Practical behavioural solutions to COVID-19: Changing the role of behavioural science in crises.

For a very long time in the COVID-19 crisis, behavioural change leading to physical distancing behaviour was the only tool at our disposal to mitigate virus spread. In this large-scale naturalistic experimental study we show how we can use behavioural science to find ways to promote the desired physical distancing behaviour. During seven days in a supermarket we implemented different behavioural interventions: (i) rewarding customers for keeping distance; (i) providing signage to guide customers; and (iii) altering shopping cart regulations. We asked customers to wear a tag that measured distances to other tags using ultra-wide band at 1Hz. In total N = 4, 232 customers participated in the study. We compared the number of contacts (< 1.5 m, corresponding to Dutch regulations) between customers using state-of-the-art contact network analyses. We found that rewarding customers and providing signage increased physical distancing, whereas shopping cart regulations did not impact physical distancing. Rewarding customers moreover reduced the duration of remaining contacts between customers. These results demonstrate the feasibility to conduct large-scale behavioural experiments that can provide guidelines for policy. While the COVID-19 crisis unequivocally demonstrates the importance of behaviour and behavioural change, behaviour is integral to many crises, like the trading of mortgages in the financial crisis or the consuming of goods in the climate crisis. We argue that by acknowledging the role of behaviour in crises, and redefining this role in terms of the desired behaviour and necessary behavioural change, behavioural science can open up new solutions to crises and inform policy. We believe that we should start taking advantage of these opportunities.

Introduction

Behaviour is often mentioned in relation to crises: the trading of mortgages that resulted in the financial crisis in 2008, the shaking of hands in the latest COVID-19 crisis, or the consuming of goods in the climate crisis. In most of these cases, behaviour is primarily considered as a factor that causes or sustains a crisis, but when it comes to solving a crisis, behaviour is less likely to be considered. For solutions to a crisis, we often turn to experts from the respective discipline—in the financial crisis we turn to economists [1], in the COVID-19 crisis we turn to epidemiologists [2], and in the climate crisis we turn to climatologists [3]. We seem to overlook the power of behavioural change, and the field of behavioural science, in our battle against these crises [4]. We should consider people’s behaviour not only as causes and sustaining factors, but also turn to the solutions that behavioural change can provide.

The most recent COVID-19 crisis clearly demonstrates the role of behaviour and behavioural change [5, 6]. The virus transmitted through behavioural contacts and it was pivotal that we found ways to alter people’s behaviour and promote physical distancing and hygiene to mitigate virus spread [7]. This need for behavioural change resulted in numerous studies into factors that determine behaviour [8-10], and worldwide regulations such as lockdowns, school closures, and travel restrictions, all directed to reduce the number of behavioural contacts. Interestingly, while these studies and regulations focused on behavioural change, their effectiveness was assessed in terms of psychological constructs such as intentions and motivations and epidemiological parameters like the reproduction number, number of cases and deaths [11]. Behavioural criteria to express the effectiveness of these regulations in directly observed behaviour (e.g., to what extent did people keep their distance) were largely unavailable [12, 13] and there was little information on whether these interventions did also successfully accomplish the desired behavioural changes. The lack of direct assessments is especially problematic given the discrepancy between people’s intentions and motivations and their actual behaviour, also known as the intention-behaviour gap [14, 15].

We have seen similar patterns in other crises. Take for example the global financial crisis of 2008, where behaviour clearly played a role, e.g., through the creation of hedge funds and taking excessive risks [16]. In finding a solution, however, governments were inclined to turn to economical solutions and e.g., lowered the interest rates [17]. At the same time, there are important questions that need to be asked, such as why did people engage in this excessive risk taking behaviour, and how could other behaviour be stimulated? [18] Such questions are crucial to understand and prevent other financial crises, and are intrinsically of behavioural nature. In the climate crisis too, behaviour is considered as cause while predominantly technological solutions are being proposed. However, it has been increasingly vocalized that the only way to combat climate change is through behavioural change [19, 20]. These examples underscore the importance of behaviour and behavioural change, not only as causing and sustaining factors of crises, but precisely also as solutions to crises.

Behavioural science offers many models into the determinants of behaviour, and thereby offers leveraging points on how to instantiate behavioural change. The widely used Capability, Opportunity, Motivation, Behaviour (COM-B) model, for example, posits that a particular behaviour occurs when someone has the capability (i.e., psychological and physical), opportunity (i.e., contextual factors that facilitate the behaviour), and motivation (i.e., processes that energize and direct behaviour) to enact the behaviour [21]. By outlining the factors that influence particular behaviours to occur, the model also provides the opportunity to identify different points of engagement to bring about the desired behavioural change.

We advocate that behavioural science plays a prominent role in finding solutions to crises, together with scientists from other fields [22]. In the context of the COVID-19 crisis, we need scientists from many different disciplines [23]: virologists and micro-biologists to understand how the virus works [24], epidemiologists on how the virus spreads [25], and medical scientists on how to treat the virus [26]. But we also need behavioural scientists to understand how we can successfully change our behaviour and combat the virus spread. Behavioural science provides a way to link science and society, with behavioural change running like a thread out of crises.

In the current paper we demonstrate how we can put this idea into practice, and use behavioural science to provide concrete answers on how to promote the desired behaviour of physical distancing during the COVID-19 crisis, similar to our previous work in an art fair [27]. In this research we focused on physical distancing behaviour in public spaces, specifically in a supermarket. We identified different ways to stimulate physical distancing by considering psychological processes, crowd management, and practical solutions (see Methods for details): (i) rewarding customers; (ii) providing signage to guide customers; and (iii) altering shopping cart regulations. Importantly, we directly measured the desired behaviour using wearable sensors that recorded the distance between customers. We subsequently systematically evaluated the effectiveness of these interventions on physical distancing in an experimental design where we varied the proposed interventions across seven days. The goal of this research was to use behavioural science to help find concrete solutions to the COVID-19 crisis by clearly defining the desired behaviour (physical distancing), implementing behavioural interventions to stimulate this behaviour, and using a direct and objective measurement of this behaviour to assess its effectiveness.

Materials and methods

Participants and design

Participants in our naturalistic study were customers of the supermarket PLUS André and Joyce van Reijen in Veldhoven, the Netherlands. Veldhoven is a town of approximately 45,000 inhabitants in the southern Netherlands, located in the Metropoolregio Eindhoven. All customers older than 16 years could participate in the study, and there were no other in- or exclusion criteria.

The experiment took place during seven days in a supermarket. We varied three interventions: reward [28], signage [29], and adjusting the shopping cart regulations. These interventions were chosen for varying reasons. First, rewarding people for displaying the desired behaviour is well-established to be effective in promoting that desired behaviour [30], and also advised during the COVID-19 pandemic in particular [31]. In our study, participants received the reward upon handing in their tag (see Procedure) and consisted of cookies on Saturday March 24th and chocolate on Friday March 26th. Second, signage is commonly used to change behaviour [29] and often used in traffic to, for example, avoid collisions [32]. We aimed to investigate whether clear signage would facilitate pedestrian flows and thereby physical distancing. We included arrows signaling unidirectional walking directions in part of the supermarket and footprints in the queue for the register (see Materials below). Third, we changed the shopping cart regulations from mandatory (which was standard in the Netherlands at that time) to optional. The mandatory shopping carts were implemented in the Netherlands in March 2020 [33] to (1) keep track of the number of participants in the supermarket as national regulations allowed a maximum of 1 customer per 10 m2, and (2) as the shopping carts were thought to facilitate physical distancing. At the same time it could be argued that the mandatory shopping carts take up a lot of space within the supermarket and, as such, hinder the opportunity for physical distancing.

We varied the interventions across days, resulting in a unique set of interventions for each day, see Table 1. Note that there is some redundancy in the experimental conditions we implemented: for example, the effect of reward could be assessed by comparing day four to day three or two, but also by comparing day 7 to day 6. We did so to minimise the influence of factors such as day and time of day. Both the day and time of day are likely to affect the type of customers and crowdedness in the supermarket. Customers during workdays may differ from customers during the weekend, and customers during the morning may differ from customers after work-hours. Similarly, the crowdedness in the supermarket is likely to affect the number of contacts made. Since these factors are difficult to control in a naturalistic study, we implemented some redundancy in the design so that we could select time intervals post-hoc that would minimise differences in these factors for an optimal comparison across conditions.

Table 1

Experimental design.

Day		Intervention				Comparison
		Reward	Signage		Shopping cart
			Arrows	Footprints
1	17 March 2021					shopping cart
2	18 March 2021				✓
3	19 March 2021				✓	shopping cart, signage
4	20 March 2021	✓			✓
5	24 March 2021			✓
6	25 March 2021		✓	✓	✓	signage, reward
7	26 March 2021	✓	✓	✓	✓	reward

Materials

Physical distance

Participants wore a SafeTag developed by KINEXON (https://kinexon.com/technology/safetag/). SafeTags are wearable tags that measure the distance to other SafeTags at a frequency of one Hz using ultra-wideband (UWB) technology with an accuracy up to 10 cm. A tag automatically turns on and starts measuring when taken out of its charging unit, and turns back off when placed back into the unit. The SafeTags measure the distance to all other tags that are active, and it is technically not possible to link multiple SafeTags when they belong to members of the same group, so to exclude contacts made between group members. Each tag has a unique tag id and locally stores measured distances until the data is read out on a laptop running the management software. The tags were solely used to measure physical distance, and did not provide any form of feedback.

Shopping experience

We asked participants to rate their shopping experience on an iPad. Participants were asked to rate three questions on a five-point scale from satisfied to dissatisfied (indicated by five emoticons): (1) How did you experience the corona regulations in the supermarket? (‘regulations’); (2) How pleasant was it to do groceries like this? (‘pleasantness’); (3) Did you feel you were helped to keep a distance? (‘help’).

Camera and traffic light

National regulations specified a maximum number of customers inside the supermarket (i.e., one customer per 10 m2). The mandatory shopping cart allowed to keep track of the number of participants inside. In order to be able to relax the mandatory shopping cart but still adhere to national regulations, we installed a camera at the entrance of the supermarket that counted all incoming customers. The camera was linked to a traffic light that indicated to incoming customers how crowded the supermarket was (green: few people inside; orange: quite busy, be aware of your distance; red: full capacity reached, wait until someone exits). The camera registered all incoming and outgoing customers which was saved to a database and could be retrieved per hour.

Signage

We had two types of signage in the supermarket: waiting signage in the queue for the register and arrows depicting unidirectional walking directions in part of the supermarket where the aisles were too narrow to keep 1.5 m distance. The waiting signage consisted of round stickers with footprints on them and were placed 1.5 m from one another. The arrows were also placed 1.5 m from one another, to both signal the unidirectional walking directions as well as the appropriate physical distance.

Procedure

In the week prior to our experiment (week 10, 2021), flyers were passively distributed alongside the ad brochure in the supermarket, to inform customers of the upcoming experiment on the effectiveness of physical distancing regulations. The flyer informed participants when the study would take place, and stressed its voluntary character. See S1 Appendix for the flyer.

The experiment itself took place in week 11 from Wednesday March 17th until Saturday March 20th and week 12 from Wednesday March 24th until Friday March 26th 2021. Each day we handed out sensors between 12:00 and 17:00. The supermarket communicated via posters that every day between 14:00 and 15:00 was intended for older and vulnerable people to do their groceries, but this was not actively enforced.

Outside of the supermarket, a team member informed customers that a study would take place inside the supermarket, for which they could participate on a voluntary basis. On days that shopping carts were not mandatory, posters were present that informed customers that they could enter without taking a shopping cart. Inside, customers were asked to participate in the experiment. In case customers had additional questions about the study, a team member took the customer aside to avoid queuing and explained the study in more detail and provided the original information flyer. If customers agreed to participate, we handed them a SafeTag to wear on a lanyard around their neck, and registered their implicit informed consent. If customers of the same household participated, they each received a SafeTag. According to regulations, household members did not need to keep a distance, but we were unable to register their group membership upon handing out the tag to avoid congestion. We did exclude contacts between group members when processing the data and only kept contacts with individuals outside of the group (see Pre-processing).

In conditions in which a reward was handed out, we also informed participants that they would receive a reward for their effort to keep their distance upon handing in their SafeTag. Participants could then proceed to do their groceries like they would otherwise do. Thus, as soon as the participants entered the supermarket they did not encounter any study personnel. After paying at the register, participants could rate their satisfaction with the supermarket visit. A desk to hand back the tags was located at the exit of the supermarket. Both the tags and lanyards were thoroughly cleaned before they were handed out again to other customers.

The ethics review board of the University of Amsterdam approved the study and implicit informed consent because of the voluntary and anonymous nature of the study (2021-PML-13247).

Analysis

Pre-processing

The raw data from the SafeTags contains for each assessment (frequency of 1 Hz) a time stamp, the tag id of the reporting and opposing tag, and the distance between them in centimetres. Since tags were handed out multiple times during one day, we first determined the start and end time of each participant wearing the tag to construct unique participant id’s. We then checked if participants entered the supermarket as a household by investigating three criteria: at least 10 contacts within 80 cm, being within 1.5 m of each other for at least 25% of their visit duration, and exiting the supermarket at most 60 seconds after each other. We assigned participants as belonging to the same group if they met at least two of these criteria and removed all contacts between the respective group members. Note that any contacts to other participants present in the supermarket were retained. We considered two participants to be in contact with one another if (a) they did not belong to the same group and (b) were within 1.5 m from one another, in accordance with the physical distance regulations at the time in the Netherlands. To compute the contact duration between two customers, we summed the duration of all registered contacts between two participants.

Descriptive analyses

We first performed a simple linear regression to assess the effect of crowdedness on the median number of unique contacts per hour. As was found in other studies [34] it is conceivable that the busier it is, the more difficult it becomes to keep a physical distance. If this is the case, then we should control for this effect by comparing conditions at times that were similar in crowdedness.

To evaluate the effectiveness of each intervention in isolation, we select two conditions that differ only in regard to whether the behavioural intervention of interest is implemented and that are similar in crowdedness and time of day. This way we can isolate the effect of the intervention while keeping other factors constant. For each intervention, we tested the difference in the number of unique contacts between participants with a Bayesian logistic regression model (see Contact networks), and in contact duration with a Mann-Whitney U test. In addition, we tested differences in ratings of shopping experience (i.e., regulations, pleasantness, and help) with two sample t-tests using all available data from that day.

Contact networks

To compute differences in number of contacts between participants, we analyzed the contact networks with a Bayesian logistic regression model, developed for our previous study, called the b2 model [27]. The b2 model is a reduced version of the multilevel p2 model [35], for undirected (i.e., a contact is always shared between two people) and unweighted (i.e., a contact is binary and duration is not taken into account) networks. The model omits reciprocity parameters, dyadic predictors, and random effects at the network level, and contains identical random sender and receiver effects. We modeled differences in the number of unique contacts between two contact networks with actor-level dummy variables. Estimating the b2 model was done in a similar manner as the j2 model [36, 37], using Markov Chain Monte Carlo simulation and similar prior distributions. For all comparisons, we report the posterior means as point estimates for the odds ratios accompanied by the corresponding 95% credible interval. The credible interval describes the range where the true odds ratio lies with 95% certainty.

All analyses were performed in R (version: 4.1.1) and the dyads (1.1.4) package [38] was used to compare the contact networks.

Results

Visitors

A total of N = 4, 232 customers participated in our study. The number of customers inside the supermarket varied over time, and we first investigated whether there was an effect between crowdedness and the number of contacts per hour. Fig 1 shows that there exists a relation between the number of customers inside as registered by the camera at the entrance, and the median number of contacts per hour (F(1, 28) = 10.89, p = .003, R2 = .28). To minimize the influence of sample size and customer type on the effects of interventions, we selected comparable hours in terms of the time of day, number of customers inside the supermarket, and number of customers participating in our study (compliance), see Table 2. This resulted in a selection of six time slots distributed over four days.

Fig 1

Crowdedness and contacts.

Relationship between the number of participants and the median number of unique contacts in one-hour time windows across six days.

Table 2

Descriptives.

Condition		Day		n reg	npart(%)	Number of contacts				Contact duration				Experience
						Range	M ± SD	Median	IQR	Range	M ± SD	Median	IQR	n	Regulation	Pleasantness	Help
reward	no	6	16–17h	275	188 (68%)	0–24	8.4 ± 5.3	8	4–12	2–22	7.0 ± 4.0	5.9	4.0–8.5	240	4.3 ± 0.8	4.0 ± 0.9	4.1 ± 0.9
	yes	7	15–16h	316	200 (63%)	0–26	6.7 ± 5.1	6	3–9	2–28	5.6 ± 4.1	4.2	3.2–6.6	238	4.2 ± 0.8	4.0 ± 0.9	4.1 ± 0.8
signage	no	3	15–16h	237	170 (72%)	0–33	9.5 ± 6.3	9	4–13	2–44	6.8 ± 4.7	5.6	4.3–8.0	194	4.1 ± 0.9	3.7 ± 1.0	3.9 ± 0.9
	yes	6	15–16h	222	152 (68%)	0–27	6.1 ± 4.5	6	3–8	2–36	7.0 ± 5.9	4.8	3.6–8.0	240	4.3 ± 0.8	4.0 ± 0.9	4.1 ± 0.9
shopping carts	mandatory	1	15–16h	204	147 (72%)	0–27	7.7 ± 5.7	6	4–10	2–48	7.4 ± 5.9	6.0	4.0–8.7	324	4.2 ± 0.8	3.8 ± 1.0	3.9 ± 1.0
	optional	3	15–16h	237	170 (72%)	0–33	9.5 ± 6.3	9	4–13	2–44	6.8 ± 4.7	5.6	4.3–8.0	194	4.1 ± 0.9	3.7 ± 1.0	3.9 ± 0.9

Table notes n indicates the number of incoming customers, as registered by the camera at the entrance, and n indicates the number of customers who agreed to wear a tag and participate in the study.

Contact network

To evaluate the effectiveness of behavioural interventions on physical distancing we followed the experimental framework proposed by Blanken et al. (2021) [27] in which participants and their contacts are represented in a contact network. This representation allows to take the network structure into account when comparing two conditions [12].

Fig 2 shows the contact network of the n = 624 participants included on the first day of our study. Each participant is represented as a node, and whenever two participants came within 1.5 m of each other, they are connected by a link. The highlighted nodes indicate n = 147 participants present between 15:00 and 16:00 that were included in the analysis.

Fig 2

Contact network in the supermarket.

The contact network of n = 624 participants on March 17th is shown on the left. All participants are represented as nodes, and two participants are linked when they came within 1.5 m. The links are weighted by their contact duration. The highlighted nodes indicate the participants present between 15:00 and 16:00, the time slot we selected for the comparison. A detailed view of the contact network of these included participants is shown on the right.

Interventions

Reward

To examine the psychological effect of rewarding participants on their physical distancing behaviour, we compared the contact network on day 6 (no reward) with the contact network of day 7 (reward). As can be seen in Table 2 and Fig 3, participants who received a reward had a median of 6 unique contacts, whereas without a reward participants had a median of 8 unique contacts. Detailed analysis taking the network structure of the data into account showed that the probability of forming contacts was lower when participants received a reward (OR = 0.83, 95% Credible Interval (CI) [0.71, 0.97]). The CI indicates some uncertainty about the size of the effect, but rewarding participants for their effort to keep a distance improved physical distancing. In addition, participants who received a reward had slightly shorter contacts (median of 4.2 seconds) than participants who did not receive a reward (median of 5.9 seconds; U = 21606, p < 0.001). Finally, participants’ ratings of the regulations, pleasantness, and help did not differ between the two conditions (all p > 0.2).

Fig 3

Contacts per experimental condition.

The number of unique contacts (<1.5 m) in each of the six conditions. The solid line represents the median, and the two dashed lines the 25th and 75th percentiles, such that 50% of the observations fall within the two dashed lines.

To examine the effect of signage on physical distancing, we compared the contact network on day 3 without signage, to the contact network on day 6, when footprints and arrows were provided. The median number of unique contacts of participants was 9 without signage, and 6 when signage was provided. The probability of participants forming contacts was lower when providing signage compared to the situation where no signage was provided (OR = 0.85, 95% CI [0.71, 1.00]). The CI around this estimate shows that there is some uncertainty about the size of the effect and the upper limit is equal to 1, but indicates that signage is likely to have a positive effect on physical distancing. Finally, contact duration did not differ significantly between conditions (U = 13038, p = 0.07), but participants in the signage condition rated their experience regarding regulations (t(383.24) = 2.18, p = .03), pleasantness (t(375.51) = 3.41, p <.001) and help (t(373.54) = 2.54, p = .01) more satisfactory than participants in the no signage condition.

Shopping cart

To examine the effect of altering shopping cart regulations, we compared the contact network on day 1, when a shopping cart was mandatory, with the contact networks on day 3, when the shopping cart was optional. On day 1 with mandatory shopping carts, participants had a median of 6 unique contacts, compared with a median of 9 unique contacts on day 3, when the shopping carts were optional. Despite these numerical differences, detailed analyses taking the network structure into account indicate that the probability of participants forming contacts was about the same in these two conditions (OR = 1.07, 95% CI [0.90, 1.26]), indicating that if the number of contacts are different between mandatory and optional shopping carts, these differences are likely to be small. Thus, mandatory shopping carts do not appear to facilitate nor inhibit physical distancing. In addition, the contact duration did not differ significantly between conditions (U = 11724, p = 0.68), and mandatory or optional shopping carts did not change participants’ ratings (all p > 0.1).

Discussion

In this paper we aspired to use behavioural science in response to crises by evaluating the effectiveness of interventions on directly observed behaviour. We applied this idea to the COVID-19 crisis and performed a behavioural experiment to investigate the effectiveness of behavioural interventions to promote physical distancing. We did so by implementing three interventions (i.e., reward, signage, shopping cart regulations) and evaluated their effect on the contacts between customers (i.e., a distance within 1.5 m). Our results demonstrate that rewarding customers for keeping their distance and providing signage both improved physical distancing and reduced the number of contacts. Interestingly, rewarding customers not only reduced the number of contacts, but also shortened the duration of the remaining contacts. In contrast, mandatory or optional shopping cards did not appear to improve or worsen physical distance behaviour. Overall, participants rated their shopping experience as satisfactory, and most of the regulations did not impact these ratings. Only signage was rated more positively on all accounts (regulations, pleasantness, perceived help) than no signage.

The current study showed how we can use behavioural science to find practical behavioural solutions to crises by formulating six key steps. First, starting out with an existing problem, we defined a desired behaviour to combat this problem. Second, and importantly, we found a way to directly measure this behaviour such that we can evaluate the effectiveness of interventions on a directly observed behavioural outcome measure. Third, based on the desired behaviour, we identified different interventions designed to stimulate this behaviour. These interventions can be based on psychological mechanisms (e.g., rewarding participants), but can also be informed by other fields, like in our case crowd management. Ultimately, the aim is to promote the desired behaviour, and any intervention targeted at this can be tested within the experimental design. Fourth, in an experimental design we systematically varied the interventions, such that we could in a fifth step analyse the effect of each of the interventions on the desired behaviour. Investigating the identified interventions in an experimental design is a crucial step, as behavioural solutions have to be tested (in the crisis situation) before they can be translated into policy [39]. Sixth and finally, the insights derived from the experiment can directly inform policy recommendations, making behavioural science a central link connecting science with society.

In Table 3 we outline the steps described above with concrete examples that we took in identifying behavioural solutions to COVID-19. Crucially, these steps transcend the COVID-19 crisis and can be applied much broader to other crises as well. For the financial crisis as well as the climate crisis these steps can similarly shed light on possible practical behavioural solutions for which we give an illustrative example and accompanying reference in Table 3. These are just a few examples, and the (behavioural) factors involved in these crises are much more complicated than can be captured in a simple table. In addition, in our current experiment we primarily focused on the context in which behaviour occurs, but clearly there are much more factors that influence behaviour such as biology and cognitive processes, social influences, and culture. Nonetheless, the steps that we outlined, and particularly directly observing behaviour, can serve as an avenue to use behavioural change as solution out of a crisis.

Table 3

Examples on how behavioural science can be used to develop effective interventions to stimulate desired behaviour.

	COVID-19 crisis	Global Financial Crisis	Climate crisis
problem desired behaviour	virus transmission physical distancing	housing bubble reduce risk taking	greenhouse gas emission increase vegetarian diets
direct measurement	physical distances between people measured using UWB	propensity to sell assets	number of vegetarian dishes sold
intervention	psychological mechanism: reward [28]pedestrian behaviour: follow signage [29]practical: adjust shopping cart regulations	psychological mechanism: salience [40], disposition effect [16]	psychological mechanism: salience [40]practical: visibility
experimental design and analysis	see current paper	see Frydman & Rangel (2014) [41]	see Kurz (2018) [42]
policy recommendation	reward people for keeping their distance	decrease salience of information related to capital gains	increase salience of vegetarian dishes

While our experimental study provides indications for practical behavioural solutions, there are some limitations that warrant attention. First, because of the naturalistic nature of the study it was not possible to randomize the participants over experimental conditions. As a result, there are several factors that we could not control, like the type of customers or the number of customers in each experimental condition. To limit this variation, we measured during the same days over a two-week time period. In addition, to compare the effectiveness of interventions we selected times during the day that were similar in crowdedness and time of day. Still, it could be possible that some interventions work differently depending on the crowdedness. For example, mandatory shopping carts could potentially facilitate physical distancing in quiet times, but actually crowd the supermarket even more during rush-hours. In the current naturalistic experiment, these factors are hard to disentangle.

Second, it might be challenging to implement some of our findings into practice. For example, we showed that rewarding participants improved physical distancing, but it might be challenging to implement these rewards structurally as supermarkets are frequently visited places. The one-off rewards could be promising for locations that people visit less often (e.g., a cinema or festival), but a different reward scheme might be necessary to achieve long term effects in a supermarket. Third, we did not include any assessments on whether the interventions were adhered to (e.g., whether participants followed the signage or refrained from using a shopping cart). However, we choose to limit the number of questions to maximize the number of participants completing the questionnaire. Even so, only between 25–51% of participants completed the questionnaires, possibly introducing some selection bias. Fourth, since the actual virus spread depends on more than behaviour alone, our study could be extended by collaborating with epidemiologists to quantify the reduction in risk of spread in each of the scenarios.

Last, all behavioural interventions we investigated were implemented in the context of a supermarket. To translate these findings to other situations it is important to consider relevant characteristics of different contexts. For example, in supermarkets people are likely to move constantly, and often visit a fixed set of locations corresponding to one’s shopping list. These characteristics may be substantially different from other situations, such as using public transport (where people may have to wait for their train to arrive), or visiting a museum (where people often visit an entire exhibition). In a previous study we showed that walking directions also facilitated physical distancing at an art fair [27], indicating that walking directions may be applicable both in situations where people plan their own stops (as in the supermarket) and in situations where people follow pre-specified routes and stops (as in the art fair).

Conclusion

To conclude, in this paper we have shown how we can find practical behavioural solutions to a crisis by defining and directly observing the desired behaviour in an experimental design. Of course, behavioural science alone will not offer the complete package to combat an entire crisis by itself. We need multidisciplinary collaborations to battle the multifaceted and complex problems of our time [22]. In these collaborations, behavioural science and behavioural change can provide new ways to look at existing (and new) challenges. We should start to take advantage of the opportunities offered by behavioural science.

Information flyer.

The translated flyer (originally Dutch) that was distributed to inform people of the upcoming experiment in the supermarket.

(PDF)

Click here for additional data file.

28 Mar 2022

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Gabriel Hoh Teck Ling, PhD

Academic Editor

PLOS ONE

Journal requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf.

2. Please update your submission to use the PLOS LaTeX template. The template and more information on our requirements for LaTeX submissions can be found at http://journals.plos.org/plosone/s/latex.

3. Thank you for stating the following in the Acknowledgments Section of your manuscript:

“We would like to thank André and Joyce van Reijen for opening up their supermarket to run this experiment, and PLUS for

printing all signage. We thank our team that helped collecting the data: Frederike Meijer, Sonja van Meerbeek, Zuzana Wilms,

Henk Nieweg, Nina Leach, and Lander Arteaga. The research project was supported by the Ministry of Economic Affairs and

Climate Policy. CT and TB were supported by an Innovation Exchange Amsterdam UvA Proof of Concept Fund”

We note that you have provided funding information that is not currently declared in your Funding Statement. However, funding information should not appear in the Acknowledgments section or other areas of your manuscript. We will only publish funding information present in the Funding Statement section of the online submission form.

Please remove any funding-related text from the manuscript and let us know how you would like to update your Funding Statement. Currently, your Funding Statement reads as follows:

“The research project was supported by the Ministry of Economic Affairs and Climate Policy. CT and TB were supported by an Innovation Exchange Amsterdam UvA Proof of Concept Fund. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.”

Please include your amended statements within your cover letter; we will change the online submission form on your behalf.

4. In your Data Availability statement, you have not specified where the minimal data set underlying the results described in your manuscript can be found. PLOS defines a study's minimal data set as the underlying data used to reach the conclusions drawn in the manuscript and any additional data required to replicate the reported study findings in their entirety. All PLOS journals require that the minimal data set be made fully available. For more information about our data policy, please see http://journals.plos.org/plosone/s/data-availability.

Upon re-submitting your revised manuscript, please upload your study’s minimal underlying data set as either Supporting Information files or to a stable, public repository and include the relevant URLs, DOIs, or accession numbers within your revised cover letter. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories. Any potentially identifying patient information must be fully anonymized.

Important: If there are ethical or legal restrictions to sharing your data publicly, please explain these restrictions in detail. Please see our guidelines for more information on what we consider unacceptable restrictions to publicly sharing data: http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions. Note that it is not acceptable for the authors to be the sole named individuals responsible for ensuring data access.

We will update your Data Availability statement to reflect the information you provide in your cover letter.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Partly

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: I Don't Know

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: No

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: This is an interesting experimental study examining the impact of rewards and cues, one component of behavioural interventions, on distancing behaviours inn a supermarket in the Netherlands during the initial wave of the COVID-19 pandemic. The study was well done and the analyses appear sound. I do not feel sufficiently expert to comment on the statistical approaches used; a statistical reviewer might be helpful.

While I do like this paper, I think that it overstates its purpose. A reader not familiar with behavioural science would conclude that this study is groundbreaking in introducing behavioural sciences to crises. This is simply not true and misleading on 2 counts. First, behavioural sciences have been quite active in the pandemic response. I did a quick PubMed search with the search terms "behavioural science" and "COVID" and got 106 papers from 2020 - 2022, including the following papers directly on behavioural science approaches:

Using social and behavioural science to support COVID-19 pandemic response

Mental health and clinical psychological science in the time of COVID-19: Challenges, opportunities, and a call to action

Research priorities for the COVID-19 pandemic and beyond: A call to action for psychological science

Psychological science and COVID-19: An agenda for social action

Infected by Bias: Behavioral Science and the Legal Response to COVID-19

Trust in Science, Perceived Media Exaggeration About COVID-19, and Social Distancing Behavior

The cognitive science of COVID-19: Acceptance, denial, and belief change

Intervening on Trust in Science to Reduce Belief in COVID-19 Misinformation and Increase COVID-19

Preventive Behavioral Intentions: Randomized Controlled Trial

Ending the Pandemic: How Behavioural Science Can Help Optimize Global COVID-19 Vaccine Uptake

Applying relationship science to evaluate how the COVID-19 pandemic may impact couples' relationships

Individual health behaviours to combat the COVID-19 pandemic: lessons from HIV socio-behavioural science

How behavioural science data helps mitigate the COVID-19 crisis

Lessons From the UK's Lockdown: Discourse on Behavioural Science in Times of COVID-19

Covid-19: What we have learnt from behavioural science during the pandemic so far that can help prepare us for the future

The Science of Persuasion Offers Lessons for COVID-19 Prevention

Can Behavioral Science Help Us Fight COVID-19

Fear of Covid-19: Insights from Evolutionary Behavioral Science

Process-based functional analysis can help behavioral science step up to novel challenges: COVID - 19 as an example

The Dynamics of Fear at the Time of Covid-19: A Contextual Behavioral Science Perspective

Effect of Targeted Behavioral Science Messages on COVID-19 Vaccination Registration Among Employees of a Large Health System: A Randomized Trial

Harnessing behavioural science in public health campaigns to maintain 'social distancing' in response to the COVID-19 pandemic: key principles

It falls to the authors to place their study into the context of the behavioural sciences. Second, the authors focus on one very specific aspect of the behavioural sciences; environmental context. They, in essence, draw on the sub-area of behaviour modification, primarily using cues and rewards to shape behaviour. There is nothing wrong with this, except behavioural science is broader, to include social influences, culture, biology, and the full range of cognitive processing characteristics. Again, nothing wrong with what the authors have done but they should inform the readers of the specific aspects of the behavioural science approach they are taking. For instance, great gains have been made by framing behavioural sciences within what is called the Theoretical Domains Framework, an integration of 33 behavioural change theories, that has identified 14 domains of behaviour change intervention and has been effectively summarized with the COM-B model; behaviour is the result of Capability, Opportunity, and Motivation. This paper falls within the Opportunity domain. For the author' information, this model has been developed from University College London UK, under the guidance of the behavioural scientist Dr. Susan Michie, who is a member of the UK COVID Response team at the highest level of government (the point being it is not accurate to say behavioural science has been left out of the response to COVID - other countries also have behavioural science teams offering advice).

The methodology of this study is very interesting and appears sound. I can see how this methodology can be useful for specific questions, I am a bit confused, however, by Table 1. We see 7 days of intervention but the analyses only involve comparing days 1, 3 (Shopping cart), 6 (signage) and 7 (rewards). What is the purpose of days 2, 4, 5? On that note the Table lists 'space' as an intervention but this is labelled shopping cart in the text.

In the discussion I wonder if the authors have any comment about this study being conducted at the beginning of lockdown experience, where most of the population was experiencing threat. Now that we are almost 2 years in, and many in the population are experiencing demoralization of outrage (the Netherlands has made international coverage of protests recently) do the authors think the study would yield the same results?

I look forward to the contribution of this work to the field, once the study is appropriately contextualized.

Reviewer #2: Summary

This was an interesting and timely naturalistic experimental study that examined the efficacy of three behavioural interventions for promoting physical distancing behaviour in grocery stores during the covid-19 pandemic: (i) rewarding customers for keeping distance; (i) providing signage to guide customers; and (iii) altering shopping cart regulations. They recruited 4323 participants and the main outcome was number of contacts less than 1.5 between customers measured using network analysis. Results showed that both rewards and signage increased physical distancing, but shopping cart regulations did not. Rewards also reduced the duration of contacts. The authors concluded that incorporating behavioural science approaches and interventions into pandemic management should be strengthened and emphasized to improve pandemic outcomes.

Comments

• The introduction of this paper was very compelling – the fact that in times of crisis we turn to crisis-specific experts (e.g., economists during financial crises), the authors did an outstanding job of asking why, given the importance of engaging in preventive behaviours (from distancing to vaccination) during the covid-19 pandemic crisis, did we not turn to behavioural science experts?

• The study was also generally well reasoned in terms of exploring the efficacy of different behavioural interventions to promote distancing behaviour. However, the choice of specific interventions was not described or articulated. Authors did not justify their underlying theoretical rationale (why would they be expected to change behaviour in this context and why these interventions over others?). Tying each intervention to an established behaviour change theory or model would strengthen the paper and highlight the importance of doing this in general. For example, the rationale for the shopping cart intervention is not obvious to me.

• Could the authors clarify what participants were told about the objective of the study – for example: did they know what each intervention was and what outcome was being measured? The authors described this as a naturalistic experiment, but if they knew what was being measured and why, this could have influenced their behaviour more than just being exposed to the intervention (without details).

• Could the authors also clarify if they delivered the interventions the same way they would have been delivered were they implemented in ‘real life’? For example, there were study personnel present to explain the study, hand out tags, and answer questions. Would these resources be available if we were deliver the interventions in real life? Would these roles be assumed by store personnel? The use of an implementation science approach to intervention design an delivery was not explicit.

• The use of objective measures of distancing (SafeTag) was judged to be a strength based on the non-intrusiveness and validity of the measures.

• The authors described how they treated shoppers who were shopping together (as they would likely have close contact throughout the intervention that needed to be accounted for). They described how they accounted for this (based on contact metrics), but the potential for misclassification seems high. For example, many people or families shopping together ‘split up’ in the interests of time – these shopping patterns may have been miscalculated for these groups. Why not just ‘tag’ people shopping together when they enter the store and receive their tags, so irrespective of their shopping patterns, they would not be counted in distancing measures (because we don’t expect those living together or family groups to distance). Could the authors clarify this?

• Given that national regulations at the time of the study limited the number of customers in the store (max one customer per 10m2), store access had to be monitored and controlled. The authors seemed to account for this by comparing conditions at times that were similar in terms of crowdedness, which is appropriate.

• For the signage intervention, the authors did not appear to assess how many people viewed the signs (eg, during exit interviews or surveys). It is difficult to attribute behavioural changes to this intervention in the absence of verifying the extent to which the intervention was ‘received’ by shoppers.

• The authors conducted experience assessments (though only 25-51% completed them), though these assessments did not appear to validate receipt of the interventions (e.g., viewing the signs) or the extent to which the decision to maintain distancing with automatic or reflexive (as per COM-B model)? This would have pointed to the mechanism of action of the interventions (which is how we think interventions are working), and so not assessing these things seems like a missed opportunity.

• The discussion could be strengthened by a discussion of the effect size of their findings and the extent to which results, some of which were statistically significant, were also clinically significant in terms of ability to translate into reduction of virus transmission. For example, on a population level, is reducing contacts from 8 to 6 clinically significant? Is reducing contact time by 1.7 seconds enough to prevent virus transmission? While I wholeheartedly agree that more behavioural interventions need to be designed, evaluated and implemented, we need to ensure we are demonstrating the value of these interventions for addressing the crisis at hand, which is going to strengthen the credibility of the argument to turn to behavioural science for solutions to the pandemic crisis.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: Yes: Kim L. Lavoie

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

15 Jun 2022

Note: mark-up was lost when copying the text in this field. A version in which we added mark-up to facilitate the review process has been added as file.

Reviewer #1: This is an interesting experimental study examining the impact of rewards and cues, one component of behavioural interventions, on distancing behaviours inn a supermarket in the Netherlands during the initial wave of the COVID-19 pandemic. The study was well done and the analyses appear sound. I do not feel sufficiently expert to comment on the statistical approaches used; a statistical reviewer might be helpful.

While I do like this paper, I think that it overstates its purpose. A reader not familiar with behavioural science would conclude that this study is groundbreaking in introducing behavioural sciences to crises. This is simply not true and misleading on 2 counts. First, behavioural sciences have been quite active in the pandemic response. I did a quick PubMed search with the search terms "behavioural science" and "COVID" and got 106 papers from 2020 - 2022, including the following papers directly on behavioural science approaches: . It falls to the authors to place their study into the context of the behavioural sciences.

We thank the reviewer for raising this important point and list of references. Upon rereading the manuscript with this comment in mind, we understand that it might be read as if we aim to introduce behavioural science to crisis, so we very much agree that we should stress the contributions of behavioural science to crises more, as we absolutely do not wish to claim that we are the first ones to do so. We do argue that we can utilize behavioural science more, specifically by investigating the effectiveness of interventions on *directly observed behaviour*. To place our study in the context of the behavioural sciences, we have now added additional references, especially throughout the introduction, and make a clearer distinction on what has already been done and what is new in this manuscript. See for example page 2 lines 16-19:

“This need for behavioural change resulted in numerous studies into factors that determine behaviour [8-10], and worldwide regulations such as lockdowns, school closures, and travel restrictions, all directed to reduce the number of behavioural contacts.”

Second, the authors focus on one very specific aspect of the behavioural sciences; environmental context. They, in essence, draw on the sub-area of behaviour modification, primarily using cues and rewards to shape behaviour. There is nothing wrong with this, except behavioural science is broader, to include social influences, culture, biology, and the full range of cognitive processing characteristics. Again, nothing wrong with what the authors have done but they should inform the readers of the specific aspects of the behavioural science approach they are taking. For instance, great gains have been made by framing behavioural sciences within what is called the Theoretical Domains Framework, an integration of 33 behavioural change theories, that has identified 14 domains of behaviour change intervention and has been effectively summarized with the COM-B model; behaviour is the result of Capability, Opportunity, and Motivation. This paper falls within the Opportunity domain. For the author' information, this model has been developed from University College London UK, under the guidance of the behavioural scientist Dr. Susan Michie, who is a member of the UK COVID Response team at the highest level of government (the point being it is not accurate to say behavioural science has been left out of the response to COVID - other countries also have behavioural science teams offering advice).

We thank the reviewer for this excellent comment. Again, we never meant to imply that behavioural science was left out of the response to COVID-19 altogether. Additionally, we agree that the manuscript is improved by placing this study within the COM-B model. We have now included an entire paragraph on the COM-B model in the introduction:

Behavioural science offers many models into the determinants of behaviour, and thereby offers leveraging points on how to instantiate behavioural change. The widely used Capability, Opportunity, Motivation, Behaviour (COM-B) model, for example, posits that a particular behaviour occurs when someone has the capability (i.e., psychological and physical), opportunity (i.e., contextual factors that facilitate the behaviour), and motivation (i.e., processes that energize and direct behaviour) to enact the behaviour [21]. By outlining the factors that influence particular behaviours to occur, the model also provides the opportunity to identify different points of engagement to bring about the desired behavioural change.

We also added the following sentence in the discussion:

In addition, in our current experiment we primarily focused on the context in which behaviour occurs, but clearly there are much more factors that influence behaviour such as biology and cognitive processes, social influences, and culture.

The methodology of this study is very interesting and appears sound. I can see how this methodology can be useful for specific questions, I am a bit confused, however, by Table 1. We see 7 days of intervention but the analyses only involve comparing days 1, 3 (Shopping cart), 6 (signage) and 7 (rewards). What is the purpose of days 2, 4, 5? On that note the Table lists 'space' as an intervention but this is labelled shopping cart in the text.

To isolate the effect of each intervention, we chose to compare days that only differed on that factor while keeping the remaining factors constant. In the design of the study, we build in some redundancy so that we could minimise the effect of factors such as day of the week and time of day, which are likely to have an effect on the type of customers present in the supermarket and could, as such, potentially affect the comparisons that were made. . Due to this redundancy, we did not actually include all days in our analysis. To clarify these decisions, we have now added the following text when describing the table:

Note that there is some redundancy in the experimental conditions we implemented: for example, the effect of reward could be assessed by comparing day four to day three or two, but also by comparing day 7 to day 6. We did so to minimise the influence of factors such as day and time of day. Both the day and time of day are likely to affect the type of customers and crowdedness in the supermarket. Customers during workdays may differ from customers during the weekend, and customers during the morning may differ from customers after work-hours. Similarly, the crowdedness in the supermarket is likely to affect the number of contacts made. Since these factors are difficult to control in a naturalistic study, we implemented some redundancy in the design so that we could select time intervals post-hoc that would minimise differences in these factors for an optimal comparison across conditions.

In addition, we extended the following sentences in the Descriptive analyses section:

To evaluate the effectiveness of each intervention in isolation, we select two conditions that differ only in regard to whether the behavioural intervention of interest is implemented and that are similar in crowdedness and time of day. This way we can isolate the effect of the intervention while keeping other factors constant.

Finally, we thank the reviewer for pointing out the inconsistency in the phrasing of “shopping cart” and “space”. We have updated the text in the table to also refer to this intervention as “shopping cart”.

In the discussion I wonder if the authors have any comment about this study being conducted at the beginning of lockdown experience, where most of the population was experiencing threat. Now that we are almost 2 years in, and many in the population are experiencing demoralization of outrage (the Netherlands has made international coverage of protests recently) do the authors think the study would yield the same results?

This is an interesting point and upon careful consideration we actually think that the study would yield similar results. Specifically, when we conducted the study in March 2021, we were already one year into the pandemic. During that time there were some large-scale demonstrations (e.g., there were some severe curfew riots in The Netherlands January 23-26 2021) and demorilazation due to ongoing lockdowns. So, we do think that results between March 2021 and later on in the pandemic would be more comparable than if our study was conducted right at the start of the pandemic. Moreover, the effectiveness of the reward intervention also fits in a larger psychological context, where threat does not need to be present for rewards to promote desired behaviour.

I look forward to the contribution of this work to the field, once the study is appropriately contextualized.

Reviewer #2: Summary

This was an interesting and timely naturalistic experimental study that examined the efficacy of three behavioural interventions for promoting physical distancing behaviour in grocery stores during the covid-19 pandemic: (i) rewarding customers for keeping distance; (i) providing signage to guide customers; and (iii) altering shopping cart regulations. They recruited 4323 participants and the main outcome was number of contacts less than 1.5 between customers measured using network analysis. Results showed that both rewards and signage increased physical distancing, but shopping cart regulations did not. Rewards also reduced the duration of contacts. The authors concluded that incorporating behavioural science approaches and interventions into pandemic management should be strengthened and emphasized to improve pandemic outcomes.

Comments

• The introduction of this paper was very compelling – the fact that in times of crisis we turn to crisis-specific experts (e.g., economists during financial crises), the authors did an outstanding job of asking why, given the importance of engaging in preventive behaviours (from distancing to vaccination) during the covid-19 pandemic crisis, did we not turn to behavioural science experts?

We thank the reviewer for judging the introduction as compelling.

• The study was also generally well reasoned in terms of exploring the efficacy of different behavioural interventions to promote distancing behaviour. However, the choice of specific interventions was not described or articulated. Authors did not justify their underlying theoretical rationale (why would they be expected to change behaviour in this context and why these interventions over others?). Tying each intervention to an established behaviour change theory or model would strengthen the paper and highlight the importance of doing this in general. For example, the rationale for the shopping cart intervention is not obvious to me.

We agree with the reviewer that the paper is strengthened by providing the rationale of choosing our set of interventions. We have now extended the following paragraph in the Participants and Design section to elaborate on our choices:

The experiment took place during seven days in a supermarket. We varied three interventions: reward [28], signage [29], and adjusting the shopping cart regulations. These interventions were chosen for varying reasons. First, rewarding people for displaying the desired behaviour is well-established to be effective in promoting that desired behaviour [30], and also advised during the COVID-19 pandemic in particular [31]. In our study, participants received the reward upon handing in their tag (see Procedure) and consisted of cookies on Saturday March 24th and chocolate on Friday March 26th. Second, signage is commonly used to change behaviour [29] and often used in traffic to, for example, avoid collisions [32]. We aimed to investigate whether clear signage would facilitate pedestrian flows and thereby physical distancing. We included arrows signaling unidirectional walking directions in part of the supermarket and footprints in the queue for the register (see Materials below). Third, we changed the shopping cart regulations from mandatory (which was standard in the Netherlands at that time) to optional. The mandatory shopping carts were implemented in the Netherlands in March 2020 [33] to (1) keep track of the number of participants in the supermarket as national regulations allowed a maximum of 1 customer per 10 m2, and (2) as the shopping carts were thought to facilitate physical distancing. At the same time it could be argued that the mandatory shopping carts take up a lot of space within the supermarket and, as such, hinder the opportunity for physical distancing.

• Could the authors clarify what participants were told about the objective of the study – for example: did they know what each intervention was and what outcome was being measured? The authors described this as a naturalistic experiment, but if they knew what was being measured and why, this could have influenced their behaviour more than just being exposed to the intervention (without details).

We thank the reviewer for raising this point. To elaborate on which information was provided to participants we have now added the information flyer as Supporting Information. The participants were aware that we measured distance to other visitors in the supermarket. Thus, it is likely that handing out the sensors (without any additional intervention) already has an effect on the distance that visitors maintained while shopping. Therefore we included two baseline conditions (Wednesday 17 March, and 24 March) in which we merely handed out the sensors, without any additional intervention. This allows us to evaluate the effect of a single intervention, while taking the potential effect of measuring the behaviour (also known as ‘mere measurement effect’) into account.

• Could the authors also clarify if they delivered the interventions the same way they would have been delivered were they implemented in ‘real life’? For example, there were study personnel present to explain the study, hand out tags, and answer questions. Would these resources be available if we were deliver the interventions in real life? Would these roles be assumed by store personnel? The use of an implementation science approach to intervention design an delivery was not explicit.

Extra personnel was only present to hand out and hand in the sensors, but the shopping experience itself was the same as usual. If evidence was found in favour of the effectiveness of an intervention, that intervention could then be implemented after the conclusion of the study. At that point, no sensors (or extra personnel) would be needed. For example, finding that providing signage works, then signage could be placed inside the store without needing to measure distances again. To clarify how the interventions were implemented, and which extra personnel was present we added the following to the Procedures section:

In conditions in which a reward was handed out, we also informed participants that they would receive a reward for their effort to keep their distance upon handing in their SafeTag. Participants could then proceed to do their groceries like they would otherwise do. Thus, as soon as the participants entered the supermarket they did not encounter any study personnel.

• The use of objective measures of distancing (SafeTag) was judged to be a strength based on the non-intrusiveness and validity of the measures.

We thank the reviewer for this comment and we agree that the SafeTag is a great non-intrusive way to measure distance. Another option would be to use cameras, but that brings quite some privacy issues. Using the SafeTags, we never need to know which person wears which tag.

• The authors described how they treated shoppers who were shopping together (as they would likely have close contact throughout the intervention that needed to be accounted for). They described how they accounted for this (based on contact metrics), but the potential for misclassification seems high. For example, many people or families shopping together ‘split up’ in the interests of time – these shopping patterns may have been miscalculated for these groups. Why not just ‘tag’ people shopping together when they enter the store and receive their tags, so irrespective of their shopping patterns, they would not be counted in distancing measures (because we don’t expect those living together or family groups to distance). Could the authors clarify this?

We understand that the rules we have for tagging people as belonging from the same household might seem quite cumbersome. The reason why we chose this approach was that we needed to be able to hand out the sensors as quickly as possible. Especially during COVID-19, we could not risk creating congestion at the entrance of the store. For that reason, we did not have time to register upon entering which visitors came in together. Whenever two participants were tagged as belonging to the same group, only the contacts between them were not counted in analysis. Any contact made outside of the group was included. To clarify these choices we have now included the following in the Procedure section:

If customers of the same household participated, they each received a SafeTag. According to regulations, household members did not need to keep a distance, but we were unable to register their group membership upon handing out the tag to avoid congestion. We did exclude contacts between group members when processing the data and only kept contacts with individuals outside of the group (see Pre-processing).

In addition we expanded the Pre-processing section:

We then checked if participants entered the supermarket as a household by investigating three criteria: at least 10 contacts within 80 cm, being within 1.5 m of each other for at least 25% of their visit duration, and exiting the supermarket at most 60 seconds after each other. We assigned participants as belonging to the same group if they met at least two of these criteria and removed all contacts between the respective group members.

• Given that national regulations at the time of the study limited the number of customers in the store (max one customer per 10m2), store access had to be monitored and controlled. The authors seemed to account for this by comparing conditions at times that were similar in terms of crowdedness, which is appropriate.

We thank the reviewer for appreciating the way in which we chose to compare conditions.

• For the signage intervention, the authors did not appear to assess how many people viewed the signs (eg, during exit interviews or surveys). It is difficult to attribute behavioural changes to this intervention in the absence of verifying the extent to which the intervention was ‘received’ by shoppers.

• The authors conducted experience assessments (though only 25-51% completed them), though these assessments did not appear to validate receipt of the interventions (e.g., viewing the signs) or the extent to which the decision to maintain distancing with automatic or reflexive (as per COM-B model)? This would have pointed to the mechanism of action of the interventions (which is how we think interventions are working), and so not assessing these things seems like a missed opportunity.

We agree that it would have been ideal to also assess how many people viewed the signs. We have now included this point as a limitation in the Discussion section:

Third, we did not include any assessments on whether the interventions were adhered to (e.g., whether participants followed the signage or refrained from using a shopping cart). However, we choose to limit the number of questions to maximize the number of participants completing the questionnaire. Even so, only between 25-51% of participants completed the questionnaires, possibly introducing some selection bias.

• The discussion could be strengthened by a discussion of the effect size of their findings and the extent to which results, some of which were statistically significant, were also clinically significant in terms of ability to translate into reduction of virus transmission. For example, on a population level, is reducing contacts from 8 to 6 clinically significant? Is reducing contact time by 1.7 seconds enough to prevent virus transmission? While I wholeheartedly agree that more behavioural interventions need to be designed, evaluated and implemented, we need to ensure we are demonstrating the value of these interventions for addressing the crisis at hand, which is going to strengthen the credibility of the argument to turn to behavioural science for solutions to the pandemic crisis.

We again agree with the reviewer that this is a very interesting point. As we do not ourselves have a background in epidemiology we feel uncomfortable to directly translate these effects into clinically meaningful statements. However, we have now included the following as a limitation in the Discussion section:

Fourth, since the actual virus spread depends on more than behaviour alone, our study could be extended by collaborating with epidemiologists to quantify the reduction in risk of spread in each of the scenarios.

Submitted filename: 220610_response-reviewers.pdf

Click here for additional data file.

1 Aug 2022

Practical behavioural solutions to COVID-19: Changing the role of behavioural science in crises

PONE-D-21-32828R1

Dear Dr. Tanis,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Gabriel Hoh Teck Ling, PhD

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: I Don't Know

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: (No Response)

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

**********

14 Sep 2022

PONE-D-21-32828R1

Practical behavioural solutions to COVID-19: Changing the role of behavioural science in crises

Dear Dr. Tanis:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Gabriel Hoh Teck Ling

Academic Editor

PLOS ONE