Proximity Contacts and Self-perceived Vulnerability to Diseases Among High School Students – A Preliminary Study

Abstract

In recent years there has been much research regarding the extent to which behavioural immune system (BIS; behavioural responses that evolved to detect and avoid pathogens) is related to a variation in social profile and personality traits such as extroversion among individuals. However, most research on the relationship between BIS and sociality is based on short-term experimental studies and thus little is known about the degree to which BIS is related to the actual frequency of day-to-day proximity contacts. We used data obtained from secondary students (n = 84, age 16-19yo, n schools/classes = 3) to determine whether the variation in BIS values is associated with the frequency of proximity contacts. Proximity data were derived from sensors registering the proximity of other classmates within a 2 m radius. Students belonging to the same class were wearing proximity sensors within the school environment for the period of 4 weeks. BIS activity was assessed using a total score of Perceived Vulnerability to Diseases (PVD) questionnaire. Our findings, based on real-world face-to-face proximity patterns, do not support the assertions of previous studies suggesting that BIS activation, as reflected in higher PVD values, is associated with a lower rate of social interactions. However, this does not rule out the existence of the relationship under investigation. Future studies should include larger samples and analyses of other measures of integration within close social networks.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1007/s12110-026-09525-0.

Introduction

In recent years, there has been much focus on the extent to which proximity contacts are associated with health indices, such as physiological stress levels and depression (Fedurek et al., 2023; Teo et al., 2015). Some studies suggest that face-to-face social interactions with close ones, such as friends and family members, are related to better indices of self-reported wellbeing as well as psychological and physiological markers of health (Fedurek et al., 2023; Fiorillo & Sabatini, 2011, 2015; Teo et al., 2015). However, it has also been showed that proximity contacts greatly influence individuals’ exposure to pathogens, and the role an individual has in disease transmission (Toroczkai & Guclu, 2007).

Due to the potential risks related to contracting diseases through face-to-face contacts, humans have developed various defence strategies to either minimise infection risk or reduce pathogen-induced symptoms (Ackerman et al., 2018). One of the most effective, complex, and specialised sets of biological mechanisms protecting organisms against pathogens is the physiological immune system (PIS) (Langman, 2014). However, the activation of physiological immune defence may be physiologically costly (Sandland & Minchella, 2003). For example, prolonged inflammation resulting from hyper-reactive PIS has been linked to numerous cardiovascular, musculoskeletal, and immune diseases (Chung et al., 2009). The activation of PIS is also related to oxidative stress and ageing (Conner & Grisham, 1996; Gassen et al., 2018; Khansari et al., 2009). Therefore, throughout history, selective pressures from various pathogens and infectious diseases have influenced many aspects of social behaviour and led to the evolution of a set of pre-emptive behavioural mechanisms that mitigate the potential shortcomings related to PIS activation responses even before the disease onset (Fumagalli et al., 2011; Barreiro & Quintana-Murci, 2010; Brinkworth & Alvarado, 2020; Fumagalli et al., 2011; Karlsson et al., 2014).

Among humans, this type of pathogen avoidance mechanism is referred to as the pathogen avoidance system (Tybur & Lieberman, 2016) or, alternatively, the behavioural immune system (BIS) (Murray & Schaller, 2016; Schaller & Duncan, 2007; Schaller & Park, 2011). In short, the behavioural immune system (BIS) is a set of cognitive mechanisms that affect behavioural responses (Murray & Schaller, 2016) helping to minimise infection and contagion risk (Schaller, 2006). BIS is believed to be the first line of defence against pathogens and is designed to minimise the necessity of engaging metabolically costly physiological immune responses by avoiding initial exposure to pathogens (Schaller & Duncan, 2007). For instance, one study reports that BIS activity allows the body to downregulate basal inflammation, preventing resultant oxidative damage to DNA and promoting longevity (Gassen et al., 2018). Such psychological mechanisms often involve a feeling of disgust towards any potential pathogen spreaders (Curtis, 2014; Oaten et al., 2009) leading to adjustments in social behaviour of an individual towards both in-group (e.g., conformism, prejudice against sick, homosexuals and other minorities), and out-group members (e.g., xenophobia), but see Fan et al. (2022) whose results do not support the hypothesis that ethnic outgroup membership alone could indicate infection risk).

Although BIS plays an important role in protecting an individual against pathogens, it might come with some costs. It has been suggested, for instance, that the activation of BIS is related to a reduced dispositional tendency to gregariousness and, ultimately, a decreased level of socialisation (Schaller et al., 2015). There is also a solid body of empirical evidence showing that activation of BIS is related to adjustments in gregariousness by decreasing the number of both social interactions and social partners, especially with strangers (Schaller et al., 2015). Moreover, it has been shown that the sensitivity of BIS varies not only in terms of sex (women report higher PVD) but also among individuals and that this variance has been linked to one of the BIS markers, such as perceived vulnerability to diseases (PVD) (Díaz et al., 2016; Duncan et al., 2009).

PVD reflects a subjective evaluation of individuals’ susceptibility to infectious diseases and the emotional distress they experience due to the perceived risk of infection (Duncan et al., 2009; Park et al., 2007). PVD consists of two domains, Perceived Infectability (i.e., perceived susceptibility to getting infectious disease) and Germ Aversion (i.e., aversive reactions to situations posing high probability of pathogen transmission; Duncan et al., 2009), both of which may translate into disease avoidance behaviours and influence individual social interactions. For instance, individuals who perceive themselves as more vulnerable to diseases show lower levels of extraversion (Duncan et al., 2009). These inter-subjective differences in the sensitivity to PVD could be heightened by salient cues of a pathogen (Schaller, 2014). For example, some studies demonstrate that individuals with high PVD express preferences towards healthier-looking faces (Welling et al., 2007) and prefer to interact socially with self-perceived friends compared to non-friend acquaintances (Mortensen et al., 2010).

Still, individuals exhibiting higher levels of self-perceived vulnerability to diseases have been reported to experience higher levels of PVD even in social contexts that are, in terms of pathogen threat, relatively benign to others (e.g., avoiding interaction with strangers who were not necessarily showing signs of a disease or infection; Schaller & Park, 2011). Moreover, some diseases propagate through asymptotic transmission, and, therefore, the presence of salience cues of a pathogen is not necessarily a prerequisite for the activation of BIS. Asymptotic transmission not only underlines some common infectious diseases, such as common flu, but also some emergent diseases, such COVID-19 caused by the SARS-CoV-2 virus (Li et al., 2020).

One would therefore expect that individuals exhibiting high levels of PVD will be generally less inclined to engage in social interactions and, thus, have a smaller number of social contacts compared to individuals exhibiting low indices of PVD. However, little is known regarding the extent to which PVD and social activity are associated with the level of day-to-day social contacts. This is because so far, studies looking at the relationship between sociality and PVD, including studies looking at social behaviour in live social interactions, have been mainly based on either experimental design or self-reported measures of sociality (Faulkner et al., 2004; Mortensen et al., 2010; Sawada et al., 2018). Such studies have usually measured short behavioural responses to social stimuli resulting from BIS activation following exposure to salient cues of disease (Mortensen et al., 2010). However, very few studies have investigated how the activation of BIS affects the actual in-person social contacts. For example, one experimental study showed that experimentally elevated BIS was associated with decreased attraction and affiliative behaviour (Sawada et al., 2018). However, the extent to which the BIS profile of an individual and the resulting inter-personal variance in PVD scores are related to actual day-to-day social contacts in natural (i.e., non-experimental) settings, especially with familiar in-group members, has not been investigated. This is important, given that throughout most of their evolutionary past, humans lived in small-scale, hunter-gatherer societies, where people mostly interacted with other group or camp members rather than strangers. Therefore, the adaptive values of BIS in relation to infection risks resulting from actual in-person social interaction in natural settings are poorly understood.

In this study, we used proximity contacts collected from secondary school students over four weeks to determine the extent to which proximity is related to PVD indices. Previous research has shown that heightened BIS is associated with lower levels of extroversion (Mortensen et al., 2010) and decreased affiliative behaviour (Sawada et al., 2018). Moreover, individuals scoring higher on PVD reported fewer face-to-face interactions, even after controlling for Big Five personality traits (Makhanova & Shepherd, 2020). Crucially, lower levels of social interactions among individuals exhibiting higher PVD have been reported to occur especially in crowded environments (Kim & Kang, 2024; Szymkow et al., 2021). Thus, we hypothesised that individuals exhibiting high PVD values will be less engaged in proximity contacts while those with low PVD values will show higher frequencies of proximity contacts. This follows that individuals who perceive themselves as more vulnerable to diseases will narrow the number of social contacts in order to minimise the risk of contracting infectious diseases.

Methods

Study Site and Target Sample Size

Using a convenience sampling method, we selected three high schools located in different regions of Poland. During the selection process, we targeted schools and teachers known to the authors or their colleagues. As a result, we recruited students from Dynów (southeastern Poland), Toruń (central Poland), and Wrocław (southwestern Poland).

In the study conducted in LO Dynów we recruited 23 students [aged 17 (n = 6), 18 (n = 15) and 19 (n = 2) years old] belonging to one class of the last (4th) year/grade of their secondary education. The study was conducted between 04.02.2023 and 06.03.2023. In the study conducted in LO VI Wrocław we recruited 30 students [aged 15 (n = 15), 16 (n = 16) and 17 (n = 1) years old] belonging to one class of the second (2) year/grade of their secondary education. The study was conducted between 20.11.2023 and 20.12.2023. In the study conducted in LO V Toruń we recruited 31 students [aged 17 (n = 2), 17 (n = 23) and 18 (n = 6) years old] belonging to one class of the fourth (last) year/grade of their secondary education. The study was conducted between 14.11.2024 and 07.02.2025 (this includes a few-week break, e.g., Christmas break). One student in Dynów, one student Toruń and three students in Wroclaw did not agree to participate in the study.

Recruitment and Ethics procedures are provided in the Supplementary Material.

Behavioural Immunity – Perceived Vvulnerability to Disease

We used the Perceived Vulnerability to Disease scale (PVD (Duncan et al., 2009); as a proxy for estimating individual levels of behavioural immunity). The PVD scale measures perceptions of infection resistance and concerns about chronic pathogens. The scale is widely used in behavioural immune research (Faulkner et al., 2004; Navarrete & Fessler, 2006; Prokop et al., 2010; Welling et al., 2007). In essence, the 7-point scale (1 = “strongly disagree”; 7 – “strongly agree”) consists of two subscales: (1) Perceived Infectability (seven items), assessing beliefs about one’s susceptibility to infectious diseases and (2) Germ Aversion (eight items), assessing emotional discomfort in contexts that connote especially high potential for pathogen transmission. In the current study, we combined scores from all items of two subscales and calculated the overall score as a general index of perceived vulnerability (following a similar approach as Faulkner et al. (2004). Higher total PVD scores indicate greater perceived vulnerability to disease. Study participants completed the PVD questionnaire on the last day of the experiment. This sequence minimised the risk that the questionnaire items would influence participants’ proximity interactions during the whole study. Students who were absent on the final day of the experiment completed the questionnaire upon their return to school.

The total PVD scale demonstrated acceptable internal consistency (Cronbach’s α = 0.78; standardised α = 0.79). The subscale consistency was excellent for the Perceived Infectability subscale (Cronbach’s α = 0.90; standardised α = 0.91) and marginal for the Germ Aversion subscale (Cronbach’s α = 0.65; standardised α = 0.65).

Assessing Proximity

Proximity Sensors

Proximity was measured using B8 Social Distancing Wristbands proximity sensors: beacon tags based on Bluetooth^® Low Energy 5.0 technology (Minew corporation). B8 Social Distancing Wristbands are small, wearable devices that aim to provide proximity data. These sensing devices communicate with one another and store all communications within a specified distance. Each device sends a message that contains its unique ID, a timestamp and the signal strength at a programmed interval of three seconds. This message is picked up and stored by any other B8 Social Distancing Wristband within a specified radius around the emitting wristband. Proximity strength between two participants was operationalized as the totalnumber of Bluetooth detections recorded by all wristbands within a dyad, producing a symmetric measure of interaction strength.

Proximity Data Acquisition

Proximity sensors were worn at all times, both during school hours and outside of school (e.g., during break times when students typically left the school premises for lunch). During the study, proximity data was downloaded for analysis at the end of a school day. As opposed to proximity sensors used in previous studies conducted in a school environment (e.g., Cattuto et al., 2010; Mastrandrea et al., 2015; Salathé et al., 2010; Stehlé et al., 2011; Toth et al., 2015), B8 Social Distancing Wristbands operate in an infrastructure-free fashion thus they do not require supporting stationary base stations or other instrumentation of measurement spaces (e.g., LAN or other types of installed in fixed locations gateways that support the sensors). Therefore, B8 Social Distancing Wristbands were able to collect proximity data within the school and outside the school environment. The latter was critical for our study, especially in terms of collecting proximity contacts during school breaks, when students freely interact with their peers, and during such breaks, students often leave the school property (e.g., to get lunch, although students in LO Dynów were not allowed to leave school premises during school breaks). A two-metre cut-off for proximate interactions, a standard distance threshold (Fedurek et al., 2021, 2022, 2023, 2024), was used in interaction studies to denote dyadic exchanges. For instance, when one tag comes within a 2 m range of another tag, both devices collect and store contact information of one another. This 2 m threshold has been shown to capture close-range proximity contacts, which are necessary for important face-to-face interactions between students (Fedurek et al., 2022), and studies show that individuals perceive interpersonal distances closer than 2 m as psychologically and biologically threatening (Sorokowska et al., 2017). Research using proximity sensors (e.g., RFID, Bluetooth, or GPS) often employs 1.5 to 2-meter thresholds to define meaningful social encounters or transmission-relevant contacts (Fedurek et al., 2023; Salathé et al., 2010; Stehlé et al., 2011). The B8 Social Distancing Wristbands were worn on both hands to achieve a better range of Bluetooth reception.

Preparation of Proximity Data

During 4 weeks of study, each student wore wristbands for 4 to 8 h daily, depending on the school timetable. However, only data collected during breaks between lessons were considered during analyses. This is because during lessons, students’ in-person interactions were constrained by classroom seating arrangements. In contrast, collecting data during lesson breaks, students were able to freely interact with other classmates, and close-range in-person contacts were more likely to reflect their social preferences. In total, we collected data during approximately 45 h of school breaks across the three schools. The proximity data from lesson breaks were filtered individually for each student, using the timestamps recorded by the wristbands. The average number of minutes per day of all lesson breaks in Dynów, Wrocław and Toruń was 75.

Calculating Strength Parameter

Proximity strength between 2 individuals was calculated as a number of Bluetooth signals detected between these two individuals being within 2 m radius divided by all Bluetooth signals (including signals of the B8 strapped on the other hand of the individual) received by these 2 individuals. Proximity strength of an individual was calculated as a sum of all dyadic proximity strengths with other study participants belonging to the same class. Simply put, individuals with higher proximity scores spent relatively larger amount of time within 2-meter proximity to (one or more) classmates.

Statistical Analysis

We used multiple linear regression to estimate the relationship between proximity strength (the dependent variable) and two predictors, namely, perceived vulnerability to diseases (the total score) and the participant’s sex (as sex is an important factor influencing both PVD and proximity strength; Ackerman et al., 2018; Duncan et al., 2009; Fedurek et al., 2023). Since perceived disease vulnerability may differ between sexes (Díaz et al., 2020) we included the interaction term between PVD and sex in the regression model. We also included study site (i.e., school) as a fixed variable to account for potential non-independence due to clustering within schools and for school-level differences. Sex (a dichotomous variable) was entered as a simple contrast where females were coded as -0.5 and males as + 0.5. In addition, we also calculated two separate multiple regression analyses with the two subscales of PVD (GA and PI) included as predictors (instead of PVD) and reported the results in the Supplementary Material.

Descriptive statistics and regression analysis (lm function) were conducted using the R, version 4.4.0. The marginal effects were calculated and plotted using emmeans and ggeffects (Lüdecke, 2018) packages. In addition, we used the following R packages in the analysis: psych (Revelle, 2020), car (Fox, 2006), while ggplot2 (Wickham et al., 2016), hrbrthemes (Rudis, 2020), extrafont (Chang, 2014) were used to create figures.

Results

Table 1 presents descriptive statistics for the variables included in the model.

Table 1.

Descriptive statistics for the variables in the regression model

Variable	Mean	SD	Min	Max	n; Proportion (%)
Proximity strength	1.320	0.269	0.611	1.996
PVD	3.503	0.890	1.400	6.533
Sex
				Females	58; (69%)
				Males	26; (31%)

The multiple regression model (F(5,78) = 1.577, p = 0.176, adjusted R² = 0.034) showed that none of the predictors significantly predicted proximity strength (Table 2; Fig. 1).

Table 2.

Results of the multiple regression model

Predictor	Estimate (β)	Std. error	t-value (df = 80)	p-value	95% CI
Intercept	1.103	0.138	7.995	< 0.001	[0.829, 1.378]
Sex	-0.318	0.269	-1.179	0.242	[-0.854, 0.219]
PVD	0.039	0.038	1.017	0.312	[-0.037, 0.115]
School 2	0.141	0.073	1.923	0.058	[-0.005, 0.288]
School 3	0.145	0.077	1.887	0.063	[-0.008, 0.298]
Interaction Sex x PVD	0.107	0.081	1.325	0.189	[-0.054, 0.267]

Fig. 1.

Relationship between proximity strength (predicted values adjusted for other variables in the model) and PVD scores as a function of participants’ sex

When the two PVD subscales, Perceived Infectability (PI) and Germ Aversion (GA), were considered separately in the model as predictors (instead of PVD), both PI and GA were non-significant predictors of the proximity strength (PI: β = 0.025, SE = 0.028, t(80) = 0.901, p = 0.370; GA: β = 0.020, SE = 0.030, t(80) = 0.644, p = 0.522, Supplementary Material). Both models were non-significant; PI model: F (5,78) = 1.423, p = 0.225, adjusted R² =0.025; GA model: F(5,78) = 2.234; p = 0.246, adjusted R² = 0.021 Table 1S and 2S).

To further investigate the obtained null results, we conducted post hoc power and sensitivity analyses. Our sample of 84 participants provided approximately 56% power to detect the observed overall effect (R² = 0.092), indicating limited sensitivity to small effects. Sensitivity analyses showed that, with this sample size, the study was adequately powered (80%) to detect medium-to-large effects at the model level (minimum detectable R² = 0.141). For individual predictors, the minimum detectable effect size corresponded to a partial R² = 0.092 (partial r = 0.30), indicating that only medium-sized effects could be reliably detected. Smaller effects would likely remain undetected with the present sample size. Consistent with this, power calculations indicated that detecting the observed overall effect (R² = 0.092) with 80% power would require a substantially larger sample (approximately n = 133), highlighting that the current study may have been underpowered to detect effects of this magnitude.

Discussion

Contrary to our predictions, we found no robust evidence for an association between perceived vulnerability to disease (PVD) and the frequency of proximity contacts. As such, the obtained results are not reliable or strong enough to support a relationship between behavioural immune system (BIS) activation and day-to-day social interactions. In other words, our findings, based on real-world face-to-face proximity patterns, do not provide support for the commonly assumed link between heightened pathogen avoidance and reduced sociality.

From an evolutionary perspective, pathogen-avoidance behaviour is expected to play an important role in shaping social interactions, as individuals who adjust their behaviour in response to disease risk may gain adaptive advantages. Previous research has suggested that BIS activation operates, at least in part, through the reduction of direct social contact (Duncan et al., 2009). Individuals who perceive themselves as more vulnerable to infectious diseases have been shown to exhibit lower levels of extraversion, agreeableness, and openness to experience (Duncan et al., 2009; Murray & Schaller, 2016), and experimentally induced disease salience has been linked to reduced affiliative behaviour (Sawada et al., 2018). However, the absence of a clear association in the present study suggests that these relationships may not generalize straightforwardly to naturally occurring social environments. It is possible that behavioural immune system responses are more sensitive to salient or immediate pathogen cues than to routine, everyday interactions, or that their effects are relatively subtle and therefore difficult to detect in ecologically valid settings.

Our findings also highlight a potential discrepancy between experimental and naturalistic approaches to studying the behavioural immune system. Much of the existing literature relies on laboratory-based manipulations or short-term behavioural responses elicited by pathogen-related stimuli, often involving unfamiliar individuals or hypothetical scenarios (Mortensen et al., 2010; Sawada et al., 2018). In contrast, the present study examines patterns of proximity among familiar individuals in a real-world setting over an extended period of time. The lack of a detectable association between PVD and proximity contacts in this context suggests that the influence of pathogen-avoidance mechanisms on social behaviour may be more context-dependent than previously assumed.

One possible explanation for this discrepancy is that prior studies often prime participants to attend to disease-related cues when assessing PVD, thereby increasing the salience of pathogen threat. Under such conditions, behavioural responses may be amplified, leading to clearer associations between perceived vulnerability and social avoidance. In contrast, our study captures everyday social interactions in the absence of experimentally induced disease salience. This distinction may be important, as previous research has shown that PVD is sensitive to environmental factors such as morbidity and mortality rates (Stefanczyk et al., 2024), particularly during periods of heightened pathogen threat, such as the COVID-19 pandemic (Szymkow et al., 2021). Our findings therefore suggest that, under typical day-to-day conditions, variation in social activity may not be strongly associated with perceived vulnerability to disease.

Although we predicted in our analysis that participants’ sex may moderate the relationship between PVD and proximity contacts, this effect was statistically nonsignificant. Prior work has suggested that men and women may differ in risk-taking behaviour and social network structure (Arch, 1993; Byrnes et al., 1999; Friedl et al., 2020; Vigil, 2007), which could, in principle, influence how pathogen-avoidance mechanisms are expressed. However, the present data do not provide clear evidence for sex-specific effects, and further research with larger samples is needed to examine this possibility.

Several limitations of the present study should be acknowledged. First, collecting proximity data in natural settings is logistically arduous and resulted in a relatively small sample size, particularly among males. As we have shown in the sensitivity analyses, the small sample size resulted in limited statistical power, which reduced our ability to detect small or moderate effects. As such, the absence of a significant association should be interpreted with caution and requires replication in larger samples. Second, the study was restricted to a limited number of school classes, which may constrain the generalizability of the findings. Although proximity contacts in school settings are shaped by structured environments, prior research suggests that most interactions still occur within classes or among students of the same grade (Hallinan & Smith, 1989; Stehlé et al., 2011). Third, the proximity data used in this study reflect physical closeness rather than confirmed social interaction, meaning that some recorded contacts may not correspond to meaningful social engagement. Finally, the use of a convenience sample further limits generalizability, and future studies should aim to include more diverse populations and settings.

Despite these limitations, the present study demonstrates the feasibility of linking real-world proximity data with measures of perceived disease vulnerability, thereby contributing to the growing literature on the behavioural immune system in naturalistic contexts. By moving beyond laboratory-based paradigms, this approach provides a more ecologically valid perspective on how pathogen-avoidance mechanisms may operate in everyday life.

At the same time, it is possible that simple measures of proximity strength may not fully capture the aspects of social behaviour most relevant to pathogen-avoidance processes. Future research could benefit from examining individuals’ integration within proximity networks, such as their centrality, clustering, or embeddedness, which may provide a more sensitive indicator of how face-to-face interactions relate to perceived vulnerability to disease. Such network-level measures may better reflect the structure and quality of social exposure and thus offer a more nuanced understanding of potential links between social interaction patterns and behavioural immune system activation.

In conclusion, our findings suggest that the relationship between perceived vulnerability to disease and everyday social proximity may be weaker or more context-dependent than previously assumed. While behavioural immune system processes likely play an important role in regulating social behaviour, their effects may be contingent on situational factors, such as the salience of pathogen threat. Future research should aim to integrate experimental and naturalistic approaches, using larger samples and diverse contexts, to better understand the conditions under which pathogen-avoidance mechanisms influence human social behaviour.

Supplementary Information

Below is the link to the electronic supplementary material.

Biographies

Piotr Fedurek, PhD,

is an evolutionary anthropologist affiliated with Hirszfeld Institute of Immunology and Experimental Therapy, Polish Academy of Sciences, Wroclaw, Poland. His research focuses on social networks, stress physiology, and wellbeing in small-scale and largescale societies. He combines behavioural, physiological, psychological and network-based approaches to study human sociality and health.

Dariusz P. Danel, PhD, DSc,

is an Associate Professor at the Hirszfeld Institute of Immunology and Experimental Therapy, Polish Academy of Sciences, Wrocław, Poland. His research integrates human evolutionary biology, evolutionary medicine, and behavioural ecology, with particular interests in social cognition, facial and ocular morphology, maternal health, and clinical research methodology. He combines interdisciplinary approaches to investigate questions related to human behaviour, health, and biomedical research. Polish Academy of Sciences.

Author Contributions

PF and DPD equally contributed to the manuscript. PF and DPD designed the study. PF and DPD collected the data. PF conducted preparation of the proximity data. DPD analysed the data. PF wrote the original manuscript. PF and DPD all contributed to writing and editing the manuscript.

Funding

The study was funded by Narodowe Centrum Nauki (ID: DEC-2022/06/X/NZ8/00184) awarded to PF.

Data Availability

Due to the relatively small sample size and concerns about participant anonymity, synthetic data thatreliably reflected the dependencies present in the original dataset could not be generated without increasing the risk of re-identification. Thedata and R code used in this study are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.