Not all exclusions are created equal: effects of physical and social components of exclusion on well-being, prosocial, and antisocial behavioral intentions

ABSTRACT

Over decades, research has shown that people are negatively affected by social exclusion. However, no experimental research has examined the effects of physical exclusion and how it might combine and interact with social exclusion. Across two studies (N = 1,238), we manipulated both components of exclusion separately via an imaginative scenario paradigm and measured their impact on established social exclusion outcomes. Social exclusion, irrespective of being paired with physical in- or exclusion, affected individuals more negatively than mere physical exclusion. Social inclusion was not able to reduce the impact of physical exclusion per se but to protect overall levels of well-being to some extent. Our findings add to a more nuanced understanding of social and physical aspects of exclusionary experiences.

KEYWORDS:

• Social exclusion
• ostracism
• physical exclusion
• physical contact
• well-being

Since its outbreak in 2019, the COVID-19 pandemic has been related to a range of government controls and restrictions necessary to slow and inhibit the spread of the virus. To a large extent, these strategies consisted of physical isolation and distancing; that is, people were urged to physically isolate themselves as much as possible. Since then, a range of scholarly work has linked these public health measures to increased social isolation and loneliness, negatively impacting individuals’ well-being (e.g., Banerjee & Rai, 2020; Pietrabissa & Simpson, 2020).

However, as has been noted, physical isolation can be but is not necessarily connected to social isolation (Aminnejad & Alikhani, 2020). Specifically, it was recommended to use technology-mediated communication (e.g., phone, social media) to maintain social connection aside from physical contact (World Health Organization, 2020). Moreover, it was suggested to use the term ‘physical distancing’ rather than ‘social distancing’ to prevent people from feeling left alone, ignored, and societally excluded (Wasserman et al., 2020). Indeed, social use of interpersonal media during the pandemic was related to reduced feelings of loneliness and increased life satisfaction (Choi & Choung, 2021). Additionally, the frequency of computer-mediated communication in early phases of the pandemic was related to more positive affect (Shufford et al., 2021), and individuals with more satisfactory online contact showed lower levels of emotional distress and feelings of isolation (Juvonen et al., 2021). Thus, it seems that the extent of negative consequences of physically limited interactions might depend on how individuals feel socially included outside the scope of direct physical contact.

Threats to belonging – physical vs. social exclusion

As social beings, humans possess a basic need to belong (Baumeister & Leary, 1995), and threats to this basic need, that is, exclusion, can take different forms. In general, exclusion can be defined as being kept apart from others physically or emotionally (Riva & Eck, 2016). While both types of exclusion have been similarly referred to as social exclusion (see Wesselmann, Grzybowski, et al., 2016), research related to this term has focused heavily on the emotional component, which covers behaviors that communicate individuals they are not wanted or valued as social relationship partners or group members, whether in the form of direct negative attention (i.e., rejection) or being ignored (i.e., ostracism; Riva & Eck, 2016). Accordingly, to be able to draw a better distinction between the two components, we will term emotional forms of exclusion (being rejected or ignored) as social exclusion in the narrower sense and refer to the physical component (being kept physically apart from others) as physical exclusion.

Regarding the social component of exclusion, a long and rich line of research exists. This line of research routinely shows that social exclusion is highly aversive for humans, as illustrated by decrements in well-being. Following the temporal need-threat-model (Williams, 2009), individuals initially respond with social pain reactions in a first reflexive phase, such as a decline in mood (e.g., Gerber & Wheeler, 2009) and a deprivation of the basic psychological needs for belonging, control, self-esteem, and meaningful existence (e.g., Williams, 2009; Zadro et al., 2004). In a subsequent reflective phase, individuals are assumed to engage in a deeper evaluation of the exclusionary experience, allowing strategies to fortify threatened needs to develop (Williams, 2009). Within this process, social exclusion can imply destructive consequences not only on an individual but also interpersonal level. Substantial evidence suggests that social exclusion can fuel antisocial tendencies toward the source of exclusion but also uninvolved third parties (e.g., Twenge & Baumeister, 2005; Twenge et al., 2001). While social exclusion seems to heighten affiliative tendencies such as increased mimicry (Lakin et al., 2008) or conformity (DeWall, 2010), research suggests that it has a dampening effect on prosocial behavior in the narrower sense, that is, behaviors aimed at providing a benefit to others (see Quarmley et al., 2022). Finally, if social exclusion pertains chronically, individuals might enter a last stage of resignation, associated with long-term effects such as alienation or depression (Riva et al., 2017; Rudert et al., 2021; Williams, 2009).

Research that explicitly focuses on the physical component of exclusion, in contrast, seems scarce. In organizational research, qualitative and correlational studies indicate that physically separated workplace arrangements and limited face-to-face interactions can be associated with negative work-related attitudes. For example, physical isolation (i.e., having no physical contact with coworkers) is related to lower perceived respect and organizational identification (Bartel et al., 2012). However, by their very nature, most of these studies have focused on work-related effects rather than individual, socio-emotional consequences of physical exclusion. Within social exclusion research, the physical component has most often been subsumed under research on social isolation and loneliness (i.e., the subjective feeling of being socially isolated). For this special type of exclusion, similar adverse effects have been documented. For example, (perceived) social isolation is connected to a higher risk for depression and levels of hostility (e.g., J. T. Cacioppo et al., 2006). Moreover, socially isolated individuals show immune impairments, less inflammatory control (Hawkley & Cacioppo, 2010), and have higher mortality rates, even after controlling for other health-related risk factors (Holt-Lunstad et al., 2015). One article reviewed the effects of perceived social isolation, ostracism, and rejection on the brain, suggesting that at least neural correlates seem to differ for these forms of exclusion (S. Cacioppo & Cacioppo, 2016). However, although social isolation might incorporate the physical component of exclusion, it usually goes beyond that. Despite its varying definitions, a key point of (perceived) social isolation is an objectively given or subjectively experienced general lack of meaningful social contact and support (e.g., De Jong Gierveld et al., 2006; Holt-Lunstad et al., 2015), making it unclear whether the negative effects of (perceived) social isolation stem from a lack of physical contact, a lack of social contact or a combination of both.

It seems reasonable that being physically separated from others against one’s preference could negatively affect well-being in itself. In evolutionary terms, social affiliation was tied to physical copresence. Similar to social exclusion, physical exclusion might trigger distress within an evolutionary anchored warning system (Williams & Nida, 2011), as being physically isolated from others automatically translated into social isolation as well, a condition in which there was little chance of survival. Nowadays, however, this is no longer necessarily the case. The advent of various communication channels such as telephony or the Internet has allowed social exchange with distant others such that one’s social inclusionary status is not entirely bound to physical copresence. Certainly, physical interactions can create more ‘social presence’ (i.e., the salience of interaction partners and social relationships; Short et al., 1976) and are related to higher levels of intimacy (Scott et al., 2006). Therefore, face-to-face interactions probably add in a unique way to one’s belonging needs (see Sacco & Ismail, 2014) and will affect individuals negatively when they are missing. However, without the socially threatening component, physical exclusion may elicit negative but still less severe reactions compared to those known from social exclusion research.

In line with this, scholars have recently called for a more nuanced distinction between physical and socio-emotional components of exclusionary experiences (e.g., Goh et al., 2022; Peng et al., 2022; Sørensen et al., 2021). Indeed, when considering both components of exclusion, it becomes evident that different combinations of their qualities create unique exclusionary experiences. For example, one can be physically separated from others but still feel socially included or vice versa; merely being physically surrounded by people does not protect one from being socially excluded. A recent study from organization research even found physical and socio-emotional workplace isolation to be uncorrelated in a sample of telecommute workers (Wang et al., 2020). However, we are unaware of studies that have tested how both components of exclusion or their combinations compare in their effect on well-being.

Additionally, studies conducted during the COVID-19 pandemic give ground to reason that effects of physical exclusion could be dependent on feelings of social connectedness. One study that differentiated social isolation into a physical and digital component found that loneliness only rose among those with limited digital contact to close others. In contrast, those with low levels of digital isolation experienced no increase in loneliness during the pandemic (Peng et al., 2022). During a phase of severe lockdown restrictions, online contact could further protect well-being of those lacking face-to-face contact (Marinucci et al., 2022). Additionally, phone communication during the pandemic was related to higher well-being via feelings of relatedness, though, as found in this study, providing no compensation for face-to-face interactions (Rudert & Janke, 2022). Thus, in line with our reasoning that physical exclusion might affect individuals negatively but less so when the threatening social component is missing, it seems that the availability of social contact could buffer negative effects of physical exclusion, at least to some extent.

Although insightful, these studies did measure the absence or presence of technology-mediated communication rather than assessing levels of physical and social exclusion. The pandemic was further a global state of emergency. Individuals were exposed to numerous stressors besides a lack of physical contact (e.g., health-related worries, financial concerns, social tensions) that presumably affected well-being, suggesting a large pool of potentially confounding variables. Moreover, previous findings in this area are correlative, making it difficult to draw causal conclusions.

The current research

Taken together, although the definition of social exclusion incorporates both socio-emotional and physical forms of exclusion, it seems that the physical component has received rare attention in social exclusion research. Within the current research, we aim to address two research gaps resulting from this. First, as studies conducted during the COVID-19 pandemic suggest that the negative impact of restricted physical contact might depend on how socially included people feel, we aimed to experimentally test the assumption of such an interaction effect under basic circumstances. Building on the findings of the COVID-19-related studies (e.g., Peng et al., 2022; Rudert & Janke, 2022), we reasoned that physical exclusion negatively affects well-being but less so when the threatening social component is missing (i.e., if individuals are not socially excluded). Accordingly, we predicted that physical exclusion (vs. physical inclusion) would lower well-being but that the effect would be less pronounced when individuals have neutral social contact or are socially included (vs. socially excluded; Hypothesis 1).

While testing this interaction provides insights whether the difference in well-being between physical in- and exclusion varies across individuals’ social inclusionary status, it remains unclear how different combinations of physical and social exclusion affect overall levels of well-being. Yet, specific combinations of both components should create unique exclusionary experiences and might therefore have differential effects on well-being. Thus, as a second major goal, we aimed to assess how different combinations of social and physical exclusion would compare in their effect on well-being. Given our reasoning that the social component should be more threatening than the physical component, we expected that mere social exclusion (i.e., physical inclusion paired with social exclusion) would have more negative effects on well-being than mere physical exclusion (physical exclusion paired with neutral or inclusionary social contact; Hypothesis 2). Additionally, building on the COVID-19-related studies (e.g., Marinucci et al., 2022; Rudert & Janke, 2022), we expected that physical exclusion would have more negative effects on well-being when individuals are additionally socially excluded (i.e., combined exclusion), compared to when they have neutral social contact or are socially included (Hypothesis 3).

Across two studies, we experimentally manipulated both components of exclusion separately within a scenario-imaginative paradigm and measured their impact on established outcomes from social exclusion research. To capture the relevant facets of well-being, we assessed impairments in mood and basic need-threat as well-known reflexive pain reactions from social exclusion research (Bernstein, 2016; Gerber & Wheeler, 2009). We further included loneliness as another facet of well-being to cover previous research on social isolation, which has included the physical component of exclusion to a greater extent.

Additionally, on an explorative basis, we included prosocial and antisocial behavioral intentions to explore whether physical exclusion and its interaction with social exclusion would not only affect well-being but also might further have interpersonal consequences known from the reflective stage of social exclusion research (Baumeister et al., 2007; Quarmley et al., 2022).

The project was acknowledged by the University’s ethical board (approval number 2022–034). Materials and data are openly available at: https://osf.io/7gph9/?view_only=3eb45d46d0e9439f92d94da4e8037577

Study 1

Method

Participants and design

Assuming a small to medium effect size (f = 0.17), an a priori power analysis (α = .05; 80% power) via G*Power (ANOVA: Fixed effects, special, main effects, and interactions) suggested a sample size of 337 participants to detect an interaction effect in a 2 × 3ANOVA. We overrecruited participants to be able to compensate for dropouts via the professional online panel Respondi. Of 454 participants who accessed the study, 50 were immediately screened out due to a failed attention check. 27 participants dropped out of the study prematurely. Beyond this, no datasets were excluded. 377 participants were included in the data analysis (185 females, 191 males, one non-binary, M_age = 43.79, SD = 14.08). A sensitivity analysis via G*Power suggested our sample provided 80% power (α = .05) to detect an interaction effect of f = 0.16 (η_p² = .025). All participants were randomly assigned to one of six conditions of a 2 (physical inclusionary status: physical exclusion vs. physical inclusion) by 3 (social inclusionary status: social exclusion vs. neutral social contact vs. social inclusion) factorial design.

Materials and procedure

After obtaining informed consent and assessing demographics, physical and social inclusionary status was manipulated via an imagined scenario task using an established manipulation paradigm in social exclusion research (Wirth, 2016; e.g.; Büttner & Rudert, 2022; Uskul & Over, 2014). We chose this paradigm as it allows for great flexibility in manipulating different types of exclusion under controlled conditions in an ethically acceptable manner (Wirth, 2016). In the current study, all participants read a scenario about starting a new job and were asked to put themselves in the employee’s situation and mentally visualize the situation (for similar scenarios, see, e.g., Aydin et al., 2017; Park & Baumeister, 2015; Sjåstad et al., 2021). Depending on physical inclusionary status, participants either read that they were placed in an open-plan office in the company’s main building and had a lot of physical contact with others (physical inclusion) or that, due to low space capacity, they were placed alone in an otherwise unoccupied, secluded building without any physical contact to others (physical exclusion). Additionally, the scenario differed in how they were socially included by their coworkers.¹ In the social inclusion condition, the colleagues made a conscious effort to get to know them and showed inclusionary behavior (e.g., always answered e-mails immediately). We further employed a neutral social contact condition where the coworkers seemed open to getting to know the participant and showed rather neutral (though still relatively friendly) behavior (e.g., answered e-mails relatively promptly). Thus, this condition reflected a state in which participants were also socially included (i.e., they were not socially excluded) but experienced a milder form of inclusion compared to the social inclusion condition. In the social exclusion condition, in contrast, the colleagues seemed to consciously avoid social contact and showed exclusionary behaviors (e.g., responded considerably delayed to e-mails). Next, participants completed measures on positive and negative affect (including three exclusion-related items serving as a manipulation check), need-threat, loneliness,² and prosocial and antisocial behavioral intentions toward their coworkers. Finally, participants were asked how well they could put themselves in the scenario and were debriefed.

Manipulation and plausibility checks

As a manipulation check, participants rated how accepted, excluded, and lonely they felt in the situation described from 1 = ‘not at all’ to 5 = ‘extremely’. These items were aggregated into a manipulation check index on exclusionary feelings (Cronbach’s α = .82). To test the plausibility of our manipulation, participants indicated how well they were able to put themselves in the scenario, ranging from 1 = ‘not at all’ to 7 = ‘very well’.

Mood

Mood was assessed using the 20-item Positive and Negative Affect Schedule (PANAS; Watson et al., 1988), measuring negative (e.g., ‘nervous’; Cronbach’s α = .92) and positive affect (e.g., ‘happy’; Cronbach’s α = .91) with ten items each from 1 = ‘not at all’ to 5 = ‘extremely’. To create an index score for mood, we deduced aggregated negative from aggregated positive affect items, such that higher positive values indicate more positive and higher negative values indicate more negative mood.

Basic need-threat

To evaluate basic need-threat, participants responded to eleven items (Pfundmair et al., 2015; based on; Zadro et al., 2004) related to the scenario, with three items each on the need to belong (e.g., ‘I felt poorly accepted by my colleagues’; Cronbach’s α = .81), control (e.g., ‘I had control over my life’; Cronbach’s α = .72), self-esteem (e.g., ‘I felt somewhat inadequate’; Cronbach’s α = .84), and two items on meaningful existence (e.g., ‘I felt as if I was non-existent’; Cronbach’s α = .88) from 1 = ‘Not at all true’ to 7 = ‘Entirely true’. All items were aggregated into a need-threat index (α = .94), with higher values indicating more need-threat.

Loneliness

For loneliness, participants responded to the six-item short scale for overall, emotional, and social loneliness (De Jong Gierveld & Van Tilburg, 2006) from 1 = ‘fully disagree’ to 7 = ‘fully agree’ (e.g., “I experience a general sense of emptiness; Cronbach’s α = .80), with higher values indicating more loneliness.

Prosocial and antisocial intentions

Participants were asked how likely they would display three prosocial (e.g., ‘Change vacation schedule, workdays, or shifts to accommodate the needs of colleagues,’ Cronbach’s α = .82) and three antisocial (e.g., ‘Withholding work-related information from colleagues’; Cronbach’s α = .87) behaviors toward their coworkers from 1 = ‘not at all likely’ to 7 = ‘Very likely’.

Analysis plan

To test the effect of physical and social inclusionary status on well-being, we employed separate 2 × 3ANOVAs for each well-being variable (i.e., negative mood, need-threat, and loneliness). As we had predicted a specific interaction pattern in Hypothesis 1, we used a set of planned contrasts via multiple regression (Judd et al., 2017, pp. 205–228) to follow up these omnibus tests. For an overview of all contrast codes in Study 1, see Table 1. For the effect of social inclusionary status, we coded the difference between social exclusion and no social exclusion (neutral social contact and social inclusion; C1) and the difference between neutral social contact and social inclusion (C2). For the effect of physical inclusionary status, we coded the difference between physical exclusion and physical inclusion (C3). Based on these contrasts (C1-C3), we formed two interaction contrasts to test the prediction that the effect of physical exclusion would be less pronounced for individuals with neutral social contact or being socially included. The first interaction contrast (C4) tested whether the effect of physical exclusion (vs. physical inclusion) would differ from the social exclusion condition to those involving no social exclusion (i.e., neutral social contact or social inclusion; C4). The second interaction contrast tested whether any differences in the effect of physical exclusion (vs. physical inclusion) would occur between neutral social contact and the social inclusion condition (C5).

Table 1. Contrast codes for testing hypothesis 1 to 3 in Study 1.

Contrasts		Experimental conditions
		Physical inclusion			Physical exclusion
		Social exclusion	Social neutral	Social inclusion	Social exclusion	Social neutral	Social inclusion
Hypothesis 1: Interaction effect of social and physical inclusionary status
C1	Social exclusion vs. No social exclusion	−2	1	1	−2	1	1
C2	Social neutral vs. Social inclusion	0	−1	1	0	−1	1
C3	Physical exclusion vs. Physical inclusion	1	1	1	−1	−1	−1
C4	Interaction C1 X C3	−2	1	1	2	−1	−1
C5	Interaction C2 X C3	0	−1	1	0	1	−1
Hypothesis 2: Mere social exclusion (physical inclusion/social exclusion) vs. Mere physical exclusion
C1	Mere social exclusion vs. Physical exclusion with neutral contact	−1	0	0	0	1	0
C2	Mere social exclusion vs. Physical exclusion with social inclusion	−1	0	0	0	0	1
Hypothesis 3: Combined exclusion (physical exclusion/social exclusion) vs. mere physical exclusion
C1	Combined exclusion vs. Physical exclusion with neutral contact	0	0	0	−1	1	0
C2	Combined exclusion vs. Physical exclusion with social inclusion	0	0	0	−1	0	1

Note: No social exclusion = Neutral social contact and social inclusion combined.

To test whether mere social exclusion would lead to lower overall levels of well-being than mere physical exclusion (H2), we ran a set of two simple contrasts per well-being variable. The first contrast tested the difference between mere social exclusion (i.e., physical inclusion/social exclusion) and physical exclusion paired with neutral social contact. The second contrast tested the difference between mere social exclusion and physical exclusion paired with social inclusion.

To test whether a combination of physical and social exclusion would lead to overall lower levels of well-being than mere physical exclusion (H3), we ran a set of two simple contrasts per well-being variable. The first contrast tested the differences between combined exclusion (i.e., physical exclusion/social exclusion) and physical exclusion paired with neutral social contact. The second contrast tested the difference between combined exclusion and physical exclusion paired with social inclusion.

As we had no clear predictions for our explorative analyses on prosocial and antisocial behavioral intentions, we ran two 2 × 3ANOVAs on behavioral intentions followed by post-hoc tests. Across all analyses, we adjusted p-values and confidence intervals for multiple dependent variables per individual H₀ being tested (see, e.g., Rubin, 2021; Wilson, 1962) using the package ‘RHSDB’ (Zhicheng & Hailin, 2022) via the R-Plugin for SPSS.³

Results

Descriptive statistics for dependent variables in Study 1 can be found in Table 2.

Table 2. Means and standard deviations (in parentheses) of dependent variables in study 1.

Variables	Experimental conditions
	Physical inclusion			Physical exclusion
	Social exclusion(n = 69)	Social neutral (n = 55)	Social inclusion (n = 58)	Social exclusion (n = 62)		Social neutral (n = 66)	Social inclusion (n = 67)
Mood	−0.25 (1.30)	2.00 (1.05)	2.10 (0.98)	−0.03 (1.46)		0.98 (1.27)	1.23 (1.23)
Need-threat	4.91 (1.39)	2.60 (1.16)	2.12 (1.00)	4.33 (1.53)		3.62 (1.30)	3.16 (1.36)
Loneliness	4.08 (1.36)	2.92 (1.09)	2.60 (0.95)	4.13 (1.37)		3.71 (1.21)	3.52 (1.29)
Prosocial intentions	4.34 (1.59)	5.04 (1.28)	5.34 (1.15)	4.33 (1.55)		4.34 (1.34)	4.83 (1.25)
Antisocial intentions	2.78 (1.76)	1.30 (0.67)	1.51 (0.78)	2.68 (1.60)		2.11 (1.18)	1.82 (1.20)

Note: For need-threat, loneliness, pro-, and antisocial intentions, higher values correspond to higher levels of the respective variable. Mood was calculated by subtracting aggregated negative from aggregated positive affect items, with higher positive values reflecting more positive and higher negative values reflecting more negative mood. Range of possible numeric values for dependent variables from minimum to maximum: Mood = −4 to + 4; Need-threat, loneliness, prosocial and antisocial intentions = 1 to 7.

Manipulation and plausibility checks

Participant’s mean rating (M = 5.33, SD = 1.49) of their ability to put themselves in the scenario ranged significantly above the seven-point scale’s mid-point (m = 4), t(374) = 17.28, p < .001, d = 0.89, 95%CI = [1.18, 1.49]. Based on a series of Bonferroni-Holm corrected equivalence tests using the R-package ‘TOSTER’ (Caldwell, 2022; via the R-Plugin for SPSS), we rejected differences in participants’ ability to put themselves in the scenario greater than one scale point across conditions containing different combinations of exclusionary experiences, all ps ≤ .048 (equivalence range: −1, +1; see supplementary).

Bonferroni-Holm adjusted contrasts further suggested that our manipulation was effective: participants that were socially, physically, or simultaneously socially and physically excluded (coded as −1, respectively) each expressed more exclusionary feelings than those who were socially and physically included (coded as + 1), all ps ≤ .001, and those who were physically included and had neutral social contact (coded as + 1), all ps ≤ .023 (see supplementary).

Interaction effects of social and physical inclusionary status on well-being

2×3 ANOVAs (Bonferroni-Holm corrected for three dependent variables) revealed a main effect of social inclusionary status on mood F(2, 371) = 83.69, p < .001, η_p² = .311, need-threat, F(2, 371) = 79.75, p < .001, η_p² = .301 and loneliness, F(2, 371) = 25.21, p < .001, η_p² = .120. Bonferroni-Holm corrected contrasts (Table 3) revealed that socially excluded individuals expressed more negative mood (M = −0.15, SD = 1.38 to M = 1.54, SD = 1.24), need-threat (M = 4.63, SD = 1.48 to M = 2.91, SD = 1.34), and loneliness (M = 4.11, SD = 1.36 to M = 3.22, SD = 1.23) compared to those who did not experience social exclusion (neutral social contact or social inclusion). Neutral social contact and social inclusion did not significantly differ in mood (M = 1.45, SD = 1.27 to M = 1.64, SD = 1.20) and loneliness (M = 3.35, SD = 1.22 to M = 3.09, SD = 1.23). However, those with neutral social contact (M = 3.15, SD = 1.33) reported higher levels of need-threat than socially included individuals (M = 2.68, SD = 1.31).

Table 3. Set of planned contrast for probing the interaction effect of social and physical inclusionary status on well-being in study 1.

Contrast		Variable	β	t(371)	p	95% CI
Social status
C1	Social exclusion vs. No social exclusion (neutral/inclusion)
		Mood	.54	12.88	<.001	[0.46, 0.69]
		Need-threat	−.52	−12.28	<.001	[−0.70, −0.46]
		Loneliness	−.33	−6.89	<.001	[−0.42, −0.20]
C2	Social neutral vs. Social inclusion
		Mood	.05	1.11	.266	[−0.07, 0.24]
		Need-threat	−.12	−2.80	.016	[−0.44, −0.03]
		Loneliness	−.08	−1.63	.208	[−0.31, 0.05]
Physical status
C3	Physical exclusion vs. Physical inclusion
		Mood	.18	4.37	<.001	[0.14, 0.42]
		Need-threat	−.15	−3.65	<.001	[−0.38, −0.11]
		Loneliness	−.22	−4.62	<.001	[−0.45, −0.14]
Social status X Physical status
C4	Social exclusion vs. No social exclusion (C1) XPhysical status (C3)
		Mood	.18	4.34	<.001	[0.08, 0.31]
		Need-threat	−.24	−5.69	<.001	[−0.39, −0.14]
		Loneliness	−.14	−2.99	.012	[−0.24, −0.02]
C5	Social neutral vs. Social inclusion (C2) XPhysical status (C3)
		Mood	−.02	−0.46	>.999	[−0.23, 0.15]
		Need-threat	−.003	−0.07	>.999	[−0.17, 0.16]
		Loneliness	−.02	−0.39	>.999	[−0.21, 0.15]

Note. P-values and confidence intervals were adjusted with Bonferroni-Holm for the number of well-being variables per H₀ being tested (i.e., six comparisons for the effect of social inclusionary status, three comparisons for the effect of physical inclusionary status, and six comparisons for the interaction).

Similarly, there was a main effect of physical inclusionary status on mood F(1, 371) = 19.09, p < .001, η_p² = .049, need-threat, F(1, 371) = 13.31, p < .001, η_p² = .035 and loneliness, F(1, 371) = 21.32, p < .001, η_p² = .054. Physically excluded individuals reported less positive mood (M = 0.75, SD = 1.42 to M = 1.18, SD = 1.59), more need-threat (M = 3.69, SD = 1.47 to M = 3.32, SD = 1.74), and more loneliness (M = 3.78, SD = 1.31 to M = 3.26, SD = 1.33) compared to physical inclusion.

These effects, however, were all qualified by a significant interaction, mood: F(2, 371) = 9.52, p < .001, η_p² = .049, need-threat: F(2, 371) = 16.17, p < .001, η_p² = .080, loneliness: F(2, 371) = 4.57, p = .011, η_p² = .024. The interaction contrasts (Table 3) revealed that the effect of physical exclusion (vs. physical inclusion) on mood, need-threat, and loneliness was different for socially excluded individuals compared to those that experienced no social exclusion but did not significantly differ between neutral social contact and social inclusion. However, the effect of physical exclusion (vs. physical inclusion) on well-being was less pronounced in the social exclusion condition compared to the neutral social contact and the social inclusion condition (for a graphical overview of the interaction effects on well-being, see Figure 1(a-c).

Figure 1. Effects of social and physical inclusionary status on self-reported levels of negative mood, need-threat, loneliness, prosocial and antisocial intentions in study 1.

Note: Displayed are estimated marginal means with 95% Confidence-Intervals.

Effects of mere social vs. mere physical exclusion on well-being

Bonferroni-Holm corrected⁴ contrasts suggested that the physical inclusion/social exclusion condition (mere social exclusion) elicited more negative mood, F(1, 371) = 49.11, p < .001, η_p² = .117, 95%CI = [0.93, 2.03], need-threat, F(1, 371) = 60.59, p < .001, η_p² = .140, 95%CI = [−2.34, −1.15], and loneliness, F(1, 371) = 7.11, p = .016, η_p² = .019, 95%CI = [−1.04, −0.09], than physical exclusion paired with social inclusion. Similarly, the physical inclusion/social exclusion condition (mere social exclusion) led to more negative mood, F(1, 371) = 33.77, p < .001, η_p² = .083, 95%CI = [0.70, 1.77], and need-threat, F(1, 371) = 32.73, p < .001, η_p² = .081, 95%CI = [−1.83, −0.75], than physical exclusion paired with neutral social contact. However, feelings of loneliness did not significantly differ between the physical inclusion/social exclusion condition (mere social exclusion) and physical exclusion paired with neutral social contact F(1, 371) = 3.02, p = .083, η_p² = .008, 95%CI = [−0.78, 0.05].

Effects of combined vs. mere physical exclusion on well-being

Bonferroni-Holm corrected⁴ contrasts revealed that the physical exclusion/social exclusion condition (combined exclusion) led to more negative mood, F(1, 371) = 33.87, p < .001, η_p² = .084, 95%CI = [0.69, 1.84], need-threat, F(1, 371) = 25.52, p < .001, η_p² = .064, 95%CI = [−1.76, −0.57], and loneliness, F(1, 371) = 8.11, p = .010, η_p² = .021, 95%CI = [−1.10, −0.13] than physical exclusion paired with social inclusion. Similarly, the physical exclusion/social exclusion condition (combined exclusion) led to more negative mood, F(1, 371) = 21.72, p < .001, η_p² = .055, 95%CI = [0.47, 1.56], and need-threat, F(1, 371) = 9.33, p = .007, η_p² = .025, 95%CI = [−1.26, −0.15] than physical exclusion paired with neutral social contact. However, individuals did not significantly differ in loneliness between the physical exclusion/social exclusion condition (combined exclusion) and physical exclusion paired with neutral social contact, F(1, 371) = 3.77, p = .053, η_p² = .010, 95%CI = [−0.85, 0.01].

Exploratory analyses on behavioral intentions

2×3 ANOVAs (Bonferroni-Holm corrected for two dependent variables) revealed a main effect of social inclusionary status on prosocial F (2, 371) = 9.47, p < .001, η_p² = .049, and antisocial intentions, F (2, 371) = 28.52, p < .001, η_p² = .133. For prosocial intentions, Bonferroni-Holm corrected⁴ post-hoc tests indicated that socially excluded participants (M = 4.34, SD = 1.57) held less prosocial intentions than socially included (M = 5.07, SD = 1.22), p < .001, 95%CI = [−1.16, −0.30]. Prosocial intentions for neutral social contact (M = 4.66, SD = 1.35) ranged in between, as they did not significantly differ from both the social exclusion, p = .132 95%CI = [−0.07, 0.71], and the social inclusion condition, p = .059, 95%CI = [−0.83, 0.01].

Bonferroni-Holm corrected post-hoc tests⁴ for antisocial intentions revealed that socially excluded individuals (M = 2.74, SD = 1.68) held more antisocial intentions than those having neutral social contact (M = 1.74, SD = 1.06), p < .001, 95%CI = [0.58, 1.42], and those being socially included (M = 1.67, SD = 1.03), p < .001, 95%CI = [0.64, 1.49]. Individuals did not significantly differ in antisocial intentions when having neutral social contact and being socially included, p = .698, 95%CI = [−0.26, 0.39].

Additionally, a main effect of physical inclusionary status emerged on prosocial, F(1, 371) = 8.08, p = .009, η_p² = .021, and antisocial intentions, F (1, 371) = 6.62, p = .010, η_p² = .018. Physically excluded participants indicated less prosocial (M = 4.51, SD = 1.39 to M = 4.87, SD = 1.43) and more antisocial intentions (M = 2.19, SD = 1.37 to M = 1.93, SD = 1.40) than physically included.

While no significant interaction effect emerged for prosocial intentions, F(2, 371) = 2.15, p =.118, η_p² = .011, the main effects on antisocial intentions were qualified by a significant interaction, F(2, 371) = 3.92, p = .041, η_p² = .021. To probe this interaction, we employed three Bonferroni-Holm corrected interaction contrasts to test whether the effect of physical exclusion (vs. physical inclusion) on antisocial intentions would differ across individuals’ social inclusionary status using the/LMATRIX command in SPSS (see Table 4 for contrast codes).

Table 4. Interaction contrast for the strength of simple effects of physical exclusion (vs. physical inclusion) on antisocial tendencies across social inclusionary status conditions in study 1.

Contrasts		Experimental conditions
		Physical inclusion			Physical exclusion
		Social exclusion	Social neutral	Social inclusion	Social exclusion	Social neutral	Social inclusion
C1	Social neutral vs. Social exclusion	1	−1	0	−1	1	0
C2	Social inclusion vs. Social exclusion	1	0	−1	−1	0	1
C3	Social neutral vs. social inclusion	0	−1	1	0	1	−1

Results revealed that the effect of physical exclusion (vs. physical inclusion) on antisocial intentions was more pronounced for individuals that had neutral social contact compared to socially excluded individuals, F(1, 371) = 7.83, p = .016, η_p² = .021, 95%CI = [0.13, 1.69]. The effect for socially included individuals ranged in between, as it did not significantly differ from both, the social exclusion condition, F(1, 371) = 1.66, p = .267, η_p² = .004, 95%CI = [−0.22, 1.05], and the neutral social contact condition, F(1, 371) = 2.26, p = .267, η_p² = .006, 95%CI = [−1.23, 0.25]. For an illustration of the effects of physical and social inclusionary status on behavioral intentions in Study 1, see Figure 1(d,e).

Discussion

Study 1 revealed first experimental evidence for the interplay of physical and social components of exclusion. In line with our expectations, there was an interaction between social and physical inclusionary status on well-being (i.e., mood, need-threat, and loneliness). However, concerning the specific pattern, results pointed in a different direction than was initially anticipated. Specifically, while we did find that the effect of physical exclusion on well-being was different in the social exclusion condition as compared to the neutral social contact or social inclusion condition, this was not driven by a weaker effect of physical exclusion if the threatening social component was missing. In contrast to our expectations, the impact of physical exclusion was less pronounced if individuals were socially excluded compared to when they had neutral social contact or were socially included. Physical exclusion altered well-being more strongly when individuals were socially included or held neutral social contact. However, excluded individuals generally held lower levels of well-being, and whether they had physical contact (mere social exclusion) or not (combined exclusion) played less of a role in how they felt.

In line with our predictions, overall levels of well-being were higher under mere physical than combined exclusion. Thus, although neither neutral nor socially including contact could reduce the impact of physical exclusion on well-being per se, they could protect overall levels of well-being at least to some extent (except for loneliness, where only social inclusion was able to protect well-being compared to combined exclusion). Additionally, mere social exclusion triggered more negative effects on well-being than mere physical exclusion (again, for loneliness, this was only true when physical exclusion was paired with social inclusion), suggesting that social exclusion might indeed be a more painful experience than physical exclusion.

Concerning our explorative investigation on behavioral intentions, preliminary findings suggest that social and physical exclusion might interact in shaping antisocial intentions. While the general level of antisocial tendencies was rather low across all conditions, social and physical exclusion, on average, heightened antisocial intentions. However, as indicated by the interaction contrasts, the extent to which physical exclusion affected antisocial intentions was most evident in the neutral social contact condition. For prosocial intentions, we found no interaction effect, but that physical and social exclusion additively lowered prosocial intentions. However, the effect size of the interaction (η_p² = .011) was smaller than the one on which we had built our power analysis. Thus, there may be a small interaction effect for prosocial intentions that the current study was not powerful enough to detect.

Study 2

While Study 1 provided first insights into the interplay of physical and social components of exclusion, one could argue that social exclusion as used in our scenario (i.e., explicit excluding behavior from coworkers) is a rather hostile form of social exclusion that presumes negative intentions by others, which is different from circumstances where people lack social contact, as often described in the social isolation literature. Here, rather than experiencing negative treatment by others, people are excluded because they lack social interactions without someone necessarily treating them hostile. In Study 2, we aimed to replicate the pattern observed in Study 1 in a larger sample and further investigate whether it would also occur for this less hostile type of social exclusion. Thus, in Study 2, we dropped the neutral social contact condition and instead added a no social contact condition as a second form of social exclusion.

As observed in Study 1, we expected the effects of physical exclusion on well-being to be dependent on individuals’ social inclusionary status. Specifically, we expected that physical exclusion (vs. physical inclusion) would have less of an effect for individuals that are socially excluded (i.e., hostile exclusion or no social contact) compared to when they are socially included (Hypothesis 1). Additionally, we expected that social forms of exclusion, irrespective of being paired with physical in- or exclusion, would affect overall levels of well-being more negatively than mere physical exclusion (physical exclusion paired with social inclusion; Hypothesis 2). We again included behavioral intentions on an exploratory basis.

Methods

Participants and design

Based on the small interaction effect size for prosocial intentions observed in Study 1 (η_p² = .011; f = 0.11), an a priori power analysis (α = .05; 80% power) via G*Power (ANOVA: Fixed effects, special, main effects, and interactions) suggested a sample size of 800 participants to detect an interaction effect within a 2 × 3ANOVA. We overrecruited participants to be able to compensate for dropouts via Respondi. Of 1132 participants who accessed the survey, 134 dropped out of the study prematurely. 133 participants who failed an attention check and one under 18 years old were screened out within the survey. Additionally, one person who reported experiencing technical issues and two reporting they did not read the scenario were excluded from data analysis. This resulted in a final sample size of N = 861 (432 females, 428 males, one non-binary, M_age = 48.92, SD = 16.16). A sensitivity analysis via G*Power suggested our sample provided 80% power (α = .05) to detect an interaction effect of f = .10 (η_p² = .011). All participants were randomly assigned to one of six conditions of a 2 (physical inclusionary status: physical exclusion vs. physical inclusion) by 3 (social inclusionary status: hostile social exclusion vs. no social contact vs. social inclusion) factorial design.

Procedure and materials

The procedure and materials were identical to Study 1 (Cronbach’s α: positive affect = .92; negative affect = .93; belonging = .83; self-esteem = .82; control = .75; meaningful existence = .89; need-threat = .95; loneliness = .84; prosocial intentions = .87; antisocial intentions = .86), except for two changes. First, we replaced a phrase within our physical inclusionary manipulation, ‘a lot of/no direct contact with others’ with ‘a lot of/no physical contact with others’ to avoid ambiguity that might have arisen from the term ‘direct contact’. Additionally, the neutral social contact condition was replaced by a no social contact condition. Participants read that their coworkers do not approach them, and no opportunities for exchange have arisen so far. They were told that their area of work does not require communication with their colleagues and that they would thus be working on their own. Accordingly, this condition reflected a state where participants were socially excluded, but no hostile behavior was directed against them.

Analysis plan

As in Study 1, we employed 2 × 3ANOVAs followed by a set of planned contrasts to test the interaction pattern between social and physical inclusionary status on well-being. For an overview of all contrast codes in Study 2, see Table 5. For the effect of social inclusionary status, we coded the difference between social exclusion (hostile social exclusion and no social contact) and social inclusion (C1), as well as the difference between hostile social exclusion and no social contact (C2). For the effect of physical inclusionary status, we coded the difference between physical exclusion and physical inclusion (C3). Based on these contrasts (C1-C3), we formed two interaction contrasts to test whether the effect of physical exclusion (vs. physical inclusion) would be less pronounced for socially excluded individuals. The first interaction contrast (C4) tested whether the effect of physical exclusion (vs. physical inclusion) would differ from the social inclusion condition to those involving social exclusion. The second interaction contrast tested whether any differences in the effect of physical exclusion (vs. physical inclusion) would occur between hostile social exclusion and no social contact (C5).

Table 5. Contrast codes for testing hypothesis 1 and 2 in study 2.

Contrasts		Experimental conditions
		Physical inclusion			Physical exclusion
		Hostile exclusion	No social contact	Social inclusion	Hostile exclusion	No social contact	Social inclusion
Hypothesis 1: Interaction effect of social and physical inclusionary status on well-being
C1	Social exclusion vs. Social inclusion	−1	−1	2	−1	−1	2
C2	Hostile exclusion vs. No social contact	−1	1	0	−1	1	0
C3	Physical exclusion vs. physical inclusion	1	1	1	−1	−1	−1
C4	Interaction C1 X C3	−1	−1	2	1	1	−2
C5	Interaction C2 X C3	−1	1	0	1	−1	0
Hypothesis 2: Social forms of exclusion (with or without physical contact) vs. Mere physical exclusion (physical exclusion/social inclusion)
C1	Hostile exclusion with physical exclusion vs. Mere physical exclusion	0	0	0	−1	0	1
C2	Hostile exclusion with physical inclusion vs. Mere physical exclusion	−1	0	0	0	0	1
C3	No social contact with physical exclusion vs. Mere physical exclusion	0	0	0	0	−1	1
C4	No social contact with physical inclusion vs. Mere physical exclusion	0	−1	0	0	0	1

Note: Social exclusion = Hostile social exclusion and no social contact combined.

To test whether both forms of social exclusion, irrespective of being paired with physical in- or exclusion, would affect overall levels of well-being more negatively than mere physical exclusion (H2), we employed four simple contrasts for each of the well-being variables. The first contrast tested the difference between mere physical exclusion (i.e., physical exclusion/social inclusion) and hostile social exclusion paired with physical exclusion. The second contrast tested the difference between mere physical exclusion and hostile social exclusion paired with physical inclusion. The third contrast tested the difference between mere physical exclusion and no social contact paired with physical exclusion. The fourth contrast tested the difference between mere physical exclusion and no social contact paired with physical inclusion.

To explore effects of physical and social inclusionary status on behavioral intentions, we ran two 2 × 3ANOVAs followed by post-hoc tests. As in Study 1, we adjusted p-values and confidence intervals for multiple dependent variables per H₀ being tested.

Results

Descriptive statistics for dependent measures in Study 2 can be found in Table 6.

Table 6. Means and standard deviations (in parentheses) of dependent variables in study 2.

Variables	Experimental conditions
	Physical inclusion			Physical exclusion
	Hostile exclusion(n = 138)	No social contact (n = 139)	Social inclusion (n = 151)	Hostile Exclusion (n = 138)		No social contact (n = 157)	Social inclusion (n = 138)
Mood	−0.11 (1.52)	0.34 (1.32)	2.06 (1.29)	−0.01 (1.58)		0.30 (1.42)	1.31 (1.28)
Need-threat	4.72 (1.56)	4.33 (1.36)	2.13 (1.13)	4.52 (1.59)		4.20 (1.51)	3.01 (1.30)
Loneliness	4.10 (1.58)	3.91 (1.38)	2.49 (1.01)	4.06 (1.62)		4.14 (1.39)	3.29 (1.15)
Prosocial intentions	4.31 (1.77)	4.47 (1.56)	5.36 (1.30)	4.19 (1.88)		4.21 (1.55)	4.98 (1.39)
Antisocial intentions	2.67 (1.74)	2.32 (1.43)	1.82 (1.11)	2.81 (1.71)		2.17 (1.20)	2.00 (1.30)

Note: For need-threat, loneliness, pro-, and antisocial intentions, higher values correspond to higher levels of the respective variable. Mood was calculated by subtracting aggregated negative from aggregated positive affect items, with higher positive values reflecting more positive and higher negative values reflecting more negative mood. Range of possible numeric values for dependent variables from minimum to maximum: Mood = −4 to + 4; Need-threat, loneliness, prosocial and antisocial intentions = 1 to 7.

Manipulation and plausibility checks

Participant’s mean rating (M = 5.40, SD = 1.48) of their ability to put themselves in the scenario ranged significantly above the seven-point scale’s mid-point (m = 4), t(860) = 27.78, p < .001, d = 1.48, 95%CI = [1.30, 1.50]. Based on a series of Bonferroni-Holm corrected equivalence tests (‘TOSTER’ package; Caldwell, 2022), we rejected differences in participants’ ability to put themselves in the scenario greater than one scale point across conditions containing different combinations of exclusionary experiences, all ps < .001 (equivalence range: −1, +1; see supplementary).

Supporting the effectiveness of our manipulation, Bonferroni-Holm adjusted contrasts revealed that each condition containing physical, social, or both forms of exclusion (coded as −1, respectively) elicited more exclusionary feelings than the physical and social inclusionary condition (coded as + 1), all ps < .001 (see supplementary).

Interaction effects of physical and social exclusion on well-being

2×3 ANOVAs (Bonferroni-Holm corrected for three dependent variables) revealed a main effect of social inclusionary status on mood F(2, 855) = 121.39, p < .001, η_p² = .221, need-threat, F(2, 855) = 171.11, p < .001, η_p² = .286, and loneliness, F(2, 855) = 68.96, p < .001, η_p² = .139. Bonferroni-Holm corrected contrasts (Table 7) revealed that compared to socially included individuals, socially excluded expressed less positive mood (M = 0.14, SD = 1.47 to M = 1.70, SD = 1.34), more need-threat (M = 4.43, SD = 1.52 to M = 2.55, SD = 1.29) and loneliness (M = 4.06, SD = 1.50 to M = 2.87, SD = 1.15). Individuals in the hostile social exclusion and no social contact condition did not significantly differ in loneliness (M = 4.08, SD = 1.60 to M = 4.03, SD = 1.39). However, those under hostile social exclusion reported significantly higher levels of negative mood (M = −0.06, SD = 1.55 to M = 0.32, SD = 1.37) and need-threat (M = 4.62, SD = 1.57 to M = 4.26, SD = 1.44) than those with no social contact.

Table 7. Set of planned contrast for probing the interaction effect of social and physical inclusionary status on well-being in study 2.

Contrast		Variable	β	t (855)	p	95% CI
Social status
C1	Social exclusion (hostile/no contact)vs. Social inclusion
		Mood	.46	15.30	<.001	[0.43, 0.61]
		Need-threat	−.52	−18.31	<.001	[−0.71, −0.53]
		Loneliness	−.37	−11.74	<.001	[−0.47, −0.31]
C2	Hostile social exclusion vs. No social contact
		Mood	.10	3.24	.004	[0.05, 0.33]
		Need-threat	−.09	−2.99	.006	[−0.31, 0.04]
		Loneliness	−.02	−0.50	.617	[−0.14, 0.08]
Physical status
C3	Physical exclusion vs. Physical inclusion
		Mood	.07	2.44	.030	[0.01, 0.23]
		Need-threat	−.05	−1.90	.058	[−0.19, 0.003]
		Loneliness	−.11	−3.50	.001	[−0.28, −0.05]
Physical status X Social status
C4	Social exclusion vs. social inclusion (C1) XPhysical status (C3)
		Mood	.11	3.84	.001	[0.04, 0.22]
		Need-threat	−.15	−5.09	<.001	[−0.26, −0.08]
		Loneliness	−.11	−3.57	.001	[−0.20, −0.04]
C5	Hostile social exclusion vs. No social contact (C2) XPhysical status (C3)
		Mood	.02	0.58	>.999	[−0.10, 0.17]
		Need-threat	−.01	−0.35	>.999	[−0.14, 0.10]
		Loneliness	−.04	−1.18	.719	[−0.21, 0.07]

Note. P-values and confidence intervals were adjusted with Bonferroni-Holm for the number of well-being variables per H₀ being tested (i.e., six comparisons for the effect of social inclusionary status, three comparisons for the effect of physical inclusionary status, and six comparisons for the interaction).

Similarly, there was a main effect of physical inclusionary status on mood F(1, 855) = 5.97, p = .030, η_p² = .007, and loneliness, F(1, 855) = 12.25, p = .001, η_p² = .014. Compared to physical inclusion, physically excluded individuals reported less positive mood (M = 0.52, SD = 1.53 to M = 0.80, SD = 1.67) and more loneliness (M = 3.84, SD = 1.45 to M = 3.47, SD = 1.52). Need-threat was higher under physical exclusion (M = 3.92, SD = 1.60) but did not significantly differ from physical inclusion (M = 3.68, SD = 1.78), F(1, 855) = 3.59, p = .058, η_p² = .004.

These effects, however, were all qualified by a significant interaction, mood: F(2, 855) = 7.49, p = .001, η_p² = .017, need-threat: F(2, 855) = 13.00, p < .001, η_p² = .030, loneliness: F(2, 855) = 7.00, p = .001, η_p² = .016. The interaction contrasts (Table 7) revealed that the effect of physical exclusion (vs. physical inclusion) on mood, need-threat, and loneliness was different for socially included individuals compared to those that experienced a form of social exclusion (hostile exclusion or no social contact) but did not significantly differ between these forms of social exclusion. The effect of physical exclusion (vs. physical inclusion) was less pronounced when individuals experienced a form of social exclusion compared to when they were socially included. For a graphical overview of the interaction pattern on well-being in Study 2, see Figure 2(a-c).

Figure 2. Effects of social and physical inclusionary status on self-reported levels of negative mood, need-threat, loneliness, prosocial and antisocial intentions in study 2.

Note. Displayed are estimated marginal means with 95% Confidence-Intervals.

Effects of social exclusion vs mere physical exclusion on well-being

As depicted in Table 8, Bonferroni-Holm corrected contrasts⁵ revealed that both forms of social exclusion, irrespective of being paired with physical in- or exclusion, led to higher impairments in mood, more need-threat and loneliness than physical exclusion paired with social inclusion (mere physical exclusion).

Table 8. Effects of social exclusion (with or without physical contact) vs. mere physical exclusion on well-being in study 2.

	Contrast	F(1, 855)	p	ηp^2	95%CI
C1	Physical exclusion with hostile social exclusion vs. Physical exclusion with social inclusion
	Mood	60.91	<.001	.067	[0.85, 1.79]
	Need-threat	78.34	<.001	.084	[−1.99, −1.02]
	Loneliness	21.78	<.001	.025	[−1.14, −0.40]
C2	Physical inclusion with hostile social exclusion vs. Physical exclusion with social inclusion
	Mood	69.84	<.001	.076	[0.94, 1.89]
	Need-threat	101.15	<.001	.106	[−2.20, −1.22]
	Loneliness	24.31	<.001	.028	[−1.21, −0.42]
C3	Physical exclusion with no social contact vs. Physical exclusion with social inclusion
	Mood	37.75	<.001	.042	[0.57, 1.44]
	Need-threat	52.39	<.001	.058	[−1.64–0.75]
	Loneliness	28.09	<.001	.032	[−1.25, −0.45]
C4	Physical inclusion with no social contact vs. Physical exclusion with social inclusion
	Mood	32.66	<.001	.037	[0.53, 1.40]
	Need-threat	60.05	<.001	.066	[−1.78, −0.85]
	Loneliness	14.22	<.001	.016	[−0.94, −0.30]

Note: Physical exclusion with social inclusion = Mere physical exclusion. P-values and confidence intervals were Bonferroni-Holm adjusted for 12 comparisons in total.

Exploratory analysis on behavioral intentions

2×3 ANOVAs (Bonferroni-Holm corrected for two dependent variables) revealed a main effect of social inclusionary status on prosocial, F(2, 855) = 29.39, p < .001, η_p² = .064, and antisocial intentions, F(2, 855) = 24.03, p < .001, η_p² = .053. Bonferroni-Holm corrected⁴ post-hoc tests indicated that compared to social inclusion (M = 5.18, SD = 1.36), individuals held less prosocial tendencies under hostile social exclusion (M = 4.25, SD = 1.82), p < .001, 95%CI = [−1.28, −0.58], and no social contact (M = 4.33, SD = 1.56), p < .001, 95%CI = [−1.18, −0.52]. Prosocial intentions did not significantly differ for individuals under hostile social exclusion and no social contact, p = .547, 95%CI = [−0.34, 0.18].

Bonferroni-Holm corrected⁴ post-hoc test for antisocial intentions revealed that compared to social inclusion (M = 1.91, SD = 1.20), individuals expressed more antisocial tendencies under hostile social exclusion (M = 2.74, SD = 1.72), p < .001, 95%CI = [0.52, 1.14], and no social contact (M = 2.24, SD = 1.31), p = .009, 95%CI= [0.07, 0.60]. Those under hostile social exclusion indicated higher levels of antisocial intentions compared to those with no social contact, p < .001, 95%CI = [0.21, 0.78].

Additionally, a main effect of physical inclusionary status emerged on prosocial, F (1, 855) = 5.43, p = .040, η_p² = .006, but not antisocial intentions, F (1, 855) = 0.40, p = .529, η_p² <.001. Physically excluded individuals indicated less prosocial intentions (M = 4.45, SD = 1.65) than physically included (M = 4.73, SD = 1.62) but did not significantly differ in antisocial intentions (M = 2.32, SD = 1.45 to M = 2.26, SD = 1.48). No significant interaction emerged for prosocial, F(2, 855) = 0.45, p = .637, η_p² = .001, nor antisocial intentions, F(2, 855) = 1.18, p = .614, η_p² = .003.

Discussion

Within Study 2, we could replicate the main pattern from Study 1 in a larger sample and further extend these findings for exclusion in the form of no, rather than hostile social contact. Under social inclusion, a lack of physical contact lowered well-being to some extent; however, both forms of social exclusion, irrespective of being paired with physical contact or not, led to lower levels of well-being than mere physical exclusion. In Study 2, no interactive effects on prosocial and antisocial behavioral intentions emerged.

General discussion

While past exclusion research has concentrated mainly on social forms of exclusion, the current work aimed to shed distinct light on the physical component and its combination with the social component of exclusion. Across both studies, the results provide first experimental evidence that physical and social exclusion can add and interact in a meaningful way in their effects on well-being and interpersonal behavioral intentions.

Effects on well-being

For well-being (i.e., mood, basic need-threat, and loneliness), it was consistently shown that a lack of physical contact was related to more negative effects among socially included (Studies 1 and 2) or those who experienced neutral social contact (Study 1). However, as soon as a form of social exclusion came to play, whether in the form of hostile social exclusion (Study 1 and 2) or a lack of social contact (Study 2), the physical component receded into the background (i.e., its effect became considerable smaller). Individuals who were socially excluded in some way showed relatively similar levels of well-being when they were physically included or excluded. Although contrary to our initial hypothesis, it seems reasonable that a lack of physical contact did not substantially worsen well-being of socially excluded individuals. Social exclusion is a particularly strong experience, and humans are sensitive to even subtle signs of social exclusion (e.g., Kouzakova et al., 2010; Wirth et al., 2010). While physical contact and face-to-face interactions are generally considered important for well-being (e.g., Kroencke et al., 2022), those aspects assumed as positive mechanisms of action get undermined by social exclusion. For example, face-to-face interaction facilitates social belongingness (Sacco & Ismail, 2014) and others’ mere physical presence provides a signal for safety (Chou & Nordgren, 2017; Tedeschi et al., 2021). However, if individuals are rejected or failed to get acknowledged, such positive aspects cannot take effect as they are levered out from the outset. Thus, without its implicit (expected) positive social notation, physical contact might get meaningless in a social sense, as also expressed in the saying of one that can be in a room surrounded by people and still feel lonely.

However, once people were socially excluded in some way, they experienced worse consequences than under mere physical exclusion, irrespective of whether they had physical contact or not. Mere physical exclusion thus lowered well-being to some extent, but levels were still higher than under mere social or combined exclusion. This suggests that while neutral social contact or even explicit social inclusion could not reduce the negative effects of physical exclusion per se, it prevented individuals from the considerably more negative consequences that occurred when these (positive) social components were lacking (i.e., if they were additionally socially excluded). That also rather neutral social contact was able to protect well-being fits research showing that even brief social encounters can serve as cues for belongingness and thus add to positive feelings and happiness (Sandstrom & Dunn, 2014a, 2014b). Importantly, one exception occurred for this pattern. In contrast to the other well-being outcomes, neutral social contact was insufficient to cushion feelings of loneliness induced by physical exclusion. Physically separated individuals with only neutral, rather than inclusionary social contact indicated similar levels of loneliness as those under mere social or combined exclusion. Thus, explicit inclusionary behavior from others seemed to be important, adding experimental support to the notion that not social interactions per se but meaningful social interactions could play a key role in shaping feelings of loneliness (e.g., Wheeler et al., 1983).

In general, our findings align with findings from technology-mediated communication research and studies conducted during COVID-19 that in the absence of physical contact or compared to non-socializing, social interactions via alternative routes are associated with greater well-being (e.g., Esposito et al., 2021; Shufford et al., 2021) but cannot fully substitute physical interactions (e.g., Kroencke et al., 2022; Rudert & Janke, 2022), as positive face-to-face contact nonetheless produces the most positive outcomes. Similarly, in our study, individuals who were both physically and socially included showed the highest levels of well-being. Clearly, face-to-face contact plays an essential role in human interaction, as it creates higher levels of ‘social presence’ (Short et al., 1976) and transports social cues such as nonverbal behaviors important for trust and relationship building that cannot be or are more difficult to detect in nonphysical communication settings (e.g., Powell et al., 2004; Sproull & Kiesler, 1986). Additionally, social interactions such as affectionate touch considered central to human well-being (Jakubiak & Feeney, 2017) can only occur in physical copresence. However, in times when physical contact is limited or restricted, social inclusionary interaction via alternative routes might protect levels of well-being, at least to some degree.

Explorative findings on behavioral intentions

Concerning behavioral intentions, the pattern was different. No interaction effect emerged across both studies for prosocial intentions, but social and physical exclusion additively lowered prosocial intentions. However, the fact we did not find a significant interaction should not necessarily be interpreted as the absence of any interaction, as the interaction effect size observed in Study 1 was rather small. Given that our sample size in Study 1 was not designed for such small effects, it could be that a small interaction effect for prosocial tendencies exists, which we were not able to detect. Yet, while our findings replicated the negative effect of social exclusion on prosocial intentions, they further hint at the possibility that this might also be observed under physical exclusion. Some studies have already revealed that prosocial actions like donations are lower for physically distant targets (e.g., Touré-Tillery & Fishbach, 2017) and that physical interactions like touch can increase giving and helping (e.g., Crusco & Wetzel, 1984; Guéguen & Fischer-Lokou, 2003). However, these studies assessed effects of geographical distance or physical interactions rather than the influence of others’ physical presence per se. Reasons for why a lack of physical contact might lower prosocial intentions could align with some of those discussed for social exclusion. Physical contact could influence how individuals consider the future consequences of their actions (Van Doesum et al., 2018). For example, when lacking physical contact, chances for future (re)inclusion, mutual support, and reciprocity might seem lower compared to when others are physically present, and if individuals weigh up whether they should sacrifice their own time and resources for the benefit of others, these aspects are likely to weigh negatively in the decision-making process (e.g., Baumeister et al., 2007; Maner et al., 2007). Interacting with someone face-to-face might also make that person appear more vivid and identifiable, a factor that has been connected to higher levels of empathy (Jenni & Loewenstein, 1997). Thus, individuals might possess lower levels of empathic concern when they are physically separated from others, which has already been identified as a driving factor for decreased prosociality following social exclusion (Twenge et al., 2007). Alternatively, physical presence in social interactions might also facilitate the emergence of a sense of ‘we-ness’ (e.g., Hertel et al., 2005). Research suggests that a physically present other is sufficient to evoke a sense of social mindfulness – to shift one’s attention to social interdependence and others’ needs (Van Doesum et al., 2018).

For antisocial intentions, an interaction effect emerged in Study 1 but not in Study 2. However, the interaction effect in Study 1 was mainly driven by the neutral social contact condition, such that the effect of physical exclusion on antisocial intentions was most evident for those who experienced neutral social contact. Of note, Study 2 did not include a neutral social condition, and findings thus align with those observed in Study 1: Irrespective of physical contact, both forms of social exclusion (i.e., hostile exclusion and no social contact) heightened antisocial responses, while we found no evidence that physical exclusion itself impacted antisocial intentions. Acting actively hostile against others is a rather extreme response, and it seems reasonable that the effect of the social component (i.e., how other people treat one) tends to prevail here. Different reasons for why social exclusion triggers antisocial tendencies have been discussed in the literature and might give hints as to why physical exclusion could play a less significant role in the emergence of antisocial intentions. Part of aggression following social exclusion may simply reflect retaliation by getting back at those who have previously treated one poorly, a facet that will usually be missing in instances of mere physical exclusion. However, retaliation cannot provide an exhaustive explanation, as aggression following social exclusion occurs toward perpetrators but also uninvolved third parties (Twenge et al., 2001). Experimental research suggests that being socially excluded induces a cognitive bias making one prone to perceive others as hostile (DeWall et al., 2009). However, this only occurred when exclusion was due to a reason individuals could attribute to themselves (i.e., a partner refused to interact with the participant) compared to a situational reason (i.e., the partner had to leave early). Being physically separated from others may cause some degree of discomfort; however, it is less likely that it will be attributed to personal features and might therefore play a less central role in antisocial intentions.

Additionally, it has been argued that aggression following social exclusion, although hindering reconnection, might at least provoke a form of acknowledgment (e.g., Wesselmann, Ren, et al., 2016). While such desires for acknowledgment could also play a role after physical exclusion, it seems reasonable that social exclusionary or inclusionary treatment by others would override these. When being socially excluded, mere physical presence of others cannot provide individuals with acknowledgment, while social inclusion itself is a form of acknowledgment that can exist unrelated to physical contact. In Study 1, effects of physical exclusion (vs. physical inclusion) on antisocial intentions were indeed most evident under rather neutral social contact. Although these findings should be treated with caution, as only one of the studies included a neutral social condition (thereby, no statements about the robustness of the effect can be made), one might speculate whether neutral social contact cannot provide the same level of acknowledgment explicit inclusionary social contact does, allowing effects of physical exclusion on antisocial tendencies to unfold. However, given our findings are explorative, future research is needed to confirm these patterns.

Limitations, conclusion, and future directions

Summarizing our findings, the current research suggests that while physical exclusion negatively affected well-being, social exclusion might represent a stronger injury to well-being, as indicated by more negative effects of mere social compared to mere physical exclusion. This reasoning is also supported when inspecting the effect sizes. While social exclusion produced medium effects on loneliness and strong effects on social pain reactions (i.e., negative mood and need-threat) consistent with former research (e.g., Gerber & Wheeler, 2009), the effect sizes for physical exclusion ranged from small to small-to-medium across both studies (consistent with the majority of effect sizes in social psychology; Richard et al., 2003). Social exclusion might even be such a strong experience that the component of physical contact becomes secondary, as indicated by our finding that physical contact played a minor role in how socially excluded individuals felt. Though not being able to reduce the impact of physical exclusion per se, social inclusion or even neutral friendly contact was able to protect overall levels of well-being to some extent compared to a state where individuals were physically and socially excluded (except for loneliness, where explicit inclusionary contact was needed). For prosocial intentions, explorative findings found evidence only for an additive negative impact of social and physical exclusion. Although no interaction effect emerged for prosocial intentions in our studies, this suggests that social inclusion might still help foster levels of prosocial intentions when physical contact is restricted. Finally, for antisocial intentions, explorative findings suggested that the physical component might play a subordinate role. Physical exclusion seemed to impact antisocial intentions only under neutral social contact. Overall, these findings propose that both the physical and social dimensions of social interactions represent important constituents for well-being and interpersonal behavior but that the social component could play a predominant role most of the time.

When interpreting these findings, some limitations must be considered. Across both studies, an imaginative scenario paradigm was used to manipulate both forms of exclusion. Research suggests that participants’ emotional appraisals of imagined or simulated situations generally align with those observed under actual conditions (e.g., Robinson & Clore, 2001). Studies on social exclusion have further documented reactions to imagined scenarios similar to those triggered by in-vivo manipulations (e.g., Aydin et al., 2010; Giesen & Echterhoff, 2018). It must be stated, though, that scenarios will arguably not be able to achieve the same level of psychological realism and seriousness as the manipulation of ‘real-life’ experiences. However, in contrast to short episodes of hostile social interactions, simulating a prolonged lack of social or physical contact in experimental settings seems difficult under ethical considerations (Wirth, 2016). Using a scenario-imaginative paradigm allowed us to separately manipulate both components of interpersonal exclusion in an ethically acceptable way. As ‘real’ physical presence might nevertheless matter (e.g., Zajonc, 1980), future research, such as quasi-experimental field studies, should explore whether our findings hold in a more naturalistic setting, thereby strengthening external validity.

Additionally, our exclusion paradigm referred to a work scenario, which is clearly different from situations where exclusion touches more or even all areas of life, as has been the case during the pandemic. In such circumstances, the effects of both types of exclusion might be even worse and social inclusion even more important. However, using this established scenario from former experimental exclusion research made it possible to assess effects isolated from other possibly confounding factors. Specifically, this allows to draw conclusions that can be considered in the light of the COVID-19 pandemic but might further contribute to fundamentals of social exclusion literature in general. Nevertheless, future studies could help to evaluate the robustness of effects outside of work-related settings.

Further, within our study, we separately manipulated physical and social exclusion. While this allowed us to gain a better understanding of the differential effects of these two components of exclusion, there may also be situations where their relationship is more complex. For example, avoiding physical contact could be used as a means to communicate an individual they are not wanted. Here, the difference between physical and social exclusion would be less clear-cut, as being physically separated from others would simultaneously represent a form of socio-emotional exclusion. In such cases, we would expect a similar pattern to emerge as we found for combined exclusion (i.e., physical and social exclusion). However, future research could further explore the relationship between both components, such as situations where they do not only occur additively but within the same act of exclusion. Future research might also take a different perspective on the interaction by focusing on how being physically separated from others alters experiences of social exclusion.

Finally, while both of our studies showed a trend in the same direction, adding some degree of replicability, the effect sizes in Study 2 tended to be smaller. As larger samples increase precision in effect size estimates (e.g., Asiamah et al., 2017), these should be regarded as more reliable. Although our studies provided conventional power to detect small-to-medium (Study 1) and small (Study 2) interaction effects, it is further important to note that our power and sensitivity analyses were oriented toward detecting omnibus interaction effects but did not include follow-up tests and alpha correction rates, which has likely affected power for these tests. Additionally, we corrected for multiple dependent variables per individual H₀ being tested (e.g., Rubin, 2021; Wilson, 1962). Thus, in correcting family-wise error rate, we defined analyses used to test a common underlying statement as a family, aiming to balance power and type 1 error concerns. However, there are no universal rules on what should be regarded as a family of tests (e.g., Miller, 1981, pp. 31–35), and some researchers have also argued for more restrictive approaches (e.g., correcting simultaneously for all tests made across an experiment; e.g., Ryan, 1962). Thus, future research is warranted to confirm the specific interaction pattern as observed in our studies.

However, while keeping these limitations in mind, to the best of our knowledge, our studies represent first experimental research on the effects of the physical component of exclusion together with its social component. Although future research is needed to evaluate whether our findings generalize over workplace-related and hypothetical exclusion paradigms as well as different contexts, we hope our findings can act as a starting point and contribute to an emerging line of research on a more nuanced understanding of social and physical aspects of exclusionary experiences.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary data

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15534510.2023.2242617

Data availability statement

Materials and data of both studies can be accessed openly at: https://osf.io/7gph9/?view_only=3eb45d46d0e9439f92d94da4e8037577

Additional information

Funding

This research did receive funding from the University of Klagenfurt’s internal support grant for doctoral dissertations.

Notes

1. Descriptions of the coworkers’ behavior were designed in a way that they were equally plausible in a scenario with or without physical contact.

2. The scale on loneliness contained an attention-check item asking participants to mark a specific option within the Likert-scale answering format.

3. Original p-values and confidence intervals next to corrected ones as well as details on the number of corrections for all analyses can be observed in the supplementary material.

4. Per number of comparisons and dependent variables.