Psychological and situational effects on social distancing and well-being during the COVID-19 pandemic: not a question of real risk

Keeping a large interpersonal space (IPS) is one of the most important measures to counter the COVID-19 pandemic. The IPS is automatically modulated according to primary affiliation and defence needs aimed at protecting our physical and psychological well-being. Through a multicentric online survey (1293 respondents) in six Italian regions during the lockdown (April-May 2020), we investigated the psychological and situational factors that influenced the regulation of IPS and psychological well-being. The results showed that the IPS was modulated according to perceived rather than actual risk of COVID-19 infection. This perception was influenced by institutional communication and citizen satisfaction in local Healthcare Systems. Higher levels of anxiety, stress and feeling of insecurity, exacerbated by situational factors linked to the context of life, led to an increase in IPS at the expense of psychological well-being. Instead, the possibility to go out and seeing other people wearing a face-mask reduced the IPS. The findings provide useful insights into pandemic management to bring human behaviour in line with the recommendations of public health experts.


Introduction
Since its first appearance in China, the COVID-19 pandemic has spread worldwide. One of the first countries to be hit after China was Italy. To contrast the pandemic, the whole country was submitted to a period of severe lockdown from March 9 th to May 4 th 2020 (Decree of the Prime Minister #iorestoacasa, 9 March 2020). In line with the World Health Organization (WHO, 2020) guidelines, the Italian government prescribed to keep interpersonal social distances of at least one meter in public spaces, along with wearing face-masks and frequent hand washing. In those days, marked by the number of dead and infected people that the Civil Protection communicated daily, there was a social climate of anxiety, concern and awareness of the need for confinement. The measure has been quite effective since the epidemic curve showed an increasing trend until March 20 th 2020 followed by a constant decrease (National updating, ISS, May 14 th 2020).
What can this lockdown tell us about how to manage a pandemic?
We are now aware that avoiding crowds while maintaining a wide social distance is a fundamental behaviour to slow down the pandemic (Arslan et al., 2020;López & Rodó, 2020) (for a review, Chu et al., 2020). The regulation of social distance from other people is automatically modulated in our daily life both by primary affiliation needs (e.g. individuals approach people if they feel lonely or need care) and by defensive needs (individuals keep risks or threats at a distance). Social psychology uses the term "personal space" to define an emotionally tinged area around our body that individuals feel like their private space and where any unwanted intrusion by others causes us discomfort, anger or fear (Hall, 1966;Hayduk, 1983;Lourenco et al., 2011). The regulation of this space between individuals is called "interpersonal space" (IPS; e.g., Hall, 1966;Hayduk, 1983;Sommer, 2002). Typically, individuals regulate IPS through two basic behaviours: people extend the distance when they feel in dangerous and uncomfortable situations (avoidance behaviour) and, vice-versa, people shrink the distance when they feel in friendly and safe situations (approach behaviour) (Hall, 1966;Holt et al., 2014;Kennedy et al., 2009;Lourenco et al., 2011;Sawada, 2003).
Proxemics literature has shown that IPS can be modulated by several psychological and situational factors (for reviews Bell et al., 2005;Hayduk, 1983;Lloyd, 2009;Sommer, 2002). In general, according to Hall's classic model (1966), the distance during more formal interactions (or social distance, up to 300 cm) is larger than the distance during more intimate interactions (or personal distance, up to 120 cm). Moreover, people prefer a larger IPS from males than females, and a larger IPS as the age of the interactants increases (Gifford, 1997;Gioiella, 1977;Hayduk, 1983;Iachini et al., 2016;Latta, 1978;Mishara et al., 1974;Winogrond, 1981). The size IPS also increases with higher anxiety levels (Brady & Walker, 1978;Iachini et al., 2015;Kaitz et al., 2004;Layden et al., 2018) (for a review see Hayduk, 1983). Interestingly, the IPS is larger when people feel a high perceived risk in the presence of novel threats that cannot be controlled (e.g., Adams, 1995;Bish & Michie, 2010;Kappers et al., 2018;Wise et al., 2020). In this framework, how do individuals regulate the IPS if the threat is invisible? What are the consequences for our psychological well-being of having to maintain a larger IPS than we naturally implement?
To answer these questions, we tried to identify which psychological and situational factors contributed to determine the size of the IPS during the lockdown to counter the COVID-19 outbreak in Italy (in the period between April 23 rd and May 2 nd 2020). The Italian case is also interesting because in that period the severity of the pandemic changed strongly depending on the territorial areas. Here we focus on six regions that reflected different levels of severity of the pandemic: very high (Lombardy), medium-high (Veneto and Emilia-Romagna), medium-low (Campania and Apulia), very low (Sardinia). As shown in a report on patients who died in Italy based on data of July 22 nd (Surveillance Group, ISS, 2020), the total number of deaths was 34.142, of which: Lombardy 49.1%, Emilia-Romagna 12.5%, Veneto 5.8%, Apulia 1.6%, Campania 1.3%, Sardinia 0.4%.
Our study therefore explores how the feeling of threat and perceived risk generated by the outbreak of COVID-19 infection affected the regulation of IPS. More specifically, we aimed at understanding: 1) the effect of people's age and gender, and of the face-mask wearing on the IPS; 2) the effect of perceived and actual COVID-19 risk in the six regions on IPS; 3) the factors that may have influenced perceived risk (e.g. actual risk, institutional communication on social media, satisfaction in the Healthcare Systems); 4) the effect of psychological reactions of anxiety and stress, sense of safety and social isolation on the IPS; 5) the effect of various situational factors (such as confinement days, cohabitants, housing situation and outings) on the IPS and psychological well-being.
To meet these aims, we presented an on-line survey to a sample of female and male participants (N= 1293) of different ages from the six regions representative of the Italian situation. The survey comprised a projective test on interpersonal social distance, that is the Interpersonal Visual Analogue Scale (IVAS, Iachini et al., 2016), standardized tests of anxiety (State-Trait Anxiety Inventory, STAI) and perceived stress (Perceived Stress Scale, PSS), and ad hoc questionnaires on various psychological reactions and situational factors during lockdown. The survey data were combined with objective data obtained from government sources, such as the actual risk of infection by region, institutional communication on social media and the degree of satisfaction of local health systems.
another person (Confederate). The confederate could be a male or female child, or young adult or elderly. Participants had to indicate, by moving a slider, their preferred interpersonal distance from the confederates (from 100 (max distance) to 0 (no distance)).
Interpersonal distance with or without face-mask. Participants had to indicate what was the distance from others that made them feel comfortable with or without face-mask. For this measure, two items were used: 1) "With mask"; 2) "Without mask". Participants had to indicate the distance on an 8-point Likert scale, ranging from 1 = ".50 meter" to 8 = "4 meters".

Perceived risk
Four questions investigated how risky people perceived the COVID-19 virus. The questions concerned: (i) the General perceived risk, i.e. how much dangerous people considered the COVID-19 per se, independently of their contingent situation, and (ii) the Local perceived risk, i.e. how much dangerous people considered the COVID-19 in the place where they lived. In both cases people had to relate their beliefs to two times: in the present and in the future. Scores could vary from 100 (max risk) to 0 (no risk)

Psychological factors
STAI. The State-Trait Anxiety Inventory (STAI, Spielberger, 2010; see also Julian, 2011) was used to measure State (S-Anxiety) and Trait anxiety (T-Anxiety) of participants. Range of scores for each sub-scale is 20-80, the higher score indicating higher anxiety. A cut-off of 39-40 is indicated to detect clinically significant symptoms on the S-Anxiety scale (Addolorato et al., 1999;Knight et al., 1983).
Perceived Stress Scale. The perceived stress of participants was assessed using the Italian version of the Perceived Stress Scale (PSS) (Cohen et al., 1983;Cohen, 1988) (Italian translation Fossati, 2010). Individual scores can range from 0 to 40: scores ranging 0-13 would indicate low stress, scores ranging 14-26 moderate stress, scores ranging 27-40 high perceived stress.
Psychological reactions to lockdown. Participants' perceived isolation and feeling of safety inside home and outside in the external space was measured with three ad hoc questions on a five point Likert-type scale.

Situational factors
As regards the living situation during the lockdown, participants had to indicate their marital status (without or with a partner), having or not children, the context of residence (city, village, countryside), the number of cohabitants and the number of rooms in their house. Moreover, they had to indicate how many days they had been locked up at home until the date of filling in the survey and how many times they had left home.
All further details about the tests and items used in this study are provided in Appendix 1.

Procedure
PsyToolkit was used for the online survey (Stoet, 2010(Stoet, , 2017. After reading the instructions and digitally signing the informed consent, participants started the survey. All participants answered the items and filled in the questionnaires in the following order: 1) demographic information; 2) situational factors; 3) psychological reactions to the lockdown experience and perceived risk; 4) State Anxiety Inventory; 5) IVAS; 6) Perceived Stress Scale; 7) Items about social distance with or without face-mask; 8) Trait Anxiety Inventory.
The time taken to complete the questionnaire ranged from 20 to 25 minutes.

RESULTS
A series of Analyses of Variance (ANOVAs) and regression analyses were carried out. In five sessions, the analyses assessed: 1) The determinants of IPS (the effects of gender and age, the effect

1) The determinants of IPS a) The effects of gender and age
A 2X2X3 mixed ANOVA was used to analyze the distance in mm (as dependent variable) with Participants' Gender (M/F) as between factor, Confederates' Gender (M/F) and Confederates' Age (Children/Young Adults/Elderly) as within factors. comparisons were significant, at least p < .0001. These main effects were qualified by a significant participants' Gender x Age interaction, F(2, 2582) = 4.02, p < .05; ƞ 2 p = .003. The preferred distance from elderly people of female participants was larger than all other conditions (at least p < .0001) (descriptive statistics in Appendix 2- Table 8). Moreover, a confederates' Gender x Age interaction (F(2, 2582) = 5.64, p < .005; ƞ 2 p = .004) revealed that IPS was shorter from young females than males (p < .0001) and there was a similar tendency in elderly people (p = .053), whereas there was no difference between female and male children (p = 10) (descriptive statistics in Appendix 2- Table 9).
Finally, the reliability of the measure was assessed: Cronbach's alpha= .97, inter-item correlation= .87. Considering the strong inter-item correlation and reliability of the six interpersonal distance dimensions, a mean interpersonal distance was calculated and used for subsequent analyses.

b) The effect of face-mask wearing
A further ANOVA was carried out to see the effect of face-mask on social distancing in meters (i.e. no-mask vs mask). The results showed that wearing a face-mask affected interpersonal distance: A Pearson correlation analysis showed a strong positive correlation between the preferred distances in meters and IVAS: all IVAS distances increased with the increase of preferred distances in meters with and without face-mask (with at least r > .28 and p < .0001).

2) The effect of perceived and actual COVID-19 risk in the six regions on IPS a) IPS in the six regions
A one-way ANOVA with a 6-level between-factor Region on mean IVAS distance showed a main effect of this factor, F(5, 1287) = 5.32, p < .0001, ƞ 2 p = .02. As illustrated in Figure 1, interpersonal distance was shorter in Emilia-Romagna and Veneto (i.e. regions with medium-high actual risk of  and similar in other regions (further descriptive statistics in Appendix 2- Table 10). The post-hoc analysis showed that IPS was unexpectedly and significantly shorter in Emilia Romagna and Veneto than two low-risk regions (Puglia and Sardinia) (at least p< .05). We reasoned that this result could be affected not so much by the actual risk in the various regions but by how people perceive the risk related to COVID-19.

b) The effect of Actual/Perceived COVID-19 risk on IPS
We first computed a comprehensive index of the Actual risk of COVID-19 in the six regions where the respondents lived. The scores of seven dimensions of the severity of the infection (number of positive subjects, number of patients in intensive care, number of all hospital discharges (inverted), number of deaths, total cases, increase of total cases compared to previous day, number of swabs) (Source: Istituto Superiore di Sanità, 2020) for each day included in the data collection period were recorded. These values were averaged across days in relation to each region. The correlations between the seven dimensions were high, ranging between .75 and .98. To obtain a composite and unique measure of the impact, a Principal Component Analysis was conducted. A single-factor solution was extracted (which explains 94% of the variance, with Cronbach's alpha of .98) and a factor score (method: regression) was calculated. Distances expressed by the factor score can be interpreted as z scores. The composite indexes of Actual risk were the following: Lombardy= 3.830, Emilia-Romagna= .977, Veneto= .661, Campania= -.331, Apulia= -.366, Sardinia= -.531. Clearly, higher indexes indicate a stronger impact of COVID-19. These indexes were used as predictor in regression analyses (see Figure 1).
To understand the effect of Actual and Perceived risk on IPS, a regression was performed with mean interpersonal distance as criterion and the following predictors: Actual risk and the four dimensions of the Perceived risk (Present-General, Present-Local, Future-General, Future-Local).
The overall model was significant: F(5,1287) = 17.703, p < .000, R = .25, R 2 = .06. However, as shown in Table 1, only the Present-General Perceived risk gave a significant contribution to the model: the higher the perceived risk of COVID-19 in the present at general level the larger the social distance (see also Figure 1). CIs are reported.

3) The determinants of Perceived risk in the six regions
Different regression analyses explored the impact of Actual risk, Social Media institutional communication, Healthcare System Satisfaction on the Perceived risk.

a) The effect of Actual risk
A regression analysis model with Actual risk in the six regions as predictor and the four dimensions of Perceived risk as criteria was carried out. The multivariate analysis showed that the overall model was significant F(4,1288) = 54.848, p < .0000, Wilks = .854. The Actual COVID-19 risk had a significant effect on all dimensions of Perceived risk, but it was dramatically stronger on the perception of the dangerousness of COVID-19 in the Present at Local level. Therefore, the more severe the actual risk the higher the local perceived risk of COVID-19 (see Table 2). Beta Coefficients, 95% CIs, t and significance levels for predictor of each outcome variable. Rsquared is also reported.

b) The effect of Social Media communication
The previous regression suggested that the Actual risk played only a marginal role in the way people perceived the risk of COVID-19 in general that, instead, seems to play a fundamental role in IPS regulation. We hypothesized that the social representation of General Perceived risk could be influenced by media communication. During the lockdown, each regional government dedicated a specific space for pandemic-related communication on its institutional websites and made extensive  Table 3).  Errors, Standardized Beta Coefficients, 95% CIs, t and significance levels for predictor for each outcome variable. R-squared is also reported.

c) The effect of Healthcare Satisfaction
The Italian Healthcare Service has a regional structure. General level in the Present and Future, and at Local level only in the Future (see Table 4). In these cases, the more respondents were satisfied with their Healthcare system the lower the perceived COVID-19 risk. variable. R-squared is also reported.  Table 5).

4) The effect of psychological reactions on IPS
The more people felt anxious and stressed for the present situation the larger the IPS. Instead, Trait anxiety had a weak effect with higher levels linked to shorter IPS. Finally, the more people felt safe outside in the external space and the more they felt isolated, the shorter their preferred IPS.   Table 6).  We also explored the possible effect of the situational factors on the psychological reactions. We used the same factors above described as predictors and perceived Safety Outside and Inside, perceived Isolation, State and Trait Anxiety and Perceived Stress as criteria.
In particular, people with a partner reported lower levels of state and trait anxiety and perceived stress, felt safer at home and less isolated. People with children reported less anxiety, perceived stress, and felt less isolated. State anxiety and perceived stress increased and the feeling of safety inside and outside decreased with more cohabitants and fewer rooms available. Instead, the more people went outside the home, the lower the levels of stress and anxiety and the higher sense of safety in the external space. The context of residence and the days of confinement at home did not exert significant effects.

Discussion
Our work focused on the relationship between the sense of threat and perceived risk of  and the regulation of interpersonal social distance. For this reason, we investigated how the dangerousness of the virus was individually and socially represented. Moreover, to understand to what extent our behaviour was influenced by risk perception rather than factual information (Bavel et al., 2020), we conducted a multicentric study in six Italian regions with different degrees of COVID-19 infection severity.

The determinants of IPS
The regulation of IPS in normal social life is particularly sensitive to the socio-emotional content and the potential threats that people perceive in their context (Argyle & Dean, 1965;Dosey & Meisels, 1969;Graziano, 2017;Ruggiero et al., 2017). The fear generated by the pandemic influenced the regulation of IPS. In fact, the interpersonal distance reported by respondents was dramatically larger (i.e. 54.86 mm) than that reported in a pre-COVID-19 study (frontal approach: 17.76 mm) that used the IVAS test .
In line with previous proxemics literature, the results showed that IPS was affected by people's age and gender (Camperio & Malaman, 2002;Gioiella, 1977;Hayduk, 1983;Iachini et al., 2016;Latta, 1978;Remland et al., 1995;Sommer, 2002;Uzzell & Horne, 2006;Winogrond, 1981). The distance was wider from male than female people, and from elderly people than young adults and children. Women preferred a larger distance than men, especially from elderly people. This might reflect a higher sensitivity of women to the risk-factor represented by elderly age in the current pandemic.
Another effect related to the pandemic is that IPS was reduced by about 1 meter when other people wore a face-mask compared to when they did not. Presumably, wearing face-masks allowed people to feel safer and this led them to reduce their interpersonal distance. This reveals a worrying aspect of people's behaviour, as the fact of wearing the face-mask does not prevent contagion if it is not associated with social distancing (WHO, 2020).
Finally, the strong correlation between all IVAS distances and the metrics related to interpersonal distances with/without face-mask reinforced the reliability of the IVAS projective test.

COVID-19 risk and IPS in the six regions
As illustrated in Figure 1, a comparison among the six Italian regions shows that IPS was shorter in Emilia-Romagna and Veneto (i.e. regions with medium-high actual risk of COVID-19) and similar in other regions. The difference was statistically significant with regions characterized by mediumlow (Apulia) and low (Sardinia) COVID-19 severity index. This result is explained by the fact that IPS was not predicted by the actual risk of COVID-19 infection in the regions, but by how people perceived at general level the risk of COVID-19. Precisely, the higher the perceived risk in the present at a general level, the larger the IPS. In line with proxemics literature, the finding confirms that perceived threat is one of the most salient factors in mediating equilibrium between interpersonal distance and social interaction (Argyle & Dean, 1965;Dosey & Meisels, 1969;Horowitz et al., 1964).
The results related to perceived risk are important for two main reasons. First, they suggest that people perceive the local threat less immediate than the general threat. This could be a matter of psychological distance (McDonald et al., 2015) or reflect a kind of optimistic bias of a spatialgeographical nature, as emerged in studies where the perception of environmental problems increases from the local to the global level (Gifford et al., 2009;Schultz et al., 2005). Consistently, shocking uncontrolled events (e.g., terrorism or earthquake) (Sjöberg, 2000; see also Becker & Janz, 1987) polarize people's attention towards general risks that are "far" from them (e.g., Adams, 1995;Bish & Michie, 2010;Kappers et al., 2018;Spence et al., 2011;Wise et al., 2020). Second, they show that it is the way people perceive the risk of COVID-19 in general and in the present time to influence IPS modulation. This perception reflected an "abstract" representation of the dangerousness of the COVID-19 regardless of the local context in which the respondents lived. For this reason, it was important to understand which factors could influence the perception of the COVID-19 risk.

The determinants of Perceived risk in the six regions
During the lockdown in Italy, the entire media system and the communication by the institutions was largely focused on the outbreak of COVID-19. In the areas with the highest impact of the virus, the media communication was accompanied by direct experience of the pandemic. We thus explored the possible impact of actual risk, social media institutional communication and Healthcare System satisfaction on the perception of the COVID-19 risk.
As regards the actual risk in the six regions, the results showed that it strongly influenced the perceived COVID-19 risk at a local level in the present: the higher the actual severity, the higher the perceived local risk. On the other hand, social media communication only influenced the COVID-increased, the general perceived risk increased. Moreover, higher levels of satisfaction in Healthcare Systems led to a reduction of the general perceived risk. Notably, satisfaction was higher in Veneto and Emilia-Romagna where social distances were reduced compared to other regions. In addition, a qualitative analysis of social media content showed that the Emilia-Romagna Facebook page supported an optimistic view ("#coronavirusinsiemesipuòbattere") compared to other regions.
Moreover, the Campania Facebook page had more comments than other regions. A qualitative analysis of its content revealed a considerable effort to provide updates on the pandemic situation and clarification of what was allowed and what was not, often accompanied by a communicative style that highlighted the risk of COVID-19.
In short, the results show that the general perceived risk in the present was influenced by institutional communication on social media and satisfaction in regional Healthcare Systems. The combination of these factors contributed to the individual and social perception of the dangerousness of the virus. In turn, this general perception influenced the modulation of interpersonal distance.

Psychological reactions
During a crisis, people normally feel psychological reactions of fear, stress and anxiety (Adhanom Ghebreyesus, 2020;Bavel et al., 2020). Not surprisingly, 73% of respondents exceeded the cut-off of 39 for state anxiety (about 29% of respondents the cut-off of 54), while about 60% of respondents showed moderate perceived stress and 16% a high perceived stress.
The results showed that the more anxious, stressed and unsafe in outside space people felt, the larger the IPS was. In line with the literature, these negative feelings induced an increase in social distance during potentially threatening interpersonal situations (Brady & Walker, 1978;Iachini et al., 2015;Kaitz et al., 2004;Layden et al., 2018; for a review see Hayduk, 1983). Instead, the sense of isolation had the opposite effect: the more isolated people felt, the shorter the IPS. This can be explained by the fact that people who feel more isolated have a greater need for social affiliation (Taylor, 2006(Taylor, , 2011 which is a basic human response to many stressful circumstances (Taylor, 2009).

Situational factors
As regards the situational factors, we found that having a partner and children made the lockdown more bearable and therefore mitigated the levels of anxiety, stress, insecurity and the feeling of isolation. Instead, having more cohabitants and fewer rooms available increased anxiety, stress and the feelings of insecurity inside and outside home. This highlights the importance of having living spaces that are not overcrowded but adapted to the needs of the inhabitants (Cohen 2020).
Finally, we found that the more people went outside the home, the lower the levels of stress, anxiety and insecurity in external space, and consistently the size of the IPS decreased. It is possible that the frequent outings allowed people to feel that the threats in the external world could be controlled and this reduced their anxiety and fear for the present situation. Conversely, the less people went out the more the external world looked threatening and interpersonal distances reflected this higher sense of unsafety.

Conclusions
IPS regulation during the Italian lockdown was based on a balance between defence needs generated by feelings of anxiety and threat that led to enlargement, and affiliation needs generated by social isolation that led to shrinkage. This mechanism was oriented by the degree of general perception of risk: a low degree due to confidence in Healthcare Systems was associated to shrinkage, while a high degree due to clear and timely institutional communication was associated to enlargement.
The lockdown to contrast the COVID-19 pandemic contributed to exacerbate the levels of anxiety, stress, insecurity and isolation to the detriment of individuals' psychological well-being (Brooks et al., 2020). On the other hand, the possibility of going outside, in compliance with safety requirements, improved people's psychological reaction. This could be very important for social categories that resulted particularly fragile: people without partners and without children, people who lived in small houses and were forced to share their living space with more cohabitants.
In conclusion, the institutional communication should provide clear information on the dangerousness of the infection and at the same time highlight that it can be controlled by following the health instructions (Kazak, 2020;Bavel et al., 2020). In this way, they could contribute to building a perception of risk that is functional to the adoption of effective mitigation strategies (Harper et al., 2020). From a theoretical point of view, it is the first time that the defensive function of IPS encoding is studied when other people are seen as a potential threat regardless of their willingness to harm, as possible virus spreaders. This confirms that the defensive function of the IPS is alerted by a potential threat. This early alert allows us to prepare appropriate behavioural reactions to protect our physical and psychological well-being. 2. Interpersonal distance with or without face-mask. Participants had to indicate what was the distance from others that made them feel comfortable with or without face-mask. For this measure, two items were used: 1) "With mask"; 2) "Without mask". Participants had to indicate the distance on an 8-point Likert scale, ranging from 1 = "0.50 meter" to 8 = "4 meters".

Perceived risk
Four questions investigated how risky people perceived the COVID-19 virus: a) How dangerous do you consider the coronavirus in general in the present? b) How dangerous do you consider the coronavirus in general in the future? c) How dangerous do you consider the coronavirus in the area where you live in the present? d) How dangerous do you consider the coronavirus in the area where you live in the future? Participants had to indicate, by moving a slider, their degree of risk perception. Scores could vary from 100 (max risk) to 0 (no risk).

Psychological factors
STAI. The State-Trait Anxiety Inventory (STAI, Spielberger, 2010; see also Julian, 2011) is a selfreport inventory based on a 4-point Likert scale to measure the presence and severity of current symptoms of anxiety and a generalized propensity to be anxious. More specifically, State anxiety (S-Anxiety) regards our feeling about an event, and Trait anxiety (T-Anxiety) is concerned with a personal characteristic. The S-Anxiety scale consists of twenty items that assess the current state of anxiety by asking how participants feel "right now, at this moment", using items that measure feelings of apprehension, tension, nervousness, worry and so forth. The T-Anxiety scale assesses relatively stable aspects of "anxiety proneness," such as general calmness, confidence and security. It has proven useful for identifying persons with high levels of neurotic anxiety. Each STAI item is given a weighted score of 1 to 4. Range of scores for each sub-scale is 20-80, the higher score indicating higher anxiety. A cut-off of 39-40 is indicated to detect clinically significant symptoms on the S-Anxiety scale (Addolorato et al., 1999;Knight et al., 1983). Some studies with elderly subjects suggested a higher cut-off of 54-55 (Kvaal et al., 2005). Perceived Stress Scale. Perceived stress was assessed using the Italian version of the Perceived Stress Scale (PSS), the most widely used psychological instrument for measuring the perception of stress (Cohen et al., 1983;Cohen, 1988) (Italian translation Fossati, 2010). The instrument was developed to measure the degree to which situations in one's life are perceived as stressful.
Psychological stress can be defined as the extent to which persons perceive that their demands exceed their ability to cope. The PSS consists of 10 items rated on a 5-point Likert scale, ranging from 0 = "never" to 4 = "very frequently". These items are related to one's feelings and thoughts during the last month and participants have to indicate how often they felt or thought a certain way. Individual scores can range from 0 to 40 with higher scores indicating higher perceived stress in response to stressful situations: scores ranging 0-13 would indicate low stress, scores ranging 14-26 moderate stress, scores ranging 27-40 high perceived stress. Psychological reactions to lockdown. With regard to the psychological reactions to the experience of lockdown, an ad hoc self-report questionnaire was devised with a five-point Likert-type scale (1= not at all, 5= very much). The questions were: a) How isolated do you feel? b) How safe do you feel at home? c) How safe do you feel outside home?)

Situational factors
As regards the living situation during the lockdown, participants had to indicate their marital status (without or with a partner), having or not children, the context of residence (city, village, countryside), the number of cohabitants and the number of rooms in their house. The latter were asked by two open-ended questions: a) "How many people live in your home, including you?"  Table 9. The Table reports the preferred average interpersonal distance (IPS) of the participants measured through the IVAS according to the gender of the confederates and the age of the confederates. M= mean; SD= standard deviation; +/-95% CI= confidence interval; N= number of participants.