Consumer Response To Negative Media Information

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 23

J Consum Policy (2015) 38:387–409

DOI 10.1007/s10603-015-9299-z

O R I G I N A L PA P E R

Consumer Response to Negative Media Information


About Certified Organic Food Products

Christoph Emanuel Müller 1 & Hansjörg Gaus 1

Received: 30 October 2014 / Accepted: 28 July 2015 /


Published online: 12 August 2015
# Springer Science+Business Media New York 2015

Abstract When fraudulent mislabelling of organic food products on the part of producers or
label misapprehension on the part of consumers is revealed by mass media sources, this may
have negative effects on consumers’ evaluations of and behaviour towards the purchase of
certified organic food products. However, even if this poses a threat to the functioning of
organic food labelling systems, there is a lack of empirical evidence. Therefore, drawing on
behavioural models and literature on the impact of negative publicity on brand evaluations and
consumer behaviour related to eco-labels, an experimental study was conducted in which
members of the manipulation group watched a documentary containing exemplary information
about fraud and misapprehension. The results show significant negative effects of the manip-
ulation, both directly after exposure and 2 weeks later, on behavioural intentions, five
attitudinal constructs, and trust in organic food labels. However, there were no significant
effects on self-reported behaviour. Moreover, we found that the influence of the documentary
on behavioural intentions was almost completely mediated by attitudes towards organic food
products, whereas we did not find any significant mediator effects on self-reported behaviour.

Keywords Consumer response . Negative media information . Organic food . Eco-labels .


Randomized controlled trial

As a consequence of consumers’ increasing environmental concern and health awareness, both


the demand for and the supply of organic food are rising continuously in many countries
(Willer 2011). Yet there is an asymmetry of information between producers and consumers

* Hansjörg Gaus
h.gaus@ceval.de
Christoph Emanuel Müller
c.mueller@ceval.de

1
Department of Sociology, Center for Evaluation (CEval), Saarland University, P.O. Box 151150,
66041 Saarbrücken, Germany
388 C.E. Müller, H. Gaus

because organic is a credence quality (Darby and Karni 1973). This means that producers
normally know whether their products are truly organic or not, while consumers are not
usually capable of verifying that (Giannakas 2002). In order to reduce this asymmetry, organic
food labelling has become a common strategy for business self-regulation (Amstel et al. 2008;
Thøgersen 2010). With the use of organic food labels, the intention is to assure consumers that
the production of food is consistent with a defined set of criteria. Consequently, organic food
labels are supposed to provide a reliable informational basis for the decisions consumers make
when purchasing organic food (Sønderskov and Daugbjerg 2011), and increase their willing-
ness to pay for organically produced food products (Rousseau and Vranken 2013).
A good deal is already known about various determinants of buying organic food products
(Aertsens et al. 2009; Hughner et al. 2007; Makatouni 2002; Padel and Foster 2005), with
consumers’ trust in the information of an organic food label playing a key role in their
decision-making process (Sønderskov and Daugbjerg 2011). However, organic food labels
are affected by two types of problem that could undermine that trust, namely, fraud on the part
of the producers and misapprehension on the part of the consumers.
First, fraud through the intentional mislabelling of conventionally produced food products
as organic is widely discussed. Examples of fraud can be found both in scholarly literature
(e.g., Giannakas 2002; Siderer et al. 2005) and in a huge body of anecdotal evidence in the
media.1 Consequently, strategies on how to avoid fraud have been suggested (e.g., Jahn et al.
2012). However, one cannot expect fraud to cease to exist, because suppliers can
(illegitimately) increase their profits by mislabelling lower-priced conventional food as more
expensive organic food (Anania and Nisticò 2004).
Second, by the term Blabel misapprehension,^ we denote the phenomenon of consumers’
erroneously attributing higher standards of environmental friendliness to certified organic
products than are actually required to obtain a specific organic food label. Frequently, a
product can legitimately be labelled as certified organic but still have significant environmental
impacts (e.g., organically produced fruit or vegetables that are shipped over large distances or
the production of which has substantial ecological side effects).
In line with this reasoning, we argue that when producers’ mislabelling of products or
consumers’ label misapprehension is revealed by mass media sources, this may have negative
effects on consumers’ evaluation of and behaviour towards the purchase of certified organic
food products. Yet even if it ought to be of high interest whether and to what degree such
negative information reduces trust in organic food labels or negatively influences consumers’
attitudes, behavioural intentions, and purchasing behaviour related to certified organic food
products, there is a lack of empirical evidence.
We therefore conducted an experimental study applying two analytical strategies. First, we
assessed whether and how strongly negative media information that highlights producers’
mislabelling of food products and consumers’ label misapprehension directly influences
consumers’ trust in and attitudes towards labelled products and their behavioural intentions
and self-reported behaviour. In a second step, we investigated whether the influence of the
negative information on behavioural intentions and behaviour is mediated by attitudes and
trust.

1
In Germany, for example, the mislabelling of conventionally produced eggs as organic recently received
extensive coverage in the media.
Consumer Response to Negative Media Information About Organic Food 389

In the section that follows, we first outline our analytical framework. Next, we describe the
method and results of our empirical study. Finally, we discuss the meanings of our findings and
their implications.

Theoretical Framework

Because research on consumer response to negative media information on eco-labelled


products is sparse, the theoretical framework of our study—beyond Fishbein and Ajzen’s
(2010) reasoned action model and Thøgersen’s (2000) model for predicting the purchase of
labelled products—draws on literature on the impact of negative publicity on brand evalua-
tions. Even if there are differences between eco-labels and brands of consumer goods, in line
with Larceneux et al. (2012), we argue that well-established organic food labels like the
German Biosiegel2 have effects on the labelled product which are similar to those of a brand
(for example, by triggering consumer associations related to product attributes, purchase
consequences, and symbolic meaning). Thus, we consider it legitimate to apply theory on
brand evaluation to consumer evaluation of organic food labels.
We further argue that being exposed to negative information on certified organic
food products leads to an increase in knowledge. Fishbein and Ajzen’s (2010)
reasoned action model, which summarizes their previous work on the theory of
reasoned action (Ajzen and Fishbein 1980) and the theory of planned behaviour
(Ajzen 1988), treats such exposure to information as a background factor which
potentially influences behavioural beliefs that form attitudes. In Thøgersen’s (2000)
model for predicting the purchase of labelled products, knowledge about eco-labels is
seen to be a determinant not only of attitudes towards environment-friendly buying
but also of trust in eco-labels.
Generally, negative publicity is considered an important factor in influencing
consumers’ brand evaluations (Ahluwalia et al. 2000). Reasons often cited for this
assumption are that publicity is a relatively credible source of information (Bond and
Kirshenbaum 1998), and the so-called Bnegativity effect,^ which is defined as Bthe
greater weighting of negative as compared with equally extreme positive information
in the formation of evaluative judgments^ (Ahluwalia 2002, p. 271). A common
explanation for the negativity effect is that Bnegative information is perceived as more
useful or diagnostic than positive information for categorizing targets into evaluative
categories^ (Ahluwalia 2002, p. 271; Herr et al. 1991). Given that an organic food
label has effects on a labelled product which are similar to those of a brand, we
contend that negative media information negatively affects consumers’ attitudes to-
wards consequences included in the message, such as the impact of food products
labelled as organic on the environment, human health, and animal welfare. However,
there is evidence that negative information not only affects brand characteristics that
are included in the message. BSpillover^ effects—Bthe extent to which information
provided in messages changes beliefs about attributes that are not mentioned in the
messages^—may also influence the evaluation of other brand attributes (Ahluwalia

2
http://www.oekolandbau.de/bio-siegel/ (last accessed June 6, 2015); as of March 31, 2015, the Biosiegel, which
is complementary to the European Union (EU) organic farming regulation, was used by 4486 companies for 70
393 organic food products.
390 C.E. Müller, H. Gaus

et al. 2001, p. 458). Therefore, we propose that negative media information on


products labelled as organic even has negative effects on consumers’ attitudes towards
product characteristics that are not contained in the message, such as product quality
or naturalness. In line with Fishbein and Ajzen’s (2010) reasoned action model—
leaving aside the perceived norm and perceived behavioural control—we hypothesize
that negative media information on certified organic products negatively affects
attitudes towards buying organic food, which in turn affect behavioural intentions
related to the purchase of organic food (i.e., intentions to buy organic food and
intentions to pay attention to organic food and organic food labels).
Trust in the message an organic food label conveys is a key determinant of
consumers’ use of label information in purchase decisions (Thøgersen 2000). With
regard to brands, there is evidence that negative publicity negatively influences brand
trust (Dahlén and Lange 2006). Consequently, we contend that trust in organic food
labels is also negatively affected by negative media information. Moreover, Thøgersen
(2000) regards trust in the information a label conveys as a variable that directly
influences the attention paid to eco-labels, which is a prerequisite for a purchase
decision. Similarly, we assume that a decrease in trust in organic food labels nega-
tively affects the behavioural intentions related to the purchase of organic food (i.e.,
intentions to buy labelled food and intentions to pay attention to organic food and
organic food labels).
Furthermore, Thøgersen’s (2000) model proposes a direct influence of knowledge
on the attention paid to eco-labels and the decision to buy labelled products. This in
turn leads us to propose that negative media information influences behavioural
intentions and actual purchasing behaviour not only indirectly, that is, mediated by
attitudes or trust, but also directly. Finally, in line with the reasoned action model, we
assume that behavioural intentions influence purchasing behaviour related to organic
food.
This reasoning results in the following theoretical model (Figure 1), explaining the effects
of received negative media information on purchasing behaviour, both directly and indirectly,
that is, mediated by attitudes, trust, and behavioural intentions.

Attitudes

(-) (+)
(-)

Negative (-) (+)


Behavioural
Media Behaviour
Intentions
Information

(-)
(+)

Trust

Fig. 1 Theoretical model


Consumer Response to Negative Media Information About Organic Food 391

Method

Manipulation and Control

In order to test our assumptions, we conducted a randomized controlled trial (RCT) with a
manipulation condition that provided critical information related to organic food production
and organic food labels. We chose the investigative 30-min TV documentary BWie billig kann
Bio sein?^ (How cheap can organic food be?), produced by the major German public-service
TV station Allgemeiner Rundfunk Deutschland (ARD). Earlier studies had shown that such TV
programmes can influence consumers’ attitudes, behavioural intentions and behaviour, for
example, in the areas of environmental protection (Syme et al. 1987) or health (Chew et al.
2002). In addition, information distributed by a public source is perceived to be more credible
than information from private-sector providers (Craig and McCann 1978). Finally, the duration
of the intervention represented a reasonable compromise between having enough time to
acquire and process relevant information on the one hand and not getting bored or losing
concentration on the other.
In general, the documentary aims to demonstrate that certified organic food products are not
necessarily produced under the conditions one would expect. To achieve that, the documentary
investigates exemplarily whether there have been any violations of European Union (EU)
organic farming standards in Germany. Furthermore, the documentary emphasizes that even if
the EU standards are met, organic food production may not necessarily lead to a significant
decrease in negative impacts on the environment and human health.
For illustration purposes, the documentary focuses on three areas of organic food produc-
tion, namely, pig rearing, egg production, and potato farming. Three different methods for
communicating its messages are applied. First, the documentary uses provocative pictures and
video clips. Second, there are interviews with researchers, experts, producers, and representa-
tives of activist groups in the area of organic food production. Third, the narrator provides the
audience with diverse facts and statistics about the production of organic food, the develop-
ment of the food production sector concerned, and its challenges and opportunities.
Instead of the documentary about certified organic food, the control group watched a film
that had absolutely nothing to do with organic food, health or environmental issues. By
contrast, the control group film covered the emergence of BDie Piratenpartei^ (The Pirate
Party) in Germany, a new political party mainly concerned with civil rights issues, privacy
protection, and freedom of information. Entitled BDer Piraten-Hype^ (The Pirate Hype), this
film was of the same length as the documentary presented in the manipulation condition, and
was also produced by one of the two major German public-service TV stations, the Zweites
Deutsches Fernsehen (ZDF).

Data Collection

As participants in our study, we recruited 145 students who responded to information flyers in
the central cafeteria and on the notice boards of a university. Participants enrolled voluntarily
via an online registration portal, where they were also informed about organizational and data
security issues. However, they did not receive any information about the topic of the study.
We assigned participants randomly to a manipulation (n=72) and a control condition (n=
73). Before participants in the manipulation group watched the documentary about certified
organic food, they were asked to complete a baseline questionnaire in order to provide pre-
392 C.E. Müller, H. Gaus

intervention information regarding the outcome variables of interest. Subsequently, having


completed the questionnaire, participants watched the film described above in a university
classroom. Directly after the screening of the film, they completed a second questionnaire
featuring the same items as the baseline questionnaire (except for the demographics and self-
reported behaviour). Furthermore, 2 weeks after exposure to the documentary, participants
were sent an online questionnaire and requested to complete the measures for the outcome
variables again. This follow-up survey enabled us to observe potential change in self-reported
behaviour, which was not possible directly after participation. Additionally, we were able to
observe effects on non-behavioural outcome variables 2 weeks after the subjects had watched
the documentary, and compare them with the immediate effects observed directly after the
screening of the film.
The control group was subjected to exactly the same procedure as the manipulation group
with the sole exception that the former watched the film about the Pirate Party. Having
completed all the tasks, the participants in both groups received a monetary incentive of 18
euros (about 23.6 USD).

Sample

Table 1 shows the distributions of some basic demographic and socioeconomic characteristics
among the members of the manipulation and control groups in our sample. As can be seen,
randomization led to a relatively well-balanced distribution between the two groups as regards
place of residence, sex, nationality, and net income. Only with regard to age did the standard-
ized effect size indicate a moderate baseline difference.3

Measures

Drawing on the literature on organic food purchase and environmental labelling, we developed
nine multi-item constructs to measure attitudes, behavioural intentions, self-reported behav-
iour, and trust in organic food labels as dependent variables for our study. An overview of all
the constructs and their indicator items is included in the Appendix. Almost all the items were
statements that had to be evaluated on seven-point rating scales, ranging from 1 (disagree
absolutely) to 7 (agree absolutely).
Because we expected that watching the documentary would lead to negative effects on
diverse attitudes towards organic food, we developed five attitudinal constructs which were all
measured by three or four items. Three of these recorded aspects of the perceived conse-
quences of organic food purchase that were explicitly mentioned in the TV documentary
presented in the manipulation condition. The first one (perceived environmental impact)
focused on the positive impacts of organic food production on the ecological environment
(Honkanen et al. 2006). Because many people believe that the consumption of organic food is
good for their health (Grankvist and Biel 2001), another construct (perceived impact on human
health) referred to human health issues. Further, as organic food production is frequently
considered as being promotive of animal welfare (Harper and Makatouni 2002), a third

3
In order to exclude distortions of estimated manipulation effects due to baseline differences in age, we re-
estimated all direct manipulation effects on the outcome variables controlling for age. Results indicated that the
re-estimated manipulation effects did not differ from those estimated without controlling for age.
Consumer Response to Negative Media Information About Organic Food 393

Table 1 Distribution of demographic and socioeconomic covariates

Characteristic Total Manipulation Control Baseline


group group difference

Place of residence Village (up to 2000) 13 (9.0 %) 6 (8.3 %) 7 (9.6 %)


Small town 30 (20.7 %) 15 (20.8 %) 15 (20.6 %) Cramér’s V=.05
(up to 20000)
Large town 51 (35.2 %) 24 (33.3 %) 27 (37.0 %)
(up to 150000)
Big city (>150000) 51 (35.2 %) 27 (37.5 %) 24 (32.9 %)
Sex Female 86 (59.3 %) 44 (61.1 %) 42 (57.5 %) Cramér’s V=.04
Male 59 (40.7 %) 28 (38.9 %) 31 (42.5 %)
Nationality German 126 (86.9 %) 63 (87.5 %) 63 (86.3 %) Cramér’s V=.02
Other 19 (13.1 %) 9 (12.5 %) 10 (13.7 %)
Age Mean 23.48 22.61 24.34
SD 3.55 3.27 3.62 Cohen’s d=.51
n 145 72 73
Net income Mean 517.50 511.15 523.48
(€ per month) SD 263.30 263.47 264.94 Cohen’s d=.05
n 134 65 69

construct measured the perception of the influence of organic food production on animal
welfare (perceived impact on animal welfare).
The last two attitudinal constructs relate to quality aspects of organic food that were not
explicitly mentioned in the TV documentary about certified organic food products. Because
consumers frequently believe that organic food products contain more natural ingredients than
conventional products (Lockie et al. 2004), another construct (perceived degree of naturalness)
measured these beliefs. Finally, because organic food products may be regarded as goods of
high quality (Grankvist and Biel 2001), we included a construct Bperceived quality of organic
food.^
As we have already said, behavioural intentions are considered to be important medi-
ators between attitudes and actual behaviour (Fishbein and Ajzen 2010). Thus, a three-
item construct recorded behavioural intentions with regard to the purchase of organic food
products and the attention paid to organic food and organic food labels (Thøgersen 2000)
in the future.
Beyond that, two constructs measured self-reported behaviour related to the purchase of
organic food. The first (self-reported behaviour 1) was a four-item construct that quantified the
amount of organic food purchased by the participants. Hence, the construct score equalled the
average share of organic food products with regard to total purchases. The other construct
(self-reported behaviour 2) also consisted of four items. These, however, were measured on
rating scales and formulated in a more general manner.
Finally, trust in environmental labels has been found to be a key determinant in deciding
whether to buy eco-labelled products or not (Thøgersen 2000). Thus a three-item construct
denoted as Btrust in organic food labels^ was also included.
Table 2 depicts the mean values of all the variables, their standard deviations, and
numbers of cases. Except for Bself-reported behaviour 1,^ all the constructs represent the
394 C.E. Müller, H. Gaus

average item scores on the respective seven-point rating scales. As the results suggest, the
pre-intervention values are fairly well balanced between the conditions, except for the
measure Bperceived impact on animal welfare,^ where the standardized difference in
means indicates a small baseline difference. Yet, because we controlled for the pretest
measures in all our analyses, this baseline difference does not pose a problem for
estimating manipulation effects. Furthermore, the Cronbach’s alpha coefficients suggest
internal consistency of all the constructs because they clearly exceed the conventionally
applied threshold of alpha≥.7.

Data Analysis

In order to investigate the effects of exposure to critical information about organic food
products and labels, we applied two analytical strategies. First, we assessed the direct
effects on the nine variables presented in the previous section. Second, we investigated
whether manipulation effects on behavioural intentions were mediated by attitudes to-
wards organic food products and trust in organic food labels. Moreover, we also inves-
tigated whether manipulation effects on self-reported behaviour were mediated by
attitudes towards organic food products, trust in organic food labels, and behavioural
intentions.

Direct Effects

We tested the direct effects of the manipulation by estimating two mixed multivariate analysis
of variance (MANOVA) models, one with a three-level within-subjects factor (time) and a two-
level between-subjects factor (manipulation) for all the variables measured at three points in
time (behavioural intentions, trust in organic food labels, perceived environmental impact,
perceived impact on human health, perceived impact on animal welfare, perceived degree of
naturalness, and perceived quality of organic food), and one with a two-level within-subjects
factor (time) and a two-level between-subjects factor (manipulation) for assessing the effects
on the two behavioural variables.
For all the variables measured at three points in time, we also investigated manipulation
effects on each individual variable, both directly after exposure to the documentary and
2 weeks later, by calculating the differences between manipulation and control group in terms
of their change scores from pretest to posttest (immediate effects) and from pretest to follow-up
test (fortnight effects) respectively. Subsequently, we calculated the differences between the
immediate effects and the fortnight effects and applied bootstrap simulations in order to
determine whether these differences significantly differed from zero. In doing so, we were
able to assess whether manipulation effects decreased or increased from the posttest to the
follow-up test.

Path Analysis

Because analysing direct effects did not tell us how the effects of exposure to the
critical information presented in the ARD documentary occurred, we tried to identify
mechanisms that underlie the potential effects of the manipulation on behavioural
intentions and self-reported behaviour. In order to do this, we derived two testable
path models (see Figure 2) from the theoretical model presented in Figure 1. In
Table 2 Descriptive statistics of measures

Construct M (Mpre) M (Mpost) M (Mfollow-up) M (Cpre) M (Cpost) M (Cfollow-up) Cronbach’s alpha Baseline differences
(Cohen’s d)
(SD/n) (SD/n) (SD/n) (SD/n) (SD/n) (SD/n) αpre αpost αfollow-up

Self-reported behaviour 1 13.73 – 13.21 16.18 – 16.67 .80 – .82 .14


(16.60/72) – (16.98/69) (18.94/73) – (21.00/73)
Self-reported behaviour 2 3.95 – 3.74 4.15 – 4.07 .90 – .87 .13
(1.54/72) – (1.36/69) (1.65/73) – (1.60/69)
Behavioural intentions 3.93 3.52 3.41 3.96 3.98 3.81 .90 .90 .92 .02
(1.70/72) (1.57/72) (1.58/69) (1.57/73) (1.55/73) (1.52/72)
Trust in organic food labels 4.03 2.39 2.80 4.00 3.85 3.90 .78 .77 .83 .02
(1.46/72) (1.10/72) (1.39/68) (1.38/73) (1.34/72) (1.55/70)
Perceived environmental impact 4.94 3.25 3.43 4.90 4.82 4.59 .79 .88 .88 .04
(1.09/72) (1.11/71) (1.26/64) (1.02/73) (1.23/73) (1.34/71)
Perceived impact on human health 4.38 3.61 3.60 4.44 4.39 4.20 .79 .78 .81 .04
(1.36/72) (1.28/72) (1.44/69) (1.39/73) (1.30/73) (1.29/72)
Perceived impact on animal welfare 5.18 2.63 2.87 4.66 4.79 4.48 .81 .91 .92 .38
Consumer Response to Negative Media Information About Organic Food

(1.32/72) (1.33/72) (1.48/69) (1.42/73) (1.41/73) (1.63/73)


Perceived degree of naturalness 3.96 3.39 3.31 3.80 3.75 3.69 .82 .90 .91 .12
(1.48/71) (1.41/72) (1.53/68) (1.29/71) (1.45/72) (1.51/71)
Perceived quality of organic food 4.49 3.02 3.26 4.38 4.31 4.10 .83 .90 .89 .09
(1.18/72) (1.16/72) (1.35/69) (1.24/72) (1.11/73) (1.24/73)

Self-reported behaviour 1 represents the average score of four items measuring the proportions of purchased organic food products (see Appendix). All other constructs represent the
average scores of the respective seven-point rating scales
Mpre mean value of manipulation group in the pretest, Mpost mean value of manipulation group in the posttest, Mfollow-up mean value of manipulation group in the follow-up test, Cpre
mean value of control group in the pretest, Cpost mean value of control group in the posttest, Cfollow-up mean value of control group in the follow-up test
395
396 C.E. Müller, H. Gaus

Model 1 Self-reported
Behaviour 1 (PRE)

(+)
(+) Behavioural
Attitudes (PRE) Attitudes (PT) Intentions (PRE)
Self-reported
(+) Behaviour 1 (FU)
(-)
(+) (+)
(-)
Manipulation
(-) Behavioural
(Negative Media
Intentions (PT)
Information) (-) (+)
(-)

Self-reported
(+)
Behaviour 2 (FU)
(+)
Trust (PRE) Trust (PT)
(+)

Self-reported
Behaviour 2 (PRE)

Model 2

(+) Behavioural
Attitudes (PRE) Attitudes (FU) Intentions (PRE)

(+)
(-)
(+)

Manipulation
(-) Behavioural
(Negative Media
Intentions (FU)
Information)
(-)

(+)
(+)
Trust (PRE) Trust (FU)

Note. PRE = Pretest measure. PT = Posttest measure. FU = Follow-up test measure.

Fig. 2 Path models

addition to the direct manipulation effects, both of these models depict mediated (i.e.,
indirect) manipulation effects on behavioural intentions and self-reported behaviour.
Model 1 depicts the direct manipulation effects on attitudes, trust, behavioural intentions,
and self-reported behaviour. Moreover, model 1 includes the mediated manipulation effect on
behavioural intentions (via attitudes and trust) and the mediated effect on self-reported
behaviour (via attitudes, trust, and behavioural intentions). As can be seen in Figure 2, model
1 contains endogenous4 constructs measured at two different points in time. Although this may
seem somewhat unconventional at first sight, it was the only way of testing a path model
containing information on all the constructs included in our theoretical model (Figure 1). The
reason for this has to do with the temporal focus of the behavioural intentions measure and the
two self-reported behavioural variables. Whereas the intentions measure focuses on respon-
dents’ intentions to perform certain kinds of behaviour in the future, the behavioural measures

4
Endogenous constructs are variables that are predicted by at least one other construct in the path model.
Consumer Response to Negative Media Information About Organic Food 397

capture behaviour that was performed in the past. Therefore, we predict self-reported behav-
iour measured 2 weeks after exposure to the ARD documentary (at the follow-up test) by
intentions measured directly after the intervention (at the posttest).
In model 2, we could only estimate direct and mediated manipulation effects on trust,
attitudes, and behavioural intentions that occurred 2 weeks after the intervention (at the follow-
up test). The reason for this is that we did not collect data on subjects’ self-reported behaviour
after the follow-up test that could have been predicted by constructs measured at the follow-up
test. Thus, model 2 replicates parts of model 1 and allows us to assess whether potentially
mediated effects of the manipulation on behavioural intentions were stable over a period of
2 weeks.
Further, in Figure 2, one can see that model 1 and model 2 have some things in common,
such as the fact that pretest information was used in both models as a means of reducing error
variance and enhancing precision. Moreover—in contrast to the analysis of direct effects
presented in the previous section—in both models, we included one overall attitudinal
construct instead of individual constructs for all five attitudes. The overall attitudinal construct
was operationalized by the 16 items of perceived environmental impact, perceived impact on
human health, perceived impact on animal welfare, perceived degree of naturalness, and
perceived quality of organic food. Finally, in both models, we incorporated the exposure to
the organic food TV documentary as an exogenous dummy variable (Bagozzi et al. 1991).
The models were tested by using the partial least squares approach (PLS) to
structural equation modelling (SEM). Relying on SEM enabled us to estimate the
relations between non-observable latent constructs and the items employed to
operationalize these latent constructs (denoted as Bmeasurement models^) as well as
the relations between the latent constructs (denoted as Bstructural model^) simulta-
neously (e.g., Haenlein and Kaplan 2004). Consequently, using SEM allowed us to
assess the quality of the specified measurement models as well as the direction,
strength, and significance of relations between the measured constructs at the same
time.
We used PLS-SEM instead of the more common covariance-based SEM because it is less
restrictive as regards data requirements, particularly with respect to sample size and distribu-
tional assumptions (Hair et al. 2011). Although PLS-SEM has also been criticized (e.g.,
Rönkkö and Evermann 2013), extensive research on the performance of PLS-SEM (e.g.,
Henseler et al. 2014; Reinartz et al. 2009) as well as major advancements—such as the
availability of a global measure of goodness of fit in the form of the standardized root mean
square residual (SRMSR)—suggests that PLS-SEM is a viable tool for analysing relations
between latent constructs. Taking into account the size of our sample and the fact that
differences between PLS and covariance-based SEM are mostly at very low levels (Reinartz
et al. 2009), we believe that choosing PLS was appropriate.
Before estimating the PLS models, we had to decide whether the measurement
models of our constructs should be specified as reflective or formative. According to
Jarvis et al. (2003, p. 200), in a reflective measurement model, Bcovariation among
the measures is caused by, and therefore reflects, variation in the underlying latent
factor.^ In formative measurement models, Bthe measures are hypothesized to cause
changes in the underlying construct^ (p. 201). In our models, all the measurement
models were specified as reflective because we assumed covariation among all the
items that were supposed to measure the same construct. For all PLS estimations, we
used the software SmartPLS 3.0 (Ringle et al. 2014).
398 C.E. Müller, H. Gaus

Results

Direct Effects

Effect Estimates

The developments of all the dependent variables over the three points of measurement in both
the manipulation and the control group are presented in Figure 3. With the first repeated
measures MANOVA, we analysed the direct manipulation effect on behavioural intentions, trust,
perceived environmental impact, perceived impact on human health, perceived impact on
animal welfare, perceived degree of naturalness, and perceived quality of organic food. The
result of the multivariate test showed a significant interaction effect between point in time and
manipulation group, Wilks’ lambda=.40, F(14, 109)=11.86, p<.001, ηp2 =.60.
Results of follow-up analyses (Table 3) indicate that exposure to the ARD documentary led
to strong and negative immediate effects on trust in organic food labels, perceived environ-
mental impact, perceived impact on animal welfare, and perceived quality of organic food.
Moreover, we found a negative immediate effect of moderate size on perceived impact on
human health, and small negative immediate effects on perceived degree of naturalness and
behavioural intentions. As regards the effects occurring 2 weeks later in the follow-up test, we

Fig. 3 The development of scores over time


Consumer Response to Negative Media Information About Organic Food 399

Table 3 Follow-up analyses of direct effects

Dependent variable τPost (d) τFollow-up (d) Δ (τ) |t(Δ)| p(Δ)

Behavioural intentions −0.43* (0.43) −0.39* (0.37) −0.04 0.15 .878


Trust in organic food labels −1.47*** (1.31) −1.17*** (0.84) −0.31 1.02 .308
Perceived environmental impact −1.60*** (1.59) −1.19*** (1.08) −0.42 1.61 .110
Perceived impact on human health −0.73*** (0.78) −0.54** (0.51) −0.18 0.77 .440
Perceived impact on animal welfare −2.67*** (2.20) −2.14*** (1.48) −0.53 1.70 .092
Perceived degree of naturalness −0.51* (0.43) −0.57* (0.44) 0.06 0.19 .847
Perceived quality of organic food −1.40*** (1.48) −0.94*** (0.86) −0.45 1.86 .065

Standard errors for calculating |t(Δ)| were obtained via bootstrap resampling (10000 replications). p(Δ) based on
two-tailed tests
τPost difference in change scores between manipulation and control group from pretest to posttest, τFollow-up
difference in change scores between manipulation and control group from pretest to follow-up test, d standard-
ized mean difference (Cohen’s d), Δ (τ) difference between τPost and τFollow-up
***p<.001; **p<.01; *p<.05

still observed strong negative effects on trust in organic food labels, perceived environmental
impact, perceived impact on animal welfare, and perceived quality of organic food. A negative
effect of moderate size was found on perceived impact on human health. Finally, in the follow-
up test, we found small negative effects on perceived degree of naturalness and behavioural
intentions. Table 3 also depicts the results as regards the development of effects between the
posttest and the follow-up test and reveals that—except for the effects on perceived degree of
naturalness—the magnitudes of effects decreased from the posttest to the follow-up test with
regard to all the dependent variables. Yet all these changes were insignificant.
With the second repeated measures MANOVA, we tested whether exposure to the documen-
tary directly affected the two self-reported behavioural variables. The result of the multivariate
test did not show a significant interaction effect between time and manipulation, Wilks’
lambda=.99, F(2, 135)=0.60, p=.550, ηp2 =.01. Because we did not find a significant
multivariate effect, there was no need for any further analyses in this case.5

Discussion of Direct Effects

First of all, analysing the direct effects of exposure to the ARD documentary revealed that
respondents’ behavioural intentions, trust in organic food labels, and five attitudinal variables
were significantly affected directly after exposure and 2 weeks later.
An important question in this context is whether these effects changed over time, because
one could argue that when the time elapsing after exposure increases, memorization (and thus
the presence) of the documentary’s contents might decrease. Hence, one would expect a
decrease in effects on intentions, trust, and attitudes from the posttest to the follow-up test.
This expectation would be in line with previous research on the effects of media information
(e.g., Sampei and Aoyagi-Usui 2009) and informational interventions (e.g., Staats et al. 2000).

5
Because of the right-skewed distribution of self-reported behaviour 1, however, we replicated all analyses
involving self-reported behaviour 1 throughout this article with a log-transformed version of self-reported
behaviour 1. Log-transforming self-reported behaviour 1 dramatically decreased the skewness of the distribution.
Yet the results of effect estimations did not change substantially in terms of significance or effect size when
compared with the original estimations. We therefore chose to stick with the original models.
400 C.E. Müller, H. Gaus

However, one could also argue that effects might become even stronger over time because of
respondents’ increased interest in fraud connected with certified organic food and the actual
impacts of organic food production on the environment, human health, and animal welfare.
Accordingly, respondents could lose even more confidence in the reliability of certified organic
food products. The absolute numbers presented in Table 3, however, show that in the majority
of cases, the effects got smaller over time. Yet all these decreases in effects are insignificant.
Thus, it seems that the effects of negative information about certified organic food were
relatively stable over a period of 2 weeks. Having said that, the effects might well have
decreased further in size if more time had elapsed since exposure to the ARD documentary due
to a diminishing presence of the knowledge transmitted.
Looking at the consequences for actual behaviour, we found no significant direct manip-
ulation effects on respondents’ self-reported behaviour 2 weeks after exposure. One reason for
the absence of an effect in this case may be the complexity of organic food purchase behaviour,
which is determined by many different factors such as emotions, attitudes, values, intentions,
socio-demographic characteristics, and macro-level determinants such as cultural, technolog-
ical or economic factors (Aertsens et al. 2009). Consequently, changing organic food purchase
behaviour is not easy. Additionally, organic food purchase behaviour is highly habitual
(Grankvist and Biel 2001), which makes it even more resistant to change.

Path Analysis

Model Evaluation

Both estimated PLS models largely meet standard quality criteria. First, the values of the
SRMSR are .106 for model 1 and .107 for model 2. These values only marginally exceed the
suggested threshold6 of SRMSR <.1, indicating a satisfying overall goodness of model fit.
Second, the explanation of variance of all endogenous constructs exceeded the suggested
threshold of R2 =.33 (Chin 1998) in both models. In model 1, we observed values of adjusted
R2 of .65 for attitudes, .56 for trust, .70 for behavioural intentions, .74 for self-reported
behaviour 1, and .69 for self-reported behaviour 2. In model 2, adjusted R2 was .53 for
attitudes, .38 for trust, and .66 for behavioural intentions. Moreover, the Stone–Geisser
criterion (Q2) for assessing predictive relevance exceeded the value of zero for all the
constructs in both models (Fornell and Cha 1994). Finally, variance inflation factors in both
models well below the conventionally accepted threshold of 10 suggest that multicollinearity
was not a problem.
In the evaluation of the individual measurement models, results of the PLS estimations
showed that standard criteria of convergent and discriminant validity were met in almost every
case. In both models, all the constructs possessed composite reliability with values above .8.
Moreover, analysing the average variance explained (AVE) by each construct revealed that
almost all the latent constructs explained more than 50 % of the variance of their indicators.
Only the attitudinal construct measured in the pretest fell below the suggested threshold of .5
(Fornell and Larcker 1981) with AVE=.4 in both models. Furthermore, the constructs in both
models fulfil the Fornell–Larcker criterion of discriminant validity (Fornell and Larcker 1981),
which states that the average variance explained by a single construct should exceed the
variance shared with other constructs measuring something different. We did not apply this
criterion for assessing the discriminant validity between constructs measured in the posttest/
6
See http://www.smartpls.de/documentation/srmr (last accessed June 4, 2015).
Consumer Response to Negative Media Information About Organic Food 401

follow-up test and their respective pretest measures because, by definition, they do not measure
something different but the same entity at a different point in time.

Effect Estimates

Table 4 depicts the results of the estimated PLS models. As expected, there were significant
and strong direct effects of the manipulation on attitudes and trust in both models. By contrast,
the PLS models did not show any significant direct effect of the manipulation on behavioural
intentions. However, we did find small significant indirect and total effects of the manipulation
on behavioural intentions. Thus, as hypothesized, the influence of exposure to negative
information about organic food products is effectively mediated. In order to investigate this
in more detail, we had to consider the effects of attitudes and trust on behavioural intentions.
As is clearly visible in Table 4, attitudes moderately affected behavioural intentions in both
models (although the effect was stronger in model 1), whereas there was no significant
relationship between trust and intentions at all. Therefore, our models suggest that the effects

Table 4 Estimates for the hypothesized structural model

Path β |t| p

Model 1 Model 2 Model 1 Model 2 Model 1 Model 2

Direct effects
Manipulation→attitudes −.58 −.42 12.21 7.52 <.001 <.001
Manipulation→trust −.48 −.31 8.64 4.38 <.001 <.001
Manipulation→intentions .04 −.02 0.77 0.36 .442 .722
Manipulation→behaviour 1 −.01 0.22 .829
Manipulation→behaviour 2 −.05 0.98 .329
Attitudes (PRE)→attitudes .60 .62 9.64 10.36 <.001 <.001
Attitudes→intentions .37 .20 3.84 2.41 <.001 .016
Trust (PRE)→trust .57 .54 9.13 7.19 <.001 <.001
Trust→intentions −.03 .02 0.39 0.23 .698 .815
Intentions (PRE)→intentions .65 .69 11.21 11.03 <.001 <.001
Intentions→behaviour 1 .11 2.44 .015
Intentions→behaviour 2 .19 2.99 .003
Behaviour 1 (PRE)→behaviour 1 .84 18.95 <.001
Behaviour 2 (PRE)→behaviour 2 .73 13.25 <.001
Indirect effects of manipulation
Manipulation→intentions −.20 −.09 4.29 2.50 <.001 .013
Manipulation→behaviour 1 −.02 1.84 .066
Manipulation→behaviour 2 −.03 1.90 .058
Total effects of manipulation
Manipulation→intentions −.15 −.11 2.91 2.00 .004 .045
Manipulation→behaviour 1 −.03 0.60 .548
Manipulation→behaviour 2 −.08 1.59 .112

Standard errors for calculating | t |were obtained via bootstrap resampling (10000 replications). p value based on
two-tailed tests
402 C.E. Müller, H. Gaus

of the ARD documentary on behavioural intentions were mainly mediated by attitudes towards
organic food products.
When we consider the direct and mediated effects of the manipulation on self-reported
behaviour in model 1, the estimates of path coefficients show that the manipulation did not
significantly affect self-reported behaviour directly. Moreover, there were no significant
indirect and total effects of the manipulation on self-reported behaviour 1 or 2 either. However,
as expected, we found significant (yet weak) direct relationships between intentions and both
behavioural constructs.
Finally, a closer look at Table 4 reveals that the pretest measures were good predictors for
the posttest and follow-up test measures used in models 1 and 2. The resulting reduction in
error variance is demonstrated by the high values of adjusted R2 presented in the previous
section and by the high beta coefficients of the pretest measures.

Discussion of Path Analysis

In accordance with our theoretical model (and the effects presented in Table 3), the results of
the PLS estimations showed that the manipulation directly affected attitudes and trust, both in
the posttest (model 1) and a fortnight later (model 2). When controlling for attitudes and trust,
we did not find any significant direct manipulation effect on behavioural intentions in either
model. The absence of direct effects on behavioural intentions in both PLS models may appear
to contradict the results presented in Table 3. However, a closer look at Table 4 reveals that the
reason for this contradiction lies in the nature of the manipulation effect on behavioural
intentions in both models. More precisely, the significant indirect manipulation effects on
intentions in combination with the close-to-zero coefficients for the direct effects of the
manipulation on intentions reveal an almost complete mediation (MacKinnon et al. 2007).
In contrast to our previous assumptions, the negative information about organic food products
thus did not affect behavioural intentions directly but was almost fully mediated. This is true
for both models 1 and 2, implying that the mediated effect of the manipulation on behavioural
intentions was stable over a period of 2 weeks.
Another particularity of our model evaluation is that we did not find any significant effects
of trust on behavioural intentions in either model 1 or model 2. Therefore, these results
contradict our assumptions and previous research (e.g., Krystallis and Chryssohoidis 2005;
Yin et al. 2009), where trust was found to be an important predictor of consumers’ willingness
to purchase and pay for certified organic food products. One explanation for this finding may
be seen in the nature of the effect of trust on intentions. For example, Thøgersen (2000) found
that trust in eco-labels moderates the effect of attitudes towards labels on attention paid to
labels. Yet follow-up analyses revealed that there was no such interaction effect in our study. A
further explanation may be found in the interrelatedness of the constructs trust and attitudes.
More specifically, the correlations between the constructs trust and overall attitudes were
strong with r=.65 in model 1 and r=.61 in model 2. Thus, including the overall attitudes
construct in the PLS model led to the disappearance of the effect of trust on intentions, which is
revealed by the fact that trust has small but significant direct effects on intentions in model 1
and model 2 when the attitudinal construct is removed.
Furthermore, with regard to self-reported behaviour, we found neither significant direct nor
indirect manipulation effects in model 1. These results are congruent with the results of the
estimated MANOVA model reported previously. Because there were no direct manipulation
effects on self-reported behaviour at all, it is not surprising that we did not find any mediated
Consumer Response to Negative Media Information About Organic Food 403

effects either. An obvious reason for the absence of indirect effects is that behavioural
intentions are only weakly related to self-reported behaviour in our study. Moreover, the effect
of the manipulation on behavioural intentions is almost exclusively mediated by attitudes,
which means that the manipulation effect on trust does not contribute to any change in
behavioural intentions at all. Therefore, as long as the influence of attitudes towards organic
food products on behavioural intentions and the effect of behavioural intentions on actual
behaviour are not considerably stronger than in our study, it seems questionable whether a
short-term increase in consumers’ knowledge due to received critical information alone is
capable of substantially changing behaviour. This interpretation is in line with previous
research about the positive effects of increased knowledge on pro-environmental consumer
behaviour, where it is stated that the transmission of information on its own is not a very
effective way of changing behaviour (e.g., Stern 1999).
However, it is conceivable that the indirect effects on self-reported behaviour would have
been stronger with a more representative sample. Because our sample consisted solely of
university students with rather low monthly income, it is reasonable to assume that the effect of
behavioural intentions on actual behaviour may have been underestimated. This is because
even if the respondents in our sample intended to purchase more organic food and pay more
attention to organic food (labels) in the future, this may not have affected self-reported
behaviour due to a lack of financial means, which renders the purchase of more expensive
organic food more unlikely. Moreover, we admit that the indirect manipulation effects on self-
reported behaviour might also become stronger if additional mediators were included in the
model. After all, important variables such as emotions, personal moral norms, social norms, or
perceived behavioural control (Aertsens et al. 2009; Chen 2007; Fishbein and Ajzen 2010),
which may have been affected by negative information and may in turn have had effects on
behavioural intentions, were not considered in our model. Finally, the way the behavioural
intentions were measured may also have contributed to the underestimation of manipulation
effects on behavioural intentions and self-reported behaviour. This issue, however, is discussed
in more detail in the subsequent section.

Limitations and Implications for Future Research

There are some limitations of our study that should be noted, and these call for further research
efforts. One limitation concerns the restricted external validity of the findings. Because study
participants were university students, our sample is not representative of the target population
of the documentary. Moreover, our experiment took place in an artificial research environment.
Respondents did not watch the documentary in a private space as they probably would do in
real life. Instead, they watched it in a classroom together with other students, which may have
influenced their response behaviour. Consequently, in order to enhance external validity, in
future studies, participants should be sampled randomly from the populations of interest and
receive the stimulus in more natural settings.
Further, we did not find any evidence that the ARD documentary affected self-reported
behaviour. Yet we only investigated the effect of a particular source of information at one point
in time and, as is known from advertising research (Bettman 1979), information works much
more effectively when consumers are exposed to it several times. Consequently, it would be
interesting to find out whether critical information about certified organic food products has
stronger effects if study participants are exposed to it at several points in time. In addition to
404 C.E. Müller, H. Gaus

that, we only investigated the effects of watching a film. Therefore, future studies should
compare the consequences of different kinds of medium such as newspapers, radio, or the
Internet.
One should also bear in mind that our manipulation was readily available and not tailor-
made for our research. It therefore contained information on fraud on the part of the producers
of certified organic food and on consumers’ misapprehension related to the consequences of
purchase. However, a third and potentially harmful kind of information was not included,
namely, deception about the standards of certification. Both fraud and misapprehension could
be less likely than deception to damage the generic image of organic produce: Fraud can be
attributed to incidental commercial parties and misapprehension to the respondent him or
herself. Deception about the standards of certification is attributable to the certifying institu-
tion. The ARD documentary may have contributed to this distinction by contrasting Bcheap
substandard supermarket produce^ and Bexpensive real organic produce,^ which could trans-
late into fraud being attributed to specific producers and supermarkets and damage being
deflected from the label. It is at least conceivable that a well-educated student sample could
draw similar inferences. Thus, future research should work with tailor-made manipulations that
allowed differentiating between these diverse kinds of problem and their specific impact on
consumer reactions.
Another limitation is that the impact of the manipulation depends to some degree on the
credibility of the documentary and its producer. Because trust in organic produce and trust in
(mainstream) media may be inversely correlated (e.g., Autio et al. 2009; McDonald et al.
2012), this could be an influence on the strength of respondents’ reaction to watching the
documentary. Even if ARD, as a German public-service TV station, is generally believed to be
highly credible (Grunenberg and Kuckartz 2003), this need not be the case in our specific
sample. Hence, future research should include measures of the credibility of the source of
information.
Moreover, there are uncertainties with regard to the variables employed. First, the behav-
ioural intentions construct contained one item capturing the respondents’ intention to buy more
organic food in the future and two items capturing their intention to pay more attention to
organic food (labels). Basically, it is conceivable that the manipulation group was less prepared
to buy organic products in the future after having watched the film, but more prepared to pay
attention to organic food labels in order not to be fooled. Yet we found that the mean values of
all three items in the manipulation group decreased after they had watched the documentary,
although the decrease in the intention of future purchase was greater than the decrease in the
items related to paying attention. Consequently, we admit that the manipulation effects might
have been stronger if we had only considered items related to the intention to purchase more
organic food. Accordingly, they might have been weaker if we had only considered items
related to paying more attention to organic food and labels.7
Second, the wording of the items of the intentions construct might be problematic. If
respondents disagree with the items, does this mean they will purchase less and/or pay less
attention in the future because of negative information? Or does it mean they will purchase
less/pay less attention because they already purchase a lot of organic food anyway? In order to
answer this question, we conducted subgroup analyses and found a negative correlation (r=
−.24) between self-reported behaviour 1 and intentions for respondents who claimed to

7
Both these interpretations are supported by re-estimations of the two PLS models when considering only the
intention to purchase more organic food or only the intention to pay attention to organic food (labels).
Consumer Response to Negative Media Information About Organic Food 405

purchase more than 20 % organic food (representing 25 % of our sample) and a positive
correlation (r=.34) for all other respondents. We then re-estimated all the direct effects and the
PLS models without considering participants who purchased more than 20 % organic. The
results revealed that all the effect sizes increased only marginally. Therefore, we infer that our
original estimates were quite robust against this type of bias. However, we admit that if there
had been more participants with high proportions of organic food purchases in our sample, we
might have had a bigger problem. Therefore, future studies should employ items for measuring
behavioural intentions whose scores are interpretable in a less ambiguous way.
Third, we collected information on self-reported behaviour which is known to be prone to
measurement bias (Chao and Lam 2011). Particularly with regard to self-reported behaviour 1,
measurements may have been affected by recall bias. We tried to assess the degree of validity
of self-reported behaviour 1 by calculating its correlation with self-reported behaviour 2. We
found that the correlation was r=.67 for the pretest and r=.72 for the follow-up test. Assuming
that self-reported behaviour 2 was less prone to recall bias, these high correlations may be a
hint that measuring self-reported behaviour 1 was not affected by recall bias. In our PLS model
too, both behavioural constructs met all the criteria of convergent and discriminant validity.
Fourth, our measures did not differentiate between organic food offered at supermarkets
and organic food offered at specialty organic stores. Yet the ARD documentary explicitly
makes this distinction by showing that organic food products offered at supermarkets are
particularly prone to fraud and misapprehension. Consequently, watchers of the documentary
might have felt less inclined to purchase organic food at supermarkets, but more inclined to do
so at organic specialty stores. When rating the items, members of the manipulation group had
to decide whether they were referring to attitudes, trust, and intentions in respect of organic
products offered in supermarkets or organic products offered in specialty stores. Because we
only found negative average manipulation effects in our study, most respondents seem to have
rated the items with regard to organic food products offered in supermarkets. Yet we cannot
exclude the possibility that some respondents may have referred to organic food offered in
specialty stores when rating the items. Assuming that the ratings of these subjects were more
positive than those of respondents who referred to organic food at supermarkets, we may thus
have underestimated the strengths of the manipulation effects.
Finally, although there is evidence that social desirability may be fairly unproblematic when
measuring environmental attitudes and behaviour (e.g., Kaiser et al. 1999; Milfont 2009), we
cannot rule out the possibility that there was some bias of that kind in our study. Thus, future
studies should employ methods for controlling or eliminating social desirability bias (Nederhof
1985).

Conclusion

The purpose of this research was to investigate the consequences of critical media
information about organic food products. In our study, we found significant direct
negative effects of exposure to the documentary on behavioural intentions, five
attitudinal constructs, and trust in organic food labels. However, we did not find
any significant effects on self-reported behaviour. Moreover, we found that the
influence of the documentary on behavioural intentions was almost completely medi-
ated by attitudes towards organic food products, whereas we did not find any
significant mediator effect on self-reported behaviour.
406 C.E. Müller, H. Gaus

What do these findings mean for the producers of truly organic food and the institutions that
issue organic food labels? Leaving aside ethical considerations, they may at first glance be
tempted to conclude that they have nothing to fear from negative media information because
watching the documentary had no effects on self-reported behaviour. However, our study did
reveal negative effects on attitudes towards organic food products and (to a lesser degree)
behavioural intentions, which are generally believed to be important antecedents of actual
behaviour. Particularly, the negative effect on the attitudinal variables was relatively stable over
a period of 2 weeks and might even increase with the accumulation of incidents of fraud and
misapprehension related to certified organic food. This is in line with previous work stating
that the salience of sustainability issues depends on media coverage of those issues (Thøgersen
2005). Hence, the continuous stream of media coverage on incidents of fraud and misappre-
hension could undermine the acceptance of certified organic products and, in the long run, also
negatively affect their sales.

Acknowledgments The authors wish to thank two anonymous reviewers and the editor John Thøgersen for
their valuable feedback and helpful comments.

Author Note Christoph Emanuel Mueller is a senior researcher in the Department of Sociology at Saarland
University, Germany. His research focuses on the evaluation of interventions in the area of environmental
sociology as well as on the development and transfer of methods for causal impact evaluation. [Email:
c.mueller@ceval.de]
Hansjörg Gaus is a senior researcher in the Department of Sociology at Saarland University and a visiting lecturer
in consumer behaviour at Chemnitz University of Technology, Germany. His research interests include inter-
ventions to foster environment-friendly consumption, values research, and experiential consumption. [Email:
h.gaus@ceval.de]

Appendix

Construct/item
Self-reported behaviour 1
BWhat percentage of the food you purchased in the last 2 weeks were organic products?^
BWhat percentage of the meat products you purchased in the last 2 weeks were organic products?^
BWhat percentage of the other animal-based food you purchased in the last 2 weeks (e.g., eggs, butter, and
cheese) were organic products?^
BWhat percentage of the plant-based food you purchased in the last 2 weeks were organic products?^
Self-reported behaviour 2
BWhen purchasing food, I pay attention to it having been produced organically.^
BIf I have a choice when purchasing food, I prefer organic products to products made by conventional means.^
BWhen I am purchasing food, it makes no difference to me whether it has been produced organically or not.^ (r)
BI often purchase organic food.^
Behavioural intentions
BI will purchase organic products more often in future than I do at present.^
BWhen purchasing food in future, I will pay more attention to whether or not it has been produced organically.^
BWhen purchasing food in future, I will pay more attention to whether or not it has been certified with an organic
label.^
Trust in organic food labels
Consumer Response to Negative Media Information About Organic Food 407

BIf a product has been certified with an organic label, I trust that it has a guaranteed organic production origin.^
BI do not believe that all certified organic products really are organic.^ (r)
BOrganic labels are important because they guarantee that the products concerned really do come from organic
production.^
Perceived environmental impact
BOrganic products are better for the environment than food produced conventionally.^
BOrganic products are produced in an environmentally friendly way.^
BIt doesn’t make any difference to the environment whether a product has an organic production origin or not.^
(r)
BBuying organic products means protecting the environment.^
Perceived impact on human health
BI believe that food with an organic label is healthier than conventional food.^
BIt doesn’t make any difference to your health whether food has an organic production origin or not.^ (r)
BAnyone who sets store by healthy nutrition should give organic food preference over conventional food.^
Perceived impact on animal welfare
BI believe that purchasing meat products with an organic label makes a contribution to improving the conditions
under which animals are farmed.^
BI am convinced that animals in organic farming facilities have a better life than animals from conventional
farming.^
BIn my opinion, meat products with an organic label come from a farming facility in which the animals do not
have to suffer so much pain.^
Perceived degree of naturalness
BOrganic products do not contain any artificial additives.^
BOrganic products are free of chemical and hormonal residues.^
BOrganic products contain only natural ingredients.^
Perceived quality of organic food
BOrganic products are of high quality.^
BAnyone who sets store by high-quality nutrition should purchase organic products.^
BOrganic labels are a sign of high-quality food.^

Note. (r) = Item was recoded for analyses.

References

Aertsens, J., Verbeke, W., Mondelaers, K., & Van Huylenbroeck, G. (2009). Personal determinants of organic
food consumption, a review. British Food Journal, 111, 1140–1167.
Ahluwalia, R. (2002). How prevalent is the negativity effect in consumer environments? Journal of Consumer
Research, 29, 270–279.
Ahluwalia, R., Burnkrant, R. E., & Unnava, H. R. (2000). Consumer response to negative publicity: the moderating
role of commitment. Journal of Marketing Research, 37, 203–214.
Ahluwalia, R., Unnava, H. R., & Burnkrant, R. E. (2001). The moderating role of commitment on the spillover effect
of marketing communications. Journal of Marketing Research, 38, 458–470.
Ajzen, I. (1988). Attitudes, personality, and behavior. Chicago: Open University Press.
Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predicting social behavior. Englewood Cliffs:
Prentice-Hall.
Amstel, M., Driessen, P., & Glasbergen, P. (2008). Eco-labeling and information asymmetry: a comparison of five
eco-labels in the Netherlands. Journal of Cleaner Production, 16, 263–276.
Anania, G., & Nisticò, R. (2004). Public regulation as a substitute for trust in quality food markets:
what if the trust substitute cannot be fully trusted? Journal of Institutional and Theoretical
Economics, 160, 681–701.
Autio, M., Heiskanen, E., & Heinonen, V. (2009). Narratives of ‘green’ consumers—the antihero, the environ-
mental hero and the anarchist. Journal of Consumer Behaviour, 8, 40–53.
408 C.E. Müller, H. Gaus

Bagozzi, R. P., Yi, Y., & Singh, S. (1991). On the use of structural equation models in experimental
designs: two extensions. International Journal of Research in Marketing, 8, 125–140.
Bettman, J. R. (1979). An information processing theory of consumer choice. Reading: Addison-
Wesley.
Bond, J., & Kirshenbaum, R. (1998). Under the radar: talking to today’s cynical consumer. New
York: Wiley.
Chao, Y.-L., & Lam, S.-P. (2011). Measuring responsible environmental behavior: self-reported and other-
reported measures and their differences in testing a behavioral model. Environment and Behavior, 43, 53–71.
Chen, M.-F. (2007). Consumer attitudes and purchase intentions in relation to organic foods in
Taiwan: moderating effects of food-related personality traits. Food Quality and Preference, 18,
1008–1021.
Chew, F., Palmer, S., Slonska, Z., & Subbiah, K. (2002). Enhancing health knowledge, health beliefs, and health
behavior in Poland through a health promoting television program series. Journal of Health Communication:
International Perspectives, 7, 179–196.
Chin, W. (1998). The partial least squares approach to structural equation modeling. In G. A. Marcoulides (Ed.),
Modern business research methods (pp. 295–336). Mahwah: Lawrence Erlbaum.
Craig, C. S., & McCann, J. M. (1978). Assessing communication effects on energy conservation. Journal of
Consumer Research, 5, 82–88.
Dahlén, M., & Lange, F. (2006). A disaster is contagious: how a brand in crisis affects other brands.
Journal of Advertising Research, 46, 388–397.
Darby, M. R., & Karni, E. (1973). Free competition and the optimal amount of fraud. Journal of Law
and Economics, 16, 67–88.
Fishbein, M., & Ajzen, I. (2010). Predicting and changing behavior: the reasoned action approach.
New York: Psychology Press.
Fornell, C., & Cha, J. (1994). Partial least squares. In R. P. Bagozzi (Ed.), Advanced methods of marketing
research (pp. 52–78). Cambridge: Blackwell.
Fornell, C., & Larcker, D. F. (1981). Evaluating structural equation models with unobservable variables and
measurement error. Journal of Marketing Research, 18, 39–50.
Giannakas, K. (2002). Information asymmetries and consumption decisions in organic food product markets.
Canadian Journal of Agricultural Economics, 50, 35–50.
Grankvist, G., & Biel, A. (2001). The importance of beliefs and purchase criteria in the choice of eco-labeled
food products. Journal of Environmental Psychology, 21, 405–410.
Grunenberg, H., & Kuckartz, U. (2003). Umweltbewusstsein im Wandel: Ergebnisse der UBA-Studie
Umweltbewusstsein in Deutschland 2002. Opladen: Leske + Budrich.
Haenlein, M., & Kaplan, A. M. (2004). A beginner’s guide to partial least squares analysis.
Understanding Statistics, 3, 283–297.
Hair, J. F., Ringle, C. M., & Sarstedt, M. (2011). PLS-SEM: indeed a silver bullet. Journal of
Marketing Theory and Practice, 19, 139–151. doi:10.2753/MTP1069-6679190202.
Harper, G., & Makatouni, A. (2002). Consumer perception of organic food production and farm
animal welfare. British Food Journal, 104, 287–299.
Henseler, J., Dijkstra, T. K., Sarstedt, M., Ringle, C. M., Diamantopoulos, A., Straub, D. W., Ketchen, D. J., Hair,
J. F., Hult, G. T. M., & Calantone, R. J. (2014). Common beliefs and reality about partial least squares:
comments on Rönkkö & Evermann (2013). Organizational Research Methods, 17, 182–209.
Herr, P. M., Kardes, F. R., & Kim, J. (1991). Effects of word-of-mouth and product-attribute
information on persuasion: an accessibility-diagnosticity perspective. Journal of Consumer
Research, 17, 454–462.
Honkanen, P., Verplanken, B., & Olsen, S. O. (2006). Ethical values and motives driving organic food
choice. Journal of Consumer Behaviour, 5, 420–430.
Hughner, R. S., McDonagh, P., Prothero, A., Shultz, C. J., & Stanton, J. (2007). Who are organic
food consumers? A compilation and review of why people purchase organic food. Journal of
Consumer Behaviour, 6, 94–110.
Jahn, G., Schramm, M., & Spiller, A. (2012). The reliability of certification: quality labels as a consumer policy
tool. Journal of Consumer Policy, 28, 53–73.
Jarvis, C. B., Mackenzie, S. B., & Podsakoff, P. M. (2003). A critical review of construct indicators
and measurement model misspecification in marketing and consumer research. Journal of
Consumer Research, 30, 199–218.
Kaiser, F. G., Ranney, M., Hartig, T., & Bowler, P. A. (1999). Ecological behavior, environmental attitude, and feelings
of responsibility for the environment. European Psychologist, 4, 59–74.
Krystallis, A., & Chryssohoidis, G. (2005). Consumers’ willingness to pay for organic food: factors that affect it
and variation per organic product type. British Food Journal, 107, 320–343.
Consumer Response to Negative Media Information About Organic Food 409

Larceneux, F., Benoit-Moreau, F., & Renaudin, V. (2012). Why might organic labels fail to influence consumer
choices? Marginal labelling and brand equity effects. Journal of Consumer Policy, 35, 85–104.
Lockie, S., Lyons, K., Lawrence, G., & Grice, J. (2004). Choosing organics: a path analysis of factors
underlying the selection of organic food among Australian consumers. Appetite, 43, 135–146.
MacKinnon, D. P., Fairchild, A. J., & Fritz, M. S. (2007). Mediation analysis. Annual Review of Psychology, 58,
593–614.
Makatouni, A. (2002). What motivates consumers to buy organic food in the UK? Results from a qualitative
study. British Food Journal, 104, 345–352.
McDonald, S., Oates, C. J., Panayiota, J. A., Young, C. W., & Hwang, K. (2012). Individual strategies for
sustainable consumption. Journal of Marketing Management, 28, 445–468.
Milfont, T. (2009). The effects of social desirability on self-reported environmental attitudes and ecological
behaviour. Environmentalist, 29, 263–269.
Nederhof, A. J. (1985). Methods of coping with social desirability bias: a review. European Journal of Social
Psychology, 15, 263–280.
Padel, S., & Foster, C. (2005). Exploring the gap between attitudes and behaviour: understanding why consumers
buy or do not buy organic food. British Food Journal, 107, 606–625.
Reinartz, W. J., Haenlein, M., & Henseler, J. (2009). An empirical comparison of the efficacy of covariance-
based and variance-based SEM. International Journal of Market Research, 26, 332–344.
Ringle, C. M., Wende, S., & Becker, J.-M. (2014). SmartPLS 3. Hamburg: SmartPLS. Retrieved from http://
www.smartpls.com.
Rönkkö, M., & Evermann, J. (2013). A critical examination of common beliefs about partial least squares path
modeling. Organizational Research Methods, 16, 425–448.
Rousseau, S., & Vranken, L. (2013). Green market expansion by reducing information asymmetries: evidence for
labeled organic food products. Food Policy, 40, 31–43.
Sampei, Y., & Aoyagi-Usui, M. (2009). Mass-media coverage, its influence on public awareness of climate
change issues, and implications for Japan’s national campaign to reduce greenhouse gas emissions. Global
Environmental Change, 19, 203–212.
Siderer, Y., Maquet, A., & Enklam, E. (2005). Need for research to support consumer confidence in
the growing organic food market. Trends in Food Science & Technology, 16, 332–343.
Sønderskov, K. M., & Daugbjerg, C. (2011). The state and consumer confidence in eco-labeling:
organic labeling in Denmark, Sweden, The United Kingdom and The United States. Agriculture
and Human Values, 28, 507–517.
Staats, H. J., van Leeuwen, E., & Wit, A. (2000). A longitudinal study of informational interventions to save energy
in an office building. Journal of Applied Behavior Analysis, 33, 101–104.
Stern, P. C. (1999). Information, incentives, and proenvironmental consumer behavior. Journal of Consumer
Policy, 22, 461–478.
Syme, G. J., Seligman, C., Kantola, S. J., & Macpherson, D. K. (1987). Evaluating a television campaign to
promote petrol conservation. Environment and Behavior, 19, 444–461.
Thøgersen, J. (2000). Psychological determinants of paying attention to eco-labels in purchase decisions: model
development and multinational validation. Journal of Consumer Policy, 23, 285–313.
Thøgersen, J. (2005). How may consumer policy empower consumers for sustainable lifestyles?
Journal of Consumer Policy, 28, 143–178.
Thøgersen, J. (2010). Country differences in sustainable consumption: the case of organic food. Journal of
Macromarketing, 30, 171–185.
Willer, H. (2011). The world of organic agriculture 2011: Summary. In H. Willer & L. Kilcher (Eds.),
The world of organic agriculture: Statistics and emerging trends 2011, FIBL-IFOAM Report (pp.
26–33). Bonn: IFOAM and Frick: FIBL.
Yin, S., Wu, L., Du, L., & Chen, M. (2009). Consumers’ purchase intention of organic food in China. Science of
Food and Agriculture, 90, 1361–1367.

You might also like