Access to this full-text is provided by PLOS.
Content available from PLOS ONE
This content is subject to copyright.
RESEARCH ARTICLE
On the psychology of environmental
preferences: The influence of contextual
priming on discrete choice experiments
Sandra NotaroID
1
*, Petr Mariel
2
, Constantinos HadjichristidisID
1
1Department of Economics and Management, University of Trento, Trento, Italy, 2Department of
Quantitative Methods, Faculty of Economics and Business, University of the Basque Country (UPV/EHU),
Bilbao, Spain
*sandra.notaro@unitn.it
Abstract
This paper addresses an important gap in discrete choice experiments literature regarding
the effect of contextual priming on preferences and willingness to pay. Contextual priming
arises when the mere context in which a survey takes place–whether interviewees are
approached in areas related or unrelated to the target issue under evaluation–can sway
stated choices. We found priming to have a significant effect on one of the analyzed attri-
butes associated with managing a natural park. We recommend interviewing participants in
locations that are neutral with respect to the attributes under investigation. This procedure
would prevent researchers from communicating incorrect recommendations to policy-
makers, natural resource planners, and managers.
1. Introduction
A challenge governments and international institutions face is how to allocate their limited
resources. For example, what portion of its budget should a government allocate to improving
the air quality in its big cities, improving education, or managing protected areas? This issue is
complex because several goods of interest, such as environmental goods, are intangible and,
therefore, do not have a market value. For such non-market goods, the regulatory bodies need
a means to capture the value citizens place on them. These values, in turn, can inform political
decisions that have palpable consequences for citizens. Two widely used methods to elicit such
values are contingent valuation [1] (and discrete choice experiments (DCE) [2–4]. Both
assume that people have stable preferences that can be elicited by asking the right questions in
surveys [5].
In DCE, framing and priming refer to the observation that how information is presented to
participants can influence their choices. Framing concerns how options or attributes of the
choice problem are presented or framed. This can include the wording, the order in which
options are presented, or the reference points used [6]. Priming refers to the influence of prior
information on a participant’s choice, that is, how the situational context influences (passively
and unintentionally) the accessibility of information that comes to mind and how this, in turn,
influences how individuals think, feel, and behave [7]. Here, we examine the priming effect of
PLOS ONE
PLOS ONE | https://doi.org/10.1371/journal.pone.0312256 October 31, 2024 1 / 20
a1111111111
a1111111111
a1111111111
a1111111111
a1111111111
OPEN ACCESS
Citation: Notaro S, Mariel P, Hadjichristidis C
(2024) On the psychology of environmental
preferences: The influence of contextual priming on
discrete choice experiments. PLoS ONE 19(10):
e0312256. https://doi.org/10.1371/journal.
pone.0312256
Editor: Kenju Akai, Shimane Daigaku, JAPAN
Received: February 17, 2024
Accepted: October 4, 2024
Published: October 31, 2024
Copyright: ©2024 Notaro et al. This is an open
access article distributed under the terms of the
Creative Commons Attribution License, which
permits unrestricted use, distribution, and
reproduction in any medium, provided the original
author and source are credited.
Data Availability Statement: Data cannot be
shared publicly because we did not ask the
respondents for consent to share the data in the
written informed consent. Data are available from
the Department of Economics, University of Trento,
Via Inama 5, 30122 Trento (Italy), E-mail: research.
dem@unitn.it, for researchers who meet the
criteria for access to confidential data.
Funding: This research was funded by the
European project LIFE11/NAT/IT/000187 “TEN”—
Trentino Ecological Network (D2 action). The
funders had no role in study design, data collection
whether the mere context in which a survey takes place–whether interviewees are approached
in areas related or unrelated to the target issue under evaluation–can sway their preferences.
Psychological research has long demonstrated that preferences are not stable but con-
structed and that they can be influenced by aspects of the environment in which a response is
elicited [8,9]. The origins of priming research date back to the late sixties and were inspired by
the spreading activation model of semantic memory, according to which the activation of a
concept spreads it to other semantically related concepts [10]. For example, activating the con-
cept of nature could also activate associated concepts with it in the memory, such as environ-
mental protection.
There are different ways in which memory constructs can be primed, such as by having par-
ticipants unscramble sentences connected to a particular theme like the environment [11] or
by infusing the location where judgment takes place with a particular fragrance [12]. Here, we
are interested in a natural and inescapable priming manipulation–the location where inter-
viewees are surveyed. Consumer research has long recognized that the atmospherics of a retail
shop, including the location, de
´cor, sounds, aromas, and lightning, can systematically impact
consumer behavior [13,14]. Situational aspects are also central to nudging strategies [15]. For
example, an effective way to gently prompt or “nudge” grocery shoppers into buying more
healthy snacks is to position healthy snacks near the cash register [16]. However, location
priming has received little attention in contingent valuation and in DCE studies on environ-
mental goods, with a few exceptions [17,18].
In the current study, we tested whether survey location influences stated preferences and
willingness to pay (WTP) for the attributes of a natural park, which were elicited with a DCE.
The survey was administered to park visitors, and the case study was of Monte Baldo Local
Natural Park (MBLNP) located in northeastern Italy. Attributes were related to biodiversity
protection (flora and fauna) and sustainable tourism development (trails and local food
products).
The rest of the paper is organized as follows. In the second section, we review the literature
on priming. In the third section, we present our methodology and then introduce and discuss
the results in the fourth section. In the fifth section, we offer conclusions.
2. The priming effect
Priming concerns how context subtly and unobtrusively makes mental content accessible and
the effect that this, in turn, can have on how people think, feel, judge, and behave [7]. Priming
was inspired by theories of memory and, in particular, by the spreading activation model [10].
According to this model, knowledge can be schematically represented as a web diagram com-
posed of nodes and links between nodes. The nodes represent concepts or cognitive units, and
the links connect semantically associated concepts. Shorter links represent stronger associa-
tions between concepts. When a cognitive unit is activated, part of the activation passes
through the links to other concepts, increasing their mental availability.
Early research on priming examined semantic priming with words. For example, partici-
pants were asked to identify as quickly and accurately as possible whether a string of letters
presented on a screen did or did not form a word [19]. It was found that words such as “nurse”
were more quickly recognized as words when they were preceded by related words such as
“doctor” than when they were preceded by unrelated words such as “airplane,” thereby sup-
porting semantic priming. Such paradigms have been used in cognitive psychology as a means
of inferring the structure of semantic representations.
Since this early research, priming has captured the interest of social psychologists, who
have demonstrated that it can affect the impressions we form of others [20] and even our
PLOS ONE
On the psychology of environmental preferences: The influence of contextual priming
PLOS ONE | https://doi.org/10.1371/journal.pone.0312256 October 31, 2024 2 / 20
and analysis, decision to publish, or preparation of
the manuscript.
Competing interests: The authors have declared
that no competing interests exist.
behavior [21]. For example, in a pioneering set of studies participants were exposed to words
connected to old age, such as Florida and bingo, and found that this made participants behave
in a way consistent with the primed concept. Participants primed with old age walked slower
to an elevator than those primed with an unrelated concept [21]. However, successive studies
failed to replicate this finding [22].
Subsequent studies suggested that such failures were in part due to the presence of moderat-
ing factors that influence the magnitude and even the directionality of the priming effect. Spe-
cifically, they suggested that priming effects are rarely direct from prime to behavior, but
rather they depend on a number of factors such as the attitudes people have towards the target
stimulus [23,24]. For example it was found that the impact of an elderly prime on behavior
critically depends on the attitude people have toward the elderly [25]. For those with a positive
attitude toward the elderly, pictures of the elderly had an impact in line with the primed con-
cept (participants walked more slowly). However, for those with a negative attitude toward the
elderly, the same prime influenced behavior in the opposite direction (participants walked
more quickly). The authors suggested that people act similarly to people they like (assimila-
tion) but dissimilarly to people they do not like (contrast).
From a similar perspective, several psychological mechanisms through which primes can
affect behavior and the variables that can moderate specific paths were discussed [24]. The
most direct mechanism is when a prime activates a construct in memory that directly activates
behavior (e.g., priming old age makes people walk more slowly). A slightly less direct way is
when a prime activates a construct in the memory that activates a goal that, in turn, drives
behavior (e.g., priming old age activates the goal of either facilitating or hindering interaction
depending on the attitudes people have toward the elderly, which then affects walking speed).
Other indirect ways exist, such as when a prime acts by changing the way we perceive a target
person, situation, or the self, which in turn affects behavior. For example, a prime that makes
people construct a prisoner’s dilemma game as a competitive game ("Wall Street Game") or a
cooperative game ("Community Game") can affect how they play it [26,27]. Thus, although
prime-to-behavior effects exist, it is challenging to make specific predictions because multiple
moderators can influence their magnitude and directionality (for challenges and critiques of
priming in social psychology research, readers can refer to the 2014 special issue of the journal
Social Cognition, What is “Social Priming”?).
Beyond cognitive and social psychology, priming has also been shown to affect environmen-
tal choices. For example, people were primed with the concept of the environment by having
them unscramble sentences that either contained words associated with the environment (e.g.,
green, ecological, earth, nature) or words unrelated to the environment [11]. The task was to
choose between television sets that differed in several dimensions, including one related to the
environment. The authors found that participants who were primed with the concept of the
environment placed more weight on the environmental dimension than did their control coun-
terparts who were primed with unrelated words. Further studies have suggested that priming
acts through the activation of relevant values, such as the value of environmental protection.
Priming has also been shown to influence choices in stated preference valuation studies [28,
29]. For example, it was examined whether honesty priming would help reduce the hypotheti-
cal bias in a DCE study [29]. Hypothetical bias is the difference between the WTP for goods
elicited by hypothetical methods and the value obtained through non-hypothetical methods
[30]. Typically, participants report higher WTP values in hypothetical settings. The authors
found that priming participants with the concept of honesty by having them unscramble sen-
tences containing words related to honesty instead of neutral words reduced the hypothetical
bias. The target good of interest was almond products that differed in attributes related to their
environmental friendliness and price.
PLOS ONE
On the psychology of environmental preferences: The influence of contextual priming
PLOS ONE | https://doi.org/10.1371/journal.pone.0312256 October 31, 2024 3 / 20
More recently, in a stated-preference study, a visual priming manipulation (a short cartoon
video of a person taking a picture with a smartphone vs. a neutral condition in which this
video was not shown) was crossed with a framing manipulation (the background in the video
was either a pristine natural landscape [positive frame] or a polluted urban landscape [negative
frame]) [28]. The authors found that the natural landscape made pro-environmental attitudes
more salient (as evidenced by higher ratings of the attribute “ethical and environmental char-
acteristics”). Furthermore, participants in the nature-priming condition were willing to pay a
higher premium to buy an environmentally friendly smartphone than participants in the
urban landscape or neutral conditions.
The mentioned studies used priming manipulations that are not always easy to implement.
For example, asking participants to unscramble words just before the target task is difficult.
However, visual priming could be implemented by changing, for example, the background of
a questionnaire. Research suggests that priming can also spontaneously and naturally occur
from the location of a survey. For example, it was shown that citizens voting in schools are
more likely to fund a school initiative than citizens voting in other polling locations [31]. The
study used data from the 2000 general elections in the state of Arizona, which included a ballot
initiative that proposed increasing educational spending by raising the state tax from 5.0% to
5.6%. This initiative received more support from citizens voting in schools, and this effect
remained statistically significant even after controlling for factors that could account for this
difference, such as voters’ political preferences or distance from schools.
The role of situational aspects on behavior has long been recognized in consumer research.
For example, in a seminal article [14]) it was highlighted the importance of the atmospherics
surrounding a target consumer product, such as the location of the retail shop, the de
´cor, aro-
mas, and lightning, on consumer behavior. A similar point was made in another study [13]
that further noted that aspects of a situation, such as the physical surroundings of a target con-
sumer product, can influence consumption to a comparable or greater degree than individual
characteristics of consumers, such as their gender, race, age, and intellect. Although these early
studies did not specifically mention priming as the underlying mechanism, they emphasized
the role of situational characteristics, including location, on behavior. Situational aspects are
also central in nudging strategies [15]. Nudges refer to any aspect of the choice architecture
that systematically and predictably alters behavior while retaining people’s freedom of choice
without using economic incentives. For example, an effective nudge to increase healthy food
consumption is to place healthy snacks in a prominent place in a grocery shop either at eye
level or near the cash registers [16].
More pertinent to the present purposes, location priming has been shown to influence peo-
ple’s tendency to engage in pro-environmental behaviors. For example, it was found that the
building in which a behavior is observed can affect that behavior [32,33]. The experimenters
recorded how people disposed of their waste during lunchtime in an eating area in a building
designed with sustainability in mind and in an eating area in a less green building. Responses
to a questionnaire suggested that people who frequented the two sites were similar in terms of
demographics. Both sites had a clearly marked disposal area with three types of bins: compost,
garbage, and recycling. The experimenters found that people in the green building were more
likely to correctly dispose of their waste than people in the less sustainable building.
Similarly, the present research investigates whether the location in which a survey takes
place affects interviewees’ responses in choice tasks regarding different management actions in
a nature park. To achieve this, we surveyed people in specific park locations, each related to a
management action under evaluation. We predicted that when the survey location was seman-
tically related (as opposed to unrelated) to the target management action, interviewees would
give more weight to the attribute associated with that action.
PLOS ONE
On the psychology of environmental preferences: The influence of contextual priming
PLOS ONE | https://doi.org/10.1371/journal.pone.0312256 October 31, 2024 4 / 20
Although we anticipated a priming effect, we had no predictions about its directionality. As
previously mentioned, the magnitude and directionality of priming effects can vary depending
on various characteristics. For example, although being interviewed close to trails could high-
light the importance of trails and the need to conserve them, different individuals could have
different beliefs about how this goal is best achieved. Some might value human intervention
positively, while others might value it negatively. Furthermore, although being interviewed
close to where a particular species lives could activate concepts related to that species, the
impact of this activation on WTP judgments depends on the affective attitudes the interview-
ees have toward the species. Thus, although we anticipated that priming would influence pref-
erences for management actions, we were mute about the directionality of this effect.
3. Materials and methods
3.1 The study area
Data for this case study are derived from a questionnaire to evaluate visitors’ preferences for
management measures at a natural park, specifically, the Monte Baldo Local Natural Park, sit-
uated in the province of Trento in northeastern Italy. Rising straight from the Garda Lake,
MBLNP covers 46.5 km
2
, from a few hundred meters above sea level to an altitude of over
2,000 meters. The park contains nine protected areas: five Natura 2000 areas and four regional
or local reserves. The park is abundant in plant biodiversity. In particular, there are 28.7 spe-
cies of flora per km
2
, whereas other protected areas in the province of Trento have 2.3 species
per km
2
. Ten species of flora are protected by the European Union, and 60 species of wild
orchids are present in the park. Thanks to its extraordinary biodiversity, this area has been a
popular destination for naturalists, apothecaries, and pharmacists since 1400 and is known as
“Hortus Italiae” (Garden of Italy). Several animal species live in the park, including rare rep-
tiles and amphibians, such as the protected yellow-bellied toad (Bombina variegata).
MBLNP is one of the oldest parks in the reserve network (RN) of the Province of Trento.
The RN was set up in 2008 as part of a European Project funded under the LIFE programme.
The RN is a network of Natura 2000, regional or local protected areas, and interconnection
zones managed to protect natural resources as well as to support the sustainable development
of these areas, which are primarily dependent on agriculture and tourism [34]. RN is locally
managed along with stakeholder participation under the supervision and coordination of the
Province of Trento [35].
The name MBLNP has been used since 2013; the original name was the Brentonico Reserve
Network. The distinction of a local natural park was given to this area because it meets specific
naturalistic and territorial criteria required by the provincial law. The participation of local
stakeholders is fundamental in managing the MBLNP, which is aimed at improving the sus-
tainable development of agriculture and tourism while preserving biodiversity. An essential
goal of the park is the protection and promotion of traditional activities, in particular, agricul-
ture, for the benefit of sustainable tourism.
3.2 Survey design and administration
Four trained interviewers, two men and two women, aged between 24 and 26, collected data in
on-site, face-to-face interviews between June 17 and September 9, 2017. Interviewing lasted all
day on all weekends and two weekdays, which varied from week to week. A written informed
consent was obtained from all subjects involved in the study.
Face-to-face interviews are advantageous to encourage the attention and effort of respon-
dents. However, there are limitations to consider, such as a potential interviewer effect. Inter-
viewers can clarify questions and explain more complex issues, but inadvertently, they can also
PLOS ONE
On the psychology of environmental preferences: The influence of contextual priming
PLOS ONE | https://doi.org/10.1371/journal.pone.0312256 October 31, 2024 5 / 20
be a source of measurement error [4]. To limit the interviewer effect, interviewers were
instructed to minimize social interactions with respondents; the respondents filled in the ques-
tionnaires by themselves, while interviewers only explained the different sections of the ques-
tionnaire and assisted in filling out choice cards. In addition, on each interview day, the
interviewers moved together to the different pre-identified locations in the park, changing the
order of the locations each day so that they were not in the same location at the same time of
the day. This precludes some systematic interviewer bias that could potentially arise if the dif-
ferent interviewers were assigned to collect data at different locations. We also trained the
interviewers to standardize their behavior and asked for feedback during the entire time of the
survey. All interviewers were dressed similarly: they wore a white T-shirt and jeans.
A systematic probabilistic sampling design was used to intercept respondents because there
was no formal visitor list. Interviewers asked every second tourist they met to take part in the
survey. People were always interviewed individually, even if they were part of a group. Both
visitors and locals took part in the interviews, and the response rate was 65%.
The questionnaire was prepared following the guidelines for DCE [36,37] and consisted of
three sections. The first section contained warm-up questions and information about the RN
and the actual management of the park. The second section included the illustration of attri-
butes and levels. The survey’s core part was represented by choice cards, preceded by a script
to ensure policy consequentiality [38]. This consequentiality script informed respondents that
the results would be presented to the managers of Monte Baldo Local Natural Park and the
Province of Trento and that they could be used to improve the management policies of the
Park. Respondents were asked to answer as precisely as possible because the results must be
accurate to aid policymakers in making the best decisions. We then asked respondents to pay
attention to the cost, imagine that their choices were real, and imagine they would have to pay
the price of the ticket on the day of the interview. We concluded by stating to participants that
there were no correct or incorrect choices; we were just interested in their selections. The third
part of the survey contained standard sociodemographic questions.
Relevant attributes were tested in a consultation process with experts and scientists, manag-
ers of the RN, managers of MBLNP, and naturalists. The initial set of candidate attributes
resulted from lengthy discussions among local stakeholders regarding park management.
From this list, four attributes were selected based on their importance in the local park man-
agement and the actual possibility of their implementation. Specialists and local stakeholders
also determined the management measures associated with the specific attribute levels. The
final set of attributes and levels are presented in Table 1.
As shown in Table 1, the attributes for protection of the yellow-bellied toad and local
organic food products have two levels, while biodiversity of the meadows and restoration and
improvement actions on trails have three levels. The offered alternatives included different
combinations of these levels along with the non-action alternative that represented the aban-
donment of local management in favor of centralized management by the Province of Trento.
Central government management would imply no participation by local stakeholders and the
impossibility of implementing actions specifically designed for local conditions. In fact, the
Province of Trento would evenly undertake central management of natural areas, making it
impossible to tailor management actions according to local environmental and socioeconomic
conditions.
The monetary attribute was represented by an entrance ticket to the MBLNP, which would
be necessary for the local community to co-fund local management initiatives. This attribute
included six price levels based on the results of previous similar surveys conducted in the sur-
rounding areas [39,40]. Respondents faced 12 choice cards with three alternatives: two alter-
natives with a non-zero cost corresponding to options for local management and one non-
PLOS ONE
On the psychology of environmental preferences: The influence of contextual priming
PLOS ONE | https://doi.org/10.1371/journal.pone.0312256 October 31, 2024 6 / 20
action alternative representing the scenario in which the park is not locally managed. The
non-action alternative was cost-free because this scenario did not include any investment in
the described local management measures.
A pilot survey with 66 visitors was implemented on-site to test the design and wording of
the survey. An Optimal Orthogonal Choice Design [41] was used to generate the choice cards
in the pilot. The responses were used to set the prior values needed to generate an efficient
design [42] for the final version of the survey. A sequential D-efficient design [43,44] was used
during the implementation of the survey by employing the parameter estimates of the first 383
questionnaires to further improve the efficiency of the design. The experimental designs were
generated using NGene software [45].
Interviewers collected responses from 858 visitors. However, our final dataset consisted of
808 respondents, as 50 did not provide the complete sociodemographic information required
for the analysis. We selected specific points inside the natural park to approach interviewees to
test the priming effect. We interviewed people close to meadows to consider the priming effect
of the meadows on flora biodiversity (149 respondents), close to mountain puddles where the
yellow-bellied toad lives to test the contextual priming for the toad (59 respondents), close to
mountain trails to test the priming effect for trails (322 respondents), and in or near huts and
shelters where food was available to test priming for local organic products (218 respondents).
Additionally, a control group of 60 respondents did not receive any treatment. These respon-
dents were interviewed in a hotel in the center of the village of Brentonico. We interviewed
people other than at mealtime to ensure the control group could not be subjected to priming.
Unfortunately, the treatments are not balanced in terms of sample size because the location
elements selected to induce priming are not present in the same quantity in the park and are
not equally popular to tourists. Time and cost limitations during the data collection led to this
relatively unbalanced number of respondents in different treatments.
3.3 Econometric analysis
Our econometric approach is pretty standard and is based on the random utility theory [46],
with a linear function of the parameters’ utility defined as:
Uint ¼x0
intbþεint ;ð1Þ
where nis the individual (n= 1,2,. . .,N); iis the alternative (i= 1,2,. . .,J); tis the choice situa-
tion (t= 1,2,. . .,T), βis a vector of the parameters; and x
int
is a vector of the attributes. We
Table 1. Attributes and levels.
Attributes Description Levels No local management
Biodiversity Conservation of biodiversity through the mowing of
meadows and control of sheep grazing
1. Low (no action for biodiversity protection)
2. Medium (controlled sheep grazing)
3. High (mowing of meadows)
Low biodiversity of the
meadows, no action to protect
it
Toad Protection of the yellow-bellied toad includes
restoration and conservation of mountain puddles
where the toad lives
1. Yes (action for toad’s protection)
2. No (no protection)
No protection
Trails Restoration and improvement of the trails 1. No (no action)
2. Restoration (making trails safe and clean)
3. Restoration and enhancement (make the trails safe and
clean and add signage; availability of paper and digital
topographic maps)
No restoration or
enhancement
Organic
products
Availability of local organic products in farms, alpine
huts, markets, and catering facilities
1. Yes (there are local products)
2. No (no local products)
No presence of local organic
products
Cost Price of daily entrance ticket to the park €3, €6, €9, €12, €15, €18 €0
https://doi.org/10.1371/journal.pone.0312256.t001
PLOS ONE
On the psychology of environmental preferences: The influence of contextual priming
PLOS ONE | https://doi.org/10.1371/journal.pone.0312256 October 31, 2024 7 / 20
opted for a latent class model (LCM) that addresses the issue of individual heterogeneity
assuming a discrete mixing distribution for the parameters β, with individual parameters clus-
tered in classes [47]. Given membership of class c, the probability of respondent n’s sequence
of choices is given by:
Pnjc¼YT
t¼1
expðx0
jnt bcÞ
PJ
i¼1expðx0
int bcÞð2Þ
The unconditional probability of choosing alternative jis a weighted average of all the
parameter estimates β
c
for each class c:
Pn¼XC
c¼1pcPnjc¼XC
c¼1pcYT
t¼1
expðx0
jnt bcÞ
PJ
i¼1expðx0
int bcÞ;ð3Þ
where π
c
is the probability of belonging to the class c. The class allocation probabilities π
c
are
usually modeled using a logit structure, where the utility of a class is a function of the socio-
demographic variables SD
n
of the respondent and parameters λ
c
, in addition to a constant μ
c
:
pc¼expðmcþSD0
nlcÞ
PJ
i¼1expðmcþSD0
nlcÞ:ð4Þ
This model is estimated by the maximum likelihood method. The log-likelihood function
to be maximized is defined as:
log L¼YN
n¼1XC
c¼1pcYT
t¼1
expðx0
jnt bcÞ
PJ
i¼1expðx0
int bcÞ
" #:ð5Þ
Each attribute was interacted with a dummy variable indicating the corresponding priming
stimulus to test if the priming effect produces differences in the respondent’s choices. Biodi-
versity of the meadows was thus interacted with being interviewed close to flowery meadows,
protection of the toad with being interviewed close to mountain puddles, trails with being
interviewed close to trails, and local organic products with being interviewed in or close to
huts and shelters where food is available.
The LCM has been estimated using the Apollo package in R [48].
4. Results
Our aim is to examine whether the differences in preferences found between sites are the result
of priming. It is possible that people with different recreational tastes and preferences choose
different locations, creating an endogeneity problem. To address this, we provide a detailed
breakdown by location of the descriptive statistics of the sociodemographic variables (age, gen-
der, and education) used in the analysis. If there are no significant differences in these statis-
tics, it is likely that the potential endogeneity problem can be considered negligible.
Tables 2and 3show the descriptive statistics for the sociodemographic variables analyzed.
The average age of the respondents was 43.5 years, and almost half were male (49%). Regard-
ing educational attainment, 52% of respondents had completed secondary education, and 38%
had a university degree. These descriptive statistics are in line with the typical profile of tourists
who visit the area under study [49].
By examining the p-values from the t-test for mean differences provided in Tables 2and 3,
it can be deduced that the means associated with various priming conditions (Biodiversity,
Toad, Trails, and Organic products) significantly differ from those of the control group. By
PLOS ONE
On the psychology of environmental preferences: The influence of contextual priming
PLOS ONE | https://doi.org/10.1371/journal.pone.0312256 October 31, 2024 8 / 20
applying weights derived from entropy balancing [50], we reweighted our sample to match the
control group on these covariates.
Entropy balancing, is a multivariate reweighting method designed to create balanced sam-
ples in observational studies [50]. This method ensures that the reweighted sample matches
the characteristics of the control group across specified covariates, thereby mitigating endo-
geneity concerns. Entropy balancing operates by adjusting the weights of the treatment group
observations so that the weighted sample moments (e.g., means, variances) of the covariates
align with those of the control group.
Table 2 shows the analysis for the variable Age, and Table 3 for the variables Gender and
Education. The variable Education had five levels: mandatory schooling (1), technical school
(2), high school diploma (3), university degree (4), master’s degree, or PhD (5). As seen from
Table 3, the p-values of the χ
2
test show that the frequencies of male and female participants
are not different across the groups, with male participants making up around half of the
respondents in each group. According to the p-values of the χ
2
test for differences in propor-
tions, the participants’ educational level differs across the groups. The lowest p-value for the
Toad group seems to be due to the relatively limited sample size of this group rather than to an
endogeneity problem.
The control group was interviewed in a hotel, and to ensure that the location itself does not
introduce other contextual influences, we included a more comprehensive analysis of the con-
trol group’s responses compared to the primed groups. Specifically, similar to Table 3, we ana-
lyze in Table 4 the proportions of responses. According to the p-values of the χ
2
test for
Table 2. Descriptive statistics for the variable Age.
t-test
Mean Median Std. Dev Min Max p-value
Total 43.5 44 13.9 17 80
Priming:
Biodiversity 45.2 46 13.9 28 74 0.06
Toad 43.9 43 11.9 19 74 0.03
Trails 42.1 41 14.8 17 77 <0.01
Organic products 42.9 43 12.5 18 80 0.01
Control group 49.1 49 13.1 23 80
https://doi.org/10.1371/journal.pone.0312256.t002
Table 3. Descriptive statistics for the variables Gender and Education.
Number of observations
149 59 322 218 60
Total Biodiversity Toad Trails Organic Control
products group
Gender (Male = 0, 49% 46% 39% 54% 47% 50%
Female = 1) 51% 54% 61% 46% 53% 50%
p-value (χ
2
test) 0.74 0.31 0.66 0.82
Education Mandatory schooling 10% 11% 3% 8% 10% 17%
Technical school 10% 14% 5% 10% 8% 13%
High school diploma 42% 42% 36% 43% 47% 27%
University degree 31% 29% 46% 33% 28% 33%
Master’s or PhD degree 7% 4% 10% 7% 7% 10%
p-value (χ
2
test) 0.13 0.05 0.10 0.06
https://doi.org/10.1371/journal.pone.0312256.t003
PLOS ONE
On the psychology of environmental preferences: The influence of contextual priming
PLOS ONE | https://doi.org/10.1371/journal.pone.0312256 October 31, 2024 9 / 20
differences in proportions, the participants’ responses differ between the Control group and
the Toad and Organic products groups. Nevertheless, the proportions of the Control group and
the Biodiversity and Trails groups do not differ. We would expect significant differences
between the Control group and all other groups if some uncontrolled contextual influences
were present in this group.
Given that other contextual factors might also influence the results, we have included a
dummy variable, Sunny and not windy day, representing the weather conditions in the model.
Its value equal to one represents a sunny day (62.1%), while zero represents partly sunny,
mostly cloudy, cloudy, and rainy days (37.9%).
The specification of the number of classes in an LCM is not integrated into the maximum
likelihood criterion. Instead, it is typically established through the utilization of information
criteria. Table 5 presents the Akaike Information Criteria (cAIC and AIC), Bayesian Informa-
tion Criteria (BIC), and log-likelihood values (LogL) for two- and three-class LCM.
An increased number of classes gave rise to numerical optimization challenges during the
estimation process, primarily due to flat regions within the log-likelihood function. These flat
regions rendered unfeasible the estimation of LCM with a higher number of classes. According
to some authors [51,52], the statistical criteria and the significance of the parameter estimates
need to be tempered by the researcher’s own judgment of the suitability of the model when the
number of classes is determined, and we, therefore, estimated a three-class model.
Table 6 presents the estimates of the three-class LCM.
The first block (Attributes) includes the estimates of the attribute coefficients β
c
defined in
(2). The second block (Priming effect) presents the coefficients of the interactions of the attri-
butes with the corresponding priming effects. The third and fourth blocks (Class allocation
parameters and Class probabilities) of Table 6 present the estimates of the parameters of the
allocation function defined in (4) and the mean allocation probabilities, respectively. Parame-
ters μ
1
and λ
1
in (4), corresponding to the first class, were set to zero to ensure the
Table 4. Descriptive statistics for the variables Choice.
Number of observations
149 59 322 218 60
Total Biodiversity Toad Trails Organic Control
products group
Choice Alternative 1 66% 66% 80% 67% 63% 60%
Alternative 2 24% 24% 12% 26% 21% 35%
Alternative SQ 10% 10% 8% 6% 16% 5%
p-value (χ
2
test) 0.19 0.01 0.38 0.02
https://doi.org/10.1371/journal.pone.0312256.t004
Table 5. Information criteria.
2 Classes 3 Classes
LogL -14,619.3 -13,836.9
Number of parameters 35 55
Sample size 9,696 9,696
AIC 29,308.5 27,783.9
AIC3 29,343.5 27,838.9
BIC 29,559.8 28,178.8
CAIC 29,594.8 28,233.8
https://doi.org/10.1371/journal.pone.0312256.t005
PLOS ONE
On the psychology of environmental preferences: The influence of contextual priming
PLOS ONE | https://doi.org/10.1371/journal.pone.0312256 October 31, 2024 10 / 20
identification of the model. The vector of sociodemographic variables SD defined in (4)
includes gender, age, and education level.
The first conclusion that can be drawn from Table 6 is that the positive values of the alterna-
tive specific constants (ASCs) representing the two alternatives associated with local manage-
ment related to an additional payment indicate a general interest in the local management
measures for MBLNP. These constants are significant at 5% in the two largest Classes 1 and 2.
In all three classes, all estimated coefficients of the main effects of the non-cost attributes have
the expected positive signs, and they are statistically significant at 5%. As expected, the cost
coefficient is negative in all classes, indicating a decreasing marginal utility to the price of the
park entrance fee.
Table 6. Results of the latent class model.
Class 1 Class 2 Class 3
Coeff. Rob t Coeff. Rob t Coeff. Rob t
Attributes
ASC1 2.75 5.78 *** 1.95 7.28 *** 0.70 0.69
ASC2 2.59 5.32 *** 1.95 7.18 *** 0.66 0.55
Biodiversity Medium 0.71 6.92 *** 0.58 6.43 *** 0.75 2.46 **
Biodiversity High 0.81 7.54 *** 0.86 7.79 *** 0.73 2.65 ***
Toad 0.62 6.33 *** 1.18 8.93 *** 0.96 2.50 **
Trails Medium 1.47 8.61 *** 1.30 10.84 *** 1.39 3.02 ***
Trails High 1.91 7.77 *** 1.62 8.39 *** 1.56 1.31
Organic products 0.69 7.49 *** 0.76 7.74 *** 0.96 4.85 ***
Cost -0.24 -13.17 *** -0.05 -2.54 ** -0.45 -3.99 ***
Priming effect
Biodiversity Medium -0.09 -0.19 -0.08 -0.58 -0.22 -0.41
Biodiversity High -0.19 -0.50 0.07 0.48 -0.20 -0.16
Toad -0.44 -1.18 -0.17 -0.77 1.02 1.15
Trails Medium 0.10 0.41 -0.13 -0.92 -0.33 -0.75
Trails High -0.11 -0.36 -0.19 -1.00 -0.41 -0.70
Organic products 0.16 1.04 0.34 2.61 *** -1.19 -3.49 ***
Class allocation parameters
Constant -0.33 -0.52 -0.84 -0.79
Woman 0.23 0.91 0.11 0.28
Age 0.02 1.65 *-0.01 -0.53
Education -0.16 -1.14 -0.09 -0.32
Sunny and no windy day 0.27 0.30 0.27 0.68
Class probability
Class 1 0.37
Class 2 0.52
Class 3 0.11
Log-likelihood -13,836.94
Number of parameters 55
Observations 9,696
AIC 27,783.88
BIC 28,178.75
***,**,*: significance at the 1%, 5% and 10% level
https://doi.org/10.1371/journal.pone.0312256.t006
PLOS ONE
On the psychology of environmental preferences: The influence of contextual priming
PLOS ONE | https://doi.org/10.1371/journal.pone.0312256 October 31, 2024 11 / 20
Based on the LCM estimates in Table 6, we derived the WTP for each attribute in the esti-
mated three classes. These class-specific WTP values with and without priming treatment
interaction are presented in Table 7.
Considering the estimates in Table 6 and the WTP values in Table 7, we can characterize
the classes accordingly. The largest Class 2 stands out due to its notably high WTP values, pri-
marily for the restoration and improvement of trails. In contrast, the second-largest Class 1
exhibits approximately four times lower WTP values across all attributes, except for Toad,
which is nearly ten times lower. The smallest Class 3 presents WTP values roughly ten times
lower than those in Class 2. What sets these two classes apart is the presence of a significant
priming effect related to Organic products. While this attribute is highly regarded in Class 2, it
is not favored in Class 3. Consequently, the modeled preference heterogeneity in this LCM
reveals a distinct and opposing priming effect, serving as the defining feature between Class 2
and Class 3. It is important to highlight that an LCM is not a classification method because
class membership is probabilistic. The key objective is to compute the final WTP values, which
are derived as weighted means of the WTP within each class; an LCM is a tool to approximate
the underlying preference heterogeneity. LCM has the advantage of being a semiparametric
specification that alleviates the need for potentially stringent or unwarranted distributional
assumptions regarding individual heterogeneity, which are needed, for example, in a Random
Parameter Logit model.
To analyze the priming effect on the WTP values in greater detail, their values for individual
nare computed as the average of the ratio non-cost/cost coefficients, weighted by the probabil-
ity defined by the allocation function (4). If β
cr
is the coefficient of the r-th non-cost attribute
(r= 1,2,. . .,6) in class cand β
c7
is the cost attribute in class c, then the individual WTP values in
our three-class case can be defined as:
WTPnr ¼pn1
b1r
b17 þpn2
b2r
b27 þpn3
b3r
b37
:ð6Þ
Table 8 presents the descriptive statistics of the WTP distributions based on the estimates
included in Table 6.Fig 1 graphically presents the same information.
The differences between two WTP distributions corresponding to WTP with and without
priming effect are tested by the complete Poe’s combinatorial test [53], which is a one-tailed
Table 7. Class-specific WTP values without and with priming.
Class 1 Class 2 Class 3
Without priming effect
Biodiversity Medium 2.9 €13.0 €1.7 €
Biodiversity High 3.4 €19.1 €1.6 €
Toad 2.6 €26.3 €2.2 €
Trails Medium 6.1 €28.9 €3.1 €
Trails High 7.9 €35.8 €3.5 €
Organic products 2.9 €17.0 €2.1 €
With priming effect
Biodiversity Medium 2.6 €11.1 €1.2 €
Biodiversity High 2.6 €20.7 €1.2 €
Toad 0.7 €22.5 €4.4 €
Trails Medium 6.5 €26.1 €2.4 €
Trails High 7.5 €31.6 €2.6 €
Organic products 3.5 €24.6 €-0.5 €
https://doi.org/10.1371/journal.pone.0312256.t007
PLOS ONE
On the psychology of environmental preferences: The influence of contextual priming
PLOS ONE | https://doi.org/10.1371/journal.pone.0312256 October 31, 2024 12 / 20
Fig 1. Distributions of individual WTP values.
https://doi.org/10.1371/journal.pone.0312256.g001
Table 8. Descriptive statistics of the individual WTP values.
Mean Minimum 1st Quart. 3rd Quart. Max.
Without priming effect
Biodiversity Medium 8.1 €6.3 €7.4 €8.8 €10.2 €
Biodiversity High 11.1 €8.6 €10.2 €12.5 €14.7 €
Toad 15.0 €10.8 €13.3 €16.7 €20.0 €
Trails Medium 17.6 €13.6 €16.0 €19.3 €22.4 €
Trails High 21.9 €17.0 €19.9 €24.0 €27.8 €
Organic products 10.2 €7.7 €9.2 €11.3 €13.2 €
With priming effect
Biodiversity Medium 7.2 €5.9 €6.7 €7.7 €8.8 €
Biodiversity High 12.5 €9.8 €11.3 €13.5 €15.6 €
Toad 12.6 €9.7 €11.4 €13.5 €15.4 €
Trails Medium 16.0 €12.4 €14.4 €17.5 €20.0 €
Trails High 19.1 €14.8 €17.2 €21.0 €24.1 €
Organic products 14.1 €9.9 €12.7 €15.1 €19.2 €
https://doi.org/10.1371/journal.pone.0312256.t008
PLOS ONE
On the psychology of environmental preferences: The influence of contextual priming
PLOS ONE | https://doi.org/10.1371/journal.pone.0312256 October 31, 2024 13 / 20
test for the null hypothesis of equality of the tested distributions. The results of this test are
shown in Table 9.
The core of our study is related to the analysis of priming effects. Focusing on Fig 1 and
Table 9, only Organic products shows a positive and statistically significant priming effect at a
5% level. Respondents interviewed in or near to huts and shelters where food was available
showed higher WTP for organic products. The mean value of the distribution for these respon-
dents was €14.0, whereas for the control group it was €10.5. Food availability thus proved to
be a strong positive stimulus for the respondents.
5. Conclusions
Methods of eliciting stated preferences are founded on the assumption that individuals are
rational and have stable preferences. However, the psychological literature has shown that
preferences are unstable and can be influenced by aspects of the context in which a survey is
implemented [8]. The situational context can influence the accessibility of information that
comes to mind, including thoughts, feelings, goals, behaviors, and preferences. This aspect can
be an element of context-dependence in non-market valuation, violating the assumption of
the stability of preferences and biasing welfare estimates. In our study, we tested the effect of
the location where people were interviewed, exploring whether locations related to some attri-
bute of the choice experiment influenced choices for that attribute. Our results showed a posi-
tive priming effect for the attribute organic products, while no effect for the other attributes.
But why was there a priming effect for organic products, but not for other attributes? We
believe that several factors must be present for priming effects to occur: (1) the prime should
be potent enough to activate relevant mental constructs in memory (necessary condition), (2)
people should have strong and uniform attitudes towards the target good, and (3) people
should have strong and uniform attitudes towards the suggested interventions to improve the
provision of the target good [23,24]. Below, we detail these factors and argue that they were
likely all present only for organic products.
A prime acts by activating concepts, values, and goals in the memory. Therefore, a precon-
dition for priming to occur is the ability of the prime to activate the relevant mental constructs.
For example, it is likely that interviewing people in restaurants would make salient the concept
of food (and organic products) and interviewing people close to trails would make salient con-
cepts related to hiking. However, interviewing people close to the habitat of the yellow-bellied
toad (especially outside the toad season) would be a less potent cue for activating related men-
tal concepts.
Second, even if a location provides a strong prime, the strength and direction of the priming
effect can still vary. One determining factor is the attitude people have towards the target good
(positive/negative) and how consistent it is across people [25]. For instance, people are more
likely to have a uniformly positive attitude towards organic products, while their attitude
Table 9. Poe test for the difference of WTP distributions without and with priming.
Attribute level H
a
p-value
Biodiversity Medium WTP (no priming) >WTP (priming) 0.24
Biodiversity High WTP (no priming) <WTP (priming) 0.29
Toad WTP (no priming) >WTP (priming) 0.18
Trails Medium WTP (no priming) >WTP (priming) 0.29
Trails High WTP (no priming) >WTP (priming) 0.22
Organic products WTP (no priming) <WTP (priming) 0.05
https://doi.org/10.1371/journal.pone.0312256.t009
PLOS ONE
On the psychology of environmental preferences: The influence of contextual priming
PLOS ONE | https://doi.org/10.1371/journal.pone.0312256 October 31, 2024 14 / 20
towards the yellow-bellied toad could vary. Some individuals may have a positive feeling
toward the yellow-bellied toad, while others may have negative feelings. Studies have shown
that people’s emotions and values towards wildlife-related can vary significantly [54–56].
Therefore, even if a location provides a potent cue for activating relevant mental constructs,
the strength and the directionality of the priming effect can vary. When people’s attitudes
towards a target are uniform and strong, then the priming effects should be either strongly
positive or strongly negative. When peoples’ attitudes are more heterogeneous, the net priming
effect would be null or weak.
A third factor that may influence priming effects is the nature of the interventions men-
tioned in the levels of provision. For example, for Organic products, there were only two levels
of provision: either these products were available or not. Therefore, to the extent that people
have a positive attitude towards organic products, they are likely to favor their availability,
resulting in a positive priming effect. However, for other target goods such as trails or biodi-
versity, the interventions were more complex. For example, trail intervention included levels
such as restoration and enhancement, whereas biodiversity included different levels of human
involvement (controlled sheep grazing versus mowing). Even if people have a uniformly posi-
tive attitude towards these goods, they might disagree about how to implement changes. Some
people might favor human interventions (e.g., increasing biodiversity by mowing the grass
rather than letting cows graze on it; improving trails by adding signage and having paper and
digital topographic maps rather than simply making trails safe and clean), while others can be
against them. These opposing views concerning implementation can weaken a priming effect
even if the location prime is potent and people have a uniformly positive or negative attitude
toward the good.
It is important to recognize that the results of our study could be influenced not only by the
inherent characteristics of the survey locations but also by the self-selection of participants. It
is possible that the locations themselves, such as meadows, puddles, mountain trails, food
pavilions, and a hotel, influenced participants’ responses due to their unique environmental
contexts and appeal. Additionally, the self-selection bias, where participants choose their desti-
nations based on personal preferences, may lead to an overrepresentation of individuals who
highly value certain locations, thus impacting the survey outcomes. This dual influence of loca-
tion characteristics and participant self-selection necessitates caution in interpreting the
results, as the observed effects may reflect both the environmental context and the heteroge-
neous value individuals place on these areas within the park. Additional contextual factors,
such as the specific time of day or day of the week when interviews took place, may have
affected the results.
An additional limitation of this study is that the hotel setting can introduce several potential
contextual influences that could bias the results. Different types of hotels might attract particu-
lar socio-demographic groups. For instance, luxury hotels may attract wealthier individuals,
while budget hotels may appeal to those with lower economic status. Socioeconomic status, in
turn, can directly influence WTP. The type of hotel (luxury vs. budget) could also have an indi-
rect influence on guests by setting certain expectations and moods, which might influence
how they perceive and value environmental attributes [57,58]. Moreover, hotels located near
major attractions might attract guests more interested in those specific attractions, which can
be related to some of the analyzed environmental attributes. Additionally, proximity to natural
or environmentally significant sites might attract guests who are more environmentally con-
scious, thereby influencing their responses. However, hotels might also attract individuals for
different reasons than the target attributes (such as extreme sports), which may lead to
decreased WTP for the target attributes.
PLOS ONE
On the psychology of environmental preferences: The influence of contextual priming
PLOS ONE | https://doi.org/10.1371/journal.pone.0312256 October 31, 2024 15 / 20
To mitigate these effects, future studies should conduct surveys across a variety of hotel
types and price ranges to capture a diverse range of economic backgrounds and minimize bias
from any one type of hotel. Furthermore, surveys should be conducted at different times of the
day and on different days of the week to reach a broader range of respondents and avoid tem-
poral biases. Finally, surveys could ex post include debriefing questions to ask respondents if
they believed that the survey setting influenced their responses.
Finally, like many survey-based studies, the current research relies on self-reported data,
which can be subject to several biases such as the social desirability bias (respondents may pro-
vide answers that they believe are socially acceptable or favourable rather than their true feel-
ings or behaviors), the recall bias (respondents may not accurately remember past behaviors or
experiences) or the acquiescence bias (a tendency for respondents to agree with statements or
questions regardless of their actual opinions). Although face-to-face interviews help mitigate
some of these issues, they cannot eliminate them entirely. It is recommended to apply some of
the possible strategies to tackle this problem, such as anonymity assurance, careful design of
questionnaires, recall aids (timelines, calendars), conducting pilot studies, or collecting data
through multiple methods or sources [59].
We know of very few studies that have used a DCE to investigate the effect of location on
stated preferences [17,18]. Tinch and colleagues investigated the effects of variations in the
timing and location of choice experiment questions about conserving a UK national park. The
same participants responded to the same choice scenarios on four different occasions: off-site
just before visiting the park, on-site, off-site immediately after the visit, and off-site four
months later. They found that participants gave very different answers during the on-site visit
than in any off-site conditions (these took place in a community center). In particular, the on-
site visit increased the variance of the error term (participants found it harder to choose
between the alternatives) and removed attention to the price associated with each alternative.
In our study, all the treatments of interest were on-site but in different locations. We found
that for one particular target good (organic products) the on-site location in which participants
were surveyed mattered: participants were more WTP when surveyed in the location that was
most associated with the target product (e.g., a restaurant) than in other on-site locations.
The role of interview location on travel-time estimates has been investigated by interview-
ing participants using an internet panel, an email register, or during an actual journey [17]. It
was found that those who answered while traveling assigned, on average, higher values to
travel time than those who did not. The researchers suggested that this could be because the
benefits of saving time are more salient when traveling. Our study and the proposed explana-
tion are similar in that we also believe that location can affect choices by making certain associ-
ated characteristics more salient. However, we manipulated the physical location rather than
the type of activity (travel) during the survey. In addition, when traveling, people may con-
sciously consider the benefits of saving time, whereas in our case, the impact of location may
have been more subtle. We also drew links between such effects and location priming, which
was not considered in these previous papers.
Governmental agencies and international institutions spend millions to survey people’s
preferences. These surveys are thought to reveal people’s real preferences about the target
issues and thus provide input for litigation and political decisions and, ultimately, to guide pol-
icy. One of the best-known examples was the assessment of natural resource damage due to
the Exxon Valdez oil spill in Prince William Sound (Alaska) in 1989. The estimation was used
in formulating the claim for compensation to the court in the case of the State of Alaska vs.
Exxon [60]. We provide preliminary evidence that people’s responses in surveys may be sys-
tematically swayed by the subtlest of cues: the location where the interviewees happened to be
surveyed.
PLOS ONE
On the psychology of environmental preferences: The influence of contextual priming
PLOS ONE | https://doi.org/10.1371/journal.pone.0312256 October 31, 2024 16 / 20
Since our results confirm the influence of priming on stated preferences, practitioners must
be cautious when designing surveys for environmental valuation. For instance, if a survey
about conservation funding is conducted in a setting with a potent cue for activating related
mental constructs, respondents might be more inclined to support higher funding due to the
positive priming effect. Concretely, a survey estimating the WTP for the protection of endan-
gered species might produce more accurate responses, if conducted in a neutral location as
opposed to a nature reserve, where the presence of animals could influence responses. Simi-
larly, a survey estimating the WTP to improve water quality may produce more accurate WTP
if respondents are not surveyed nearby water bodies. To obtain more accurate valuation esti-
mates, surveys should ideally be conducted in neutral locations to minimize the influence of
environmental cues. This is important to ensure that public funding is allocated based on rep-
resentative WTP estimates.
Nevertheless, finding truly neutral locations can be challenging and may require pre-testing
and focus group discussions to identify areas that do not prime respondents towards certain
attitudes. This can increase the logistic complexity of the survey implementation and its eco-
nomic cost. It is noteworthy to mention that locations are not either neutral or not neutral in
an absolute sense, but rather their neutrality depends on the attributes being studied. For
example, although a mid-range hotel may provide a fairly neutral context for evaluating prefer-
ences towards an animal in danger, it might be non-neutral with regards to evaluating inter-
ventions that support touristic activities. This adds to the complexity of identifying neutral
contexts.
In summary, recognizing and mitigating priming effects in environmental valuation and
broader social research can lead to more accurate data collection and better-informed policy
decisions. By identifying and implementing neutral survey locations and considering demo-
graphic variations, researchers can enhance the reliability and relevance of their findings and
policymakers may allocate funds based on representative findings.
The present research has highlighted many questions in need of further investigation. Prim-
ing effects appear to depend on a number of factors, including the ability of a prime to activate
relevant mental constructs, the attitudes people have toward the target good and the unifor-
mity of such attitudes across people, as well as the attitudes people have toward the suggested
interventions (levels of provision). To get a clearer picture of the mechanisms underpinning
priming effects, future studies could include measures of these factors (strength of a prime,
attitudes towards target good, attitudes towards proposed interventions) and measure their
impact on responses. The findings of these studies could help refine current theoretical models
and enable them to capture context effects.
Supporting information
S1 Questionnaire.
(DOCX)
S1 File. Design of the choice experiment.
(PDF)
Author Contributions
Conceptualization: Sandra Notaro, Constantinos Hadjichristidis.
Data curation: Sandra Notaro, Petr Mariel.
Formal analysis: Sandra Notaro, Petr Mariel.
PLOS ONE
On the psychology of environmental preferences: The influence of contextual priming
PLOS ONE | https://doi.org/10.1371/journal.pone.0312256 October 31, 2024 17 / 20
Funding acquisition: Sandra Notaro.
Investigation: Sandra Notaro.
Methodology: Sandra Notaro, Petr Mariel.
Writing – original draft: Sandra Notaro, Petr Mariel, Constantinos Hadjichristidis.
Writing – review & editing: Sandra Notaro, Petr Mariel, Constantinos Hadjichristidis.
References
1. Mitchell RC, Carson RT. Using surveys to value public goods: The contingent valuation method. Wash-
ington, DC: Resources for the future. RFF Press; 1989. https://doi.org/10.4324/9781315060569
2. Louviere JJ, Hensher DA, Swait J.D. Stated choice methods: analysis and applications Cambridge:
Cambridge University Press; 2000. https://doi.org/10.1017/CBO9780511753831
3. Bateman IJ, Carson RT, Day B, Hanemann WM, Hanley N, Hett T, et al. Economic valuation with stated
preference techniques: A manual. Massachussets: Edward Elgar; 2002.
4. Mariel P, Hoyos D, Meyerhoff J, Czajkowski M, Dekker T, Glenk K, et al. Environmental valuation with
discrete choice experiments: Guidance on design, implementation and data analysis. Cham: Springer
Nature; 2021. https://doi.org/10.1007/978-3-030-62669-3
5. Hanley N, Barbier E. Pricing nature: Cost-benefit analysis and environmental policy. Cheltenham:
Edward Elgar Publishing; 2009.
6. Faccioli M, Glenk K. More in good condition or less in bad condition? Valence-based framing effects in
environmental valuation. Land Econ. 2022; 98(2): 314–336. https://doi.org/10.3368/le.98.2.051920-
0067R12022
7. Bargh JA, Chartrand TL. The mind in the middle: A practical guide to priming and automaticity research.
In: Reis H T, Judd CM, editors. Handbook of research methods in social and personality psychology.
Cambridge: Cambridge University Press; 2000. pp. 253–285.
8. Slovic P. The construction of preference. Am Psychol. 1995; 50(5):364–371. https://doi.org/10.1037/
0003-066X.50.5.364
9. Graffeo M, Ritov I, Bonini N, Hadjichristidis C. To make people save energy tell them what others do but
also who they are: A preliminary study. Front Psychol. 2015; 6: 1287. doi.org/https://doi.org/10.3389/
fpsyg.2015.01287 PMID: 26379603
10. Collins AM, Loftus EF. A spreading-activation theory of semantic processing.” Psychol Rev. 1975; 82:
407–428. https://doi.org/10.1037/0033-295X.82.6.407
11. Verplanken B, Holland RW. Motivated decision making: Effects of activation and self-centrality of values
on choices and behavior. J Pers Soc Psychol. 2002; 82(3): 434–447. https://doi.org/10.1037//0022-
3514.82.3.434 PMID: 11902626
12. Bonini N, Graffeo M, Hadjichristidis C, Perrotta V. The effects of incidental scents in the evaluation of
environmental goods: The role of congruity. PsyCh J. 2015; 4(2): 66–73. https://doi.org/10.1002/pchj.
76 PMID: 26261906
13. Belk RW. Situational variables and consumer behavior. J Consum Res. 1975; 2(3): 157–164. https://
doi.org/10.1086/208627
14. Kotle P. Atmospherics as a marketing tool. J Retailing. 1973; 49 (4): 48–64.
15. Thaler RH, Sunstein CR. Libertarian paternalism. Am Econ Rev. 2003; 93(2): 175–179.
16. Kroese FM, Marchiori DR, de Ridder DTD. Nudging healthy food choices: a field experiment at the train
station. J Public Health. 2016; 38(2): e133–e137. https://doi.org/10.1093/pubmed/fdv096 PMID:
26186924
17. Halse A, Flu¨gel S, Kouwenhoven M, Jong G, Sundfør H, Hulleberg N, et al. A minute of your time: The
impact of survey recruitment method and interview location on the value of travel time. Transportation.
2023; 1–32. https://doi.org/10.1007/s11116-022-10287-8
18. Tinch D, Colombo S, Hanley N. The impacts of elicitation context on stated preferences for agricultural
landscapes. J Agric Econ. 2015; 66: 87–107. https://doi.org/10.1111/1477-9552.12080
19. Neely JH. Semantic priming and retrieval from lexical memory: Roles of inhibition less spreading activa-
tion and limited capacity attention. J Exp Psychol Gen. 1977; 106: 226–254. https://doi.org/10.1037/
0096-3445.106.3.226
PLOS ONE
On the psychology of environmental preferences: The influence of contextual priming
PLOS ONE | https://doi.org/10.1371/journal.pone.0312256 October 31, 2024 18 / 20
20. Srull TK, Wyer RS. The role of category accessibility in the interpretation of information about persons:
Some determinants and implications. J Pers Soc Psychol. 1979; 37: 1660–1672. https://doi.org/10.
1037/0022-3514.37.10.1660
21. Bargh JA, Chen M, Burrows L. Automaticity of social behavior: Direct effects of trait construct and ste-
reotype activation on action. J Pers Soc Psychol. 1996; 71(2): 230–244. https://doi.org/10.1037//0022-
3514.71.2.230 PMID: 8765481
22. Doyen S, Klein O, Pichon CL, Cleeremans A. Behavioral priming: It’s all in the mind, but whose mind?
PloS One. 2012; 7(1), e29081. https://doi.org/10.1371/journal.pone.0029081 PMID: 22279526
23. Loersch C, Payne BK. The situated inference model: An integrative account of the effects of primes on
perception, behavior, and motivation. Perspect Psychol Sci. 2011; 6(3): 234–252. https://doi.org/10.
1177/1745691611406921 PMID: 26168515
24. Wheeler SC, DeMarree KG. Multiple mechanisms of prime-to-behavior effects. Soc Personal Psychol
Compass. 2009; 3, 566–581. https://doi.org/10.1111/j.1751-9004.2009.00187.x
25. Cesario J, Plaks JE, Higgins ET. Automatic social behavior as motivated preparation to interact. J Pers
Soc Psychol. 2006; 90(6): 893–910. https://doi.org/10.1037/0022-3514.90.6.893 PMID: 16784341
26. Kay AC, Ross L. The perceptual push: The interplay of implicit cues and explicit situational construal in
the Prisoner’s Dilemma. J Exp Soc Psychol. 2003; 36: 634–643. https://doi.org/10.1016/S0022-1031
(03)00057-X
27. Liberman V, Samuels SM, Ross L. The name of the game: Predictive power of reputations versus situa-
tional labels in determining prisoner’s dilemma game moves. Pers Soc Psychol Bull. 2004; 30(9):
1175–1185. https://doi.org/10.1177/0146167204264004 PMID: 15359020
28. Bimonte S, Bosco L, Stabile A. Nudging pro-environmental behavior: Evidence from a web experiment
on priming and WTP. J Environ Plan Manag. 2020; 63(4): 651–668. https://doi.org/10.1080/09640568.
2019.1603364
29. Magistris Tde-, Gracia A, Nayga RM. On the use of honesty priming tasks to mitigate hypothetical bias
in choice experiments. Am J Agric Econ. 2013; 95(5): 1136–54. https://doi.org/10.1093/ajae/aat052
30. Carlsson F, Martinsson P. Do hypothetical and actual marginal willingness to pay differ in choiceExperi-
ments? Application to the valuation of the environment. J Environ Econ Manage. 2001; 41: 179–192.
https://doi.org/10.1006/jeem.2000.1138
31. Berger J, Meredith M, Wheeler SC. Contextual priming: Where people vote affects how they vote. P
Natl Acad Sci. 2008; 105(26): 8846–8849. https://doi.org/10.1073/pnas.0711988105 PMID: 18574152
32. Wu DWL, DiGiacomo A, Kingstone A. A sustainable building promotes pro-environmental behavior: An
observational study on food disposal. PloS One. 2013; 8(1): e53856. https://doi.org/10.1371/journal.
pone.0053856 PMID: 23326521
33. Wu DWL, DiGiacomo A, Lenkic PJ, Wong VK, Kingstone A. Being in a “green” building elicits “Greener”
recycling, but not necessarily “better” recycling. PloS One. 2016; 11(1): e0145737. https://doi.org/10.
1371/journal.pone.0145737 PMID: 26731651
34. Provincial Law n. 11/2007. Legge Provinciale sulle Foreste e sulla Natura (Provincial Forest and Nature
Act). Available from: https://www.consiglio.provincia.tn.it/leggi-e-archivi/codice-provinciale/Pages/
legge.aspx?uid=16530
35. Martini U, Buffa F, Notaro S. Community participation, natural resource management and the creation
of innovative tourism products: Evidence from Italian networks of reserves in the Alps. Sustainability.
2017; 9(12): 2314. https://doi.org/10.3390/su9122314
36. Johnston RJ, Boyle KJ, Adamowicz WV, Bennett J, Brouwer R, Cameron TA, et al. Contemporary guid-
ance for stated preference studies. J Assoc Environ Resour Econ. 2017; 4(2): 319–405. https://doi.org/
10.1086/691697
37. Riera P, Signorello G, Thiene M, Mahieu PA, Navrud S, Kaval P, et al. Non-market valuation of forest
goods and services: Good practice guidelines.” J For Econ. 2012; 18: 259–270. https://doi.org/10.
1016/j.jfe.2012.07.001
38. Carson RT, Groves T. Incentive and informational properties of preference questions. Environ Resour
Econ. 2007; 37: 181–210. https://doi.org/10.1007/s10640-007-9124-5
39. Scarpa R, Notaro S, Raffaelli R, Louviere J. Exploring Scale Effects of Best/Worst Rank Ordered
Choice Data to Estimate Benefits of Tourism in Alpine Grazing Commons. Am J Agric Econ. 2011; 93
(3): 813–828. https://doi.org/10.1093/ajae/aaq174
40. Notaro S, Paletto A, Raffaelli R. Economic Impact of Forest Damage in an Alpine Environment. Acta Sil-
vatica et Lignaria Hungarica. 2009; 5: 131–143.
41. Street DJ, Burgess L. The construction of optimal stated choice experiments: theory and methods.
New Jersey: John Wiley & Sons; 2007.
PLOS ONE
On the psychology of environmental preferences: The influence of contextual priming
PLOS ONE | https://doi.org/10.1371/journal.pone.0312256 October 31, 2024 19 / 20
42. Rose JM, Bliemer MCJ. Constructing efficient stated choice experimental designs.” Transp Rev. 2009;
29: 587–617. https://doi.org/10.1080/01441640902827623
43. Bliemer MCJ, Rose JM, Hess S. Approximation of Bayesian efficiency in experimental choice designs.
J Choice Model. 2008; 1(1): 98–126. https://doi.org/10.1016/S1755-5345(13)70024-1
44. Ferrini S, Scarpa R. Designs with a priori information for non-market valuation with choice experiments:
A Monte Carlo study. J Environ Econ Manage. 2007; 53(3): 342–363. https://doi.org/10.1016/j.jeem.
2006.10.007
45. ChoiceMetrics. Ngene 1.1.2 User Manual & Reference Guide. 2014.
46. Manski CF. The structure of random utility models. Theor Decis. 1977; 8 (3): 229–254.
47. Greene WH, Hensher DA. A latent class model for discrete choice analysis: Contrasts with mixed logit.
Transp Res Part B- Meth. 2003; 37(8): 681–698. https://doi.org/10.1016/S0191-2615(02)00046-2
48. Hess S, Palma D. Apollo: A flexible, powerful and customisable freeware package for choice model esti-
mation and application. J Choice Model. 2019; 32: 100170. https://doi.org/10.1016/j.jocm.2019.100170
49. P.A.T. Provincia Autonoma di Trento. Turismo in Trentino, Rapporto 2015 (Tourism in Trentino, Report
2015). Trento: P.A.T. 2016.
50. Hainmueller J. Entropy Balancing for Causal Effects: A Multivariate Reweighting Method to Produce
Balanced Samples in Observational Studies. Polit Anal. 2012; 20(1): 25–46. https://doi.org/10.1093/
pan/mpr025
51. Scarpa R, Thiene M. Destination choice models for rock climbing in the Northeastern Alps: A latent-
class approach based on intensity of preferences. Land Econ. 2005; 81(3): 426–444. https://doi.org/10.
3368/le.81.3.426
52. Hynes S, Hanley N, Scarpa R. Effects on Welfare Measures of Alternative Means of Accounting for
Preference Heterogeneity in Recreational Demand Models. Am J Agric Econ. 2008; 90 (4): 1011–
1027. https://doi.org/10.1111/j.1467-8276.2008.01148.x
53. Poe GL, Giraud KL, Loomis JB. Computational methods for measuring the difference of empirical distri-
butions. Am J Agric Econ. 2005; 87(2): 353–365. https://doi.org/10.1111/j.1467-8276.2005.00727.x
54. Jacobs MH, Vaske JJ, Roemer JM. Toward a mental systems approach to human relationships with
wildlife: The role of emotional dispositions. Human Dimensions of Wildlife. 2012; 17: 4–15. https://doi.
org/10.1080/10871209.2012.645123
55. Grilli G, Notaro S, Campbell D. Including Value Orientations in Choice Models to Estimate Benefits of
Wildlife Management Policies. Ecol Econ. 2018; 151: 70–81. https://doi.org/10.1016/j.ecolecon.2018.
04.035, ISSN 0921-8009.
56. Notaro S, Grilli G How much fear? Exploring the role of integral emotions on stated preferences for wild-
life conservation. Environ Manage. 2022; 69: 449–465. https://doi.org/10.1007/s00267-022-01593-z
PMID: 35032185
57. Ibanez L, Roussel S. The effects of induced emotions on environmental preferences and behavior: An
experimental study. PloS One. 2021; 16(9): e0258045. https://doi.org/10.1371/journal.pone.0258045
PMID: 34591912
58. Hanley N, Boyce C, Czajkowski M, Tucker S, Noussair C, Townsend M. Sad or Happy? The Effects of
Emotions on Stated Preferences for Environmental Goods. Environ Resour Econ. 2017; 68 (4): 821–
846. https://doi.org/10.1007/s10640-016-0048-9
59. Lupi F, Phaneuf DJ, von Haefen RH. Best Practices for Implementing Recreation Demand Models. Rev
Environ Econ Policy. 2020; 14 (2): 302–323. https://doi.org/10.1093/reep/reaa007
60. Carson R, Mitchell RC, Hanemann M, Kopp RJ, Presser S, Ruud PA. Contingent valuation and lost pas-
sive use: Damage from the Exxon Valdez oil spill. Environ Resour Econ. 2003; 25: 257–286. https://
doi.org/10.1023/A:1024486702104
PLOS ONE
On the psychology of environmental preferences: The influence of contextual priming
PLOS ONE | https://doi.org/10.1371/journal.pone.0312256 October 31, 2024 20 / 20
Content uploaded by Sandra Notaro
Author content
All content in this area was uploaded by Sandra Notaro on Nov 05, 2024
Content may be subject to copyright.
