Future Foods 12 (2025) 100722

Contents lists available at ScienceDirect

Future Foods
journal homepage: [Link]/locate/fufo

Nudging effects on consumer preferences and willingness to pay for

lycopene-enriched fresh tomatoes
Maria De Salvo a,* , Isabella Trovato a, Laura Giuffrida b, Marika Cerro b, Giovanni Signorello b,
Giuseppe Cucuzza b
a
University of Messina, Department of Veterinary, Polo Universitario dell’Annunziata 98168 Messina, Italy
b
University of Catania, Department of Agriculture, Food and Enviornment, via Santa Sofia 100 95123 Catania, Italy

A R T I C L E I N F O A B S T R A C T

Keywords: Using a Discrete Choice Experiment, this study examines how informational prompts about the nutraceutical
Nudging properties of tomatoes influence consumers’ perceptions and preferences for lycopene-enriched fresh tomatoes.
Fresh tomatoes The results show that such nudges effectively increase consumer interest in functional foods by raising the
Nutraceutical properties
perceived importance of their health-related benefits. Furthermore, nudging significantly impacts Willingness To
Discrete choice experiments
Pay (WTP), with consumers exhibiting higher WTP for tomatoes with enhanced nutraceutical attributes when
Attribute Non-Attendance
exposed to the informational prompts. These findings offer valuable insights for promoting healthier dietary
choices and for guiding the development of food products that align with shifting consumer values and health
consciousness.

1. Introduction Weimer et al., 2022; Yi et al., 2022). However, the impact of nudging on
Willingness To Pay (WTP) in a Discrete Choice Experiment (DCE) setting
In recent years, behavioural interventions such as nudging have remains relatively unexplored.
gained increasing attention in public policy and consumer research for This study contributes to the literature by investigating the effect of
their potential to improve individual decision-making without limiting informational nudges on consumer preferences and WTP for fresh to­
freedom of choice (Thaler and Sunstein, 2021). Nudging is especially matoes, with a particular focus on nutraceutical content, specifically
effective in situations involving routine, low-engagement decisions, lycopene enrichment. Tomatoes are among the most widely consumed
such as those related to food purchasing and dietary habits (Gravert and vegetables in Italy and represent a key segment of the national agri-food
Kurz, 2021; Hansen et al., 2021; Vella et al., 2024). Within this context, system (ISMEA, 2017). They are also a significant source of lycopene, a
informational prompts represent a non-intrusive form of nudging that carotenoid with demonstrated antioxidant properties and potential
can enhance awareness of nutritional or functional food attributes, anticancer effects (Poojary and Passamonti, 2015). Lycopene is known
potentially steering consumers toward healthier and more sustainable for its role in reducing oxidative stress and inflammation and contributes
choices. to the prevention of cardiovascular diseases and other chronic condi­
A growing body of literature has examined how nudges affect con­ tions (Collins et al., 2022; Tan et al., 2018). In addition, tomatoes
sumer behaviour in food-related contexts (Franceschinis et al., 2024). contain a wide range of other antioxidants, vitamins, and phenolic
Systematic reviews by Čop and Njavro (2024), Lindstrom et al. (2023), compounds, some of which exhibit antimicrobial effects (Ibrahim et al.,
Pandey et al. (2023), and Sapio and Vecchio (2024) underscore the 2023). The nutritional content of tomatoes is influenced by both agro­
effectiveness of nudges in guiding food choices, especially toward nomic and post-harvest practices (Giovannetti et al., 2012; Lima et al.,
increased Fruit and Vegetable (F&V) consumption. More focused studies 2022).
have investigated behavioural outcomes and choice architecture related Despite the evidence supporting lycopene’s health benefits, the Eu­
to F&V intake (Almeida et al., 2024; Bauer et al., 2021; Blom et al., ropean Food Safety Authority (EFSA NDA Panel, 2011) has not yet
2021; Ensaff, 2021; Gillebaart et al., 2023; Vecchio and Cavallo, 2019; approved any official health claims for lycopene, citing insufficient

* Corresponding author.
E-mail addresses: [Link]@[Link] (M. De Salvo), [Link]@[Link] (I. Trovato), [Link]@[Link] (L. Giuffrida), [Link]@[Link]
(M. Cerro), [Link]@[Link] (G. Signorello), [Link]@[Link] (G. Cucuzza).

[Link]
Received 26 February 2025; Received in revised form 8 May 2025; Accepted 18 July 2025
Available online 20 July 2025
2666-8335/© 2025 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license ([Link]
M. De Salvo et al. Future Foods 12 (2025) 100722

scientific consensus. However, consumer-oriented communication of the sampling strategy, a summary of key sample characteristics, and a
about lycopene’s potential benefits could increase perceived value and statistical comparison of the two subgroups. It also details the nudging
promote demand for functional tomato products. treatment and outlines the econometric models employed to estimate
While substantial literature exists on consumer preferences for consumer preferences. Section 3 reports and discusses the main empir­
traditional tomato attributes such as colour, size, origin, packaging, and ical findings. Finally, Section 4 offers concluding remarks and proposes
certifications (Carroll et al., 2018; De Salvo et al., 2020; Ellison et al., directions for future research on nudging in food choice contexts.
2016; Joya et al., 2022; Kassas et al., 2023; Mancuso et al., 2024; Maples
et al., 2018; Nishimura, 2021; Nohara, 2024; Printezis and Grebitus, 2. Materials and methods
2020; Ruth et al., 2016; Shonkwiler, 2024; Tran et al., 2023; Zapata
et al., 2025), only a limited number of studies explore consumer valu­ 2.1. Data
ation of nutraceutical attributes. To date, Verneau et al. (2019) is the
only study that has examined the impact of lycopene-related informa­ Data were collected by a specialized Italian market research com­
tion on WTP, and it focused on canned rather than fresh tomatoes. pany that manages a national panel of over 50,000 respondents. The
In addition, although DCEs are widely used to elicit consumer pref­ survey targeted a representative sample of 1,009 Italian consumers who
erences, they are subject to potential biases related to Attribute Non- are primarily responsible for purchasing F&V for their households and
Attendance (ANA), wherein respondents ignore one or more attributes who personally consume fresh tomatoes at least once per month. The
due to cognitive overload or strategic simplification (Alemu et al., 2013; sample was stratified by gender, age, and geographical region to ensure
Carlsson et al., 2018; Campbell et al., 2011; Hensher, 2006; Scarpa et al., representativeness of the Italian adult population. Data collection was
2009; Weller et al., 2014). Only a few recent studies (Jonker et al., 2018; carried out between June and July 2023 through an online question­
Kassie et al., 2020) have attempted to measure how nudging may affect naire composed of four sections. The first section gathered information
ANA, cognitive effort, or decision fatigue in DCEs. on fresh tomato consumption habits. The second section focused on the
These observations point to several research gaps. First, there is respondent’s self-reported health status. The third section explored the
limited understanding of how nudging affects valuation of nutraceutical respondent’s awareness and knowledge of the nutraceutical properties
attributes, especially in fresh produce. Second, little is known about how of tomatoes. The final section included the DCE tasks used to elicit
nudging influences the processing of attribute information in DCE set­ preferences and WTP.
tings, particularly in relation to cognitive effort and decision consis­ Table 1 presents the list of attributes and corresponding levels used
tency. Third, the literature lacks empirical evidence on whether nudging in the DCE. These were selected based on a review of the relevant
modifies the structure and distribution of marginal WTP estimates. literature and the results of a pilot survey involving 141 consumers, who
This study addresses these gaps using a DCE to estimate Italian were asked to identify the four tomato attributes they considered most
consumers’ preferences and WTP for fresh tomato attributes, with a important in their purchasing decisions (Mariel et al., 2022; Morejón
special emphasis on lycopene content. A split-sample design was Cabrera et al., 2022).1 The experimental design included a relatively
implemented: the control group received no informational prompt, large number (nine) of attributes and levels to reflect the complexity of
while the treatment group was exposed to a science-based informational real-world decision-making contexts, where the risk of ANA naturally
nudge about the health benefits of lycopene (Congiu and Moscati, 2022). increases, allowing us in investigating whether the incidence of ANA is
This design enables causal identification of nudging effects on both influenced by the use of nudging (Caraban et al., 2019; Collins, 2012).
preferences and cognitive processes. ANA phenomenon rate was detected by asking to each respondent ex
The study investigates the following research questions: post debriefing questions about their attendance of attributes (e.g.,
‘stated ANA’, Balcome et al., 2015).
- RQ1. Does nudging influence consumer preferences exclusively for A D-efficient experimental design (Rose et al., 2008) was developed
the lycopene attribute, or does it also affect the valuation of other using priors derived from the existing literature and implemented
tomato characteristics? through NGENE software version 1.2.1 to maximize the statistical effi­
- RQ2. Does nudging alter the correlation structure across product ciency of the choice tasks. The design consists of 10 blocks, each
attributes? randomly assigned to respondents and containing six choice scenarios.
- RQ3. By promoting more deliberate and rational decision-making, In each scenario, participants are presented with three alternatives: two
does nudging increase cognitive effort? If so, does this increased represent different options for a 500 g container of fresh tomatoes,
effort significantly affect the likelihood of ANA? tailored to the preferred colour and shape as reported by the respondent
- RQ4. Does nudging lead to significant shifts in the distribution of in the initial section of the questionnaire; the third is an opt-out option
marginal WTP, indicating changes in average valuation and/or
preference heterogeneity?
1
The pilot questionnaire, which included 12 questions on consumer behav­
By addressing these questions, the study contributes to the broader
iour and habits related to fresh tomatoes, was administered by the research
understanding of how informational nudges affect consumer behaviour team via social media platforms. The respondents, primarily women with an
in complex choice environments and provides practical insights for average age of 48, were mostly based in central and southern Italy. In 84% of
designing interventions that promote healthier and more sustainable cases, respondents reported that their households purchase and consume fresh
food choices. tomatoes, at least occasionally. Participants identified the most important at­
Results show that nudging significantly enhances consumer interest tributes influencing their purchasing decisions as intrinsic product character­
in functional tomato attributes, particularly lycopene content. Partici­ istics—including taste, aroma, juiciness, shape or variety, pulp firmness,
pants who received the informational prompt displayed a higher WTP external integrity, colour, and texture when cut. Other highly valued attributes
for lycopene-enriched tomatoes compared to those in the control group. included product origin, production area, and price. By contrast, some intrinsic
This effect extended beyond lycopene to influence other product attri­ factors were considered less relevant, such as skin thickness, lycopene content,
fruit size (within the same shape or type), and seed presence. Among certifi­
butes as well, suggesting a broader shift in consumer valuation once
cation types, only ethical certification was regarded as important, while oth­
health information is provided. Moreover, results suggest that infor­
ers—such as environmental sustainability, nickel-free status, and nutraceutical
mational nudging may increase perceived cognitive load, particularly properties—were viewed as less influential. Purchase location and packaging
among those already engaging in simplified decision-making strategies. method were rated as moderately important. Finally, the presence of a QR code
The remainder of the paper is structured as follows. Section 2 pre­ on packaging and product advertising were considered the least relevant factors
sents the design and implementation of the DCE, including a description in the decision-making process.

2
M. De Salvo et al. Future Foods 12 (2025) 100722

Table 1
Attributes and levels used in the DCE.
Attributes Levels

1. Nickel-free certification - Yes;

- No.
2. Origin of the product - Italian, same respondent’s region of residence;
- Italian, different respondent’s region of residence;
- EU country;
- Extra-EU country.
3. Packaging material - Plastic;
- Paper;
- 100 % biodegradable.
4. Origin Label (PGI - Protected Geographical Indication) - Yes;
- No.
5. Eco-sustainability label (products obtained while safeguarding the environment) - Yes;
- No.
6. Ethical label (products obtained while safeguarding the - Yes;
health and safety of workers, without employing child labour, etc.) - No.
7. Organic - Yes;
- No.
8. Lycopene content - Medium;
- High.
9. Price (€/500 g container) - 2.99;
- 3.29;
- 3.62;
- 3.98;
- 4.38.

(i.e., the “no choice” alternative). This structure allows for the estima­ respondents each completing six choice tasks in the DCE. The sample
tion of preferences while maintaining realism and respondent engage­ characteristics are summarized in Table 2 and detailed in the Supple­
ment. An example of a choice card is shown in Fig. 1. mentary Materials. Participants ranged in age from 18 to 80 years, with
Respondents were randomly assigned to one of two experimental a mean age of 52.99 years (standard deviation - SD: 12.97). Men rep­
groups: a control group, which received no nudge stimuli, and a treat­ resented 55 % of the sample, with an average age of 54.00 years (SD:
ment group, which was exposed to the nudge. In this study, we 12.38), average weight of 78.61 kg (SD: 12.87), and average height of
employed an informational nudging approach, following the typology 175.23 cm (SD: 7.40). Women accounted for the remaining 45%, with
proposed by Hansen and Jespersen (2013), and applied in subsequent an average age of 51.81 years (SD: 13.49), weight of 65.70 kg (SD:
work (Congiu and Moscati, 2022; Nielsen et al., 2017). Participants in 15.40), and height of 164.01 cm (SD: 6.69). Regarding socioeconomic
the treatment group were asked to read a one-page informational card (a characteristics, 54% of respondents were employed, and 32% held a
copy of which is provided in the Supplementary Materials) that pre­ university degree. In terms of household income, the largest group
sented concise, evidence-based content on the nutritional, nutraceutical, (27%) reported an annual income between 30,000 and €50,000, fol­
and health-related properties of fresh tomatoes. The card was displayed lowed by 25% in the €20,000 to €29,999 bracket. The most common
for a minimum of two minutes. The informational content emphasized household size was three members, reported by 30% of participants. The
that fresh tomatoes are composed of approximately 94 % water, contain majority of participants reported being non-smokers and not affected by
low levels of macronutrients, and are rich in vitamins (including pro­ major health conditions, including obesity, cardiovascular disease, hy­
vitamin A), minerals, and organic acids. Special emphasis was placed on pertension, or cancer. Most respondents also indicated that they do not
their antioxidant profile, which includes over 40 polyphenols and follow any specific diets and have no known allergies to Solanaceae (e.
several carotenoids, with lycopene highlighted as the most abundant g., tomatoes) or nickel. Furthermore, only a small minority identified as
and functionally significant compound due to its well-documented vegan or vegetarian.
health benefits.2 To ensure the validity of the experimental design, comparison tests
The informational card also included a list of key reasons explaining were conducted to assess both homogeneity and heterogeneity across
why tomatoes are considered a healthy food. To assess participants’ a the two subsamples. Results confirm that the groups are statistically
priori knowledge, we implemented different approaches for the two equivalent with respect to key demographic variables, including age,
subgroups. In the control group (not exposed to the nudge), respondents gender, educational attainment, and household size (see Table 2).
were asked to evaluate whether a series of statements describing the
nutraceutical properties of tomatoes were true or false. All of the
2.2. Econometric analysis
statements reflected scientifically accurate information. In contrast,
participants in the treatment group (who received the informational
As part of the preliminary analysis, we estimated both a Multinomial
nudge) were asked whether they had prior knowledge of the specific
Logit (MNL) model, which assumes fixed coefficients and homogeneous
nutraceutical properties listed on the informational card before
preferences across respondents, and a Mixed Logit (MXL) model, which
participating in the survey.
allows for preference heterogeneity by incorporating random parame­
The final dataset comprises 6,054 observations, derived from 1,009
ters, assuming a fixed distribution for the price coefficient.3
Model comparison based on the Akaike Information Criterion (AIC)
and Bayesian Information Criterion (BIC) strongly favoured the MXL
2
Lycopene is mainly present in the outermost part of the tomato and in the
skin (epicarp). In the fresh, free-range product, the lycopene content ranges
3
from 3 mg/kg in yellow tomatoes to 50 mg/kg in red tomatoes. The daily dose A fixed-price model specification was preferred to the one based on a
of lycopene our organism needs is between 5 and 15 mg, roughly equal to what lognormal distribution to obtain well-defined WTP distributions for qualitative
would be found in about half a kilo of tomatoes. The protective action of to­ attributes, facilitate post-estimation analysis through a simple visual compari­
matoes increases when cooked and seasoned with oil. son of WTP distributions across subsamples.

3
M. De Salvo et al. Future Foods 12 (2025) 100722

Fig. 1. Example of choice card.

The choice probability is conditional on βn. Respondents are assumed

Table 2
to be randomly drawn from a population which is formalized by stating
Summary statistics* and results of comparison tests.
that the individual parameter vectors are sampled from a continuous
Variables Statistics “Nudged” “No-nudged” Whole distribution f (β). This distribution is often assumed to be normal,
subsample subsample sample
lognormal, uniform or triangular (Train, 2009). In this study, we used
Age (in years) Mean 53.24 52.73 52.99 the normal distribution to describe the heterogeneity of respondents’
Standard 12.85 13.11 12.97
preferences for qualitative attributes, as already done in similar studies
deviation
MALE (equals to Mean 0.57 0.54 0.55 (De Salvo et al., 2020).
one if yes) Standard 0.50 0.50 0.50 The unconditional probability is then calculated as the weighted
deviation average of the logit formula evaluated at different values of β, with the
EDU (in years) Mean 13.90 13.98 13.94 weights given by the density f(β). Given that in this case of study each
Standard 3.33 3.18 3.25
respondent faced more than one choice occasion, the probability of
deviation
Family size (n.) Mean 2.94 2.87 2.90 observing a particular sequence of choices for each respondent is the
Standard 1.07 1.13 1.10 product of standard logit formulas. In the so-called Panel Mixed Logit,
deviation which allowed us to use a MXL specification in the context of repeated
Tests**: F Prob>F
choices by respondents assuming specific taste distributions (Revelt and
1. Allowing for
homogeneity: Train, 1998), the unconditional probability that n-th respondent choice
Wilks’ lambda 0.9975 0.647 the i-th alternative in the t-th choice occasion is computed using the
Pillai’s trace 0.0025 0.647 following integral (Train, 2009):
Lawley-Hotelling 0.0025 0.647
∫ ∫ ∏
T
trace
Roy’s largest root 0.0025 0.647 Pnit = P(i|βn )f(β)dβ = Lnit (β)f (β)dβ (2)
2. Allowing for t=1

heterogeneity
Multivariate 0.62 0.647 where:
Normality Test
Likelihood Ratio 2.50 0.647 - Pnit is the probability that the n-th respondent (with N = 1009) choice
test
the i-th alternative (with J = 3) in the t-th choice occasion (with T =
*
Summary statistics about other variables related to the respondents’ profile, 6);
habits and preferences are reported in supplementary materials. Vnit(β)
** - Lnit (β) = ∑eJ is the logit probability evaluated at parameters β;
These comparison tests assume as null hypothesis the equality of sub­ j=1
e
Vnjt(β)

samples’ means. - and Vnjt(β) is the observed (indirect) portion of the utility which in
the case of study is assumed to be linear in attribute levels (xnjt) and
specification, which consistently yielded lower values for both criteria. parameters (β): Vnjt (β) = β′xnjt.
These results support the use of the MXL model for subsequent analysis.4
MXL model assumes that each respondent has their own parameter Adopting normal distributed β parameters, Eq. (2) became:
vector βn. Therefore, the probability that the n-th respondent choices the ∫
i th alternative given the set J is: eVnit(β)
Pni = ∑J V f(β|θ)dβ (3)
j=1 e
njt(β)

Pni = P(i|βn ) (1)

where f(β|θ) is the multivariate probability of β given the distributional
parameters θ.
The vector θ represents the population-level parameters and char­
4
AIC values equal to 10,944 in the MNL model and to 9,757 in the MXL one. acterizes the properties of the assumed normal distributions. It includes
Similarly, BIC values respectively equal to 11,045 and 10,467.

4
M. De Salvo et al. Future Foods 12 (2025) 100722

the mean vector μ for the individual parameters βn, the vector of stan­ with results from previous studies conducted both in Italy and interna­
dard deviations σ, and the covariances, as the analysis accout for po­ tionally (De Salvo et al., 2020; Mancuso et al., 2024).5
tential correlation among the parameters in βn. Standard deviations and Among respondents who received the informational card (N = 505),
covariances are summarized into the variance-covariance matrix Ω the overall level of prior knowledge regarding the nutraceutical prop­
which captures both the extent of individual heterogeneity and the in­ erties of tomatoes appeared to be relatively low. Most participants re­
terdependencies between preference parameters. ported being unaware of the various health-related benefits before the
Since the unconditional choice probability defined in Eq. (3) has no interview, with the exception of a few well-known aspects, such as the
closed-form solution, simulation-based methods are required for model high lycopene content, the hydrating effect on tissues, and the digestive
estimation. In this study, we applied maximum simulated likelihood benefits. For the control group (N = 504), who did not receive the
estimation, following the approach described by Train (2009), using the informational prompt, respondents were asked to evaluate whether the
mixlogit routine in Stata (Hole, 2007). same statements included on the card were true or false. Most partici­
To assess potential attribute overlap, we conducted a series of post- pants correctly identified the statements as true; however, there was
estimation diagnostics. First, we estimated the correlation matrix of noticeable uncertainty or incorrect responses regarding the claims that
the attribute coefficients to identify possible multicollinearity. Then, we tomatoes should be cooked and seasoned with oil to enhance their nu­
re-estimated the model excluding attributes that appeared conceptually traceutical properties, and that they offer benefits for visual health.
or statistically overlapping, in order to evaluate the stability of the These findings suggest that consumer knowledge of tomato-related nu­
remaining parameter estimates. These checks confirmed that, although traceutical benefits is generally limited and fragmented.
some attributes shared conceptual similarities, the core parameter esti­ Prior to completing the DCE tasks, all respondents were asked to
mates remained stable, and the interpretability of the model was select the four most important attributes they consider when purchasing
preserved. tomatoes, from the set included in the experiment. As illustrated in
The MXL model was estimated for both the full sample and the two Fig. 2, the most frequently selected attributes were product origin,
sub-samples, corresponding to the nudged and non-nudged groups. regional quality labels (e.g., PDO/PGI), as well as organic and envi­
Comparative analysis across subsamples was conducted by examining ronmental sustainability certifications.
the significance and direction of the estimated coefficients, allowing for Coefficient estimates of DCE-MXL model are presented in Table 3.
an assessment of how informational nudging influenced preference Based on the magnitude of the estimated coefficients, it becomes evident
structures. It is not possible to compare the magnitude of the coefficients that the order of importance stated a priori by the respondents was not
across models, as they are specified in preference space and therefore confirmed by DCE, given that the four most important qualitative at­
subject to scale differences. tributes, derived from the selections made across the choice tasks, are, in
Additionally, we estimated the Cholesky decomposition L of the full sequence, the regional label, the material used to packet tomatoes, the
variance-covariance matrix Ω, where Ω = (LL’), to derive the correlation organic label and the ethical label.
matrix for the random parameters. This enabled an analysis of the The price coefficient displays the expected negative sign, indicating
nudging effect not only in terms of changes in coefficient significance, that utility decreases as price increases, and is statistically significant,
but also with respect to the significance of the correlation structure confirming its influence on consumer choice behaviour. Moreover, es­
among attributes. Estimates of the mean vector μ for the individual timates indicate that utility increases if the product is nickel-free, PGI-
parameters and the variance-covariance matrix Ω obtained for the two certified, eco-sustainable (eco), ethically produced (etico), organic (bio)
subsamples (nudged and non-nudged) were used to generate corre­ and has a high lycopene content (lico).
sponding multivariate distributions of individual-level preference pa­ Regarding the provenance of the products, compared to non-EU
rameters. These distributions enabled the derivation of marginal WTP at products, consumers prefer, in the following decreasing order, prod­
the population level and facilitated visual comparisons between the two ucts from their own region (same_reg), from other Italian regions (oth­
experimental groups. In addition, the estimated parameters served as er_reg) or, although to a lesser extent, from other EU countries.
the basis for behavioural inference, allowing us to construct probabi­ In terms of packaging, compared to plastic, the use of paper (paper)
listic demand functions. We conducted simulation exercises separately or biodegradable materials (degr) increases utility, particularly in the
for nudged and non-nudged consumers, considering products with both case of biodegradable packaging.
medium and high lycopene content. All simulations were based on Finally, choosing the option that does not involve purchasing,
10,000 random draws, ensuring robust and stable inference. generally, results in disutility. The coefficient of the corresponding
alternative specific constant (ASC_none) is in fact negative and highly
3. Results and discussion significant (p < 0.001).
Heterogeneity in preferences is highlighted by significant estimates
The data confirm that fresh tomatoes are a staple in the Italian diet. of the coefficients for the standard deviation only for some attributes,
Approximately 39% of respondents reported consuming <200 g per day, named the nickel free certification, as well as PGI and eco-sustainability,
while 37% indicated a daily intake between 200 and 300 g. Among these the provenience of the product from the same respondent’s residence
consumers, 60% stated that they consume tomatoes year-round, region and, finally, for the ASC_none.
whereas the remaining 40% limit their consumption to the spring­ To assess whether the estimated MXL was affected by potential
–summer period. Regarding frequency, 63% of participants reported attribute overlap, we computed the correlation matrix of the attribute
eating fresh tomatoes several times per week, and 20% consume them at coefficients (reported in the Supplementary Materials). We then re-
least once a week. In terms of product freshness, 65% of respondents estimated the model after removing attributes exhibiting moderate to
reported consuming tomatoes within 2–3 days of purchase. Price high correlation with others,specifically “etico” (which showed a cor­
sensitivity varied, but promotional offers emerged as the primary pur­ relation of 0.76 with “PDO”) and “bio” (which correlated at 0.49 with
chasing trigger for 42% of consumers. In terms of preferences, the most “lico”). Comparison of these results with those from the original model
favoured tomato varieties were cherry tomatoes (29%), followed by date indicates that the key mean coefficients remained stable in sign,
tomatoes (22%), beef tomatoes (15%), and Piccadilly (11%). A strong magnitude, and statistical significance. These findings suggest that the
preference for red-coloured tomatoes was evident, with 85% of re­ exclusion of overlapping attributes does not materially impact the main
spondents indicating red as their preferred colour. The three most
influential factors driving purchasing decisions were: product origin
(88%), region-linked certifications such as PDO/PGI (74%), and organic 5
Figures and Tables related to such issues are reported in the Supplementary
(62%) or environmentally sustainable (60%) labels. These findings align Materials section.

5
M. De Salvo et al. Future Foods 12 (2025) 100722

Fig. 2. A priori importance assigned by respondents to attributes included into the DCE.

results, thereby confirming the robustness and reliability of the model more pronounced and consistent effect on the significance of the
estimates. Full details of this analysis are provided in the Supplementary covariance coefficients, which reflect correlations between preferences
Materials. for different attributes. As shown in Fig. 3, many of the covariance terms
A comparison of the coefficient estimates from the MXL models, that were statistically significant in the non-nudged sample become non-
separately estimated for the nudged and non-nudged subsamples, re­ significant under the nudged condition. Specifically, 25 out of 26
veals important differences in consumer response to lycopene content. covariance coefficients lose significance in the nudged group. The only
In the non-nudged group, the coefficient for lycopene is not statistically correlation that remains significant across both models is between
significant, suggesting that, in the absence of the informational prompt, lycopene content and regional origin, suggesting a persistent link in
this attribute does not meaningfully influence choice. In contrast, in the consumer valuation of these two attributes. Interestingly, in three cases,
nudged group, the lycopene attribute becomes highly statistically sig­ covariance coefficients that were not significant in the non-nudged
nificant (p < 0.001), both in terms of the mean and the standard devi­ sample emerge as significant in the nudged group. There are at least
ation of the random parameter. This indicates that the informational two plausible explanations for this finding. First, nudging may not only
nudge not only enhances the overall importance of lycopene content in suppress existing correlations between preferences but also introduce
consumer preferences but also increases preference heterogeneity for new patterns of correlation, as observed in the emergence of a few sig­
this attribute. Overall, the informational nudge has a substantial impact nificant covariance terms in the nudged sample. Second, nudging may
on model estimates, resulting in notable differences in the significance of influence the cognitive approach participants adopt when completing
coefficients compared to the non-nudged subsample. In the non-nudged the DCE. As highlighted in the literature, informational nudges tend to
subsample, respondents appear to consider all attributes (except for promote more deliberate and rational decision-making, encouraging
lycopene content) and exhibit preference heterogeneity for several at­ respondents to focus on explicit trade-offs between attribute levels
tributes, including nickel-free (nickel), PGI certification (PGI), eco- rather than relying on heuristics or considering attributes jointly. This
labeling (eco), ethical certification (etico), organic (bio), and ASC_none. shift in decision strategy places greater emphasis on the evaluation of
By contrast, in the nudged group, respondents do not assign significant individual attributes, potentially reducing the perceived importance of
importance to the nickel-free characteristic or PGI certification and attribute interactions. As a result, responses become more precise and
perceive no meaningful difference between tomatoes sourced from EU consistent, which may explain the increased significance of the
and non-EU countries. However, preferences in this group remain het­ individual-level coefficients and the reduced significance of correlation
erogeneous for attributes such as nickel-free, eco-labels, lycopene con­ terms in the nudged group. Additional findings appear to support this
tent, paper packaging, and PGI, despite the latter not being statistically last interpretation. Specifically, the mean coefficient estimates show
significant. Importantly, the signs of significant coefficients are consis­ improved statistical significance in the nudged subsample, suggesting
tent across both subsamples, indicating that the direction of preferences more precise estimation of preferences. Moreover, for certain attributes,
remains stable, even though the salience and strength of certain attri­ such as ethical certification (etico), organic (bio), and ASC status quo,
butes are affected by the nudging intervention. preferences appear to be more homogeneous among respondents
Fig. 3 compares the results of the two models, focusing on the sig­ exposed to the nudge. This further reinforces the idea that nudging
nificance of the variance-covariance matrix estimates derived from the promotes a more rational and consistent decision-making process. By
nudged and non-nudged subsamples (see Supplementary Materials for simplifying the choice task and helping to reduce hypothetical bias,
full matrices). The findings reveal mixed effects of the informational nudging appears to enhance the overall quality of respondent choices
nudge on preference heterogeneity, as reflected in the significance of the (Caputo and Scarpa, 2022).
standard deviation coefficients. Specifically, significance increases for Fig. 4 exhibits how respondents perceived the effort required to ex­
certain attributes, such as nickel-free, PGI certification, and lycopene press their preferences during the DCE. It considers both the presence or
content, while it decreases for others, including eco-label, ethical label absence of nudging, and whether respondents evaluated all or only some
(etico), organic (bio), paper packaging, and biodegradable packaging attributes in their decision-making process. Results show that 35% of
(degr). These results indicate that nudging does not exert a uniform in­ respondents only partially considered all available attributes when
fluence on heterogeneity across attributes. In contrast, nudging has a completing the DCE, a finding that aligns with previous research in the

6
M. De Salvo et al. Future Foods 12 (2025) 100722

field (De Salvo et al., 2020). However, a deeper analysis reveals a subtle

0.112
0.112
0.155
0.147
0.158
0.169
0.276
0.200
0.238
0.391
0.254
0.257
difference between the two sub-samples: the percentage of respondents
reporting the occurrence of ANA phenomena is slightly higher in the

​
nudged group (36%) compared to the non-nudged group (34%). This

Standard deviation
suggests that nudging may have modestly increased the cognitive effort

***
***

***

***
**
required from participants. This effect becomes more pronounced when

​
​
​

​
​
​
​
examining self-reported cognitive effort. Among participants who
acknowledged using heuristic decision strategies, such as partially
0.750
0.513
0.431
0.187
0.098
0.033
1.810
0.236
0.219
0.347
0.095
1.001
considering the attributes, the proportion who found the experiment
slightly tiring increased markedly in the nudged group, rising from 30%
​

to 41%. A smaller yet notable increase was also observed in the share of
participants who reported the task as moderately tiring, which rose from
0.105
0.062
0.060
0.065
0.055
0.063
0.061
0.145
0.097
0.140
0.092
0.086
0.408
39% to 41%. These shifts suggest that informational nudging may in­
crease perceived cognitive load, particularly among those already
engaging in simplified decision-making strategies. When analysing
***
***

***
***
***
***
***

***
***
***
***
**

which attributes were most frequently considered under heuristic

*

decision-making, distinct patterns emerge between the two subsamples.

In the non-nudged group, respondents most commonly focused on a
− 0.763

− 5.519
0.244
0.142
0.291
0.307
0.211
0.317
1.851
0.306
1.443
0.376
0.527

limited set of attributes, with the most frequent heuristics involving only
Mean
All

origin (11%), origin and price (9%), origin and PGI certification (7%),
only price (5%), and origin with organic certification (5%). In contrast,
in the nudged group, the most frequent heuristic combinations were
0.147
0.161
0.170
0.209
0.192
0.171
0.622
0.331
0.309
0.296
0.271
0.762

origin and price (12%), only origin (9%), only price (6%), origin,
packaging, and price (4%), and origin, organic certification, and price
​

(4%). All other attribute combinations appeared with lower frequencies

Standard deviation

in both subsamples. These results suggest that while origin and price
***
***
***

***

consistently dominate simplified decision strategies, nudging may subtly

​

​
​
​
​

​
​

shift the salience of certain attributes (e.g., packaging or organic certi­

fication) in heuristic processing.
0.822
0.661
0.544
0.020
0.661
0.089
0.971
0.107
0.197
0.632
0.049
1.084

The combined findings from Figs. 3 and 4 offer a clearer under­

​

standing of the effects of nudging in the DCE. Fig. 3 reveals that nudging
influences the significance of the variance-covariance estimates. While
0.153
0.087
0.087
0.091
0.078
0.097
0.086
0.196
0.136
0.189
0.133
0.122
0.589

nudging improves the significance of some standard deviation co­

efficients associated with individual attributes, it simultaneously leads
to a decline in the significance of most covariance coefficients. This
pattern suggests that nudging may prompt respondents to focus more on
***

***
***
***
***

***

***
***
**

**

evaluating attributes individually, rather than considering potential

​
​

interactions between them. As a result, the estimates for individual at­

tributes become more precise and internally consistent.
− 0.819

− 5.578
Nudge

0.135
0.138
0.274
0.314
0.370
0.293
1.754
0.206
1.371
0.417
0.600
Mean

Fig. 3 contributes an important behavioural insight to these findings.

While nudging appears to encourage a more rational and structured
decision-making process, it also seems to slightly increase cognitive
burden. This is evidenced by a higher proportion of respondents in the
4.890
3.480
3.350
3.100
1.870
4.780
0.630
1.540
0.690
0.420
0.900
3.050

nudged group who reported partially considering the attributes and who
​

described the choice tasks as mentally tiring. These results support the
notion that nudging helps to simplify decision-making by directing
Standard deviation

***
***
***

***
**

**

attention to key trade-offs but may also inadvertently heighten cognitive

​

​
​
​
​
​

effort for some participants, particularly those already inclined to rely

on heuristic processing.
0.931
0.911
0.747
0.546
0.401
0.953
0.319
0.504
0.340
0.151
0.327
1.046

Interestingly, the analysis of heuristic patterns reveals that re­

spondents in both the nudged and non-nudged subsamples tended to
​

. p < 0.10, * p < 0.05, ** p < 0.01, *** p < 0.001.

prioritize similar key attributes, most notably, origin and price. However,
in the nudged group, heuristic combinations involving these attributes
4.310
4.200
1.730
3.830
3.920
1.510
3.810
7.980
2.970
6.860
2.540
3.620
8.140

(e.g., origin and price) were more frequently observed, suggesting that
the nudge effectively guided respondents’ attention toward specific,
salient dimensions of the choice task. This finding supports the inter­
***
***

***
***

***
***

***

***
***

pretation that nudging may encourage more deliberate and attribute-

**
Estimates of mixed logit model.

*
​

focused decision-making, while simultaneously reducing reliance on

attribute interactions or more complex trade-offs. Taken together, these
No nudge

− 0.794

− 6.066

results point to a dual effect of nudging. On one hand, it enhances the

0.469
0.188
0.455
0.398
0.160
0.469
2.465
0.521
1.940
0.427
0.574
Mean

significance and internal consistency of individual attribute estimates,

thereby improving the quality of the data and robustness of preference
inference. On the other hand, it appears to increase cognitive effort, as
Attributes

ASC_none

indicated by a higher incidence of self-reported fatigue and partial

other_reg
same_reg
Table 3

nickel

paper

attribute processing. This trade-off is critical when designing nudging

price

etico

degr
PGI

lico
eco

bio

UE

interventions. While nudges can support more thoughtful decisions, they

7
M. De Salvo et al. Future Foods 12 (2025) 100722

Fig. 3. Changes in the significance levels of variance–covariance coefficient estimates between non-nudged and nudged subsamples.

Fig. 4. Self-reported cognitive effort during the DCE and declared onset of Attribute Non-Attendance (ANA).

the marginal WTP for lycopene content increases markedly, rising from
Table 4
€0.19 per package in the non-nudged group to €0.43 per package in the
Summary statistics of marginal WTP estimates.
nudged group, highlighting the impact of the nudge in elevating the
Attributes No-nudged sub-sample Nudged sub-sample perceived value of this functional attribute. However, the nudge also
Mean Standard Deviation Mean Standard Deviation appears to influence WTP for competing attributes. Specifically, average
nichel 0.59 1.17 0.19 1.01 marginal WTP values for characteristics such as nickel-free, PGI certi­
PGI 0.23 1.16 0.17 0.87 fication, eco-sustainability, product origin, organic production, and
eco 0.57 1.30 0.33 0.71 ethical certification are consistently lower among nudged respondents
etico 0.49 0.98 0.38 0.41 compared to their non-nudged counterparts. This suggests a reallocation
lico 0.19 1.05 0.46 0.91
bio 0.60 1.60 0.36 0.50
of attention and perceived value toward lycopene at the expense of other
same_reg 3.11 4.54 2.14 2.92 health- or sustainability-related attributes. By contrast, the marginal
UE 0.63 1.78 0.27 0.77 WTP for packaging-related attributes, such as paper-based and 100%
other_reg 2.44 4.27 1.67 2.58 biodegradable materials, shows less pronounced differences between
paper 0.51 1.74 0.49 1.07
the two groups, indicating that nudging had a more limited impact on
degr 0.67 1.86 0.73 0.76
preferences related to environmental packaging.
Fig. 5 displays the marginal WTP distributions for nudged and non-
may also raise the risk of participant fatigue or disengagement, partic­ nudged respondents. The shape of the distributions clearly illustrates
ularly in complex choice environments. that the nudging intervention significantly alters their form, making them
Table 4 summarizes the marginal WTP distributions for both non- systematically leptokurtic in the nudged group across all attributes. This
nudged and nudged respondents. The results indicate that the infor­ result suggests that marginal WTP values among nudged participants are
mational nudge has a significant effect on mean WTP estimates. Notably, more tightly concentrated around the mean, with fewer extreme values

8
M. De Salvo et al. Future Foods 12 (2025) 100722

Fig. 5. Marginal WTP distributions.

compared to those in the non-nudged group. These findings reinforce the Fig. 6 illustrates the predicted demand curves for nudged and non-
notion that nudging promotes more consistent and focused decision- nudged consumers, focusing on tomatoes with medium and high lyco­
making. By reducing the dispersion of WTP estimates, the nudge appears pene content. The behaviour of the demand functions indicates that
to encourage respondents to engage in a more structured and systematic increasing lycopene levels shifts the demand curves upward and, in
evaluation of attribute trade-offs, thereby reducing behavioural noise and some cases, causes them to rotate, depending on both the type of tomato
improving the reliability of preference inference. and whether the consumer was exposed to the informational nudge. As

9
M. De Salvo et al. Future Foods 12 (2025) 100722

Fig. 6. Predicted demand functions in response to lycopene content across tomato types for nudged and non-nudged consumers.

expected, non-nudged consumers exhibit relatively low sensitivity to 4. Conclusions

changes in lycopene content, with minimal variation in demand across
scenarios. In contrast, nudged consumers respond more strongly, This study employed a DCE to investigate the effect of nudging on
particularly for European-sourced, organic tomatoes packaged in paper consumer behaviour regarding fresh tomatoes, with a particular focus
materials, regardless of the presence or absence of additional value at­ on its impact on cognition, preference heterogeneity, and marginal WTP
tributes such as nickel-free, eco-sustainable, or ethical certifications. for lycopene content and other intrinsic and extrinsic product attributes.
Across all simulated conditions, nudged consumers demonstrate Although the nudge was designed to provide information about the
more elastic demand curves compared to their non-nudged counter­ nutraceutical properties of tomatoes, its influence extended beyond
parts, indicating a greater responsiveness to price and attribute changes. lycopene. The results show that nudging significantly affected prefer­
However, demand responses to lycopene content are less pronounced for ences for a range of attributes, not just the one directly highlighted in the
locally produced tomatoes, even among nudged participants, suggesting informational prompt.
that local origin may act as a dominant cue, reducing the marginal in­ One key finding relates to the Cholesky decomposition of the var­
fluence of other health-related attributes. iance–covariance matrix, where nudging improved the significance of

10
M. De Salvo et al. Future Foods 12 (2025) 100722

the variance coefficients for several attributes (e.g., lycopene, nickel- powerful tool for aligning consumer behaviour with public health and
free, and PGI certification) while reducing the significance of many sustainability goals. From a policy and marketing perspective, the
covariance terms. This pattern suggests that nudging encouraged re­ findings suggest that strategically implemented informational nudges
spondents to focus more on individual attributes, diminishing attention can enhance the perceived value of health-related attributes, improve
to potential interactions among them. the quality of consumer decision-making, and potentially support the
In addition, results indicate that nudging slightly increased cognitive adoption of healthier, more sustainable diets. However, it is crucial to
effort. This was evidenced by a higher share of participants in the balance these benefits against the risk of cognitive overload, ensuring
nudged group reporting partial attribute processing and greater mental that nudging interventions remain effective without overwhelming
fatigue. While nudging supported a more rational, trade-off-based de­ consumers.
cision-making process, it was also associated with a modest increase in
ANA and heuristic-based behaviour, as more participants simplified Funding
their choices by ignoring certain characteristics.
Nudging also had a substantial impact on marginal WTP distribu­ This research was funded by Italian Ministry of University and
tions, particularly for lycopene content. The average WTP for lycopene Research (MUR), Project AGRIFOOD ARS01_00640, “POFACS - Con­
rose from €0.19 to €0.43 per package in the nudged group. However, servabilità, qualità sicurezza dei prodotti ortofrutticoli ad alto contenuto
WTP for other attributes, such as nickel-free, PGI, eco-sustainability, di servizio” (Preservability, quality safety of fruit and vegetable products
origin, organic, and ethical certification, tended to be lower among with high service content), PON R&I 2014–2020.
nudged respondents, suggesting a reallocation of perceived value to­
wards the attribute emphasized by the nudge. Moreover, the WTP dis­ Declaration of generative AI and AI-assisted technologies in the
tributions for nudged participants appeared more concentrated around writing process
the mean, indicating greater consistency and reduced variability,
consistent with more systematic and focused decision-making. During the writing process, the authors used ChatGPT 4◦ to enhance
In summary, the findings highlight a dual effect of nudging. On the grammar, improve readability, and refine the language of the manu­
one hand, nudging enhances preference consistency and improves the script. All content generated by the tool was critically reviewed and
statistical significance of individual-level estimates. On the other, it in­ edited by the authors, who take full responsibility for the accuracy,
troduces slightly greater cognitive demands, as participants become integrity, and originality of the final publication.
more focused on trade-offs but may also experience decision fatigue.
Additionally, nudging tends to narrow attention toward certain attri­
Ethical statement
butes, reducing consideration of more complex attribute interactions.
Overall, while nudging improves decision quality, its potential cognitive
The study was explained to consumers in the online questionnaire.
cost must be carefully weighed.
They were informed that all data will be de-identified and only reported
While this study offers valuable insights, several limitations must be
in the aggregate. All participants acknowledged an informed consent
acknowledged. First, phenomena such as ANA and cognitive effort were
statement in order to participate in the study.
measured through self-reported debriefing questions after the DCE. As is
well documented in the literature, self-reports are not always reliable in
capturing actual cognitive processes. Future studies should complement CRediT authorship contribution statement
this approach with objective measures, such as eye-tracking, response
time analysis, or neurophysiological tools, which offer direct insight into Maria De Salvo: Validation, Methodology, Conceptualization,
attention allocation and cognitive load. However, such methods often Writing – original draft, Writing – review & editing, Software, Formal
involve higher costs and smaller sample sizes, posing practical con­ analysis. Isabella Trovato: Writing – original draft. Laura Giuffrida:
straints. Second, the design of the DCE may have introduced some Visualization, Data curation. Marika Cerro: Visualization, Data cura­
conceptual or perceived overlap among attributes, especially given the tion. Giovanni Signorello: Methodology, Writing – original draft,
complexity of real-world tomato purchasing decisions. This overlap Writing – review & editing, Conceptualization. Giuseppe Cucuzza:
could lead to multicollinearity, which, although typically handled well Funding acquisition, Supervision.
by MXL models, may still affect the separation and interpretation of
effects. To address this, we conducted robustness checks by examining
Declaration of competing interest
attribute correlations and re-estimating models excluding highly corre­
lated attributes. Nevertheless, some residual confounding effects may
The authors declare that they have no conflicts of interest to disclose.
remain. Third, the analysis focuses exclusively on information-based
nudging, which appeals to rational, cognitive pathways. Other forms
Supplementary materials
of nudging such as emotional cues, visual prompts, or choice architec­
ture interventions, may produce different behavioural effects. As such,
Supplementary material associated with this article can be found, in
the generalizability of our findings is limited to informational nudges,
the online version, at doi:10.1016/[Link].2025.100722.
and caution is advised when extrapolating to other types of behavioural
interventions. Finally, the study is based on stated preference data,
Data availability
which reflect hypothetical choices rather than real market behaviour.
Although stated preference methods are widely accepted in the litera­
Data will be made available on request.
ture and offer robust insights, they remain subject to hypothetical bias,
whereby respondents may overstate or misrepresent their true prefer­
ences. Future research should seek to validate these findings using References
incentivized experiments or real-world behavioural data to confirm the
Alemu, M.H., Mørkbak, M.R., Olsen, S.B., et al., 2013. Attending to the reasons for
reliability of the WTP estimates. attribute non-attendance in choice experiments. Environ. Resour. Econ. 54,
Despite these limitations, this study provides compelling evidence 333–359. [Link]
that nudging can meaningfully influence consumer preferences, espe­ Almeida, C., Azevedo, J., Fogel, A., Lopes, E., Vale, C., Padrao, P., 2024. Effectiveness of
nudge interventions to promote fruit and vegetables’ selection, purchase, or
cially in relation to functional food attributes like lycopene content. By consumption: a systematic review. Food Qual. Prefer. 116, 105122. [Link]
promoting more consistent, rational decision-making, nudges can be a 10.1016/[Link].2024.105122.

11
M. De Salvo et al. Future Foods 12 (2025) 100722

Balcombe, K., Fraser, I., McSorley, E., 2015. Visual attention and attribute attendance in ISMEA. (2017). I Numeri Della Filiera del Pomodoro da Industria ([Link]
multi-attribute choice experiments. J. Appl. Econ. 30, 447–467. [Link] com/url?sa=t&source=web&rct=j&opi=89978449&url=[Link]
10.1002/jae.2383. flex/cm/pages/[Link]/L/IT/D/8%25252Fd%25252Fb%25252FD.
Bauer, J.M., Bietz, S., Rauber, J., Reisch, L.A., 2021. Nudging healthier food choices in a f7b5a2e810dad283d1a1/P/BLOB:ID%3D10110/E/pdf&ved=2ahUKEwjale_
cafeteria setting: a sequential multi-intervention field study. Appetite 160, 105106. krJGNAxUy_rsIHWowKLoQFnoECBwQAQ&usg=AOvVaw3J3N6paDEI-as
[Link] WQIxfMZiT).
Blom, S.S., Gillebaart, M., De Boer, F., van der Laan, N., De Ridder, D.T., 2021. Under Jonker, M.F., Donkers, B., de Bekker-Grob, E.W., Stolk, E.A., 2018. Effect of level overlap
pressure: nudging increases healthy food choice in a virtual reality supermarket, and color coding on attribute non-attendance in discrete choice experiments. Value
irrespective of system 1 reasoning. Appetite 160, 105116. [Link] Health 21 (7), 767–771. [Link]
[Link].2021.105116. Joya, K., Ramli, N.N., Shamsudin, M.N., Kamarulzaman, N.H., 2022. Consumers’
Campbell, D., Hensher, D.A., Scarpa, R., 2011. Non-attendance to attributes in willingness to pay for food safety attributes of tomato. Br. Food J. 124 (3), 701–717.
environmental choice analysis: a latent class specification. J. Environ. Plan. Manag. [Link]
54 (8), 1061–1076. [Link] Kassas, B., Cao, X., Gao, Z., House, L.A., Guan, Z., 2023. Consumer preferences for
Caputo, V., Scarpa, R., 2022. Methodological advances in food choice experiments and country-of-origin labeling: bridging the gap between research estimates and real-
modeling: current practices, challenges, and future research directions. Annu. Rev. world behavior. J. Choice Model. 48. [Link]
Resour. Econ. 14 (1), 63–90. [Link] Kassie, G.T., Zeleke, F., Birhanu, M.Y., Scarpa, R. (2020). Reminder nudge, Attribute
023242. nonattendance, and Willingness to Pay in a Discrete Choice Experiment, Selected Paper
Caraban, A., Karapanos, E., Gonçalves, D., Campos, P., 2019. 23 ways to nudge: a review prepared for presentation at the 2020 Agricultural & Applied Economics Association,
of technology-mediated nudging in human-computer interaction. In: Proceedings of Annual Meeting, Kansas City, MO, July 26-28, 2020. DOI: 10.22004/[Link].3042
the 2019 CHI conference on human factors in computing systems, pp. 1–15. https:// 08.
[Link]/10.1145/3290605.3300733. Lima, G.P.P., Gómez, H.A.G., Seabra Junior, S., Maraschin, M., Tecchio, M.A., Borges, C.
Carlsson, F., Kataria, M., Lampi, E., 2018. Dealing with ignored attributes in choice V., 2022. Functional and nutraceutical compounds of tomatoes as affected by
experiments on valuation of Sweden’s environmental quality objectives. Environ. agronomic practices, postharvest management, and processing methods: a mini
Resour. Econ. 47, 65–89. [Link] review. Front. Nutr. 9, 868492. [Link]
Carroll, K.A., Samek, A., Zepeda, L., 2018. Food bundling as a health nudge: investigating Lindstrom, K.N., Tucker, J.A., McVay, M., 2023. Nudges and choice architecture to
consumer fruit and vegetable selection using behavioral economics. Appetite 121, promote healthy food purchases in adults: a systematized review. Psychol. Addictive
237–248. [Link] Behav. 37 (1), 87. [Link]
Collins, A., 2012. Attribute Nonattendance in Discrete Choice models: Measurement of Mancuso, T., De Cianni, R., Di Vita, G.A., Spada, E., Brun, F., Spadaro, D.C., Zanchini, R.,
bias, and a Model For the Inference of Both Nonattendance and Taste Heterogeneity. 2024. Understanding Italian consumers’ perceptions of tomato agricultural
Univ. Sydney Aust. PhD Diss., Inst. Transp. Logist. Stud., Bus. Sch. innovation: exploring the nexus between sustainability, health and consumer beliefs.
Collins, E.J., Bowyer, C., Tsouza, A., Chopra, M., 2022. Tomatoes: an extensive review of J. Clean. Prod. 435, 140528. [Link]
the associated health impacts of Tomatoes and factors that can affect their Maples, M., Interis, M.G., Morgan, K.L., Harri, A., 2018. Southeastern
cultivation. Biology 11 (2), 239. [Link] consumers’willingness to pay for environmental production attributes of fresh
Congiu, L., Moscati, I., 2022. A review of nudges: definitions, justifications, effectiveness. tomatoes. J. Agric. Appl. Econ. 50 (1), 27–47. [Link]
J. Econ. Surv. 36, 188–213. [Link] Mariel, P., Khan, M.A., Meyerhoff, J., 2022. Valuing individuals’ preferences for air
Čop, T., Njavro, M., 2024. Nudge research in agricultural economics: two-decade quality improvement: evidence from a discrete choice experiment in South Delhi.
retrospective of empirical evidence. China Agric. Econ. Rev. 16 (4), 731–746. Econ. Anal. Policy. 74, 432–447. [Link]
[Link] Morejón Cabrera, G., Mariel, P., Abadía, L.K., 2022. Postgraduate study preferences of
De Salvo, M., Scarpa, R., Capitello, R., Begalli, D., 2020. Multi-country stated preferences business administration and economics students from Colombia, Ecuador, and Spain.
choice analysis for fresh tomatoes. Bio-Based Appl. Econ. 9 (3), 241–262. https:// Int. J. Educ. Res. 112. [Link]
[Link]/10.13128/bae-7928. Nielsen, A.S.E., Sand, H., Sørensen, P., Knutsson, M., Martinsson, P., Persson, E.,
EFSA Panel on Dietetic Products, Nutrition and Allergies (NDA), 2011. Scientific opinion Wollbrant, C., 2017. Nudging and Pro-Environmental Behaviour. Nordisk
on the substantiation of health claims related to lycopene and protection of DNA, Ministerråd. [Link]
proteins and lipids from oxidative damage (ID 1608, 1609, 1611, 1662, 1663, 1664, Nishimura, T., 2021. The effect of greenhouse pollination methods on consumers’
1899, 1942, 2081, 2082, 2142, 2374), protection of the skin from UV-induced willingness to pay for tomatoes in Japan. J. Agric. Appl. Econ. 53 (2), 186–208.
(including photo-oxidative) damage (ID 1259, 1607, 1665, 2143, 2262, 2373), [Link]
contribution to normal cardiac function (ID 1610, 2372), and maintenance of normal Nohara, K., 2024. Willingness to pay for pesticide-free vegetables in Hokkaido, Japan:
vision (ID 1827) pursuant to article 13(1) of Regulation (EC) No 1924/2006. EFSa J. the relationship between appearance and pesticide use. Humanit. Soc. Sci. Commun.
9 (4), 2031. [Link] [28 pp.]. 11 (1), 1–12. [Link]
Ellison, B., Bernard, J.C., Paukett, M., Toensmeyer, U.C., 2016. The influence of retail Pandey, S., Olsen, A., Perez-Cueto, F.J., Thomsen, M., 2023. Nudging toward sustainable
outlet and FSMA information on consumer perceptions of and willingness to pay for food consumption at university canteens: a systematic review and meta-analysis.
organic grape tomatoes. J. Econ. Psychol. 55, 109–119. [Link] J. Nutr. Educ. Behav. 55 (12). [Link]
joep.2016.05.002. Poojary, M.M., Passamonti, P., 2015. Extraction of lycopene from tomato processing
Ensaff, H., 2021. A nudge in the right direction: the role of food choice architecture in waste: kinetics and modelling. Food Chem. 173, 943–950. [Link]
changing populations’ diets. Proc. Nutr. Soc. 80 (2), 195–206. [Link] [Link].2014.10.127. Epub 2014 Oct 29. PMID: 25466110.
10.1017/S0029665120007983. Printezis, I., Grebitus, C., 2020. College-age millennials’ preferences for food supplied by
Franceschinis, C., Scarpa, R., Thiene, M., 2024. Nudging consumers’ choices for niche urban agriculture. Front. Sustain. Food Syst. 4, 48. [Link]
milk: a real purchase experiment. Food Policy. 128, 102729. [Link] fsufs.2020.00048.
10.1016/[Link].2024.102729. Revelt, D., Train, K., 1998. Mixed logit with repeated choices: households’ choices of
Gillebaart, M., Blom, S.S.A.H., Benjamins, J.S., de Boer, F., De Ridder, D.T.D., 2023. The appliance efficiency level. Rev. Econ. Stat. 80 (4), 647–657. [Link]
role of attention and health goals in nudging healthy food choice. Front. Psychol. 14, 10.1162/003465398557735.
1270207. [Link] Rose, J.M., Bliemer, M.C., Hensher, D.A., Collins, A.T., 2008. Designing efficient stated
Giovannetti, M., Avio, L., Barale, R., Ceccarelli, N., Cristofani, R., Iezzi, A., Mignolli, F., choice experiments in the presence of reference alternatives. Transp. Res. Part B 42
Picciarelli, P., Pinto, B., Reali, D., Sbrana, C., Scarpato, R., 2012. Nutraceutical value (4), 395–406. [Link]
and safety of tomato fruits produced by mycorrhizal plants. Br. J. Nutr. 107 (2), Ruth, C., Lakshmi, T.N., Gopal, K., 2016. Studies on epidemiology of major diseases of
242–251. [Link] horticultural crops. Plant Arch. 16 (1), 65–70.
Gravert, C., Kurz, V., 2021. Nudging à la carte: a field experiment on climate-friendly Sapio, S., Vecchio, R., 2024. The effect of nudges on healthy food choices of individuals
food choice. Behav. Public Policy. 5 (3), 378–395. [Link] with a low socio-economic position: a systematic literature review. Trends. Food Sci.
bpp.2019.11. Technol. [Link]
Hansen, P.G., Jespersen, A.M., 2013. Nudge and the manipulation of choice: a Scarpa, R., Gilbride, J., Campbell, D., Hensher, D.A., 2009. Modelling attribute non-
framework for the responsible use of the nudge approach to behaviour change in attendance in choice experiments for rural landscape valuation. Eur. Rev. Agric.
public policy. Eur. J. Risk. Regul. 4 (1), 3–28. Econ. 36 (2), 151–174. [Link]
Hansen, P.G., Schilling, M., Malthesen, M.S., 2021. Nudging healthy and sustainable food Shonkwiler, V.P., 2024. Exploring Co-labeling strategies in local food markets: insights
choices: three randomized controlled field experiments using a vegetarian lunch- from a choice experiment on Southern buyers’ Preferences for cherry tomatoes.
default as a normative signal. J. Public Health (Bangk.) 43 (2), 392–397. [Link] HortScience 59 (6), 736–742. [Link]
org/10.1093/pubmed/fdz154. 269403437.
Hensher, D.A., 2006. Revealing differences in willingness to pay due to the Tan, B.L., Norhaizan, M.E., Liew, W.P., Sulaiman Rahman, H., 2018. Antioxidant and
dimensionality of stated choice designs: an initial assessment. Environ. Resour. Econ. oxidative stress: a mutual interplay in age-related diseases. Front. Pharmacol. 9,
34, 7–44. [Link] 1162. [Link] PMID: 30405405; PMCID:
Hole, A.R., 2007. Fitting mixed logit models by using maximum simulated likelihood. PMC6204759.
Stata J. 7 (3), 388–401. [Link] Thaler, R.H., Sunstein, C.R., 2021. Nudge: The final Edition. Yale University Press.
Ibrahim, M.S., Mujtaba, A., Hassan, I., Nadeem, M., Sohail, A., Abbasi, K.S., Samee, A., Tran, L., McCann, L., Su, Y., 2023. Consumer preferences for produce grown with
Azeem, W., Waqar, W., Fatima, I., 2023. Antioxidant and nutraceutical potential of reduced pesticides: a choice experiment in Missouri. Renew. Agric. Food Syst. 38,
tomato for health and life quality: a review: tomato as functional fruit. Food Sci. e49. [Link]
Appl. Microbiol. Rep. 2 (1), 8–25. Train, K.E., 2009. Discrete Choice Methods With Simulation. Cambridge university press.
[Link]

12
M. De Salvo et al. Future Foods 12 (2025) 100722

Vecchio, R., Cavallo, C., 2019. Increasing healthy food choices through nudges: a Weimer, K., Ahlström, R., Esteves, F., 2022. The effect of nudging in promoting the
systematic review. Food Qual. Prefer. 78. [Link] consumption of organic fruits and vegetables. Front. Physiol. 13, 720606. https://
foodqual.2019.05.014. [Link]/10.3389/fpsyg.2022.720606.
Vella, F., Migliore, G., Schifani, G., Vecchio, R., 2024. Nudging interventions to foster the Weller, P., Oehlmann, M., Mariel, P., Meyerhoff, J., 2014. Stated and inferred attribute
reduction of pesticides in viticulture. Insights from Italian winegrowers. Environ. non-attendance in a design of designs approach. J. Choice Model. 11, 43–56. https://
Dev., 101120 [Link] [Link]/10.1016/[Link].2014.04.002.
Verneau, F., La Barbera, F., Furno, M., 2019. The role of health information in Yi, S., Kanetkar, V., Brauer, P., 2022. Nudging food service users to choose fruit-and
consumers’ Willingness to pay for canned crushed tomatoes enriched with lycopene. vegetable-rich items: five field studies. Appetite 173. [Link]
Nutrients. 11 (9), 2173. [Link] PMID: 31510086; appet.2022.105978.
PMCID: PMC6770788. Zapata, S.D., Villavicencio, X.A., Marconi, T.G., Avila, C.A., 2025. Using sensory analysis
and stated preference valuation to assess the willingness to pay for fresh tomato
attributes. Agribusiness. [Link]

13