Rasmus FriispersName, Laurits R. SkovpersName (2017date). “Comparison of Three Nudge Interventions (Priming, Default Option, and Perceived Variety) to Promote Vegetable Consumption in a Self-Service Buffet Setting.” PLOS ONEtitle·j, CopenhagenplaceName.
Object of study — Whether primingterm·prm and other nudgesterm·prm increase healthy food choice in a real buffet.
Research question — Do primingterm·prm, default optionsterm·dom and perceived varietyterm nudgeterm·prm people to eat more vegetablesterm·key in a self-service setting?
Contribution — Tests primingterm·prm as a behavioural nudgeterm·prm for vegetable consumptionterm·var, situating primingterm·prm within applied behavioural economics and public health.
| P · prime | an environmental priming cue in a buffet |
| T · target | a food-choice opportunity |
| B · baseline | vegetable intake with no cue |
| E · priming effect | increased vegetable consumption |
| A · association | congruent (positive) |
phenomenon · claim-type · cross-domain-transfer · mechanism · claim-relation · replication-status · limitation — grounded in the full text| phenomenon | nudge effects on vegetable intake — “The objective within the present study is to investigate whether nudging can be used to influence vegetable intake at a self-service buffet.” | |
| claim-type | existence — “This study suggests that the nature of a nudge-based intervention can have different effects, whether it is increasing intake of healthy components, or limiting intake of unhealthy meal components.” | |
| method — evidence | — “The study had an experimental cross-over design with three treatments corresponding to the three Nudge strategies (priming, default and perceived variety) and a control serving.” | |
| transfer | from cognitive / behavioral psychology (heuristics and biases): nudging / choice-architecture interventions (priming, defaults) drawing on heuristics-and-biases psychology, applied to influence food choice — “Nudging is attracting considerable interest due to its understanding and application of heuristic biases among consumers” | transfers[0] |
| mechanism | nudging: redesign the choice architecture (priming, defaults, perceived variety) to steer behaviour toward a target without restricting options or changing incentives — “consumer behaviour can be influenced without restricting or providing incentives for behaviour change” | mechanisms[0] |
| builds-on | the nudge concept (choice architecture, Thaler and Sunstein) — “in a predictable way without forbidding any options” | claim_relations[0] |
| extends | prior default-serving and plate-clearing research conducted with actual plates (refs 22, 23, 43, 44) — “Although the majority of research on default servings and plate clearing is conducted on actual plates [22, 23, 43, 44], our study showed similar effects although using a bowl-type container” | claim_relations[1] |
| extends | prior research on default servings and plate clearing conducted on actual plates [22, 23, 43, 44] — “Although the majority of research on default servings and plate clearing is conducted on actual plates [22, 23, 43, 44], our study showed similar effects although using a bowl-type container.” | claim_relations[2] |
| builds-on | the theory of plate clearing — “The default portion was approximately 25% higher than the average intake during the control setting and the observed increase in intake of vegetables is in correspondence with what was anticipated according to the theory of plate clearing.” | claim_relations[3] |
| sense | THE PRIMING ARM MAY HAVE STOPPED BEING A NUDGE, AND THE AUTHORS SAY SO IN THEIR OWN DISCUSSION: the green ambience `could lead to an activation of a reflective decision making, promoting a healthy-eater self-schema of food choice, hence not be an automatic process nudge, but a reflective.` The room was meant to work below awareness; it may instead have reminded people that they are the kind of person who eats well. The intervention had an effect and the CATEGORY it was meant to demonstrate did not survive it - which is a different kind of negative result from a null, and one a corpus that records only effect sizes cannot hold — “This could lead to an activation of a reflective decision making, promoting a healthy-eater self-schema of food choice [40], hence not be an automatic process nudge, but a reflective” | senses[0] |
| sense | AND THEY SAY THEY CANNOT TELL WHICH IT WAS: `However, this study did not measure the participants' level of health-schematic nor to what extent their food choice was reflective or not.` The distinction that decides whether the finding is a nudge result or its opposite was not measured. Stated plainly, immediately after the interpretation that depends on it, rather than left for a reader to notice — “However, this study did not measure the participants’ level of health-schematic nor to what extent their food choice was reflective or not” | senses[1] |
| sense | THE DEFINITION OF A NUDGE FORBIDS ITS OWN MANIPULATION CHECK, which is the sharpest thing in this paper. The variety arm had to leave the ACTUAL variety unchanged `in order to be consistent with the definition of a nudge` - and then: `ASKING THE PARTICIPANTS of their perception of the vegetable variety WOULD REQUIRE THEM TO EMBARK IN REFLECTIVE DECISION-MAKING, THUS LOSING THE INTENDED IDEA OF NUDGING their food choice.` You cannot ask whether it worked without destroying the thing you are asking about. The definition does not merely describe the intervention; it rules out the measurement that would verify it — “Asking the participants of their perception of the vegetable variety would require them to embark in reflective decision-making thus losing the intended idea of nudging their food choice” | senses[2] |
| sense | THE ARM THAT `WORKED` MOVED THE WRONG VARIABLE, TWICE OVER. Priming did not raise vegetable intake - it `significantly decreased the consumption of chilli con carne`. And the variety arm `decreased the self-served amount of meat and total energy, but the total intake of vegetables was not increased as hypothesised`. Two of three interventions produced a healthier meal by SUBTRACTION rather than by the addition they were designed to cause, so a summary reporting `nudges promoted healthier meal composition` is true and tells a reader the opposite of what happened — “The experiment of increasing the visual perceived variety decreased the self-served amount of meat and total energy, but the total intake of vegetables was not increased as hypothesised” | senses[3] |
| sense | A CONFOUND NAMED AT THE CONCEPTUAL LEVEL RATHER THAN THE STATISTICAL ONE: `From a conceptual level an important concern to address is that default and variety conditions also differ in the easiness to reach the vegetables.` In the default arm people grasped a pre-packed box; in the variety arm they served themselves. So the two conditions differ in effort as well as in the thing being manipulated, and no statistic in the paper separates them. The authors raise it against their own comparison — “From a conceptual level an important concern to address is that default and variety conditions also differ in the easiness to reach the vegetables” | senses[4] |
| control | each participant eating twice and so serving as her or his own control, once in the control setting and once in the experimental setting — held — rules out between-person differences in appetite, habitual vegetable intake and body size, which a between-groups comparison of 88 people would leave in the estimate — “Each individual was partaking twice thus serving as her/his own control, randomly assigned to start with control or experimental setting.” | controls[0] |
| control | the order of the two settings randomly allocated in a balanced manner, so that equal numbers began with the control and with the experimental setting — held — rules out an order effect - that eating in the lab a second time, whatever the setting, changes how much salad a person takes — “After recruitment, participants were randomly allocated to either start with their control setting or experimental setting in a balanced manner.” | controls[1] |
| evidence status note | prior default-serving and plate-clearing research conducted with actual plates (refs 22, 23, 43, 44) — QUOTE RECOVERED FROM THE PAPER'S OWN CHARACTERS. The recorded quote was LETTERS-ONLY IDENTICAL to this stretch of the staged text but not an exact substring of it - spaces, hyphens, ligature glyphs or mathematical operators differed, so it could never anchor. This is the file's own text between the same two letters: not produced by a rule about the damage, copied out of the file. Round-trip checked (the recovered span is letters-only equal to the original) and length-checked. The text as originally recorded is kept in evidence as extracted. Update, 15 September 2026: CLEARED evidence status='quote-not-in-current-staged-text' - the quote now locates in the current staged text, so the condition the marker recorded no longer holds. — “Although the majority of research on default servings and plate clearing is conducted on actual plates [22, 23, 43, 44], our study showed similar effects although using a bowl-type container” | claim_relations[1] |
| audit | nudge —distinct-from→ reflective decision making — Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but the span says a prime 'could lead to' reflective decision making and 'hence not be' an automatic nudge; hedged. — “This could lead to an activation of a reflective decision making, promoting a healthy-eater self-schema of food choice [40], hence not be an automatic process nudge, but a reflective” | concept_edges[0] |
| differs in | nudge —distinct-from→ reflective decision making — WHETHER THE PERSON DELIBERATED. A nudge is defined as working on automatic decision processes; if the green ambience worked by reminding people of a healthy-eater self-schema, they deliberated, and the intervention was 'NOT ... an automatic process nudge, but a reflective'. The two are alternative routes to the SAME observed behaviour change, which is why the effect cannot settle which occurred — “This could lead to an activation of a reflective decision making, promoting a healthy-eater self-schema of food choice [40], hence not be an automatic process nudge, but a reflective” | concept_edges[0] |
| note | nudge —distinct-from→ reflective decision making — WEIGHT 0.85 AND THE HEDGE IS THE FIRST CLAUSE: 'This COULD lead to'. The authors are raising this against their own priming arm, and the next sentence says they did not measure whether choice was reflective. The edge is worth recording precisely because it is the distinction that would undo the paper's own category - a reader should meet it with its uncertainty attached rather than later. — “This could lead to an activation of a reflective decision making, promoting a healthy-eater self-schema of food choice [40], hence not be an automatic process nudge, but a reflective” | concept_edges[0] |
| weight | nudge —distinct-from→ reflective decision making — 0.85 — “This could lead to an activation of a reflective decision making, promoting a healthy-eater self-schema of food choice [40], hence not be an automatic process nudge, but a reflective” | concept_edges[0] |
| outcome (stated by the tagger, no quotation) | control — held | controls[0] |
| outcome (stated by the tagger, no quotation) | control — held | controls[1] |
| replication | untested | |
| limitation | food choice is complex, determined by many interrelating factors — “food choice behaviours are highly complex and determined by several interrelating factors” | limitations[0] |
| limitation | The priming manipulation's proposed reflective mechanism was not measured. — “However, this study did not measure the participants’ level of health-schematic nor to what extent their food choice was reflective or not.” | limitations[1] |
| limitation | Only a single lunch meal was measured, so whole-day intake cannot be inferred. — “This study only looked at a singular meal (lunch) and thus cannot conclude much about a whole day dietary intake” | limitations[2] |
| limitation | Awareness of being in an experimental eating setting may itself have depressed intake. — “Likewise, participants knowing it was an experimental food setting (although not obvious at first) might have made them conscious about being measured and evaluated, leading to a decrease in intake” | limitations[3] |
| limitation | Women were over-represented relative to the Danish population. — “This current study had an over-representation of women compared to men, misrepresenting the national gender distribution of Denmark.” | limitations[4] |
| limitation | The study cannot say whether the priming effect worked through automatic or reflective decision-making, because that was never measured. — “However, this study did not measure the participants’ level of health-schematic nor to what extent their food choice was reflective or not.” | limitations[5] |
| limitation | Only a single lunch meal was observed, so whole-day dietary intake cannot be inferred. — “This study only looked at a singular meal (lunch) and thus cannot conclude much about a whole day dietary intake; however, previous research suggests that participants did not compensate for their energy intake during lunch later in the day [52], hence indicating the potential for promoting health and well being through a singular lunch serving.” | limitations[6] |
| limitation | Awareness of being in an experimental food setting may itself have depressed intake. — “Likewise, participants knowing it was an experimental food setting (although not obvious at first) might have made them conscious about being measured and evaluated, leading to a decrease in intake [51].” | limitations[7] |
| limitation | The unfamiliar laboratory setting may have affected consumption. — “These experiments were held in, for some, a new and unfamiliar setting (the Foodscape Lab).” | limitations[8] |
finding · quantity · variable · definition · key-term — grounded in the full text| finding | RETRACTED (RE-GROUNDED, AND THE CLAIM WAS WRONG, NOT ONLY THE SPAN) · In a self-service buffet, three nudge interventions (priming, default option, perceived variety) were tested for boosting vegetable consumption; THE DEFAULT ARM increased vegetable energy intake (124 vs 90 kcal, p<0.01). — “The default experiment successfully increased the energy intake from vegetables among the study participants (124 kcal vs. 90 kcal in control, p<0.01).” | content_tags.findings[0] |
| retracted | finding — RE-GROUNDED, AND THE CLAIM WAS WRONG, NOT ONLY THE SPAN. The published quote welded 'The priming arm of the experiment' to a predicate 1,985 characters away; the paper's subject there is 'The default experiment'. The tag therefore attributed the vegetable-intake result to PRIMING when the paper attributes it to the DEFAULT nudge - an error in the direction of this project's own subject. Priming's own result is the opposite outcome and is now tagged separately. | content_tags.findings[0] |
| finding | Priming and perceived variety reduced total energy intake (169 kcal and 124 kcal, both p<0.01), mainly by cutting the meat-based component. — “Both the priming condition and perceived variety reduced the total energy intake among the study participants (169 kcal, p<0.01 and 124 kcal, p<0.01, respectively), mainly through a decrease in the meatbased meal component.” | content_tags.findings[1] |
| quantity | 88 participants — initial study sample recruited in Copenhagen — “The initial sample consisted of 88 participants recruited in Copenhagen, Denmark” | content_tags.quantities[0] |
| quantity | 124 vs 90 kcal (p<0.01) — vegetable energy intake in the priming arm vs control — “successfully increased the energy intake from vegetables among the study participants (124 kcal vs. 90 kcal in control, p<0.01)” | content_tags.quantities[1] |
| quantity | 200g — pre-portioned salad in the default-option arm — “the salad was pre-portioned into a bowl containing 200g of vegetables” | content_tags.quantities[2] |
| variable | vegetable consumption (dependent) — “increased the energy intake from vegetables among the study participants” | content_tags.variables[2] |
| variable | energy intake (dependent) — “increased the energy intake from vegetables among the study participants” | content_tags.variables[3] |
| definition | EVIDENCE CORRECTED (RE-GROUNDED) · nudge — “This is done by making a choice architectural change that alters people’s behaviour in a predictable way without forbidding any options or significantly changing their economic incentives” | content_tags.definitions[0] |
| retracted | definition — RE-GROUNDED. The evidence was a substring of the paper's own title, 'three nudge interventions (priming, default option, and perceived variety) to promote vegetable consumption', which names the study rather than defining the term. The replacement is the paper's actual definition of a nudge as a choice-architectural change 'without forbidding any options or significantly changing their economic incentives' - the definitional content a definitions row is for. | content_tags.definitions[0] |
| definition | default option — “the salad was pre-portioned into a bowl containing 200g of vegetables” | content_tags.definitions[1] |
| key-terms | nudge, vegetable, consumption, priming, default, participants, intervention, choice architecture, variety |
| concept | priming — “Priming attempted to prime vegetable choice, but significantly decreased the consumption of chilli con carne” | concepts[0] |
| concept | nudge — “This study suggests that the nature of a nudge-based intervention can have different effects, whether it is increasing intake of healthy components, or limiting intake of unhealthy meal components” | concepts[1] |
| concept | default option — “The default option recommends a course of action, which requires no or little reflection, where consumers will” | concepts[2] |
| concept | perceived variety — not defined in this paper's own words; the arm designed so that the ACTUAL variety never changed - the same salad, separated into its components - which is what keeps it inside the definition of a nudge and also what makes it uncheckable — “Similarly, it has been demonstrated across studies that increased variety, or increase in perceived variety, can lead to an increase in food intake” | concepts[7] |
| concept | reflective decision making — not defined in this paper's own words; the route a nudge is defined AGAINST, and the one the authors suspect their own priming arm took. Naming it is what turns a successful intervention into an unresolved result — “This could lead to an activation of a reflective decision making, promoting a healthy-eater self-schema of food choice [40], hence not be an automatic process nudge, but a reflective” | concepts[8] |
| ANOVA F | — “Oneway ANOVA with post hoc test was performed to explore any changes between the three nudges for: total intake, total chilli con carne intake, total rice intake and total vegetable intake.” | content_tags.statistics[0] |
| chi-square χ² | — “2 With the use of the chi-square test.” | content_tags.statistics[1] |
| effect-size d | — “Comparing the effect size of the three nudges (presented as the difference between intervention and control in g) a significant difference” | content_tags.statistics[2] |
| evidence status note | effect-size — QUOTE RECOVERED FROM THE PAPER'S OWN CHARACTERS. The recorded quote was LETTERS-ONLY IDENTICAL to this stretch of the staged text but not an exact substring of it - spaces, hyphens, ligature glyphs or mathematical operators differed, so it could never anchor. This is the file's own text between the same two letters: not produced by a rule about the damage, copied out of the file. Round-trip checked (the recovered span is letters-only equal to the original) and length-checked. The text as originally recorded is kept in evidence as extracted. | content_tags.statistics[2] |
| hypothesis-testing | as written: hypothesis — “The study had hoped to have a more representative sample, but it was not a requirement of the study’s design nor a theoretical basis for the alternative hypothesis’ expected effectiveness.” | content_tags.statistics[3] |
| mean μ | — “3 Mean ± SD, all such values.” | content_tags.statistics[4] |
| p-value p | as written: p-values — “Results are presented as means ± SDs, with corresponding p-values for comparisons for both the analysis within and between nudges” | content_tags.statistics[5] |
| evidence status note | p-value — QUOTE RECOVERED FROM THE PAPER'S OWN CHARACTERS. The recorded quote was LETTERS-ONLY IDENTICAL to this stretch of the staged text but not an exact substring of it - spaces, hyphens, ligature glyphs or mathematical operators differed, so it could never anchor. This is the file's own text between the same two letters: not produced by a rule about the damage, copied out of the file. Round-trip checked (the recovered span is letters-only equal to the original) and length-checked. The text as originally recorded is kept in evidence as extracted. | content_tags.statistics[5] |
| sample-size n/N | as written: n = 24 — “Priming (n = 24) Default (n = 33) Variety (n = 31)” | content_tags.statistics[6] |
| standard-deviation SD/σ | as written: SD — “3 Mean ± SD, all such values.” | content_tags.statistics[7] |
| standard-error SE | as written: SE — “grams) Rice 2 (grams) Chilli con Carne 3 (grams) Total intake 4 (grams) Total intake 5 (kcal) Mean differenc e SE p Mean difference SE p Mean differenc e SE p Mean difference SE p Mean differen ce SE p Default Priming 80.59” | content_tags.statistics[8] |
| evidence status note | standard-error — THE QUOTE IS A FLATTENED TABLE: it starts inside the first column heading of the paper's Table 4 (vegetables, in grams) and runs across the heading row, where 'SE' heads each standard-error column, into the first data value. | content_tags.statistics[8] |
| statistical-significance | — “satisfaction and appetite levels were mainly not significant within and between participants, and if they achieved statistical significance, the difference was not meaningful” | content_tags.statistics[9] |
| evidence status note | statistical-significance — QUOTE RECOVERED FROM THE PAPER'S OWN CHARACTERS. The recorded quote was LETTERS-ONLY IDENTICAL to this stretch of the staged text but not an exact substring of it - spaces, hyphens, ligature glyphs or mathematical operators differed, so it could never anchor. This is the file's own text between the same two letters: not produced by a rule about the damage, copied out of the file. Round-trip checked (the recovered span is letters-only equal to the original) and length-checked. The text as originally recorded is kept in evidence as extracted. | content_tags.statistics[9] |
| t-test t | — “0.04 2.4±1.0 2.6±1.0 ns Are you satisfied? 8.4± 1.3 8.1±1.6 ns 8.5±1.0 8.5±1.5 ns 8.6±1.0 8.4±1.0 ns * Student t-test analysis of appetite scores from control day compared to interventi on day https://do i.org/10.1371/j ournal.” | content_tags.statistics[10] |
Full text — tagged
The reference list is not part of the text we hold for this paper. Our copy was extracted from the publisher's page and the bibliography did not survive that extraction, so the article below ends before its references. Nothing has been removed by us.
RESEARCH ARTICLE Comparison of three nudgeterm·prm interventionsterm·key (primingterm·prm, default optionterm·dom, and perceived varietyterm) to promote vegetable consumptionterm·var in a self-service buffet setting Rasmus Friis1¤a, Laurits Rohden Skov2¤b, Annemarie Olsen1, Katherine Marie Appleton3, Laure Saulais4,5, Caterina Dinnella6, Heather Hartwell7, Laurence Depezay8, Erminio Monteleone6, Agnès Giboreau4, Federico J. A. Perez-Cueto1* 1 Department of Food Science, University of Copenhagen, Frederiksberg C, Denmark, 2 Institute for Planning & Development, Aalborg University, Copenhagen SV, Denmark, 3 Research Centre for Behaviour Change, Department of Psychology, Faculty of Science and Technology, Bournemouth University, Poole, Dorset, United Kingdom, 4 Centre for Food and Hospitality Research, Institut Paul Bocuse, Chateau du vivier, Ecully Cedex, France, 5 UMR GAEL, CNRS, INPG, INRA, Universite´ Grenoble-Alpes, Saint Martin d’Hères, France, 6 Department of Management of Agricultural, Food and Forestry Systems, University of Florence, Florence, Italy, 7 School of Tourism, Foodservice and Applied Nutrition Research Group & Health and Wellbeing, Faculty of Management, Bournemouth University, Bournemouth, United Kingdom, 8 BONDUELLE Corporate Research & Communication, Food & Behaviours department, Fondation Louis Bonduelle, Villeneuve D’Ascq, France ¤a Current address: Aarhus University, Aarhus, Denmark ¤b Current address: Danish Competition and Consumer Authority, Valby, Denmark * apce@food.ku.dk Abstract Background Dietary choices in out-of-home eating are key for individual as well as for public health. These dietary choices are caused by a wide array of determinants, one of which is automatic decision-making. Nudging is attracting considerable interest due to its understanding and application of heuristic biases among consumers. The aim of this study is to test and compare three nudgesterm·prm in promoting vegetable consumptionterm·var among test persons in a food labbased experiment.
Methods The initial sample consisted of 88 participantsterm·key recruited in Copenhagen, Denmark. Each study participantterm·key was randomly assigned to one of the three experiments: primingterm·prm, defaultterm·key and perceived varietyterm. The primingterm·prm arm of the experiment consisted of creating a leafy environment with green plants and an odour of herbs. In the defaultterm·key arm of the experiment, the salad was pre-portioned into a bowl containing 200g of vegetablesterm·key. The third experiment divided the pre-mixed salad into each of its components, to increase the visual varietyterm·key of vegetablesterm·key, yet not providing an actual increase in items. Each individual was partaking twice thus serving as her/his own control, randomly assigned to start with control or experimental setting.
a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: FriispersName R, SkovpersName LR, Olsen A, Appleton KM, Saulais L, Dinnella C, et al. (2017date) Comparison of three nudgeterm·prm interventionsterm·key (primingterm·prm, default optionterm·dom, and perceived varietyterm) to promote vegetable consumptionterm·var in a self-service buffet setting. PLoS ONE 12(5): e0176028. https://doi.org/10.1371/ journal.pone.0176028 Editor: Gotthard Kunze, Leibniz-Institut fur Pflanzengenetik und Kulturpflanzenforschung Gatersleben, GERMANY Received: January 5, 2017date Accepted: April 4, 2017date Published: May 31, 2017date Copyright: © 2017date FriispersName 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: All relevant data are within the paper and its supporting information files. Data are from the VeggiEAT study whose authors may be contacted at apce@food.ku.dk.
Funding: The research leading to these results has received funding from the European Community’s Seventh Framework Programme (FP7/2007-2013) under the Grant Agreement No. FP7 - 612326 - VeggiEat.
PLOS ONE | https://doi.org/10.1371/journal.pone.0176028 May 31, 2017date 1 / 16 Competing interests: LD is employed by Louis Bonduelle Foundation. Her role in the present study has been financed by EU/Marie Curie FP7 VeggiEAT Project. This study has been performed and reported without any interference from Bonduelle Group or Louis Bonduelle Foundation.
Abbreviations: BMI, Body Mass Index; cm, Centimetres; iBuffet, Intelligent Buffet; kg, Kilogram; NCD, non-communicable diseases.
Comparison of three nudgeterm·prm interventionsterm·key to promote vegetable consumptionterm·var Results The defaultterm·key experiment successfully increased the energy intaketerm·var from vegetablesterm·key among the study participantsterm·key (124 kcal vs. 90 kcal in control, p<0.01). Both the primingterm·prm condition and perceived varietyterm reduced the total energy intaketerm·var among the study participantsterm·key (169 kcal, p<0.01 and 124 kcal, p<0.01, respectively), mainly through a decrease in the meatbased meal component.
Conclusions Considerable progress has been made with regard to understanding the use of nudging in promoting a healthier meal composition, including increasing vegetableterm·key intake. This study suggests that the nature of a nudge-based interventionterm·key can have different effects, whether it is increasing intake of healthy components, or limiting intake of unhealthy meal components. This work has demonstrated that consumer behaviour can be influenced without restricting or providing incentives for behaviour change. The present findings have promising application to the foodservice sector.
Background Poor dietary choices, such as overconsumption of energy-dense foods and low intakes of fruit and vegetablesterm·key, have been associated with an increased risk of non-communicable diseases (NCDs) such as, obesity, type II diabetes and ischemic heart diseases. According to the World Health Organization, e.g. NCDs account for 90% of all disease-related deaths in Denmark [1], and it is expected to increase by 13% by the year 2020date [2]. A diet composed of more plant origin (fruits and vegetablesterm·key) can decrease the incidence of NCD [3]. The current public health challenge is to replace high energy dense foods with fruits and vegetablesterm·key in order to reduce the energy content of the diet [4]. The majority of the Danish population have knowledge of the recommendations, which have been in place for the last 17 years, supported by strong health campaigning [5]. They also have the economic and environmental ability to follow them. Yet, there is a gap between this knowledge and actual behavioural action [5], and the intake of fruits and vegetablesterm·key neither matches the recommendations among men nor among women in Denmark [6].
Out-of-home eating is widely considered a part of the modern lifestyle and the public food environment has a central role to play in providing a sense of quality in diets and ensuring public health. Yet, research suggests that out-of-home eating is linked to more energy dense meals [7, 8], larger portion sizes [9, 10], and poor nutritional intake [11, 12]. Canteen and selfservice buffets provide customers the opportunity to assemble a meal according to their own preference. However, food choice behaviours are highly complex and determined by several interrelating factors such as availability, liking and quality. Furthermore, eating is also a source of pleasure and comfort, and reflects our personal and cultural characteristics, social status and relationships.
After decades of addressing unhealthy dietary behaviours through information campaigns, legislation and education, the public health sector now suggests targeting the food environment in order to promote desirable food choices [13]. Recent research in consumer science acknowledges that a food decision depends partly upon how the choice is presented at pointof-purchase [14–17], as well as the previously mentioned determinants. Behavioural economic PLOS ONE | https://doi.org/10.1371/journal.pone.0176028 May 31, 2017date 2 / 16 Comparison of three nudgeterm·prm interventionsterm·key to promote vegetable consumptionterm·var theories of decision-making suggest that food choices can be nudged towards healthier meal selections by changing the choice task [18, 19]. This is done by making a choice architectural change that alters people’s behaviour in a predictable way without forbidding any options or significantly changing their economic incentives [19]. As such, a person’s behaviour may be altered without having to impose or force them into actively engaging in healthier choices. One approach of nudging is primingterm·prm, that is to say, providing cues in the environment that unconsciously drive decisions. For instance, visual stimuli and odours of food successfully increase hunger by influencing dopamine’s action in certain regions of the brain [20]. Likewise, ambience affects food intake in that smell, light as well as colour have substantial impact on food choice [21]. A second approach is offering a defaultterm·key choice. The default optionterm·dom recommends a course of action, which requires no or little reflection, where consumers will ‘go with the flow’ of pre-set options such as type of food or the choice of clearing the plate where the consumer will consume what the plate holds [22, 23]. Notably, studies show that portion size (which constitutes a defaultterm·key quantity) is a significant determinant of food intake [9, 24–27]. A third approach is to increase the perceived varietyterm of healthy food choices. It has been shown that varietyterm·key is valued by consumers, as a varietyterm·key of options increases the chance of matching optimal consumer preference [28]. Meengs and colleagues found that proposing a varietyterm·key of three different vegetablesterm·key (carrots, broccoli and snap peas) increased the consumptionterm·key compared to serving only one of the vegetableterm·key items [29]. Similarly, it has been demonstrated across studies that increased varietyterm·key, or increase in perceived varietyterm, can lead to an increase in food intake [17, 29–33]. However, extending the (perceived) amount of options may also increase the cognitive burden of evaluating opportunities [28], depending on the extent and the nature of the varietyterm·key proposed [34].
The objective within the present study is to investigate whether nudging can be used to influence vegetableterm·key intake at a self-service buffet. In particular the aim was to investigate and compare the efficiency of primingterm·prm, using defaultterm·key choice options, and increasing perceived varietyterm.
Methods The study had an experimental cross-over design with three treatments corresponding to the three Nudgeterm·prm strategies (primingterm·prm, defaultterm·key and perceived varietyterm) and a control serving. The participantsterm·key were randomized to one of the three nudgeterm·prm strategies and a control serving. The experiment was conducted at the FoodScape Lab at Aalborg University, Copenhagen, reproducing the settings of a self-serving buffet [35]. To achieve as natural a lunch-situation as possible, participantsterm·key were not informed of the overall study purpose and food served was similar to what you would normally find in the campus canteen. In total there were 11 test days; the three experiments were dispersed over six days (two for each experimental setting) and there were five control days. The order of the sessions were randomized. Each participantterm·key attended one experimental and one control session. The food was provided free of charge, but no further incentives were used.
Recruitment Participantsterm·key were recruited through advertisement posters, social media and through the website forsoegsperson.dk (trial volunteers website). Recruitment posters were set up at train stations, workplaces, grocery shops and Universities in Frederiksberg, Valby, Sydhavnen and Nørrebro in Copenhagen, while on social media a word-of-mouth strategy was employed [36]. The inclusion criteria were: men and women between 18 and 55 years of age. Participantsterm·key were excluded if they were on a special diet (weight lost/gain, religious diet and vegetarian/ PLOS ONE | https://doi.org/10.1371/journal.pone.0176028 May 31, 2017date 3 / 16 Comparison of three nudgeterm·prm interventionsterm·key to promote vegetable consumptionterm·var vegan) or with the intention of changing smoking or exercising habits, and/or planning to be or already being pregnant. After recruitment, participantsterm·key were randomly allocated to either start with their control setting or experimental setting in a balanced manner.
Measures A web-based questionnaire was distributed among recruited participantsterm·key prior to their first eating session. The questionnaire contained questions about gender, age, educational achievement and self-reported height (cm) and weight (kg). The latter were used to calculate Body Mass Index (BMI) [37]. Educational achievement was grouped into one of three: Educational level I, which included people who had finished up to 3 years at University or less; Educational level II, included people who had finished a college bachelors degree (3–4 years) and lastly; Educational level III, which included people who have finished University at Masters level or Ph.D. (>4 years). The web-based questionnaire also included a short self-administered food frequency questionnaire, regarding weekly consumptionterm·key of salad, cooked vegetablesterm·key or veggies for snacking, with the following options: never/seldom, once per month, 2–3 times per month, 1–2 times per week, 3–4 times per week, 5–6 times per week, once a day, 2 or more times a day. It included also a measure of self-perceived fruit & vegetableterm·key intake, estimated on a perceived daily consumptionterm·key, calculated on the basis of given portion weighs (1 carrot 100g, 1 tomato 60g, ½ paprika 40 g and ¼ cucumber 75g) and self-perceived health (very good, good, tolerable, bad, very bad). The questionnaire was in Danish and constructed with pre-tested and validated parts from the Danish KRAM-survey, the Danish OPUS study and the American NHANES survey [38, 39].
Participantsterm·key furthermore filled out a questionnaire on appetite just before and just after each eating session. They scored their appetite on a 10-point Likert scale anchored from “I am not at all” (score: 1) to “I have never been more” (score: 10). Participantsterm·key gave two scores before eating: I) "how hungry are you?" (10 points scale, from not at all hungry = 1 to I have never been so hungry = 10) II) "how much do you think you can eat?" (10 points scale, from nothing at all = 1 to a lot = 10) and two after eating: I) can you eat more? (10 points scale, from nothing at all = 1 to a lot = 10) II) How satisfied are you (I am not at all satisfied = 1 to I can’t eat anything more = 10?
Food & serving equipment The experiment was conducted during a typical Danish lunch period from 11.30am to 1pm. There was no maximum set for the lunch duration of the study, but the observed average dining session lasted for approximately 30 minutes. Food was served at an Intelligent Buffet (iBuffet) in a FoodScape Lab that was described elsewhere [35]. This installation and its equipment allow exact measurement in grams and automatic register of each food item that the participantsterm·key took from the iBuffet, after they scan a wristband used for identification. Leftovers on the plate were weighed afterwards to determine exactly how much of each food item by the participantsterm·key. Foods were: white boiled and salted rice, chilli con carne (kidney beans, tomato, minced beef, onion, garlic, chilli, cumin, dark chocolate) and white salad (white cabbage, cauliflower, orange, rocket salad, pea sprouts, olive oil, honey and apple cider vinegar); which were served in 5L square bowls with sharp edges. Rice and chilli con carne were served using a 1 dl serving spoon. Red salad (grilled peppers, zucchini, watercress and vinegar) was also offered and served in three-litre square bowls. For the white and red salad the serving was with tongs. Salsa (tomato, onion, garlic, parsley and jalapenos) was served in a 0.8L white porcelain jar with 0.2 dl ladle. Foods were served ad libitum and continuously replenished. Participantsterm·key were allowed to help themselves at the buffet as many times as they wanted. Plate size was PLOS ONE | https://doi.org/10.1371/journal.pone.0176028 May 31, 2017date 4 / 16 Comparison of three nudgeterm·prm interventionsterm·key to promote vegetable consumptionterm·var 24cm in diameter, 25cl water glass and 1.4L transparent water carafe. All tableware was placed at on a separate cutlery table. All recipes are available as supplementary materials.
The iBuffet had eight serving surfaces, four in each side (See Fig 1) with a mix of cold and warm surfaces. Participantsterm·key had to check-in at each station where they took anything from the buffet.
Experimental design All tables, chairs, iBuffet and observation cameras had the same setup during data collection for all servings to ensure coherence and limit bias. The experimental settings were only changed as explicitly explained in the following paragraphs.
Control. The food environment was not manipulated. The buffet was organised for selfserving, and each dish component was presented in a large bowl (rice, chilli-con-carne and salads) and in a tall pot for the salsa. The two sides of the buffet were organised to avoid queuing at the buffet, but participantsterm·key were able to go around freely (Fig 1) and there was not a predetermined participantterm·key flow. Participantsterm·key had to serve themselves from the bowls, which were frequently replenished by the researchers. Participantsterm·key were allowed to refill their plates as many times as they wished.
Primingterm·prm. The food environment was tailored to accommodate a green ambience of plants, green servings bowls and herbs in the dining area as shown in Fig 2. Plants were large green leafed and herbs included: basil, thyme, watercress and oregano. The plants and herbs were placed intentionally, so study participantsterm·key would be visually exposed when interacting with the entire area I) at the cutlery section II) at the iBuffet III) at the dining tables IV) and as general de´cor in the eating area. Beside the visual effect, the fresh herbs provided a strong odour and participantsterm·key could freely use the provided fresh herbs (basil, parsley) as part of their meal. Defaultterm·key. The two salads were divided into pre-portioned individual transparent ‘takeaway’ bowls with the defaultterm·key amount of 150g of the white salad and 50g of the red salad. The bowls were placed at the iBuffet on serving area 4 and serving area 8 (see Fig 3). To replace Fig 1. Visual presentation of the control serving.
https://doi.org/10.1371/journal.pone.0176028.g001 PLOS ONE | https://doi.org/10.1371/journal.pone.0176028 May 31, 2017date 5 / 16 Comparison of three nudgeterm·prm interventionsterm·key to promote vegetable consumptionterm·var Fig 2. Visual presentation of the experimental setting: Primingterm·prm. A green ambiance was created using plants and herbs.
https://doi.org/10.1371/journal.pone.0176028.g002 where the white salad was on serving area 3, another salsa jar (containing the same salsa) was placed. Participantsterm·key were to take the individual bowls with them, compared to the control situation where they had to serve themselves on their plates. The participantsterm·key had the possibility of taking as many bowls of salad as they liked.
Perceived varietyterm. The two mixed salads served in the control setting were split into each of the ingredients in separate two-litre square bowls (Fig 4). All of the bowls were served on the same serving surface and the amounts served from each ingredient were accumulated into Fig 3. Visual presentation of the experimental setting: Defaultterm·key. The salad was pre-portioned into transparent bowls of 150g of the white salad and 50g of the red salad.
https://doi.org/10.1371/journal.pone.0176028.g003 PLOS ONE | https://doi.org/10.1371/journal.pone.0176028 May 31, 2017date 6 / 16 Comparison of three nudgeterm·prm interventionsterm·key to promote vegetable consumptionterm·var Fig 4. Visual presentation of the experiment setting: Perceived varietyterm. The salad was split up to its component, increasing the visual varietyterm·key and allowing participantsterm·key to compose their own salad.
https://doi.org/10.1371/journal.pone.0176028.g004 one measure of ‘White salad’ or ‘Red salad’. The overall selection of ingredients was exactly the same, but just divided in to more bowls, hence allowing each participantterm·key to compose their own salad.
Data analysis A complete case analysis was conducted where the difference in weighed intake (consumptionterm·key minus wastage) between the two eating sessions was considered as a proxy for how well the Nudgeterm·prm worked.
A mixed general linear model analysis was performed to explore changes within the three nudgesterm·prm for rice, chilli con carne and total vegetableterm·key intake. The intake was set as the dependent variable, time as the fixed effect, subjects as the random effect and measures for appetite were included as co-variates. Q-Q plots and histograms were used to check for normality.
Oneway ANOVAterm·stat with post hoc test was performed to explore any changes between the three nudgesterm·prm for: total intake, total chilli con carne intake, total rice intake and total vegetableterm·key intake. The intake difference in each food group was set as the dependent variable and the interventionterm·key was set as the fixed factor. Levene’s test of equality was non-significant, hence the assumption of equal variance was accepted.
The considered significant level was set at p(cid:20)0.05. Statistical analyses were performed with the software IBM SPSS Statistics version 22.0. Results are presented as means ± SDs, with corresponding p-valuesterm·stat for comparisons for both the analysis within and between nudgesterm·prm.
Ethics, consent and permissions Participantsterm·key were provided with information about VeggiEAT project (FP7–612326) and of the specific data collection that required them to attend two meals at the FoodScape Lab in convenient dates. All participantsterm·key provided a written and signed consent to participate in the study prior to its start, and were free to leave at any point. No personal or sensitive information PLOS ONE | https://doi.org/10.1371/journal.pone.0176028 May 31, 2017date 7 / 16 Comparison of three nudgeterm·prm interventionsterm·key to promote vegetable consumptionterm·var Table 1. Characteristics of participantsterm·key—Divided by choice architectureterm·dom approaches: Primingterm·prm, defaultterm·key and varietyterm·key.
Women [n (%)] Age (y) BMI (kg/m2) Education level [n (%)] III Self-reported daily intake of vegetablesterm·key [g] Primingterm·prm (n = 24) 19 (70) 27.3±6.63 23.4±5.6 9 (33) 10 (37) 8 (30) 320±170 Defaultterm·key (n = 33) 19 (58) 25.9±7.1 22.6±2.6 20 (61) 7 (21) 6 (18) 276±179 Varietyterm·key (n = 31) 21 (66) 26.3±6.5 22.5±2.8 21 (66) 10 (31) 1 (3) 343±338 ns2 ns 0.024* 1 Between primingterm·prm, defaultterm·key and varietyterm·key by using One-way ANOVAterm·stat, unless otherwise specified. 2 With the use of the chi-squareterm·stat test. 3 Mean ±term·stat SD, all such values. 4 With use of Fisher’s Exact test. * There was a significant difference between educational level I in primingterm·prm and educational level III in varietyterm·key (p<0.01).
https://doi.org/10.1371/journal.pone.0176028.t001 was obtained and data was/is stored according to Danish regulations, complying with EU Directive 95/46/EC on the protection of individuals with regard to the processing of personal data and on the free movement of such data. All required standards and criteria of The National Committee on Health Research Ethics under the Danish Ministry of Health have been met (http://www.dnvk.dk/) and complied with the Helsinki Declaration revised in 1983date. The Scientific Committee of the Capital Region of Denmark revised the study protocol, gave the consent to go ahead with the study, and cleared from further formal approval (Ref. H-2- 2013-FSP61).
Results Participantsterm·key Table 1 presents the characteristics of participantsterm·key divided by choice architectureterm·dom exposure: primingterm·prm, defaultterm·key and varietyterm·key. In total, 222 participantsterm·key were recruited. 110 were excluded, either because they did not meet the inclusion criteria or because they did not reply to further contact. After having completed one full test day, 24 of the 112 assigned participantsterm·key did not show up for second session (21% drop out). Participantterm·key characteristics are given in Table 2. No major differences were detected between the basic characteristics of the participating sample in each of the three experiments, except in the level of education, where a difference was seen between Educational level I in primingterm·prm and Educational level III in varietyterm·key, however, education Table 2. Appetite score from control day and interventionterm·key day. Measurement scale used: 10 point Likert scale anchored from “I am not at all” (score: 1) to “I have never been more” (score: 10).
How hungry are you?
How much do you think you can eat?
Can you eat more?
Are you satisfied?
Primingterm·prm Defaultterm·key Varietyterm·key Control Interventionterm·key Control Interventionterm·key 7.5±1.3 7.7±1.2 3.2±2.2 8.4± 1.3 6.3±1.7 6.7±1.1 3.3±1.8 8.1±1.6 <0.01 <0.01 6.8±1.5 7.5±1.6 2.5±1.1 8.5±1.0 6.8±1.5 7±1.6 2.1±0.7 8.5±1.5 0.08 0.04 Control Interventionterm·key 6.8±1.4 7.0±1.4 2.4±1.0 8.6±1.0 6.4±1.6 6.8±1.4 2.6±1.0 8.4±1.0 * Student t-testterm·stat analysis of appetite scores from control day compared to interventionterm·key day https://doi.org/10.1371/journal.pone.0176028.t002 PLOS ONE | https://doi.org/10.1371/journal.pone.0176028 May 31, 2017date 8 / 16 Comparison of three nudgeterm·prm interventionsterm·key to promote vegetable consumptionterm·var was not significant in the final modelling. The majority of the sample was composed of students (77%) where 67% of participantsterm·key were in the process of a higher education (>4y) and 18% of the whole sample already had a higher education. There was no difference in the distribution of students between the three experiments according to gender, age, BMI or vegetable consumptionterm·var.
Appetite score Table 2 shows the appetite score from control day and interventionterm·key day measured by a 10 points scale (from “I am not at all hungry” = 0 to “I have never been more hungry” = 10). Primingterm·prm was the only nudgeterm·prm group where a difference was found in the appetite score before eating. In the appetite score after eating, a difference was found in defaultterm·key for "Can you eat more?".
Food intake Table 3 shows the measured intake by in each of the experimental conditions and specified by food groups: vegetablesterm·key, rice and chili con carne. Primingterm·prm reduced the total food consumptionterm·key by a mean of 168.87g (p<0.01) hereof mostly by decreasing the intake of chilli con carne (p<0.01). The defaultterm·key interventionterm·key successfully increased the intake of vegetablesterm·key by a mean of 45g (p = 0.02). Providing a larger perceived varietyterm at the buffet reduced the total intake of food by a mean of 99g (p = 0.02), mostly due to a decreased intake of chilli con carne (p<0.01). No effect was found in the intake of rice. Further analysis shows that for perceived varietyterm gender had an effect on total intake, where men ate 201g less than the women in the group (p = 0.03). This is in contrast to primingterm·prm where there was a tendency for men to eat 200g more than the women (p = 0.07). For the vegetableterm·key intake primingterm·prm had an age-related effect with an 8g/year increase for the group (p = 0.01) (not tabulated).
Table 4 displays the observed energy intaketerm·var differences specified by food groups. The quantities of vegetableterm·key consumed were statistically different between interventionterm·key type (F = 4.398 & p = 0.015). Comparing the effect sizeterm·stat of the three nudgesterm·prm (presented as the difference between interventionterm·key and control in g) a significant difference was seen between primingterm·prm and defaultterm·key, with a mean difference of 201g in total intake and 81g for total vegetableterm·key intake (Tables 3 and 4).
Looking at the energy intaketerm·var, primingterm·prm decreased the total energy intaketerm·var by 169 kcal (p<0.01) and it decreased the energy from chilli con carne with 134 kcal (p<0.01). Defaultterm·key increased the energy intaketerm·var from vegetablesterm·key by 34 kcal (p<0.01) and offering perceived varietyterm decreased the total energy intaketerm·var by 124 kcal (p<0.01) and the chilli con carne with 96 kcal Table 3. Intake in grams (g) divided by food group: Vegetablesterm·key, rice and chilli con carne. Presented using means and SD. Test of significance using mixed modelling. Defaultterm·key n = 33, Primingterm·prm n = 24, Perceived varietyterm n = 31. “How hungry are you?” and “How much do you think you can eat?” were included as covariates in the analysis. “Can you eat more?” was further included as a covariate in the analysis for the Defaultterm·key nudgeterm·prm.
Nudgeterm·prm Treatment Vegetablesterm·key (grams) Rice (grams) Chilli con Carne (grams) Total intake (grams) Total intake (kcal) Mean Mean Mean Mean Mean Defaultterm·key Control 193.67 82.61 0.016 152.39 103.50 0.76 275.33 145.12 0.41 621.39 240.99 0.33 596.30 264.04 0.62 Interventionterm·key 238.88 141.49 156.97 102.90 257.79 148.78 653.64 229.82 615.13 258.45 Primingterm·prm Control 214.58 75.83 0.15 145.33 80.30 0.27 278.38 147.92 <0.01 638.29 257.94 <0.01 602.03 267.16 <0.01 Interventionterm·key 179.21 113.15 127.42 60.39 162.79 104.20 469.42 214.83 432.96 208.49 Perceived Varietyterm Control 267.77 85.68 0.56 167.97 88.24 0.063 296.61 116.62 <0.01 732.36 240.32 0.019 678.64 248.21 <0.01 Interventionterm·key 279.13 111.38 139.94 74.99 213.94 106.46 633.00 241.62 554.90 232.54 https://doi.org/10.1371/journal.pone.0176028.t003 PLOS ONE | https://doi.org/10.1371/journal.pone.0176028 May 31, 2017date 9 / 16 Comparison of three nudgeterm·prm interventionsterm·key to promote vegetable consumptionterm·var [ Passage withheld — in our stored copy of this paper these characters are set one per token and run backwards, which is what a PDF extractor returns for a table or a caption printed sideways. It is not what the page of the article looks like, so we do not reproduce it. The tags cut from this paper are unaffected. ] PLOS ONE | https://doi.org/10.1371/journal.pone.0176028 May 31, 2017date 10 / 16 Comparison of three nudgeterm·prm interventionsterm·key to promote vegetable consumptionterm·var (p<0.01) (not tabulated). No effect was found in energy intaketerm·var deriving from rice. The agerelated effect seen for vegetableterm·key intake in grams was also seen in the energy analyses with a rise of 6 kcal per year increment (p<0.01).
ANOVAterm·stat analysis with post hoc testing among the three nudgesterm·prm revealed significant differences in total energy intaketerm·var between primingterm·prm and defaultterm·key (Table 4). The 143 kcal difference in energy intaketerm·var between perceived varietyterm and defaultterm·key was significant (p<0.05), considering that the perceived varietyterm interventionterm·key successfully decreased the energy consumptionterm·key by 123 kcal (p<0.01). For energy intaketerm·var deriving specifically from vegetablesterm·key the defaultterm·key interventionterm·key increased the energy intaketerm·var deriving from vegetablesterm·key with 34 kcal (p<0.01). Additionally, the primingterm·prm interventionterm·key achieved a non-significant decrease in energy from vegetablesterm·key of 13 kcal, hence a net difference in energy intaketerm·var from vegetablesterm·key between the two experiments of 46 kcal (not tabulated).
Discussion The aim of this study was to test and compare three nudgesterm·prm in promoting vegetable consumptionterm·var among test persons in a food lab-based experiment. The three employed nudgesterm·prm overall promoted what seem to be a healthier meal composition. However, the separate analysis for each experiment suggests that various automatic decision processes might be more effective depending on what the targeted food choice behaviour is. Primingterm·prm attempted to prime vegetableterm·key choice, but significantly decreased the consumptionterm·key of chilli con carne. A potential explanation could be found in the intense green ambience functioning as a food-external stimulus indirectly influencing the participantterm·key’s cognitive reactions [21]. This could lead to an activation of a reflective decision makingterm, promoting a healthy-eater self-schema of food choice [40], hence not be an automatic process nudgeterm·prm, but a reflective. The visual presence of herbs and green ambience could remind the participantsterm·key about a green healthy lifestyle and therefore activate behaviour relevant to his/her self-image, which would result in them limiting their intake of the unhealthy food. However, this study did not measure the participantsterm·key’ level of health-schematic nor to what extent their food choice was reflective or not. Although selfreported appetite was significantly different in the interventionterm·key arm compared to its control in the primingterm·prm strategy, it did not distort the effect of the nudgeterm·prm. Increasing vegetable consumptionterm·var is a public health goal, which may contribute to dilute total energy intaketerm·var and further contribute to tackling the obesity issues in EU. Even small changes towards a dietary pattern rich in foods of plant origin have positive effects on health in the longterm [41, 42]. Another finding of this study is related to the role of achieving satisfaction with less calorie dense foods. The questions on appetite (Table 2) provided interesting insights, particularly the fact that the differences in satisfaction and appetite levels were mainly not significant within and between participantsterm·key, and if they achieved statistical significanceterm·stat, the difference was not meaningful.
The defaultterm·key interventionterm·key provided a fixed portion of salad as the status-quo or defaultterm·key bias, which successfully increased the intake of the salad, but did not decrease the total energy intaketerm·var. The defaultterm·key portion was approximately 25% higher than the average intake during the control setting and the observed increase in intake of vegetablesterm·key is in correspondence with what was anticipated according to the theory of plate clearing. Although the majority of research on defaultterm·key servings and plate clearing is conducted on actual plates [22, 23, 43, 44], our study showed similar effects although using a bowl-type container. The used bowl was transparent which might likewise have influenced the increase in intake as this has been illustrated to influence consumptionterm·key volume [45].
The experiment of increasing the visual perceived varietyterm decreased the self-served amount of meat and total energy, but the total intake of vegetablesterm·key was not increased as hypothesised.
PLOS ONE | https://doi.org/10.1371/journal.pone.0176028 May 31, 2017date 11 / 16 Comparison of three nudgeterm·prm interventionsterm·key to promote vegetable consumptionterm·var Despite the experiment promoted a healthier meal composition, it was not supportive of the aim of this study. A non-conclusive gender-specific sub analysis did show, that the nudgeterm·prm might have increased women’s intake whereas decreased men’s. It may be hypothesised for future studies, that the observed decrease found in men, may be due to the “cognitive burden” of choices between the ingredients in addition to the multitasking by handling the plate and checking-in at the iBuffet. In psychology studies it is shown that men are less efficient than women when multitasking and/or being in a choice situation [28, 46–48]. The causal link of these findings is based on the assumption, that the separation of fruit and vegetablesterm·key components did in fact increase the visual varietyterm·key. The actual varietyterm·key stayed the same in control and interventionsterm·key in order to be consistent with the definition of a nudgeterm·prm. Asking the participantsterm·key of their perception of the vegetableterm·key varietyterm·key would require them to embark in reflective decision-making thus losing the intended idea of nudging their food choice.
From a conceptual level an important concern to address is that defaultterm·key and varietyterm·key conditions also differ in the easiness to reach the vegetablesterm·key. More effortful cognitive tasks (choices) might alter the individual’s self-regulation capacity [49]. In the defaultterm·key condition participantsterm·key had to simply grasp a pre-packed box while in the varietyterm·key condition people had to self-serve and select themselves the amounts of vegetablesterm·key they wanted, hence requiring more effort from their part. In a similar study [17] modification in the order of the buffet (vegetablesterm·key first and hence easier to reach) and self-serving from different plates increased vegetable consumptionterm·var while decreasing total energy intaketerm·var without significantly altering the total quantity consumed. The present study was as close as possible to a real buffet setting where consumers are used to serve themselves, and shows a case where a promising successful nudgeterm·prm is one that made the healthier and desirable choices easier.
Limitations and strengths These experiments were held in, for some, a new and unfamiliar setting (the Foodscape Lab). During an unfamiliar eating session, the number of people may have had an impact on the generally low food consumptionterm·key [50]. Likewise, participantsterm·key knowing it was an experimental food setting (although not obvious at first) might have made them conscious about being measured and evaluated, leading to a decrease in intake [51]. This study only looked at a singular meal (lunch) and thus cannot conclude much about a whole day dietary intake; however, previous research suggests that participantsterm·key did not compensate for their energy intaketerm·var during lunch later in the day [52], hence indicating the potential for promoting health and well being through a singular lunch serving. This is underlined by the fact that an increasing number of meals are consumed outside home [53] and self-service buffet eating is associated with excess caloric intake [54, 55]. This current study had an over-representation of women compared to men, misrepresenting the national gender distribution of Denmark. The study had hoped to have a more representative sample, but it was not a requirement of the study’s design nor a theoretical basis for the alternative hypothesis’ expected effectiveness. Similar trends are seen in other Danish nutritional studies, with a general overrepresentation of women [38, 56, 57]. One reason for this is that women in general are more concerned with health and nutrition topics than men and therefore voluntarily sign up for studies of this nature [58, 59].
The major strength of this study was the iBuffet that with hidden weights measured exact amounts of food taken combined with the precise amount of food waste registered by the individual wristbands worn by the participantsterm·key. This ensured a measure of the dietary intake and not retrospective intake with little or no inconvenience for the participating subject [60]. To the knowledge of the authors this is one of the few studies that has been performed in a setting PLOS ONE | https://doi.org/10.1371/journal.pone.0176028 May 31, 2017date 12 / 16 Comparison of three nudgeterm·prm interventionsterm·key to promote vegetable consumptionterm·var that allowed complete accurate measurement of dietary intake, limiting bias from self-reporting [61, 62] and undetected plate waste.
Practical implications & Conclusion Considerable insight has been gained with regard to the complexity of the automaticities that influences food choice in a self-service setting. Nudging is being used as a theoretical and philosophical umbrella expression to understand these mechanisms. Mechanisms that are highly influential for shaping consumer behaviours in promoting healthier eating. Using nudging is a promising tool to shift the meal composition to a healthier one. This study has illustrated that serving vegetablesterm·key in a defaultterm·key portion size is effective if the aim is to promote vegetableterm·key intake. If the aim is to decrease quantities of food eaten, this study suggest that primingterm·prm vegetablesterm·key and/or increasing the visual varietyterm·key of vegetablesterm·key could be effective.
With the global increase in out-of-home eating and the casual link with buffet style eating and lifestyle diseases, the public health community ought to address the issue of obesity in this context. Existing strategies to influence choices and behaviour have been successful to a certain extent, but not all of the way. This study illustrates that behaviour can be influenced and that healthy eating can be promoted solely based on the automatic decision-making system. Future research and public policy need to promote behaviours, acknowledging the reflective rational part of the consumer, as well as the habitual [63].
Supporting information S1 Dataset. Datasets_2017date_Article.sav.zip (ZIP) Acknowledgments The authors would like to acknowledge Pernille Møller Andersen for partaking in the data collection and Associated Professor Helene Christine Reinbach for valuable input during the analysis. This research was conducted as part of RF’s and PMA’s master thesis in Human Nutrition at University of Copenhagen.
Author Contributions Conceptualization: RF AO FPC.
Data curation: FPC.
Formal analysis: RF LRS AO FPC.
Funding acquisition: HH FPC.
Investigation: RF LRS.
Methodology: RF AO FPC.
Project administration: FPC.
Resources: AO FPC.
Supervision: AO FPC.
Validation: FPC LRS.
Visualization: RF LRS FPC.
PLOS ONE | https://doi.org/10.1371/journal.pone.0176028 May 31, 2017date 13 / 16 Comparison of three nudgeterm·prm interventionsterm·key to promote vegetable consumptionterm·var Writing – original draft: RF LRS.
Writing – review & editing: RF LRS AO KMA LS CD HH LD EM AG FPC.
NOT THE PAPER'S TEXT. Added by this project on 2026-08-25: the tables below were restored from the held scrape because the staged text did not contain them, and document order is therefore not the paper's. [TABLES 1-3 RESTORED 2026-08-25 by terminal-a from fulltext.html - the staged body dropped them and THREE LIVE TAGS ON THIS RECORD ARE ALREADY CUT FROM THEM (content_tags.statistics[6], [8] and [10]), so the corpus was asserting numbers it could not verify. 14% numeric: this is a table dump and it reads like one. It is restored for GROUNDING, not for reading - which is a different reason from the one that governed moher2009-prisma and kozlowski2019, where the table was the paper's own instrument. APPENDED AT THE END so no existing anchor offset moves.] Table 1. Character istics of participantsterm·key —Divide d by choice architectu re approa ches: Primingterm·prm, defaultterm·key and varietyterm·key. Primingterm·prm (n = 24) Defaultterm·key (n = 33) Varietyterm·key (n = 31) p 1 Women [n (%)] 19 (70) 19 (58) 21 (66) ns 2 Age (y) 27.3±6.6 3 25.9±7.1 26.3±6.5 ns BMI (kg/m 2 ) 23.4±5.6 22.6±2.6 22.5±2.8 ns Educati on level [n (%)] I 9 (33) 20 (61) 21 (66) 0.02 4 * II 10 (37) 7 (21) 10 (31) III 8 (30) 6 (18) 1 (3) Self-reporte d daily intake of vegetab les [g] 320±170 276±179 343±338 ns 1 Between primingterm·prm, defaultterm·key and varietyterm·key by using One-way ANOVAterm·stat, unless otherwise specifie d. 2 With the use of the chi-squareterm·stat test. 3 Mean ±term·stat SD, all such values. 4 With use of Fisher’s Exact test. * There was a signific ant difference between educationa l level I in primingterm·prm and education al level III in varietyterm·key (p<0.01). https://do i.org/10.1371/j ournal.pone .0176028.t001 Table 2. Appetite score from control day and interventio n day. Measurem ent scale used: 10 point Likert scale anchore d from “I am not at all” (score: 1) to “I have never been more” (score: 10). Primingterm·prm Defaultterm·key Varietyterm·key Control Interven tion p* Contro l Interventionterm·key p* Control Interven tion p* How hungry are you? 7.5±1.3 6.3±1.7 <0.01 6.8±1.5 6.8±1.5 ns 6.8±1.4 6.4±1.6 ns How much do you think you can eat? 7.7±1.2 6.7±1.1 <0.01 7.5±1.6 7±1.6 0.08 7.0±1.4 6.8±1.4 ns Can you eat more? 3.2±2.2 3.3±1.8 ns 2.5±1.1 2.1±0.7 0.04 2.4±1.0 2.6±1.0 ns Are you satisfied? 8.4± 1.3 8.1±1.6 ns 8.5±1.0 8.5±1.5 ns 8.6±1.0 8.4±1.0 ns * Student t-testterm·stat analysis of appetite scores from control day compared to interventi on day https://do i.org/10.1371/j ournal.pone .0176028.t002 Comparison of three nudgeterm·prm interventionsterm·key to promote vegetableterm·key consump tion PLOS ONE | https://doi.or g/10.137 1/journal.po ne.01760 28 May 31, 2017date 8 / 16 p. 9 was not significant in the final modelling. The majority of the sample was composed of stu- dents (77%) where 67% of participantsterm·key were in the process of a higher education (>4y) and 18% of the whole sample already had a higher education. There was no difference in the distri- bution of students between the three experiments according to gender, age, BMI or vegetable consumptionterm·var. Appetite score Table 2 shows the appetite score from control day and interventionterm·key day measured by a 10 points scale (from “I am not at all hungry” = 0 to “I have never been more hungry” = 10). Primingterm·prm was the only nudgeterm·prm group where a difference was found in the appetite score before eating. In the appetite score after eating, a difference was found in defaultterm·key for "Can you eat more?". Food intake Table 3 shows the measured intake by in each of the experimental conditions and specified by food groups: vegetablesterm·key, rice and chili con carne. Primingterm·prm reduced the total food consumptionterm·key by a mean of 168.87g (p<0.01) hereof mostly by decreasing the intake of chilli con carne (p<0.01). The defaultterm·key interventionterm·key successfully increased the intake of vegetablesterm·key by a mean of 45g (p = 0.02). Providing a larger perceived varietyterm at the buffet reduced the total intake of food by a mean of 99g (p = 0.02), mostly due to a decreased intake of chilli con carne (p<0.01). No effect was found in the intake of rice. Further analysis shows that for perceived varietyterm gen- der had an effect on total intake, where men ate 201g less than the women in the group (p = 0.03). This is in contrast to primingterm·prm where there was a tendency for men to eat 200g more than the women (p = 0.07). For the vegetableterm·key intake primingterm·prm had an age-related effect with an 8g/year increase for the group (p = 0.01) (not tabulated). Table 4 displays the observed energy intaketerm·var differences specified by food groups. The quantities of vegetableterm·key consumed were statistically different between interventionterm·key type (F = 4.398 & p = 0.015). Comparing the effect sizeterm·stat of the three nudgesterm·prm (presented as the differ- ence between interventionterm·key and control in g) a significant difference was seen between prim- ingterm·prm and defaultterm·key, with a mean difference of 201g in total intake and 81g for total vegetableterm·key intake (Tables 3 and 4). Looking at the energy intaketerm·var, primingterm·prm decreased the total energy intaketerm·var by 169 kcal (p<0.01) and it decreased the energy from chilli con carne with 134 kcal (p<0.01). Defaultterm·key increased the energy intaketerm·var from vegetablesterm·key by 34 kcal (p<0.01) and offering perceived varietyterm decreased the total energy intaketerm·var by 124 kcal (p<0.01) and the chilli con carne with 96 kcal Table 3. Intake in grams (g) divided by food group: Veget ables, rice and chilli con carne. Presented using means and SD. Test of significance using mixed modelling. Defaultterm·key n = 33, Primingterm·prm n = 24, Perceive d varietyterm·key n = 31. “How hungry are you?” and “How much do you think you can eat?” were included as covariates in the analysis. “Can you eat more?” was further included as a covariate in the analysis for the Defaultterm·key nudgeterm·prm. Nudgeterm·prm Treatment Vegetablesterm·key (grams) Rice (grams) Chilli con Carne (grams) Total intake (grams) Total intake (kcal) Mean SD p Mean SD p Mean SD p Mean SD p Mean SD p Defaultterm·key Control 193.67 82.61 0.016 152.39 103.50 0.76 275.33 145.12 0.41 621.39 240.99 0.33 596.30 264.04 0.62 Interventi on 238.88 141.49 156.97 102.90 257.79 148.78 653.64 229.82 615.13 258.45 Primingterm·prm Control 214.58 75.83 0.15 145.33 80.30 0.27 278.38 147.92 <0.01 638.29 257.94 <0.01 602.03 267.16 <0.01 Interventi on 179.21 113.15 127.42 60.39 162.79 104.20 469.42 214.83 432.96 208.49 Perceived Varietyterm Control 267.77 85.68 0.56 167.97 88.24 0.063 296.61 116.62 <0.01 732.36 240.32 0.019 678.64 248.21 <0.01 Interventi on 279.13 111.38 139.94 74.99 213.94 106.46 633.00 241.62 554.90 232.54 https://do i.org/10.1371/j ournal.pone .0176028.t003 Comparison of three nudgeterm·prm interventionsterm·key to promote vegetableterm·key consump tion PLOS ONE | https://doi.or g/10.137 1/journal.po ne.01760 28 May 31, 2017date 9 / 16 p. 10 Table 4. Observe d difference in energy intaketerm·var (gram) session compared, divided by food group: Vegetablesterm·key, rice and chilli con carne. Statistical test for significanc e of the three interven tions compared to each other using one-way ANOVAterm·stat with a post hoc test. Nudgeterm·prm Nudgeterm·prm comparis on Vegetab les 1 (grams) Rice 2 (grams) Chilli con Carne 3 (grams) Total intake 4 (grams) Total intake 5 (kcal) Mean differenc e SE p Mean difference SE p Mean differenc e SE p Mean difference SE p Mean differen ce SE p Defaultterm·key Primingterm·prm 80.59 29.29 0.027 22.49 22.53 0.61 98.04 32.95 0.015 201.12 58.87 <0.01 187.89 60.62 0.011 Perceived Varietyterm 33.86 27.31 0.47 32.61 21.00 0.31 65.13 30.72 0.11 131.60 54.89 0.062 142.57 56.52 0.047 Perceived Varietyterm Primingterm·prm 46.73 29.69 0.29 -10.12 22.83 0.91 32.91 33.40 0.62 69.52 59.67 0.51 45.32 61.44 0.76 1 Observed power was 0.675.