claves del contexto. En dos estudios, primero evaluamos la ambivalencia de
los participantes hacia un alimento en particular. Una semana ms tarde
observamos la conducta alimentaria (Estudio 1), y analizamos la influencia
de las claves contextuales en dicha conducta (Estudio 2). Los resultados
mostraron que nicamente las actitudes menos ambivalentes son buenos
predictores del comportamiento real. Es importante destacar que las claves
English version: pp. 430443 / Versin en espaol: pp. 444458

References / Referencias: pp. 458461
Translated from English / Traduccin del ingls: Ainhoa Casado
Authors Address / Correspondencia con los autores: Maria Toscano Batista, ISCTE
Lisbon University Institute, Centro de Investigao e Interveno Social, Av. das Foras
Armadas, 1649026 Lisboa, Portugal. E-mail: maria.e.s.t.batista@gmail.com
2014 Fundacion Infancia y Aprendizaje
Ambivalence, attitudes and food behaviour / Ambivalencia, actitudes, comportamiento alimentario
431
contextuales slo influyeron en el comportamiento de los individuos con

actitudes ambivalentes.
Palabras clave: actitudes; ambivalencia actitudinal; comportamiento alimentario;

influencia social
The possibility of predicting behaviour based on attitudes has captivated the

interest of social psychologists for decades. The first studies on the relationship
between attitudes and behaviours showed that it was inconsistent (e.g., Kutner,
Wilkins, & Yarrow, 1952; LaPiere, 1934; Wicker, 1969). Later research sought to
identify variables that would influence when an attitude is more likely to predict
behaviours (e.g., Fazio & Towles-Schwen, 1999; Holland, Verplanken, &
Knippenberg, 2002). Among these the strength of attitude was identified as an
important moderator (e.g., Holland et al., 2002).
Attitudinal ambivalence has been identified as one dimension of attitude
strength, such that the more ambivalent the attitude, the weaker it is (e.g., Conner
& Sparks, 2002; Thompson, Zanna, & Griffin, 1995). Erber, Hodges, and Wilson
(1995) suggested that ambivalent attitudes could be highly influenced by the
contexts in which they are retrieved. For this reason, they put forward that ambivalence may play a relevant role in the consistency of the relationship between
attitudes and behaviours. Nevertheless, conclusions regarding the weak behavioural
predictability of ambivalent attitudes rest on research that used self-reported measures, not observed behaviour (e.g., Armitage & Conner, 2000). Although there is
evidence that ambivalent attitudes are more susceptible to social influence (e.g.,
Hodson, Maio, & Esses, 2001), we were unable to find studies addressing the joint
effects of attitudinal ambivalence and social contexts on behaviours.
In this article we aim to demonstrate the relevance of both attitudinal ambivalence and social contexts in the relationship between attitudes and behaviours.
Specifically, in two studies using longitudinal designs and observation of actual
behaviour, we tested that more ambivalent attitudes are worse predictors of
behaviour and more susceptible to indirect social influence than less ambivalent
attitudes. This was done by observing actual behaviour, not only by using selfreport measures as has often been the case in research (Dormandy, Hankins, &
Marteau, 2006). This strategy will allow us to analyse whether social contexts
have a relevant role in predicting behaviours when attitudes are ambivalent.
Obtaining empirical support for this claim will contribute to the development of
strategies aiming at the promotion of healthier eating.
Attitudinal ambivalence
According to a unidimensional approach, attitudes can only be positive, neutral or
negative evaluations (Eagly & Chaiken, 1993). In contrast, the bidimensional
view of attitudes allows for the existence of two independent evaluative dimensions: positive and negative (Bell, Esses, & Maio, 1996; Kaplan, 1972; Maio,
Esses, & Bell, 2000; Thompson et al., 1995). This viewpoint is helpful to
distinguish neutral (i.e., indifferent) from ambivalent (i.e., mixed) attitudes.
432
M.T. Batista et al.
In fact, drawing on a view that understands attitudes as bidimensional

constructs, the perspective of attitudinal ambivalence entails the simultaneous
existence of positive and negative evaluations of the same object (Kaplan,
1972; Scott, 1968; Thompson et al., 1995). According to Eagly and Chaiken
(1993), ambivalence may lead to low consistency between attitudes and behaviours because ambivalent attitudes are more sensitive to contextual cues than
non ambivalent ones. Jonas, Broemer, and Diehl (2000) support this explanation by focusing on the impact of the kind of information retrieved.
Specifically, Jonas et al. (2000) argue that when attitudes are ambivalent,
individuals retrieve more positive or negative characteristics, according to the
present situational cues.
Yet, the role of attitudinal ambivalence on the attitude-behaviour relationship
has not been clearly demonstrated. On the one hand, research suggests that less
ambivalent attitudes are better predictors of behaviour (e.g., Armitage & Conner,
2000; Berndsen & van der Pligt, 2004; Conner et al., 2002; Conner, Povey,
Sparks, James, & Shepherd, 2003; Povey, Wellens, & Conner, 2001; Shepherd,
1999; see also Conner & Sparks, 2002, for a review). On the other hand, the kinds
of behavioural measurements employed most of the time (i.e., self-reported
behaviour and sometimes a measure of behavioural intention) prevent us from
drawing strong conclusions. Nevertheless, even research that measures actual
behaviour (Conner, Godin, Sheeran, & Germain, 2013; Dormandy et al., 2006)
has thus far not considered the simultaneous effect of social context and attitudinal
ambivalence on the attitude-behaviour relationship.
Regarding the influence of the social context on peoples actual behaviour, we
must distinguish between direct and indirect influence. When social influence is
indirect, the source may not try to persuade an individual to change his/her
attitudes or behaviours. For instance, the mere presence of others exerts influence
by changing the attitudes or behaviours of individuals who are unlikely to
perceive that they are being influenced. There is some evidence indicating that
this effect might be especially noticeable among individuals who hold ambivalent
attitudes (Lavine, 2001; Lavine, Huff, Wagner, & Sweeney, 1998; Tourangeau,
Rasinski, Bradburn, & DAndrade, 1989). Nevertheless, this literature has not
used measures of actual behaviour.
When social influence is direct, the sources seek to persuade an individual to
change an attitude or to choose a particular behaviour. Research has shown that
individuals with more ambivalent attitudes are more easily persuaded, that is, they
more readily change their attitude according to the persuasive message (e.g.,
Armitage & Conner, 2000; Bell & Esses, 1997; Broemer, 2002; Linville &
Jones, 1980; Maio, Bell, & Esses, 1996).
Overview of the studies

In two studies using longitudinal designs we analysed the role of attitudinal
ambivalence on actual behaviour. In order to empirically address this matter we
have turned to observations across time and to the use of a behavioural measure.
433
These strategies not only allowed us to register what participants said they had
done or intended to do in the future but also what they actually did.
We present unhealthy food (specifically, chips) as attitude objects for two
reasons. First, this kind of food is likely to simultaneously activate the positive
and the negative dimensions of an attitude (Batista & Lima, 2010; Urland & Ito,
2005; see also Beardsworth, 1995; Conner & Sparks, 2002; Mischel, Shoda, &
Rodriguez, 1989; Shepherd, 1999). For instance, chips taste good but they are
fattening. This involves a pleasure/health conflict (Conner & Armitage, 2002;
Conner & Sparks, 2002) which is a socially relevant topic. In fact, research
indicates that ambivalent individuals show greater food-related concerns
(Stroebe, Mensik, Aarts, Schut, & Kruglanski, 2008) and that ambivalence is
mainly felt towards tasty but unhealthy kinds of food (Urland & Ito, 2005).
Overall, this body of research, however, has not studied the effect of those
variables on behaviour itself. Moreover, even those studies that measured behaviour did not take attitudinal ambivalence into account. For instance, Hofmann,
Rauch, and Gawronski (2007) studied the role of attitudes and self-regulatory
resources on eating behaviour (see also Hofmann, Friese, & Roefs, 2009) and
Burger et al. (2010) studied the relationship between descriptive norms and food
choices (see also Pliner & Mann, 2004).
Second, the use of unhealthy food in these studies provided a straightforward
and meaningful way to study the influence of situational cues and attitudinal
ambivalence on behaviour. In fact, eating behaviours frequently occur in social
contexts, and are often used to share or celebrate something socially (Conner &
Armitage, 2002; Ogden, 2003). In such situations, individuals eating behaviours
are mainly under indirect social influence. Yet, as already indicated, the literature
on attitude ambivalence has neglected this kind of social influence on behaviours.
In Study 1 we tested whether attitude ambivalence moderates the attitudebehaviour relationship. For that purpose we did not simply measure behavioural
intention. Instead, we measured attitudes and observed actual behaviour. Extending
this study, in Study 2 we experimentally tested the moderating role of social context
on the previous relationship. We predict that more ambivalent compared to less
ambivalent attitudes promote inconsistency of the attitude-behaviour relationship
and make peoples behaviour more susceptible to indirect social influence.
Study 1
In this study we tested the hypothesis that attitudinal ambivalence moderates the
consistency of the attitude-behaviour relationship. Based on the above-mentioned
literature, we hypothesized that among more ambivalent participants there is no
relationship between the attitude towards chips measured at one point in time and
chip eating behaviour occurring one week later. Conversely, we expect attitudes to
be good predictors of behaviour among less ambivalent participants.
434
M.T. Batista et al.
Method
Participants
One hundred and twenty-three university students, 34% males and 66% females,
aged 1746 years, M = 21.47, SD = 3.78, participated in this study.
Procedure
The study was conducted in two phases. In time 1, we measured participants
attitudes and ambivalence towards chips. One week later the same participants
went to the laboratory to individually take part in a study presented as research on
the influence of hunger on life satisfaction. In order to make this information
credible and to justify the presence of food, participants were told that the study
had two conditions: one condition would include only participants who were
allowed to eat while they performed the tasks, the other condition would include
only participants who were not allowed to eat while they performed the tasks. A
rigged draw grouped all participants in the allowed to eat condition. Participants
were handed out a questionnaire on life satisfaction and offered two kinds of food
they could eat (Pringles chips and apple slices) while they were responding. The
presence of two kinds of food was required so that participants could make a
choice as our experimental manipulation implied. Before they left, we probed as
to whether they had associated the two phases of the study. After they had left, we
counted the number of chips consumed.
Attitudinal ambivalence
We used the measure proposed by Thompson et al. (1995). We thus gauged
positive and negative evaluations of chips separately, which was counterbalanced
across participants. As regards the positive component, we asked participants to
first think about how they evaluate chips by considering only their positive
characteristics and ignoring their negative ones. The question asked How positive
is your evaluation of chips? and the response scale ranged from 1 (in no way
positive) to 4 (very positive). We then asked them to think only about their general
feelings of satisfaction when they see, eat or talk about chips, while ignoring their
feelings of dissatisfaction. The question asked How satisfied do you feel about
chips? and the response scale ranged from 1 (in no way satisfied) to 4 (very
satisfied). Finally, we asked them to think about their thoughts or beliefs about
chips when they see, eat or talk about them, by considering only their beneficial
qualities, while ignoring their harmful qualities. The question asked How beneficial do you think chips are? and the response scale ranged from 1 (in no way
beneficial) to 4 (very beneficial).
As regards the negative component, we followed the reverse procedure, that is,
we asked participants to answer the questions on chips by only considering
negative characteristics, negative feelings and negative thoughts, with responses
ranging from 1 (in no way negative) to 4 (very negative), 1 (in no way
435
dissatisfied) to 4 (very dissatisfied) and 1 (in no way harmful) to 4 (very harmful),

respectively.
We used the equation proposed by Griffin (Thompson et al., 1995) to calculate
ambivalence: ambivalence = [(P + N)/2 - |P - N| + .5)]/4.5. Where P is the average
of the three questions gauging the positive component, N is the average of the
three questions gauging the negative component. The constant .5 is used to avoid
negative values (Thompson et al., 1995). The constant 4.5 is used so that the
ambivalence indicator ranges from 0 (in no way ambivalent) to 1 (highly ambivalent) (see Conner & Sparks, 2002).
Attitude measurement
Participants attitude towards chips was measured through the positive and negative evaluative components of ambivalence. The attitude indicator results from
subtracting the negative evaluative component from the positive evaluative component, thus ranging from -3 (very negative) to +3 (very positive) (Thompson
et al., 1995).
Eating behaviour
In the laboratory setting, there were two plates containing food: one contained 10
apple slices and the other contained 10 Pringles chips. We chose these items of
food because pre-testers evaluated apple slices positively and chips ambivalently
(Batista & Lima, 2010). We chose Pringles because these chips are always of the
same size. The plates of food were placed in such a way that participants could
easily take the apple slices/chips while performing the tasks. The behaviour
indicator consisted of the number of chips participants consumed during the
study (minimum = 0; maximum = 10).
Results and discussion

To test the hypothesis that the relationship between attitude and behaviour is
moderated by attitudinal ambivalence, the number of chips consumed was
regressed
on
attitude,
ambivalence
and
the
interaction
term
1
Attitude Ambivalence . Results showed that behaviour is reliably associated
with the predictors, R2adjusted = .12, F(3, 119) = 6.49, p < .0012. Estimated
parameters can be consulted in Table 1 which indicate that only the interaction
between attitude and ambivalence is significant. As predicted and shown in
Figure 1, simple slopes revealed that for less ambivalent participants (i.e., 1 SD
below the mean), attitude significantly predicts behaviour, that is, participants
with positive attitudes ate more chips one week later, = 4.61, SE = .63, than
participants with negative attitudes, = 2.26, SE = .48; = .34, t = 3.69, p < .001.
In contrast, among participants with more ambivalent attitudes (i.e., 1 SD above
the mean), the relationship between attitude and behaviour was not significant,
that is, there were no significant differences in the number of eaten chips between
436
M.T. Batista et al.
Table 1. The moderating role of attitudinal ambivalence in attitude-behaviour relationships

(Study 1) and attitude-behaviour relationships according to contextual cues (Study 2).
Attitude
Ambivalence
Attitude Ambivalence
Contrast 1
Contrast 2
Attitude Contrast 1
Attitude Contrast 2
Ambivalence Contrast 1
Ambivalence Contrast 2
Attitude Ambivalence Contrast 1
Attitude Ambivalence Contrast 2
Note:
Study 1
Study 2
.15
.02
-.26*
.28**
.06
-.13
.37***
.06
-.03
-.18
.22*
.01
-.14
-.22*
p < .10; * p < .05; ** p < .01; *** p < .001.
Figure 1. Consistency between attitude and behaviour, moderated by attitudinal ambivalence (Study 1).
participants with positive, = 3.48, SE = .68, and negative attitudes, = 3.72,

SE = .72; = -.04, t = -0.22, ns.
In sum, in this study, we analysed the role of attitudinal ambivalence in the
relationship between attitude and behaviour. Supporting our hypotheses, results
show that attitude consistently predicts behaviour only for participants with low
437
attitudinal ambivalence. In contrast, this relationship was not significant among

more ambivalent participants.
Study 2
This study extends the previous one by analysing the moderating role of ambivalence in the attitude-behaviour relationship in different situations. For that purpose
we manipulated social contexts where a confederate was present. Our goal was to
ascertain whether or not different social contexts predict the behaviour of more
ambivalent individuals. Each participant was randomly distributed to one of three
situations which contained different contextual cues regarding (un)healthy eating.
One promoted healthy eating (the confederate only ate apple slices), another
promoted unhealthy eating (the confederate only ate chips), and the third promoted neither (the confederate ate an equal number of apple slices and chips),
which served as our control condition.
We expect that contextual cues should influence eating behaviour of more
ambivalent individuals but not that of less ambivalent individuals. Specifically,
among the less ambivalent participants we expect to replicate Study 1s results. In
other words, we expect attitudes of less ambivalent participants to be good
predictors of behaviours independent of the contexts where those behaviours
take place (Hypothesis 1). Thus, they will eat more chips when their attitudes
are positive and fewer chips when their attitudes towards chips are negative. As
far as more ambivalent participants are concerned, we expect that their eating
behaviour is more influenced by contextual cues than by their attitudes
(Hypothesis 2). Thus, they will behave according to the normative cues present
in the context in that they will eat more chips when unhealthy cues are salient, and
will eat fewer chips when healthy cues are present.
Method
Participants
Ninety-six university students, 31% males and 69% females, aged 1761 years
took part in this study, M = 22.22, SD = 7.32.
Procedure
Procedure was similar to that in Study 1. In phase 1, we gauged participants
ambivalence and attitude toward chips. One week later, the same participants took
part in another laboratory study, where they were asked to eat chips and/or apple
slices while filling out a questionnaire. They were randomly separated into one of
three experimental conditions (control condition vs. healthy cue vs. unhealthy
cue). In the control condition, participants worked together with a partner who ate
all 10 chips and apple slices. In the healthy cue (HC) condition, the partner asked:
Can I eat just apples? They are much healthier, and chips are very fattening. The
experimenter answered: You can eat whatever you want. What you eat is not
438
M.T. Batista et al.
relevant to the study. What matters is that you keep eating while you answer the
questionnaire. In this condition, the partner ate all 10 apple slices but no chips. In
contrast, in the unhealthy cue (UC) condition, the partner would ask: Can I just
eat chips? I am really hungry, and apples arent very filling. The experimenter
answered as in the previous experimental condition, and the partner ate all 10
chips but no apple slices. After completing the task, all of the participants filled
out a manipulation check regarding the manipulation of the social context. We
probed as to whether or not they had associated the two phases of the study, and in
the end we counted the number of food items consumed.
Manipulation check
To check whether or not participants had detected the presence of situational cues,
they answered the following question at the end of the procedure: While participating in this study, the other participant chose to eat, with responses given
using a scale of five points (1 = only apples; 2 = more apples; 3 = as many apples
as chips; 4 = more chips; 5 = only chips).
Attitude, ambivalence and eating behaviour

These variables were measured as in Study 1.
Results and discussion

Manipulation check
We carried out a one-way ANOVA with the perception of the partners behaviour
as the dependent variable and the cues as the between-subjects factor. The results
showed that the effect of contextual cues was significant, F(2, 93) = 206.90,
p < .001, 2p = .82, such that participants in the HC condition indicated that the
partner ate significantly more apple slices, M = 1.20, SD = .62, than participants in
the control group, M = 2.97, SD = .71, t(59) = 10.44, p < .001. Conversely,
participants in the UC condition indicated that the partner ate significantly more
chips, M = 4.71, SD = .75, than participants in the control group, t(64) = -9.70,
p < .001.
Hypotheses testing
We attributed codes to the conditions of the contextual cues variable in order to define
two orthogonal contrasts (see Judd & McClelland, 2001). Contrast 1 (unhealthy
cue = +.5; control = 0; healthy cue = -.5); Contrast 2 (unhealthy cue = -1/3;
control = 2/3; healthy cue = -1/3). In this coding scheme, Contrast 1 compared the
unhealthy cue condition with the healthy cue condition, while Contrast 2 compared
the control condition with both the healthy and unhealthy cue conditions. We then
calculated the two-way and the three-way interaction terms. Finally, we regressed
439
behaviour on attitude, ambivalence, contrast codes and these interaction terms3. The
estimated parameters are shown in Table 1.
Results showed that behaviour was significantly related to predictors,
R2adjusted = .30, F(11, 84) = 4.61, p < .001. The effect of contrast 1 was significant,
showing that participants in the unhealthy condition ate more chips, = 4.69,
SE = .50, than participants in the healthy condition, = 1.61, SE = .544. The effect
of attitude was positive and significant so that the more positive the attitude was, the
more chips the participants ate. A significant (contrast 1 ambivalence) and a
marginally significant (contrast 2 attitude) two-way interaction effects were
obtained (see Table 1). Of greater importance in testing our hypotheses, these
effects were qualified by a three-way interaction involving contrast 2, suggesting
that eating chips was better predicted by the interaction between attitudes, ambivalence and contextual cues. In order to better interpret this interaction, we analysed
the simple effects of the experimental conditions on participants with less ambivalent (i.e, 1 SD below the mean; see Figure 2a) and participants with more ambivalent attitudes (i.e., 1 SD above the mean; see Figure 2b).
As far as participants with less ambivalent attitudes are concerned, as expected
(Hypothesis 1) there were no significant differences among the experimental conditions
both when the participants attitudes were positive, |ts| < 1.54, ns, or negative |ts| < 1.00,
ns. Specifically, participants with positive attitudes (1 SD above the mean) ate more
chips than participants with negative attitudes (1 SD below the mean) in the healthy
condition, = 3.34, SE = .85 vs. = 1.81, SE = .71; = .31, t = 2.08, p < .05, in the
control conditions, = 4.59, SE = 1.64 vs. = 2.14, SE = .85; = .36, t = 1.66, p = .10,
and in the unhealthy conditions, = 5.21, SE = .86 vs. = 1.96, SE = .89; = .47,
Figure 2a. Consistency between attitude and behaviour, moderated by contextual cues
among less ambivalent participants (Study 2).
440
M.T. Batista et al.
Figure 2b. Consistency between attitude and behaviour, moderated by contextual cues
among more ambivalent participants (Study 2).
t = 3.29, p < .001. We also compared the different experimental conditions within
positive and negative attitudes separately. For participants with a positive non-ambivalent attitude, eating chips in the unhealthy condition, = 5.21, SE = .86, did not
significantly differ from the control, = 4.59, SE = 1.65; = .08, t = .32, ns, or the
healthy conditions, = 3.35, SE = .85; = .25, t = 1.54, ns. Also, the difference between
the control and the healthy conditions was not significant, = -.17, t = -.68, ns. In a
similar way, for participants with a negative non-ambivalent attitude, eating chips in the
unhealthy condition, = 1.96, SE = .89, was not significantly different from the control,
= 2.14, SE = .85; = -.03, t = -.15, ns, or the healthy conditions, = 1.81, SE = .71;
= .10, t = .56, ns. The difference between the control and the healthy conditions was
also not significant, = -.12, t = -.76, ns. Thus, these results mean that the contextual
cues did not influence the behaviour of less ambivalent participants.
As our Hypothesis 2 predicted, we found a different pattern among participants with more ambivalent attitudes (see Figure 2b). Specifically, among these
participants there was a clear influence of the cue conditions on eating chips
behaviour. In fact, more ambivalent participants who expressed a positive attitude
(+1 SD) ate more chips in the unhealthy condition, = 6.77, SE = 1.03, than
participants in both the control, = 2.66, SE = 1.23; = .56, t = 2.57, p < .05, and
in the healthy conditions, = 2.96, SE = 1.42; = .52, t = 2.18, p < .05.
Moreover, the difference in eating chips between the control and the healthy
conditions was not significant, = -.04, t = -.16, ns. Regarding more ambivalent
participants who expressed a negative attitude (-1 SD), they ate fewer chips in the
healthy condition, = .01, SE = .97, than participants in both the control,
= 4.96, SE = 1.16; = -.75, t = -3.96, p < .001, and the unhealthy conditions,
441
= 4.82, SE = 1.38; = -.59, t = -3.39, p < .001. The difference in eating chips
between the control and the unhealthy conditions was not significant, = .02,
t = .07, ns. Therefore, the amount of chips eaten by more ambivalent participants
significantly varied according to the contextual cues. In analysing these effects
from another perspective, we found that among more ambivalent participants the
difference in eating chips between the ones who expressed negative and those
who expressed positive attitudes was not significant either in the unhealthy
condition, = .28, t = .98, ns, nor in the healthy condition, = .59, t = 1.79,
p < .08. Importantly, neither was this difference significant in the control condition, = -.33, t = -1.22, ns, which replicates the results of Study 1 and shows that
attitudes did not predict the behaviour of ambivalent participants.
In sum, in this study we analysed the role of attitudinal ambivalence on the
influence of social context regarding eating behaviour. Results support our hypothesis that attitudes are better predictors of behaviour of less ambivalent participants
than of more ambivalent participants. In other words, the attitudes of less ambivalent participants predict their behaviour in all contexts. In contrast, contextual cues
influence the eating behaviour of more ambivalent participants. In fact, their
ingestion of chips increases when the contextual cues highlight unhealthy eating
and it decreases when the contextual cues highlight healthy eating.
General discussion
We presented a research programme addressing two aspects in which ambivalence
could play a central role in predicting behaviours: the attitude-behaviour relationship, and the influence of the social context on actual behaviour. In Study 1, we
tested the moderating role of ambivalence in the attitude-behaviour relationship.
As expected, results show that ambivalence moderates the consistency of the
attitude-behaviour relationship. As in Armitage and Conner (2000), Conner
et al. (2003) and Conner et al. (2002), whereas among less ambivalent participants
their attitudes predict behaviour, among highly ambivalent individuals they do
not. Our results are relevant because they show that these patterns are not limited
to self-reported behaviour, as found in previous research (e.g., Berndsen & van
der Pligt, 2004; Povey et al., 2001; Shepherd, 1999), and can be thus extended to
actual behaviour. After showing that attitudes can predict the behaviour of less
ambivalent individuals, we then sought to identify predictors of more ambivalent
individuals behaviour. In this way, Study 2 indicates that for less ambivalent
participants attitudes are again good predictors of behaviour, regardless of the
social context in which they occur. Maybe the most innovative aspect of the
present set of studies concerns the fact that it shows that the influence of the
contextual cues occurs only in more ambivalent individuals. Specifically, participants higher in ambivalence showed an increase or a decrease in chips ingestion,
according to the situational cues. These results are in accordance with research
indicating a higher sensitivity of ambivalent attitudes to the context (e.g., Hodson
et al., 2001; Lavine, 2001; Lavine et al., 1998; Tourangeau et al., 1989). More
importantly, these results are in accordance with theorizing which predicted that
442
M.T. Batista et al.
ambivalence can be the core of attitude-behaviour inconsistency (Eagly &

Chaiken, 1993; Jonas et al., 2000). Our results also suggest that ambivalent
attitudes may be more likely to amplify behaviours in the direction of the
information present in the social context (see also Briol, Petty, & Wheeler,
2006; Clark, Wegener, & Fabrigar, 2008).
The relevance of our results is not restricted to a theoretical point of view. In
fact, our results not only contribute to a better understanding of both eating
behaviour but also to how it may be influenced through interventions.
Specifically, in line with previous research (e.g., Clendenen, Herman, & Polivy,
1994; de Castro & Brewer, 1992; Herman & Polivy, 2005; Herman, Roth, & Polivy,
2003; Pliner, Bell, Hirsch, & Kinchla, 2006; Vartanian, Herman, & Wansink, 2008),
Study 2 shows that eating behaviours may be influenced by subtle messages in the
contexts where people eat. Thus, the contexts in which eating behaviours take place
(e.g., school or company cafeterias) should contain cues that make salient the
negative dimension associated with unhealthy food, and/or the benefits of healthy
food. For instance, the decoration of the cafeteria could show images of healthy
food associated with healthy people. These contexts may thus contribute to decreasing the ingestion of unhealthy food and to increasing the ingestion of healthy food.
Nevertheless, we should note that the effect of the context on eating behaviour is
limited to individuals with ambivalent attitudes. Regarding non ambivalent individuals who have positive attitudes towards unhealthy food, the strategy should focus
on attitude change. Success in transforming a non-ambivalent positive attitude
towards unhealthy food into a non-ambivalent negative one should lead to behaviour change in the desired direction (Erber et al., 1995).
Despite the relevance and consistency of our findings, we would like to point
out some limitations that should be addressed in future research. First, in our
studies we did not compare the effect of neutral versus ambivalent attitudes. This
limitation derives from the fact that it is very difficult to find food-stimuli towards
which individuals have purely neutral attitudes without some associated ambivalence. In fact, within a list comprising 256 items of food, Batista and Lima (2010)
were not able to find non-ambivalent neutral (or non-ambivalent negative) ones.
Second, we used a global index of ambivalence. Thus we could not test whether
or not ambivalence towards food results from the conflict between health and
pleasure. Future research should directly test the effects of inter-component
ambivalence, that is, the conflict between cognitions (health) and affect (taste).
Third, it may be possible that the fact that participants could choose apple slices
increased ambivalence towards chips. A way to address this matter is to replicate
these studies by adding a condition in which no apple slices are present. Also, as
suggested by Thompson et al. (1995), in both studies we used positive and
negative judgments to compute both ambivalence and attitudes. Despite the fact
that these measures are only weakly correlated in our studies, possibly a better
way to measure attitude would have been through a bipolar scale. Finally we
would like to indicate further suggestions for future studies. It would be most
interesting to replicate our studies in order to verify whether these effects generalize to other ambivalent attitudinal objects, such as environmental behaviours. If
443
our findings apply to other domains, the context where people retrieve their
ambivalent attitudes can be crucial for increasing these behaviours (and possibly
other citizenship behaviours). Also, we suggest a replication of our studies using a
longer time interval between the measurement of attitudes and the observation of
behaviour. Although Armitage and Conner (2000) used a five-/eight-month interval and found consistency, given the changes we introduced it is important to test
whether or not our findings are observed with a longer time interval. Furthermore,
in future studies it could also be interesting to ascertain the role of attitudinal
ambivalence and social norms on behaviour. According to our results and those of
Hodson et al. (2001), social norms will exert a higher influence on the behaviour
of more ambivalent individuals. Furthermore, research would benefit from including other layers of the population (i.e., besides college students) for whom norms
regulating eating behaviours may be different from those that the literature has
addressed (Batista & Lima, 2013). Furthermore, these studies should be replicated
controlling for several dimensions of attitude strength that are often confounded
with ambivalence (e.g., low certainty and accessibility). Finally, it would be very
relevant to add an attitude measure at time 2 which would allow us to test whether
(a) attitudes are only stable among less ambivalent individuals and (b) ambivalent
attitudes measured one week before do not predict behaviour because they are
unstable although ambivalent attitudes measured at the time of behaviour may
guide it. Specifically, among more ambivalent individuals, it may be the case that
the context influences the attitude which in turn will guide behaviour.
Despite these limitations, the current research brings a unique contribution to
the literature on the relationship between attitude and behaviour by showing the
importance of social contexts in predicting actual behaviour of individuals with
ambivalent attitudes. The main message of this research is that attitudes are only
good predictors of behaviour when they are less ambivalent. Different processes
occur in more ambivalent attitudes in that individuals with such attitudes are
sensitive to contextual cues which influence their behaviour accordingly.
Notes
1.
2.
3.
4.
Attitude and ambivalence were only weakly correlated, r(123) = .27, p = .002.
Due to the fact that the two evaluative items that make our attitude measure are also
elements of the P and N parts of the ambivalence formula we computed a different
composite measure of attitudes by reversing the negative item and combining it with
the positive one on a single composite measure. This measure could range between 1
(negative attitude) and 4 (positive attitude). The regression results did not change.
As in Study 1, attitude and ambivalence were only weakly correlated, r(127) = .19,
p = .029.
Simple effects also suggested that participants in the healthy condition ate fewer
chips than participants in the control condition, = 3.59, SE = .58; = -.27, t =
-2.50, p < .05. There was no significant difference between unhealthy and control
conditions, = .15, t = 1.44, ns.
444
M.T. Batista et al.
Me encantan las patatas fritas, las como y (tambin) las odio: el

papel de la ambivalencia y de las claves contextuales en el
La posibilidad de predecir el comportamiento segn las actitudes ha suscitado
durante dcadas gran inters entre los psiclogos sociales. Los primeros estudios
mostraron que la relacin entre actitudes y comportamiento era inconsistente (p.
ej., Kutner, Wilkins, y Yarrow, 1952; LaPiere, 1934; Wicker, 1969). En posteriores investigaciones se intent identificar variables que inciden en que una
actitud tenga ms probabilidad de predecir comportamientos (p. ej., Fazio y
Towles-Schwen, 1999; Holland, Verplanken, y Knippenberg, 2002). Entre estas
variables la intensidad de la actitud se identific como un moderador importante
(p ej., Holland et al., 2002).
La ambivalencia actitudinal ha sido identificada como una dimensin de la
intensidad de la actitud, de tal forma que a mayor ambivalencia menor intensidad
de la actitud (p. ej., Conner y Sparks, 2002; Thompson, Zanna, y Griffin, 1995).
Erber, Hodges, y Wilson (1995) sugirieron que las actitudes ambivalentes podran
estar muy influenciadas por los contextos en los que aparecen. Por esta razn,
propusieron que la ambivalencia puede desempear un papel relevante en la
consistencia de la relacin entre actitudes y comportamientos. Sin embargo, las
conclusiones respecto a la baja predictibilidad del comportamiento correspondiente a las actitudes ambivalentes se basan en investigaciones que emplearon
medidas de auto-informe y no del comportamiento observado (p. ej., Armitage y
Conner, 2000). Aunque hay pruebas que muestran que las actitudes ambivalentes
son ms sensibles a la influencia social (p. ej., Hodson, Maio, y Esses, 2001), no
hemos podido encontrar estudios relativos a los efectos conjuntos sobre el comportamiento de la ambivalencia actitudinal y los contextos sociales.
Este artculo tiene como objetivo mostrar la relevancia tanto de la ambivalencia actitudinal como de los contextos sociales en la relacin entre actitudes y
comportamientos. Concretamente, en dos estudios de diseo longitudinal y en los
que se observaban comportamientos reales, comprobamos que las actitudes ms
ambivalentes son peores predictores del comportamiento y son ms sensibles a la
influencia social indirecta que las menos ambivalentes. Esto se hizo observando
comportamientos reales y no slo con medidas de auto-informe como ha sido
frecuente en la investigacin previa (Dormandy, Hankins, y Marteau, 2006). Esta
estrategia nos permitir analizar si el contexto social tiene un papel relevante en la
prediccin de comportamientos cuando las actitudes son ambivalentes. La
obtencin de datos empricos que sustenten esta afirmacin, contribuir al
desarrollo de estrategias para el fomento de una alimentacin ms saludable.
445
Ambivalencia actitudinal
Segn un enfoque unidimensional, las actitudes slo pueden ser evaluaciones
positivas, neutras o negativas (Eagly y Chaiken, 1993). En cambio, la visin
bidimensional de las actitudes permite la existencia de dos dimensiones independientes de evaluacin: positiva y negativa (Bell, Esses, y Maio, 1996; Kaplan,
1972; Maio, Esses, y Bell, 2000; Thompson et al., 1995). Este punto de vista es
til para distinguir entre las actitudes neutras (de indiferencia) y las ambivalentes
(en conflicto).
De hecho, al adoptar una visin que entiende las actitudes como constructos
bidimensionales, la perspectiva de la ambivalencia actitudinal supone la existencia
simultnea de evaluaciones positivas y negativas del mismo objeto (Kaplan, 1972;
Scott, 1968; Thompson et al., 1995). Segn Eagly y Chaiken (1993), la
ambivalencia puede dar lugar a una dbil consistencia de la relacin entre
actitudes y comportamientos debido a que las actitudes ambivalentes son ms
sensibles a las claves contextuales que las que no lo son. Jonas, Broemer, and
Diehl (2000) apoyan esta explicacin centrndose en el efecto del tipo de
informacin recibida. Concretamente, Jonas et al. (2000) sostienen que cuando
las actitudes son ambivalentes los individuos encuentran ms caractersticas
positivas o negativas segn las claves situacionales del momento.
Sin embargo, todava no se ha mostrado claramente cmo incide la
ambivalencia actitudinal en la relacin actitud-comportamiento. Por una parte, la
investigacin previa sugiere que las actitudes menos ambivalentes son mejores
predictores del comportamiento (p. ej., Armitage y Conner, 2000; Berndsen y van
der Pligt, 2004; Conner, Povey, Sparks, James, y Shepherd, 2003; Conner et al.,
2002; Povey, Wellens, y Conner, 2001; Shepherd, 1999; ver tambin Conner y
Sparks, 2002, para una revisin). Por otro lado, los tipos de mediciones del
comportamiento que se usan la mayora de las veces (auto-informes y en ocasiones una medida de la intencin de comportamiento) nos impiden establecer
conclusiones slidas. Adems, incluso las investigaciones que miden comportamientos reales (Conner, Godin, Sheeran, y Germain, 2013; Dormandy et al.,
2006) no han considerado hasta el momento los efectos conjuntos del contexto
social y la ambivalencia actitudinal en la relacin actitud-comportamiento.
Respecto a la influencia del contexto social en el comportamiento real de las
personas, debemos distinguir entre la influencia directa y la indirecta. Cuando la
influencia social es indirecta, puede que la fuente no trate de persuadir al
individuo de que cambie sus actitudes o comportamientos. Por ejemplo, la mera
presencia de otras personas ejerce una influencia sobre los individuos (que stos
probablemente no perciben) y que cambia sus actitudes o comportamientos. Hay
algunas evidencias que apuntan a que este efecto es tal vez ms perceptible en los
individuos con actitudes ambivalentes (Lavine, 2001; Lavine, Huff, Wagner, y
Sweeney 1998; Tourangeau, Rasinski, Bradburn, y DAndrade, 1989). Sin
embargo, en la literatura correspondiente no se usaron medidas de comportamientos reales.
Cuando la influencia social es directa, las fuentes pretenden persuadir al
individuo para que cambie una actitud o escoja un comportamiento particular.
446
M.T. Batista et al.
Las investigaciones han mostrado que se convence ms fcilmente a los individuos con actitudes ms ambivalentes, es decir, cambian ms fcil y rpidamente
su actitud siguiendo el mensaje persuasivo (p. ej., Armitage y Conner, 2000; Bell
y Esses, 1997; Broemer, 2002; Linville y Jones, 1980; Maio, Bell, y Esses, 1996).
Descripcin de los estudios

Analizamos el rol de la ambivalencia actitudinal en relacin con el comportamiento real en dos estudios de diseo longitudinal. Para abordar de manera
emprica esta cuestin optamos por el uso de observaciones secuenciadas en el
tiempo y de medidas de comportamiento. Esta estrategia nos permiti registrar
qu decan haber hecho o pensaban hacer en el futuro los participantes, pero
tambin aquello que hicieron realmente.
Presentamos comida no saludable (concretamente, patatas fritas de bolsa)
como objeto de actitud por dos motivos. En primer lugar, es probable que este
tipo de comida active a la vez las dimensiones positivas y negativas de una actitud
(Batista y Lima, 2010; Urland y Ito, 2005; ver tambin Beardsworth, 1995;
Conner y Sparks, 2002; Mischel, Shoda, y Rodriguez, 1989; Shepherd, 1999).
Por ejemplo, las patatas fritas saben bien pero engordan. Esto supone un conflicto
entre placer y salud (Conner y Armitage, 2002; Conner y Sparks, 2002) lo que
constituye un tema socialmente relevante. De hecho, las investigaciones indican
que los individuos ambivalentes muestran ms preocupaciones en relacin con la
comida (Stroebe, Mensik, Aarts, Schut, y Kruglanski, 2008) y que la ambivalencia concierne principalmente a las comidas sabrosas pero no saludables (Urland y
Ito, 2005). Sin embargo estas investigaciones por lo general no han estudiado el
efecto de esas variables en el propio comportamiento. Adems, incluso en
aquellos estudios en los que s se meda el comportamiento no se consideraba la
ambivalencia actitudinal. Por ejemplo, Hofmann, Rauch y Gawronski (2007)
estudiaron el rol de las actitudes y los recursos de auto-regulacin en el comportamiento alimentario (vase tambin Hofmann, Friese, y Roefs, 2009) y Burger
et al. (2010) estudiaron la relacin entre las normas descriptivas y las elecciones
de comida (ver tambin Pliner y Mann, 2004).
En segundo lugar, el uso de comida no saludable en estos estudios ha
proporcionado una manera directa y significativa de examinar la influencia de
las claves situacionales y la ambivalencia actitudinal en el comportamiento. De
hecho, los comportamientos alimentarios tienen lugar frecuentemente en contextos sociales y a menudo se usan para compartir o celebrar algo socialmente
(Conner y Armitage, 2002; Ogden, 2003). En este tipo de situaciones, los
comportamientos alimentarios de los individuos estn principalmente bajo la
influencia social indirecta. Y, sin embargo, como ya hemos indicado, la literatura
sobre la ambivalencia actitudinal no ha prestado atencin a este tipo de influencia
social sobre los comportamientos.
En el Estudio 1 analizamos si la ambivalencia actitudinal modera la relacin
actitud-comportamiento. Para ello, no medimos simplemente la intencin de
comportamiento, sino las actitudes y el comportamiento real observado. Como
447
ampliacin de este estudio, en el Estudio 2 comprobamos experimentalmente el

rol moderador del contexto social en la relacin anterior. Nuestra prediccin era
que las actitudes ms ambivalentes (en comparacin con las menos ambivalentes)
favorecen la inconsistencia en la relacin actitud-comportamiento y hacen que el
comportamiento de las personas sea ms sensible a la influencia social indirecta.
Estudio 1
En este estudio pusimos a prueba la hiptesis de que la ambivalencia actitudinal
modera la consistencia de la relacin actitud-comportamiento. Basndonos en la
literatura antes mencionada, formulamos la hiptesis de que para los participantes
ms ambivalentes no existe una relacin entre la actitud hacia las patatas fritas
medida en un momento concreto y el comportamiento de comer patatas fritas que
tiene lugar una semana despus. Y, a la inversa, esperamos que las actitudes sean
buenos predictores del comportamiento en los participantes menos ambivalentes.
Mtodo
Participantes
En este estudio participaron 123 estudiantes universitarios, 34% hombres y 66%
mujeres, con edades comprendidas entre los 17 y los 46 aos, M = 21.47,
DT = 3.78.
Procedimiento
El estudio se llev a cabo en dos fases. En la primera fase medimos la actitud y
ambivalencia de los participantes hacia las patatas fritas. Una semana ms tarde
los mismos participantes fueron al laboratorio para participar de forma individual
en un estudio que se present como una investigacin sobre la influencia del
hambre en la satisfaccin vital. Para que fuera creble y justificar la presencia de
comida, se dijo a los participantes que en el estudio haba dos condiciones. Los
participantes de una condicin tenan permiso para comer mientras realizaban las
tareas y los de la otra no. Un sorteo manipulado agrup a todos los participantes
en la condicin con permiso para comer. Repartimos un cuestionario sobre
satisfaccin vital entre los participantes y les ofrecimos dos tipos de comida
(patatas Pringles y trozos de manzana) que podan comer mientras respondan.
Para que los participantes pudieran elegir, tal y como era necesario segn nuestra
manipulacin experimental, se requera la presencia de los dos tipos de alimentos.
Antes de que se marcharan, les sondeamos para ver si haban asociado las dos
fases del estudio. Una vez que se fueron, contamos cuntas patatas fritas se haban
comido.
448
M.T. Batista et al.
Ambivalencia actitudinal
Utilizamos la medida propuesta por Thompson et al. (1995). Medimos las evaluaciones positivas y negativas de las patatas fritas por separado, de manera
compensada para todos los participantes. Respecto al componente positivo,
pedimos a los participantes que pensasen primero en cmo evaluaran las patatas
teniendo en cuenta slo sus caractersticas positivas y dejando a un margen las
negativas. La pregunta era: Cun positiva es su evaluacin de las patatas fritas?
y la escala de respuestas iba del 1 (en ningn modo positiva) al 4 (muy positiva).
Luego les pedimos que pensaran nicamente en sus sentimientos generales de
satisfaccin cuando ven, comen o hablan de patatas fritas y dejaran al margen los
sentimientos de insatisfaccin. La pregunta era: Cunto te satisfacen las patatas
fritas? y la escala de respuestas iba del 1 (en ningn modo satisfecho) al 4 (muy
satisfecho). Finalmente, les pedimos que reflexionaran sobre sus pensamientos o
creencias sobre las patatas fritas cuando las ven, las comen o hablan de ellas,
teniendo en cuenta slo las cualidades beneficiosas y dejando una vez ms al
margen las cualidades perjudiciales. La pregunta era: En qu medida cree que
son beneficiosas las patatas fritas? y las respuestas iban desde 1 (no son beneficiosas de ninguna manera) hasta 4 (muy beneficiosas).
En cuanto al componente negativo, invertimos el procedimiento, esto es,
pedimos a los participantes que respondieran a las preguntas sobre las patatas
fritas teniendo en cuenta slo las caractersticas, los sentimientos y los pensamientos negativos; las respuestas iban desde 1 (en absoluto negativo) hasta 4
(muy negativo), 1 (nada insatisfecho) hasta 4 (muy insatisfecho) y 1 (de ninguna
manera perjudiciales) hasta 4 (muy perjudiciales), respectivamente.
Usamos la ecuacin propuesta por Griffin (Thompson et al., 1995) para
calcular la ambivalencia: Ambivalencia = [(P + N)/2 - |P - N| + .5)]/4.5 . Donde
P es la media de las tres preguntas que medan el componente positivo y N la
media de las tres preguntas que medan el componente negativo. Se usa la
constante.5 para evitar los valores negativos (Thompson et al., 1995).
La constante 4.5 se usa para que el indicador de ambivalencia vare desde 0
(ninguna ambivalencia) hasta 1 (muy ambivalente) (ver Conner y Sparks, 2002).
Medicin de la actitud
La actitud de los participantes hacia las patatas fritas se midi a travs de los
componentes evaluadores positivos y negativos de la ambivalencia. El indicador
de actitud resulta de restar el componente evaluador negativo al componente
evaluador positivo, variando de esta manera desde -3 (muy negativo) hasta +3
(muy positivo) (Thompson et al., 1995).
Comportamiento alimentario
En el laboratorio haba dos platos de comida: uno con 10 trozos de manzana y
otro con 10 patatas Pringles. Elegimos estos alimentos porque los participantes de
los pre-tests evaluaban las patatas fritas de forma ambivalente y la manzana de
449
forma positiva (Batista y Lima, 2010). Escogimos las Pringles porque estas
patatas son todas del mismo tamao. Los platos de comida se dispusieron de tal
modo que los participantes pudieran coger fcilmente los trozos de manzana o las
patatas mientras realizaban las tareas. El indicador de comportamiento era el
nmero de patatas fritas que coman los participantes durante el estudio
(mnimo = 0; mximo = 10).
Resultados y discusin
Para poner a prueba la hiptesis de que la relacin entre actitud y comportamiento
est moderada por la ambivalencia actitudinal, se hizo un anlisis de regresin del
nmero de patatas consumidas con respecto a la actitud, la ambivalencia y el
trmino de interaccin actitud ambivalencia1. Los resultados mostraron que el
comportamiento est asociado de manera consistente con los predictores,
R2ajustado = .12, F(3, 119) = 6.49, p < .0012. Los parmetros estimados pueden
consultarse en la Tabla 1 que indica que slo la interaccin entre actitud y
ambivalencia es significativa. Como se predijo y se muestra en la Figura 1, las
pendientes de regresin muestran que en el caso de los participantes menos
ambivalentes (1 DT por debajo de la media), la actitud predice el comportamiento
de manera significativa, es decir, los participantes con actitudes positivas
comieron ms patatas una semana despus, = 4.61, ET = .63, que los participantes con actitudes negativas, = 2.26, ET = .48; = .34, t = 3.69, p < .001. Por
el contrario, en el caso de los participantes con actitudes ms ambivalentes (1 DT
por encima de la media), la relacin entre actitud y comportamiento no era
significativa, es decir, no haba diferencias significativas en el nmero de patatas
Tabla 1. El rol moderador de la ambivalencia actitudinal en la relacin actitud-comportamiento (Estudio 1) y la relacin actitud-comportamiento segn las claves contextuales
(Estudio 2).
Actitud
Ambivalencia
Actitud Ambivalencia
Contraste 1
Constraste 2
Actitud Contraste 1
Actitud Contraste 2
Ambivalencia Contraste 1
Ambivalencia Contraste 2
Actitud Ambivalencia Contraste 1
Actitud Ambivalencia Contraste 2
Nota:
p < .10; * p < .05; ** p < .01; *** p < .001.
Estudio 1
Estudio 2
.15
.02
-.26*
.28**
.06
-.13
.37***
.06
-.03
-.18
.22*
.01
-.14
-.22*
450
M.T. Batista et al.
Figura 1. Consistencia entre actitud y comportamiento, moderada por la ambivalencia

actitudinal (Estudio 1).
Nmero previsto de patatas fritas. Menor ambivalencia (1 DT). Mayor ambivalencia
(+1 DT). Actitud Negativa (-1 DT). Actitud Positiva (+1 DT).
fritas consumidas entre los participantes con actitudes positivas, = 3.48,

ET = .68, y actitudes negativas, = 3.72, ET = .72; = -.04, t = -.22, ns.
En resumen, en este estudio analizamos el papel de la ambivalencia actitudinal
en la relacin entre actitud y comportamiento. Los resultados corroboran nuestra
hiptesis y muestran que la actitud predice el comportamiento de manera consistente slo en el caso de los participantes con baja ambivalencia actitudinal. Por
el contrario, esta relacin no era significativa entre los participantes ms
ambivalentes.
Estudio 2
Este estudio ampla el anterior analizando el papel moderador de la ambivalencia
en la relacin actitud-comportamiento en diferentes situaciones. Con este
propsito, manipulamos algunos contextos sociales contando con la presencia
de un cmplice. Nuestro objetivo era determinar si los diferentes contextos
sociales predicen el comportamiento de los individuos ms ambivalentes. Todos
los participantes fueron distribuidos al azar y asignados a una de tres situaciones
posibles en las que haba distintas claves de contexto relacionadas con la
alimentacin saludable y no saludable. En una se promocionaba la alimentacin
sana (el cmplice slo coma trozos de manzana), en otra la alimentacin no
saludable (el cmplice slo coma patatas fritas), y en la tercera, que serva como
451
control, ninguna de las dos (el cmplice coma el mismo nmero de trozos de
manzana que de patatas fritas).
Esperamos que las claves de contexto influyan el comportamiento alimentario
de los individuos ms ambivalentes y no el de aquellos que lo son menos.
Concretamente, esperamos que se repitan los resultados del estudio 1para los
participantes menos ambivalentes. En otras palabras, esperamos que las actitudes
de los participantes menos ambivalentes sean buenos predictores de los comportamientos independientemente de los contextos en los que se producen esos
comportamientos (Hiptesis 1). Segn esto, comern ms patatas fritas cuando
sus actitudes respecto a las mismas sean positivas y menos cuando esas actitudes
sean negativas. En cuanto a los participantes ms ambivalentes, esperamos que su
comportamiento alimentario est ms influenciado por las claves de contexto que
por sus actitudes (Hiptesis 2). De este modo, se comportarn de acuerdo con
claves normativas presentes en el contexto en tanto en cuanto comern ms
patatas cuando las claves no saludables estn ms presentes y menos cuando las
claves ms presentes sean las saludables.
Mtodo
Participantes
En este estudio participaron 96 estudiantes universitarios, 31% hombres y 69%
mujeres, con edades entre los 17 y 61 aos, M = 22.22, DT = 7.32.
Procedimiento
El procedimiento fue similar al empleado en el Estudio 1. En la primera fase
medimos la ambivalencia y la actitud de los participantes respecto a las patatas
fritas. Una semana despus, las mismas personas participaron en un estudio de
laboratorio, en el que se les pidi que comieran patatas fritas o trozos de manzana
mientras rellenaban un cuestionario. Se les dividi de manera aleatoria en tres
grupos experimentales (grupo de control, clave saludable, clave no saludable). En
la condicin de control, los participantes trabajaban con un compaero que se
coma todas las patatas y todos los trozos de manzana. En el grupo de la clave
saludable (CS), el compaero preguntaba: Puedo comer slo manzanas? Son
mucho ms saludables, y las patatas engordan mucho. El encargado del experimento responda: Puedes comer lo que quieras. Lo que comas no es relevante
para el estudio. Lo que importa es que comas mientras que contestas el cuestionario. En este grupo el compaero se coma los 10 trozos de manzana pero no
coma ninguna patata frita. En cambio, en la condicin clave no saludable (CNS),
el compaero preguntaba: Puedo comer slo patatas? Tengo mucha hambre, y
las manzanas no llenan mucho. El encargado del experimento responda lo
mismo que en la condicin experimental anterior y el compaero se coma
las10 patatas fritas pero ningn trozo de manzana. Despus de finalizar las tareas,
todos los participantes contestaron a una pregunta de comprobacin en relacin
con la manipulacin del contexto social. Les sondeamos para saber si haban
452
M.T. Batista et al.
relacionado o no las fases del estudio, y por ltimo contamos las patatas y los
trozos de manzana consumidos.
Comprobacin de la manipulacin
Para comprobar si los participantes haban detectado la presencia de claves de
situacin, contestaron a la siguiente pregunta al final del procedimiento: Durante
su participacin en el estudio, el otro participante opt por comer..., y las
respuestas posibles tenan una escala de cinco puntos (1 = slo manzana;
2 = ms manzana; 3 = la misma cantidad de manzana que de patatas fritas;
4 = ms patatas fritas; 5 = slo patatas fritas).
Actitud, ambivalencia y comportamiento alimentario

Estas variables se midieron del mismo modo que en el Estudio 1.
Resultados y Discusin
Comprobacin de la manipulacin
Realizamos un ANOVA unidireccional con la percepcin del comportamiento del
compaero como variable dependiente y las claves como factor inter-sujetos. Los
resultados mostraron que el efecto de las claves contextuales era significativo,
F(2, 93) = 206.90, p < .001, 2p = .82. Los participantes de la condicin CS
indicaron que el compaero comi una cantidad significativamente superior de
manzana, M = 1.20, DT = .62, en comparacin con lo que dijeron los participantes
del grupo de control, M = 2.97, DT = .71, t(59) = 10.44, p < .001. A la inversa, los
participantes de la condicin CNS indicaron que el compaero comi una cantidad significativamente superior de patatas fritas, M = 4.71, DT = .75, en
comparacin con lo que dijeron los participantes del grupo de control, t(64) =
-9.70, p < .001.
Comprobacin de las hiptesis

Asignamos cdigos a las condiciones de las claves contextuales para definir dos
contrastes ortogonales (ver Judd y McClelland, 2001). Contraste 1 (clave no
saludable = +.5; control = 0; clave saludable = -0.5); Contraste 2 (clave no
saludable = -1/3; control = 2/3; clave saludable = -1/3). Con estos cdigos, el
Contraste 1 comparaba la condicin clave no saludable con la condicin clave
saludable, y el Contraste 2 comparaba la condicin de control con las otras dos
condiciones. A continuacin calculamos los trminos de interaccin bi-direcionales y tri-direccionales. Finalmente, llevamos a cabo un anlisis de regresin del
comportamiento respecto a la actitud, la ambivalencia, los cdigos de contraste y
estos trminos de interaccin3. Los parmetros estimados se muestran en la
Tabla 1.
453
Los resultados muestran que el comportamiento estaba relacionado significativamente con los predictores, R2ajustada = .30, F(11, 84) = 4.61, p < .001. El efecto
del contraste 1 era significativo, y mostraba que los participantes en la condicin
no saludable comieron ms patatas fritas, = 4.69, ET = .50, que los participantes
en la condicin saludable, = 1.61, ET = .544. El efecto de la actitud fue positivo
y significativo; cuanto ms positiva era la actitud, ms patatas fritas coman los
participantes. Se obtuvieron los efectos de una interaccin bi-direccional, una
significativa (contraste 1 ambivalencia) y otra parcialmente significativa (contraste 2 actitud) (ver Tabla 1). Y lo que es ms importante para comprobar
nuestras hiptesis, estos efectos se matizaron mediante una interaccin tri-direccional en la que se usaba el contraste 2, y que sugera que la interaccin entre
actitudes, ambivalencia y claves de contexto predicen mejor el comportamiento de
comer patatas fritas. Para interpretar mejor esta interaccin, analizamos los efectos
simples de las condiciones experimentales en los participantes con actitudes
menos (1 DT por debajo de la media, ver Figura 2a) y ms (1 DT por encima
de la media, ver Figura 2b) ambivalentes.
En lo que se refiere a los participantes con actitudes menos ambivalentes, y
como esperbamos (Hiptesis 1) no hubo diferencias significativas entre las
condiciones experimentales ni cuando las actitudes eran positivas |ts| < 1.54, ns,
ni cuando eran negativas |ts| < 1.00, ns. En concreto, los participantes con
actitudes positivas (1 DT por encima de la media) comieron ms patatas fritas
que los participantes con actitudes negativas (1 DT por debajo de la media) en la
Figura 2a. Consistencia entre actitud y comportamiento, moderada por las claves contextuales para los participantes menos ambivalentes (Estudio 2).
Nmero previsto de patatas fritas. Figura 2a Participantes menos ambivalentes (-1 DT).
Saludable. Control. No saludable. Actitud Negativa (-1 DT). Actitud Positiva (+1 DT).
454
M.T. Batista et al.
Figura 2b. Consistencia entre actitud y comportamiento, moderada por las claves contextuales para los participantes ms ambivalentes (Estudio 2).
Nmero previsto de patatas fritas. Figura 2b. Participantes ms ambivalentes (+1 DT).
Saludable. Control. No saludable. Actitud Negativa (1 DT). Actitud Positiva (+1 DT).
condicin saludable = 3.34, ET = .85 vs. = 1.81, ET = .71; = .31, t = 2.08,

p < .05, en la de control, = 4.59, ET = 1.64 vs. = 2.14, ET = .85; = .36,
t = 1.66, p = .10, y en la no saludable, = 5.21, ET = .86 vs. = 1.96, ET = .89;
= .47, t = 3.29, p < .001. Tambin comparamos las diferentes condiciones
experimentales para los individuos de actitudes positivas y negativas por
separado. En el caso de los participantes con una actitud positiva no-ambivalente,
la cantidad de patatas fritas consumidas en la condicin no saludable, = 5.21,
ET = .86, no difera significativamente de las cantidades en la de control,
= 4.59, ET = 1.65; = .08, t = .32, ns, o en la condicin saludable, = 3.35,
ET = .85; = .25, t = 1.54, ns. Adems, la diferencia entre las condiciones control
y saludable no fue significativa, = -.17, t = -.68, ns. Asimismo, para los
participantes con una actitud negativa no ambivalente, la cantidad de patatas
consumidas en la condicin no saludable, = 1.96, ET = .89, no difera significativamente a las de la condicin control, = 2.14, ET = .85; = -.03, t = -.15,
ns, o las del grupo saludable, = 1.81, ET = .71; = .10, t = .56, ns. La diferencia
entre la condicin control y la saludable tampoco era significativa, = -.12, t =
-0.76, ns. De esta forma, los resultados muestran que las claves contextuales no
influyeron en el comportamiento de los participantes menos ambivalentes.
Como predeca nuestra Hiptesis 2, encontramos un patrn diferente para los
participantes con actitudes ms ambivalentes (vase Figura 2b). Concretamente,
en estos participantes las condiciones de claves influan claramente en el comportamiento de comer patatas fritas. De hecho, los participantes ms ambivalentes
455
que expresaron una actitud positiva (+1 DT) comieron ms patatas fritas en la
condicin no saludable, = 6.77, ET = 1.03, que los de la condicin control
= 2.66, ET = 1.23; = .56, t = 2.57, p < .05, y la condicin saludable, = 2.96,
ET = 1.42; = .52, t = 2.18, p < .05. Adems, la diferencia en comer patatas fritas
entre la condicin control y la condicin saludable no era significativa, = -.04,
t = -.16, ns. Y en cuanto a los participantes ms ambivalentes que expresaron un
actitud negativa (-1 DT), comieron menos patatas fritas en la condicin saludable,
= .01, ET = .97, que los participantes de la condicin control, = 4.96,
ET = 1.16; = -.75, t = -3.96, p < .001, y los de la condicin saludable,
= 4.82, ET = 1.38; = -.59, t = -3.39, p < .001. La diferencia en comer patatas
fritas entre la condicin control y la no saludable no era significativa, = .02,
t = .07, ns. Por tanto, la cantidad de patatas fritas consumidas por los participantes
ms ambivalentes vari significativamente en funcin de las claves contextuales.
Al analizar estos efectos desde otra perspectiva, encontramos que entre los
participantes ms ambivalentes las diferencias en patatas fritas comidas entre los
que expresaban actitudes negativas y los que expresaban actitudes positivas no
eran significativas ni en la condicin no saludable, = .28, t = .98, ns, ni en la
saludable, = .59, t = 1.79, p < .08. Es importante el hecho de que la diferencia
tampoco era significativa en la condicin control, = -.33, t = -1.22, ns, lo que
replica los resultados del Estudio 1 y muestra que las actitudes no predijeron el
comportamiento de los participantes ambivalentes.
En resumen, en este estudio analizamos el papel de la ambivalencia actitudinal
en la influencia del contexto social respecto al comportamiento alimentario. Los
resultados corroboran nuestras hiptesis de que las actitudes son mejores predictores del comportamiento para los participantes menos ambivalentes que para
aquellos que lo son ms. Es decir, las actitudes de los participantes menos
ambivalentes predicen su comportamiento en todos los contextos. Por el contrario,
las claves contextuales influyen en el comportamiento alimentario de los participantes ms ambivalentes. De hecho, su consumo de patatas fritas aumenta cuando
las claves contextuales destacan una forma de comer no saludable y disminuye
cuando destacan una forma de comer saludable.
Discusin General
Presentamos un programa de investigacin que abordaba dos aspectos en los que
la ambivalencia poda desempear un rol importante para la prediccin de los
comportamientos: la relacin actitud-comportamiento, y la influencia del contexto
social en el comportamiento real. En el estudio 1, comprobamos el papel
moderador de la ambivalencia en la relacin actitud-comportamiento. Como
esperbamos, los resultados muestran que la ambivalencia modera la consistencia
de la relacin actitud-comportamiento. Como en Armitage y Conner (2000),
Conner et al. (2003) y Conner et al. (2002), en el caso de participantes menos
ambivalentes la actitud predice el comportamiento, pero no en el caso de participantes muy ambivalentes. Nuestros resultados son relevantes porque muestran que
estos patrones no se limitan al comportamiento recogido mediante auto-informes,
456
M.T. Batista et al.
como se ha encontrado en investigaciones previas (p. ej., Berndsen y van der

Pligt, 2004; Povey et al., 2001; Shepherd, 1999), sino que puede generalizarse al
comportamiento real. Nuestra intencin, despus de mostrar que las actitudes
pueden predecir el comportamiento de los individuos menos ambivalentes, era
identificar los predictores del comportamiento de los individuos ms ambivalentes. En este sentido, el Estudio 2 indica que para los participantes menos
ambivalentes las actitudes, una vez ms, son buenos predictores del comportamiento, independientemente del contexto social. Tal vez el aspecto ms innovador
de estos estudios sea que muestran que la influencia del contexto social slo tiene
lugar en el caso de los individuos ms ambivalentes. Concretamente, para los
participantes ms ambivalentes el consumo de patatas fritas aumentaba o
disminua segn las claves situacionales. Estos resultados coinciden con las
investigaciones que indican que las actitudes ambivalentes son ms sensibles al
contexto (p. ej., Hodson et al., 2001; Lavine, 2001; Lavine et al., 1998;
Tourangeau et al., 1989). Y lo que es an ms importante, estos resultados
coinciden con los trabajos tericos que predecan que la ambivalencia puede ser
el motivo principal que subyace a la inconsistencia de la relacin actitud-comportamiento (Eagly y Chaiken, 1993; Jonas et al., 2000). Nuestros resultados
tambin sugieren que es ms probable que las actitudes ambivalentes hagan que
los comportamientos se guen ms por la informacin presente en el contexto
social (vase tambin Briol, Petty, y Wheele, 2006; Clark, Wegener, y Fabrigar,
2008).
Nuestros resultados no son solo relevantes desde un punto de vista terico. De
hecho, nuestros resultados no solo contribuyen a una mejor comprensin del
comportamiento alimentario sino tambin de cmo se pueden disear intervenciones que influyan en l. Concretamente y en lnea con investigaciones previas
(p. ej., Clendenen, Herman, y Polivy, 1994; de Castro y Brewer, 1992; Herman y
Polivy, 2005; Herman, Roth, y Polivy, 2003; Pliner, Bell, Hirsch, y Kinchla, 2006;
Vartanian, Herman, y Wansink, 2008), el Estudio 2 muestra que se puede influir
sobre los comportamientos alimentarios a travs de mensajes sutiles en los contextos en los que la gente come. Los contextos donde tienen lugar los comportamientos alimentarios (p. ej., los comedores de escuelas o de empresas) deberan
albergar claves que destacasen la dimensin negativa asociada a la comida no
saludable y/o los beneficios de la comida sana. Por ejemplo, la decoracin del
comedor podra incluir imgenes de comida sana asociadas a personas saludables.
Estos contextos podran contribuir a disminuir el consumo de comida no saludable
y aumentar el consumo de comida sana. Sin embargo, hay que aclarar que el
efecto del contexto sobre el comportamiento alimentario se limita a los individuos
con actitudes ambivalentes. En cuanto a los individuos no ambivalentes con
actitud positiva respecto a la comida no saludable, la estrategia debera centrarse
en el cambio de actitud. La transformacin de una actitud positiva no ambivalente
respecto a la comida no saludable en una tambin no ambivalente pero negativa
debera llevar a un cambio de actitud en el sentido pretendido (Erber et al., 1995).
A pesar de la relevancia y consistencia de nuestros hallazgos, quisiramos
sealar algunas limitaciones que deberan abordarse en futuras investigaciones. En
457
primer lugar, en nuestros estudios no comparamos el efecto de las actitudes

neutras con respecto a las ambivalentes. Esta limitacin se debe al hecho de que
es muy difcil encontrar estmulos (alimentos) hacia los que los individuos tengan
actitudes puramente neutras sin ninguna ambivalencia asociada. De hecho, en una
lista con 256 comidas, Batista y Lima (2010) no pudieron encontrar ninguna que
fuera no-ambivalente y neutral (o no-ambivalente negativa). En segundo lugar,
usamos un ndice global de ambivalencia. Por tanto no pudimos comprobar si la
ambivalencia hacia la comida se debe o no a un conflicto entre salud y placer. Las
prximas investigaciones deberan estudiar los efectos de la ambivalencia intercomponentes, es decir, el conflicto entre las cogniciones (salud) y el afecto
(gusto). En tercer lugar, es posible que el hecho de que los participantes pudieran
escoger trozos de manzana aumentase la ambivalencia hacia las patatas fritas. Una
manera de abordar esta cuestin es replicando estos estudios aadiendo una
condicin en la que no haya trozos de manzana. As mismo, y como sugirieron
Thompson et al. (1995), en los dos estudios usamos juicios positivos y negativos
para computar tanto las ambivalencias como las actitudes. A pesar de que las
medidas slo presentan una correlacin dbil, posiblemente se podra haber
medido mejor la actitud a travs de una escala bipolar. Finalmente, nos gustara
aadir ms sugerencias para futuros estudios. Sera muy interesante repetir los
estudios para verificar si estos efectos pueden generalizarse a otros objetos de
actitud ambivalentes, como los comportamientos medioambientales. Si nuestros
hallazgos pueden aplicarse en otros mbitos, el contexto del que las personas
extraen sus actitudes ambivalentes puede ser crucial para incrementar la frecuencia de estos comportamientos (y posiblemente otro tipo de comportamientos
cvicos). Tambin sugerimos replicar estos estudios empleando un intervalo
mayor entre la medicin de las actitudes y la observacin del comportamiento.
Armitage y Conner (2000) usaron un intervalo de cinco/ocho meses y hallaron
consistencia, pero dados los cambios que introdujimos es importante comprobar si
los hallazgos se vuelven a observar con un periodo de tiempo ms prolongado.
Adems, en futuros estudios, sera interesante determinar el rol de la ambivalencia
actitudinal y las normas sociales en el comportamiento. Segn nuestros resultados
y los de Hodson et al. (2001) las normas sociales tendrn ms influencia en el
comportamiento de los individuos ms ambivalentes. Sera beneficioso para la
investigacin incluir personas de otros sectores de la poblacin (no slo estudiantes universitarios); puede que para otros tipos de poblacin las normas que
regulan el comportamiento alimentario sean diferentes de las que se han abordado
en la literatura (Batista y Lima, 2013). Tambin deberan repetirse los estudios
controlando varias dimensiones de la intensidad de la actitud que a menudo se
confunden con la ambivalencia (p. ej., baja certeza y accesibilidad). Finalmente,
sera muy relevante aadir una medicin de la actitud en la fase 2, lo que nos
permitira comprobar si (a) las actitudes son estables slo en los individuos menos
ambivalentes y (b) las actitudes ambivalentes medidas una semana antes no
predicen el comportamiento porque son inestables, a pesar de que las actitudes
ambivalentes medidas en el momento de observacin del comportamiento s
458
M.T. Batista et al.
puedan guiarlo. Concretamente, entre los individuos ms ambivalentes puede que

el contexto influya en la actitud que a su vez guiar el comportamiento.
A pesar de estas limitaciones, esta investigacin aporta una contribucin nica
a la literatura sobre la relacin entre actitud y comportamiento al mostrar la
importancia de los contextos sociales en la prediccin del comportamiento real
de los individuos con actitudes ambivalentes. La idea principal de esta
investigacin consiste en que las actitudes son slo buenos predictores del
comportamiento cuando son menos ambivalentes. Diferentes procesos entran en
juego en el caso de las actitudes ms ambivalentes, en el sentido de que los
individuos con ese tipo de actitudes son ms sensibles a las claves contextuales y
estas a su vez influyen en su comportamiento.
Acknowledgments / Agradecimientos
This research was financed by Fundao para a Cincia e Tecnologia through [Grant
Number SFRH/BD/24758/2005]. / Esta investigacin fue financiada por la Fundao
para a Cincia e Tecnologia a travs del proyecto SFRH/BD/24758/2005.
Notas
1. La correlacin entre actitud y ambivalencia era dbil, r(123) = .27, p = .002.
2. Debido al hecho de que los dos tems de evaluacin que forman nuestra medida de
actitud tambin son elementos de las partes P y N de la frmula de la ambivalencia,
calculamos una medida compuesta diferente de las actitudes invirtiendo el tem
negativo y combinndolo con el positivo en una nica medida compuesta. Esta
medida poda variar entre 1 (actitud negativa) y 4 (actitud positiva). Los resultados
de regresin no cambiaron.
3. Como en el Estudio 1, la actitud y la ambivalencia presentaban una correlacin dbil,
r(127) = .19, p = .029.
4. Los efectos simples tambin sugeran que los participantes de la condicin saludable
comieron menos patatas fritas que los de la condicin control, = 3.59, ET = .58;
= -.27, t = -2.50, p < .05. No haba diferencia significativa entre las condiciones no
saludable y control, = .15, t = 1.44, ns.
References / Referencias
Armitage, C. J., & Conner, M. (2000). Attitudinal ambivalence: A test of three key
hypotheses. Personality and Social Psychology Bulletin, 26, 14211432.
doi:10.1177/0146167200263009
Batista, M. T., & Lima, M. L. (2010). Seleco de estmulos alimentares ambivalentes e
comparao de medidas em quatro indicadores de ambivalncia atitudinal [Selection
of food stimuli and measure comparison in four indexes of attitudinal ambivalence].
Laboratrio de Psicologia, 8, 121148.
Batista, M. T., & Lima, M. L. (2013). Quem est comendo comigo? Influncia social
indirecta no comportamento alimentar ambivalente [Who is eating with me? Indirect
social influence on food consumption]. Psicologia: Reflexo e Crtica / Psychology,
26, 113121.
Beardsworth, S. A. (1995). The management of food ambivalence: Erosion or reconstruction? In D. Maurer & J. Sobal (Eds.), Eating agendas: Food and nutrition as social
problems (pp. 117141). New York, NY: De Gruyter.
459
Bell, D. W., & Esses, V. M. (1997). Ambivalence and response amplification toward
native peoples. Journal of Applied Social Psychology, 27, 10631084. doi:10.1111/
j.1559-1816.1997.tb00287.x
Bell, D., Esses, V. M., & Maio, G. R. (1996). The utility of open-ended measures to assess
intergroup ambivalence. Canadian Journal of Behavioural Science / Revue canadienne des sciences du comportement, 28, 1218. doi:10.1037/0008-400X.28.1.12
Berndsen, M., & van der Pligt, J. (2004). Ambivalence towards meat. Appetite, 42, 7178.
doi:10.1016/S0195-6663(03)00119-3
Briol, P., Petty, R. E., & Wheeler, S. C. (2006). Discrepancies between explicit and
implicit self-concepts: Consequences for information processing. Journal of
Personality and Social Psychology, 91, 154170. doi:10.1037/0022-3514.91.1.154
Broemer, P. (2002). Relative effectiveness of differently framed health messages: The
influence of ambivalence. European Journal of Social Psychology, 32, 685703.
doi:10.1002/ejsp.116
Burger, J. M., Bell, H., Harvey, K., Johnson, J., Stewart, C., Dorian, K., & Swedroe, M.
(2010). Nutritious or delicious? The effect of descriptive norm information on food
choice. Journal of Social and Clinical Psychology, 29, 228242. doi:10.1521/
jscp.2010.29.2.228
Clark, J. K., Wegener, D. T., & Fabrigar, L. R. (2008). Attitude accessibility and message
processing: The moderating role of message position. Journal of Experimental Social
Psychology, 44, 354361. doi:10.1016/j.jesp.2006.12.001
Clendenen, V. I., Herman, C. P., & Polivy, J. (1994). Social facilitation of eating among
friends and strangers. Appetite, 23, 113. doi:10.1006/appe.1994.1030
Conner, M., & Armitage, C. J. (2002). The social psychology of food. Buckingham: Open
University Press.
Conner, M., & Sparks, P. (2002). Ambivalence and attitudes. European Review of Social
Psychology, 12, 3770. doi:10.1080/14792772143000012
Conner, M., Godin, G., Sheeran, P., & Germain, M. (2013). Some feelings are more
important: Cognitive attitudes, affective attitudes, anticipated affect, and blood donation. Health Psychology, 32, 264272. doi:10.1037/a0028500
Conner, M., Povey, R., Sparks, P., James, R., & Shepherd, R. (2003). Moderating role of
attitudinal ambivalence within the theory of planned behaviour. British Journal of
Social Psychology, 42, 7594. doi:10.1348/014466603763276135
Conner, M., Sparks, P., Povey, R., James, R., Shepherd, R., & Armitage, C. J. (2002).
Moderator effects of attitudinal ambivalence on attitude-behaviour relationships.
European Journal of Social Psychology, 32, 705718. doi:10.1002/ejsp.117
de Castro, J. M., & Brewer, E. M. (1992). The amount eaten in meals by humans is a
power function of the number of people present. Physiology and Behavior, 51, 121
125. doi:10.1016/0031-9384(92)90212-K
Dormandy, E., Hankins, M., & Marteau, T. M. (2006). Attitudes and uptake of a screening
test: The moderating role of ambivalence. Psychology and Health, 21, 499511.
doi:10.1080/14768320500380956
Eagly, A. H., & Chaiken, S. (1993). The psychology of attitudes. Fort North, FL: Harcourt
Brace Jovanovich.
Erber, M. W., Hodges, S. D., & Wilson, T. D. (1995). Attitude strength, attitude stability,
and the effects of analyzing reasons. In R. E. Petty & J. A. Korosnick (Eds.), Attitude
strength: Antecedents and consequences (pp. 433454). Mahwah, NJ: Laurence
Erlbaum Associates.
Fazio, R. H., & Towles-Schwen, T. (1999). The MODE model of attitude-behavior
processes. In S. Chaiken & Y. Trope (Eds.), Dual-process theories in social psychology (pp. 97116). New York, NY: Guilford Press.
Herman, C. P., & Polivy, J. (2005). Normative influences on food intake. Physiology &
Behavior, 86, 762772. doi:10.1016/j.physbeh.2005.08.064
460
M.T. Batista et al.
Herman, C. P., Roth, D. A., & Polivy, J. (2003). Effects of the presence of others on food
intake: A normative interpretation. Psychological Bulletin, 129, 873886.
doi:10.1037/0033-2909.129.6.873
Hodson, G., Maio, G., & Esses, V. (2001). The role of attitudinal ambivalence in
susceptibility to consensus information. Basic and Applied Social Psychology, 23,
197205. doi:10.1207/S15324834BASP2303_6
Hofmann, W., Friese, M., & Roefs, A. (2009). Three ways to resist temptation: The
independent contributions of executive attention, inhibitory control, and affect regulation to the impulse control of eating behavior. Journal of Experimental Social
Hofmann, W., Rauch, W., & Gawronski, B. (2007). And deplete us not into temptation:
Automatic attitudes, dietary restraint, and self-regulatory resources as determinants of
eating behavior. Journal of Experimental Social Psychology, 43, 497504.
doi:10.1016/j.jesp.2006.05.004
Holland, R. W., Verplanken, B., & Knippenberg, A. V. (2002). On the nature of attitudebehavior relations: The strong guide, the weak follow. European Journal of Social
Psychology, 32, 869876. doi:10.1002/ejsp.135
Jonas, K., Broemer, P., & Diehl, M. (2000). Attitudinal ambivalence. European Review of
Social Psychology, 11, 3574. doi:10.1080/14792779943000125
Judd, C. M., & McClelland, G. H. (2001). Data analysis: A model-comparison approach.
Belmont, CA: Wadsworth\Thomson Learning.
Kaplan, K. J. (1972). On the ambivalence-indifference problem in attitude theory and
measurement: A suggested modification of the semantic differential technique.
Psychological Review, 77, 361372.
Kutner, B., Wilkins, C., & Yarrow, P. R. (1952). Verbal attitudes and overt behavior
involving racial prejudice. The Journal of Abnormal and Social Psychology, 47, 649
652. doi:10.1037/h0053883
LaPiere, R. T. (1934). Attitudes vs. actions. Social Forces, 13, 230237. doi:10.2307/
2570339
Lavine, H. (2001). The electoral consequences of ambivalence toward presidential candidates. American Journal of Political Science, 45, 915929. doi:10.2307/2669332
Lavine, H., Huff, J. W., Wagner, S. H., & Sweeney, D. (1998). The moderating influence
of attitude strength on the susceptibility to context effects in attitude surveys. Journal
of Personality and Social Psychology, 75, 359373. doi:10.1037/0022-3514.75.2.359
Linville, P. W., & Jones, E. E. (1980). Polarized appraisals of outgroup members. Journal
of Personality and Social Psychology, 38, 689703. doi:10.1037/0022-3514.38.5.689
Maio, G. R., Bell, D. W., & Esses, V. M. (1996). Ambivalence and persuasion: The
processing of messages about immigrant groups. Journal of Experimental and Social
Psychology, 32, 513536. doi:10.1006/jesp.1996.0023
Maio, G. R., Esses, V. M., & Bell, D. W. (2000). Examining conflict between components
of attitudes: Ambivalence and inconsistency are distinct constructs. Canadian Journal
of Behavioural Science / Revue canadienne des sciences du comportement, 32, 7183.
doi:10.1037/h0087102
Mischel, W., Shoda, Y., & Rodriguez, M. L. (1989). Delay of gratification in children.
Science, 244, 933938. doi:10.1126/science.2658056
Ogden, J. (2003). The psychology of eating: From healthy to disordered behavior. Oxford:
Blackwell Publishing.
Pliner, P., & Mann, N. (2004). Influence of social norms and palatability on amount
consumed and food choice. Appetite, 42, 227237. doi:10.1016/j.appet.2003.12.001
Pliner, P., Bell, R., Hirsch, E. S., & Kinchla, M. (2006). Meal duration mediates the effect
of social facilitation on eating in humans. Appetite, 46, 189198. doi:10.1016/j.
appet.2005.12.003
461
Povey, R., Wellens, B., & Conner, M. (2001). Attitudes towards following meat, vegetarian and vegan diets: An examination of the role of ambivalence. Appetite, 37, 1526.
doi:10.1006/appe.2001.0406
Scott, W. A. (1968). Attitude measurement. In G. Lindzey & E. Aronson (Eds.), The
handbook of social psychology (Vol. 2, pp. 204273). Reading, MA: Addison-Wesley.
Shepherd, R. (1999). Social determinants of food choice. Proceedings of the Nutrition
Society, 58, 807812. doi:10.1017/S0029665199001093
Stroebe, W., Mensink, W., Aarts, H., Schut, H., & Kruglanski, A. W. (2008). Why dieters
fail: Testing the goal conflict model of eating. Journal of Experimental Social
Thompson, M. M., Zanna, M. P., & Griffin, D. W. (1995). Lets not be indifferent about
(attitudinal) ambivalence. In R. E. Petty & J. A. Krosnick (Eds.), Attitude strength:
Antecedents and consequences (pp. 361386). Hillsdale, NJ: Lawrence Erlbaum.
Tourangeau, R., Rasinski, K. A., Bradburn, N., & DAndrade, R. (1989). Belief accessibility and context effects in attitude measurement. Journal of Experimental Social
Psychology, 25, 401421. doi:10.1016/0022-1031(89)90030-9
Urland, G. R., & Ito, T. A. (2005). Have your cake and hate it, too: Ambivalent food
attitudes are associated with dietary restraint. Basic and Applied Social Psychology,
27, 353360. doi:10.1207/s15324834basp2704_8
Vartanian, L. R., Herman, C. P., & Wansink, B. (2008). Are we aware of the external
factors that influence our food intake? Health Psychology, 27, 533538. doi:10.1037/
0278-6133.27.5.533
Wicker, A. W. (1969). Attitudes versus actions: The relationship of verbal and overt
behavioral responses to attitude objects. Journal of Social Issues, 25, 4178.
doi:10.1111/j.1540-4560.1969.tb00619.x

Revista de Psicología Social: International Journal of Social Psychology

Încărcat de

Informații document

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Revista de Psicología Social: International Journal of Social Psychology

Încărcat de

Drepturi de autor:

Formate disponibile

Revista de Psicologa Social

International Journal of Social Psychology

ISSN: 0213-4748 (Print) 1579-3680 (Online) Journal homepage: http://www.tandfonline.com/loi/rrps20

I love chips and (h)ate them too: the role of

Published online: 15 Dec 2014.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

Date: 23 November 2015, At: 21:32

Downloaded by [Universidad Nacional Andres Be] at 21:32 23 November 2015

Revista de Psicologa Social / International Journal of Social Psychology, 2014

Instituto Universitrio de Lisboa (ISCTE-IUL), CIS-IUL; bInstituto de Cincias Sociais

English version: pp. 430443 / Versin en espaol: pp. 444458

Ambivalence, attitudes and food behaviour / Ambivalencia, actitudes, comportamiento alimentario

contextuales slo influyeron en el comportamiento de los individuos con

Downloaded by [Universidad Nacional Andres Be] at 21:32 23 November 2015

Palabras clave: actitudes; ambivalencia actitudinal; comportamiento alimentario;

The possibility of predicting behaviour based on attitudes has captivated the

Downloaded by [Universidad Nacional Andres Be] at 21:32 23 November 2015

M.T. Batista et al.

In fact, drawing on a view that understands attitudes as bidimensional

Overview of the studies

Downloaded by [Universidad Nacional Andres Be] at 21:32 23 November 2015

Ambivalence, attitudes and food behaviour / Ambivalencia, actitudes, comportamiento alimentario

M.T. Batista et al.

Downloaded by [Universidad Nacional Andres Be] at 21:32 23 November 2015

Ambivalence, attitudes and food behaviour / Ambivalencia, actitudes, comportamiento alimentario

Downloaded by [Universidad Nacional Andres Be] at 21:32 23 November 2015

dissatisfied) to 4 (very dissatisfied) and 1 (in no way harmful) to 4 (very harmful),

Results and discussion

M.T. Batista et al.

Downloaded by [Universidad Nacional Andres Be] at 21:32 23 November 2015

Table 1. The moderating role of attitudinal ambivalence in attitude-behaviour relationships

p < .10; * p < .05; ** p < .01; *** p < .001.

participants with positive, = 3.48, SE = .68, and negative attitudes, = 3.72,

Ambivalence, attitudes and food behaviour / Ambivalencia, actitudes, comportamiento alimentario

Downloaded by [Universidad Nacional Andres Be] at 21:32 23 November 2015

attitudinal ambivalence. In contrast, this relationship was not significant among

M.T. Batista et al.

Downloaded by [Universidad Nacional Andres Be] at 21:32 23 November 2015

Attitude, ambivalence and eating behaviour

Results and discussion

Downloaded by [Universidad Nacional Andres Be] at 21:32 23 November 2015

Ambivalence, attitudes and food behaviour / Ambivalencia, actitudes, comportamiento alimentario

Downloaded by [Universidad Nacional Andres Be] at 21:32 23 November 2015

M.T. Batista et al.

Downloaded by [Universidad Nacional Andres Be] at 21:32 23 November 2015

Ambivalence, attitudes and food behaviour / Ambivalencia, actitudes, comportamiento alimentario

Downloaded by [Universidad Nacional Andres Be] at 21:32 23 November 2015

M.T. Batista et al.

ambivalence can be the core of attitude-behaviour inconsistency (Eagly &

Downloaded by [Universidad Nacional Andres Be] at 21:32 23 November 2015

Ambivalence, attitudes and food behaviour / Ambivalencia, actitudes, comportamiento alimentario

M.T. Batista et al.

Downloaded by [Universidad Nacional Andres Be] at 21:32 23 November 2015

Me encantan las patatas fritas, las como y (tambin) las odio: el

Downloaded by [Universidad Nacional Andres Be] at 21:32 23 November 2015

Ambivalence, attitudes and food behaviour / Ambivalencia, actitudes, comportamiento alimentario

M.T. Batista et al.

Downloaded by [Universidad Nacional Andres Be] at 21:32 23 November 2015

Descripcin de los estudios

Ambivalence, attitudes and food behaviour / Ambivalencia, actitudes, comportamiento alimentario

p < .10; * p < .05; p < .01; * p < .001.

p < .10; * p < .05; p < .01; * p < .001.