Death of the social self? Comparing the effects of ostracism to mortality salience

ABSTRACT

The ‘social death’ metaphor is used to reflect ostracism’s severity and death-related themes often influence ostracism research. To determine its accuracy, we examined the similarity of ostracism and mortality salience (MS) outcomes. To manipulate these constructs, we used writing prompts in Study 1, and Cyberball and a novel MS manipulation, ‘Cybergrave,’ in studies 2, 3a, and 3b. In Study 4, we correlated chronic ostracism and death-thought accessibility. Ostracism uniquely threatened psychological needs, whereas MS uniquely activated death thoughts. Moreover, the correlation between chronic ostracism and death-thought accessibility was small and non-significant when controlling for related variables. Results suggest the death metaphor may not always be accurate. Death-related language may bias how ostracism is conceptualized and studied, emphasizing pain over recovery.

KEYWORDS:

• Ostracism
• social exclusion
• mortality salience
• death
• death-thought accessibility

Words breathe life into our experiences and are essential for conveying what those experiences are like to others. However, how do we convey threatening experiences in a way that is easily grasped by all? One method involves the use of metaphors that liken the threat associated with one experience to another threatening experience.

Ostracism – being ignored and excluded – is a common yet painful experience, that all individuals have likely endured. Within the literature, ostracism’s severity is often portrayed via references to death or death-related themes intended to capture the experience of feeling like one no longer exists. For example, Case and Williams (2004) have likened ostracism to the sensation of being socially dead, referring to ostracism as a metaphor for death because the sense of invisibility and meaninglessness that often results from ostracism can remind individuals of what it would be like if they died and can, thus, prime mortality. Hales (2018) maintains that death is a metaphor for ostracism and specifies ways in which ostracism and death are similar, citing, for example, their universality (e.g., everyone dies, and most of us encounter ostracism). Death-related themes are often present within the literature and shape the way researchers conceptualize and examine ostracism. Moreover, comparisons between ostracism and death are effective because death can represent the ultimate threat, often fraught with uncertainty as to how and when it will occur, yet certain to occur at some point within the lifespan.

However, what are the implications of comparing ostracism to death within literature? One way to begin addressing this question, is to determine whether ostracism and mortality salience – the awareness of our inevitable physical death – produce similar effects. If these two constructs share a similar psychological mechanism, a fruitful avenue for research would be to examine whether ways that are effective for facilitating recovery from ostracism would apply to mortality salience and vice versa. For example, if we find that ostracism produces similar effects as mortality salience by increasing death-related cognitions and causing individuals to become intolerant of those who oppose their worldviews, then researchers can examine whether methods associated with the reduction of mortality salience’s effects would also be effective for ostracism. Specifically, some research on reduction of mortality salience effects suggests the effectiveness of increasing self-esteem (e.g., Harmon-Jones et al., 1997) and affirming valued aspects of one’s worldviews (Schmeichel & Martens, 2005). The same approach applies if mortality salience produces similar effects as ostracism, threatening psychological needs (e.g., belonging, self-esteem, need for control, and meaningful existence) and worsening mood (Williams, 2009). Some research on recovery from ostracism suggests that both prayer, affirmation of one’s positive qualities, and distraction help people cope following ostracism (Hales et al., 2016). Importantly, then, our research would illuminate new and unique ways to address the effects of both constructs.

If, on the other hand, mortality salience and ostracism produce distinct effects and are uniquely threatening, it may be preferable to emphasize their differences. One notable difference is that ostracism is reversible because individuals can affiliate with others or be re-included by those who ostracized them, whereas death is not reversible. Emphasizing the reversibility of ostracism may, thus, foster hope among those who are ostracized, by suggesting that although death is inevitable, ‘social death’ is transient. Arguably, however, individuals who regularly experience ostracism might struggle with the view of ostracism as transient.

Nevertheless, the use of death-themed language may have deleterious consequences for the study of ostracism and for consumers of ostracism research. For example, comparisons to death may inspire researchers to focus on the severity of ostracism under various circumstances rather than on ways to recover. Moreover, death implies a sense of permanence and comparisons of ostracism to death may send a devastating message to those who feel ostracized and ‘stuck’ within that state of being. Emphasizing pain over recovery, thus, leaves us with a dearth of knowledge surrounding ways to overcome the pain associated with ostracism.

Additionally, the notion that metaphorical associations with death shape the way that individuals conceptualize ostracism is supported by Conceptual Metaphor Theory (Lakoff & Johnson, 1980 as cited in Gibbs, 2011), which proposes that metaphors represent the ways we think about the world by emphasizing deeper, underlying connections between different concepts (e.g., the connection between ostracism and death). Importantly, research supporting Conceptual Metaphor Theory shows that metaphors can have important consequences for human cognition by influencing processes such as decision-making and problem-solving (Gibbs, 2011). It is, thus, reasonable to suggest that likening ostracism to death has important consequences for the way it is conceptualized theoretically and for the ways that people think about and cope with ostracism. Moreover, there is evidence of the impact of metaphorical language within the ostracism literature. For example, research showing that individuals tend to describe their experiences of ostracism and rejection – a more direct way of communicating relational devaluation (Williams, 2007a) – using language related to physical pain (e.g., hurt feelings; Leary & Springer, 2001) inspired research by Eisenberger et al. (2003) on the neural overlap between physical pain and pain from being ostracized. This work, in turn, inspired further work on the beneficial use of analgesics like, Tylenol, for use in dimming the pain one feels when ostracized (e.g., DeWall et al., 2010). Although this example of the impact of metaphors within the literature has been largely positive and productive, we propose that use of the ‘social death’ metaphor may also have the negative consequences previously discussed.

Through a series of studies, four experimental and one correlational, we compared the effects of ostracism to mortality salience. In our experimental studies, we manipulated temporary ostracism, to tap into brief everyday experiences with ostracism that lead to short-term negative psychological effects, and in our correlational study, we examined chronic ostracism – a more severe form of ostracism that occurs either more frequently or for prolonged periods. We aimed to clarify the nature of ostracism as social death and to determine the appropriateness of relying on references to death both in ostracism research and generally.

Ostracism, its consequences, and the influence of the death metaphor

According to Williams’s (2009) Temporal Need Threat Model of Ostracism, ostracism negatively affects four fundamental needs: belonging, self-esteem, control, and meaningful existence. When individuals are ostracized, social ties are severed and the need for belonging is threatened. Self-esteem is threatened following ostracism because individuals often make self-deprecating attributions. Moreover, ostracized individuals are not able to reason or argue with the ostracizer(s) and thus feel they are not in control of their outcomes. Finally, the lack of acknowledgment that results from being ostracized threatens the perception of a meaningful existence, leaving individuals feeling worthless and invisible (Case & Williams, 2004; Williams, 2009).

The Temporal Need Threat model also posits that ostracism’s effects occur in stages (Williams, 2009). During the reflexive stage, ostracism is universally painful, and along with decreased need satisfaction people report negative affect like anger and sadness. After the initial sting of ostracism, individuals ponder why they were ostracized and try to restore their threatened needs during the reflective stage. Finally, in cases of chronic ostracism, the resources necessary to cope are depleted and individuals enter the resignation stage which elicits devastating effects for the person experiencing it such as depression and alienation (Riva et al., 2017; Williams, 2009).

Given the often-detrimental effects of ostracism, death-related themes are often present within the ostracism literature to illustrate the intense discomfort that targets of ostracism feel. In work describing the immediate and delayed consequences of ostracism, Williams (2007b) refers to ostracism as the ‘kiss of social death.’ Wesselmann and Williams (2017) refer to ostracism as the ‘social death penalty’ and include the phrases ‘social life’ and ‘social death’ within their title. Similarly, Ouwerkerk et al. (2005) use ‘social death’ in their title, Avoiding the Social Death Penalty: Ostracism and Cooperation in Social Dilemmas. Moreover, evidence of how the death metaphor influences the questions researchers ask within the ostracism literature is present in work by Rudert et al. (2017) who explicitly refer to ostracism as social death and examine whether simply being acknowledged (even if the acknowledgment is hostile) is preferable to being ostracized completely.

Even when the phrase social death is not used explicitly, death-related themes are often present within the ostracism literature. For example, in their work on ostracism and the benefits of prayer and self-affirmation, Hales et al. (2016) begin with the question, ‘What does it feel like to be a ghost?’ In other work, Hales et al. (2020) emphasize ostracism’s effect on meaningful existence, and discuss how it produces similar effects to mortality salience, citing Steele et al. (2015). The death metaphor is also present when examining earlier representations of ostracism within psychology. For example, in Principles of Psychology, James (1890) likened being ignored and excluded by others to being ‘cut dead’ and mentioned that even the cruelest of bodily tortures would be preferable to receiving no acknowledgment at all.

Theoretically, immediate, gross detection of ostracism has been associated with the avoidance of death among social animals. Williams (2009) maintains that ostracism likely results in death for non-human social animals, humans belonging to tribes, and may have resulted in death for our ancestors due to the breaking of social connections that enhance the likelihood of survival and reproduction. This evolutionary component of the Need Threat Model may explain why reactions to ostracism can be severe and has influenced some researchers to adopt an evolutionary approach to the study of ostracism (e.g., Shilling & Brown, 2016; Wesselmann et al., 2012). Considering death’s influence within the ostracism literature, we compared the psychological effects of ostracism to those of mortality salience by turning to the literature on Terror Management Theory.

Mortality salience and its relation to ostracism

Terror Management Theory maintains that because humans can think about death complexly, they are susceptible to the terror associated with facing an unavoidable end and inhabiting an uncontrollable universe (Solomon et al., 1991). Thus, Terror Management Theory posits that human behavior is motivated by a desire to keep nonconscious death-related cognitions at bay (Solomon et al., 1991). Individuals non-consciously primed with mortality salience also show greater activation of death-related thoughts or death-thought accessibility (for a review see Hayes et al., 2010).

Separate lines of research suggest mortality salience and ostracism produce similar effects. First, both literatures on ostracism and mortality salience discuss effects on meaning, and how strengthening one’s sense of meaning may facilitate recovery. According to Terror Management Theory, we manage the discomfort associated with facing death through our cultural worldviews, which imbue life with meaning, stability, and order (Solomon et al., 1991). For example, research shows that individuals defend themselves by becoming more punitive of those who violate cultural norms when mortality is made salient (Florian & Mikulincer, 1997). Similarly, following ostracism, individuals work to restore their need for a meaningful existence, which is operationalized as feeling acknowledged and feeling that one’s life is significant (Williams, 2009). However, the literatures on ostracism and mortality salience conceptualize meaning in distinct ways. For example, meaningful existence as discussed within the ostracism literature focuses more on the self, and meaning as discussed within the mortality salience literature focuses on a diverse range of beliefs and values. It is, thus, possible for different threats, like ostracism or mortality salience, to affect meaningful existence and worldview defense in distinct ways. Additionally, ostracism and mortality salience both involve a desire to improve self-esteem. Like attempts to restore self-esteem following ostracism in the reflective stage (Williams, 2009), mortality salience increases attempts to bolster self-esteem because of the protective function it incurs for the individual (for a review see Pyszczynski et al., 2004).

Moreover, ostracism and mortality salience both affect belonging. According to the need fortification hypothesis of the Temporal Need Threat Model of Ostracism, individuals attempt to restore belonging by affiliating with others and by increasing their social servility – or the tendency for a person to be so concerned with fitting in that they become easily influenced by others (Carter-Sowell et al., 2008; DeWall, 2010; Williams, 2009). For example, previous research demonstrated that ostracized individuals displayed increased social servility by being more likely to obey a command to take 39 unique and creative photographs outdoors amid frigid winter weather compared to those who were not ostracized (Riva et al., 2014); and ostracized college students were more likely than those who were included to comply with a request to donate money to the university’s marching band (Carter-Sowell et al., 2008). Mortality salience also seems to influence belonging. For example, Florian et al. (2002) found that mortality salience increased commitment to romantic partners. Moreover, Wisman and Koole (2003) found that mortality salience increased the desire to affiliate with strangers, with mortality salience-primed participants more likely to sit next to strangers during a group discussion.

Finally, ostracism and mortality salience both involve threats to control. Following ostracism, individuals are motivated to behave in ways that will restore control, sometimes resorting to aggressive means (Warburton et al., 2006). Terror Management Theory also maintains that individuals are cognitively and behaviorally influenced by mechanisms (e.g., cultural worldviews) that help them regain feelings of control by reducing mortality salience threats (Pyszczynski et al., 1998). With these similarities in mind, we believe that mortality salience may also threaten need satisfaction.

Despite these similarities, at least one difference exists. Ostracism involves the direct experience of being unacknowledged and feeling invisible (Case & Williams, 2004). Although mortality salience may trigger meaninglessness (Pyszczynski et al., 2004), an individual is not actively being ignored by others. Thus, the effect of ostracism on need satisfaction might be stronger than that of mortality salience. Research by Van Beest et al. (2011) suggests that ostracism and mortality salience may produce distinct results. They found that being ostracized from a symbolically life-threatening situation decreased need satisfaction and worsened mood to a lesser extent than being ostracized from a non-life-threatening situation. However, because ostracism was compared to a condition involving mortality salience and ostracism, we cannot compare both constructs using their findings.

Research comparing the effects of ostracism to mortality salience

Previous research has examined whether ostracism produces effects like those of mortality salience, but it is limited. Steele et al. (2015) found that ostracized individuals reported higher death-thought accessibility compared to included individuals. However, they did not employ a mortality salience manipulation in their study, which precluded them from directly comparing the effects of ostracism to mortality salience. Moreover, Steele et al. (2015) recruited a small sample (N = 35), which affects the replicability and generalizability of their results. Similarly, Yaakobi (2019) found that ostracism led to higher death-thought accessibility compared to inclusion and that death-thought accessibility mediated the relationship between ostracism and decreased need satisfaction. However, Yaakobi (2019) does not directly compare the effects of ostracism to mortality salience. Thus, studies directly comparing ostracism to mortality salience on outcomes important to both constructs with adequately sized samples would elucidate the similarities or differences between both constructs.

Although some research suggests that the effects of ostracism are like those of mortality salience, there is also research implying their effects differ. A meta-analysis of mortality salience studies found that when negative topics such as social exclusion were used as control groups, they did not produce the same effects as mortality salience (Burke et al., 2010). This finding suggests that there is a unique element to mortality salience that is not shared by negative or threatening control group topics such as social exclusion. We aimed to clarify the inconsistent pattern of findings in the literature.

Current studies

Given the impact of death-themed language on ostracism research, we sought to provide strong tests of the social death metaphor, by determining whether ostracism and mortality salience produce similar or distinct effects on outcomes relevant to both constructs. To do this, we used adequately powered samples and included conditions that allowed us to directly compare the effects of mortality salience to those of ostracism and make stronger conclusions about their effects. Our goals were to assess whether ostracism and mortality salience produce similar or different effects on measures of death-thought accessibility, worldview defense, need satisfaction, and mood. Considering that chronic ostracism is more severe, we also wanted to examine whether it would be associated with death-thought accessibility, an outcome of mortality salience.

Studies 1, 2, 3a, and 3b employed a single-factor (mortality salience vs. ostracism vs. control) between-subjects design. Study 4 was correlational and examined the association between chronic ostracism and death-thought accessibility. For studies 1 and 2, we had four research questions: we wondered whether ostracism and mortality salience would 1) increase death-thought accessibility, 2) decrease need satisfaction, 3) worsen mood, and 4) increase worldview defense compared to a control condition.

Study 1

For the first study, we used established methods in both literatures to explore the similarities between ostracism and mortality salience. To prime participants with mortality, we asked them to write about their own death (Rosenblatt et al., 1989), which is a method commonly used to manipulate mortality salience that yields moderate effects (Burke et al., 2010). Similarly, to prime participants with ostracism, we asked them to recall a time they felt socially excluded (e.g., Pickett et al., 2004). This ostracism manipulation also tends to yield moderate to large effect sizes (e.g., Chen et al., 2020; Godwin et al., 2014; Zhong & Leonardelli, 2008). Using this information, we selected a moderate effect size for our power analysis. As a neutral control condition, we had participants write about what they do and how they feel on a typical Wednesday.

Method

Participants

Introduction to psychology students (58.4% female; 62.5% White; M_age=19.25, SD_age = 1.39) were recruited from a large Midwestern university. An a-priori power analysis to achieve 80% power (alpha=.05; partial η² = 0.13¹) determined a desired sample size of 69 participants. Due to multiple research questions of interest and exploratory analyses we chose to continue data collection until the end of the semester, with 69 participants as the minimum. We originally recruited 303 participants. However, four participants did not provide their age, one participant was underage, and seven participants took the survey more than once, and all their attempts were deleted. This left us with a final sample of 291 participants who were randomly assigned to write about ostracism (n=97), their own death (n=100), or a neutral control (n=94). All participants received course credit following participation.

Measures²

All measures not relevant to the primary analyzes are described in the supplemental materials.

Death-thought accessibility

Participants were presented with 20 word fragments, eight of which could be completed using death-related words or non-death words (e.g., ‘ki__ed’ can be killed or kissed; Greenberg et al., 1994). The number of death words created was used to indicate death-thought accessibility. Using a two-way mixed effects model, we calculated the absolute agreement between the death-word data of three coders, and found high agreement, with an average measure intraclass correlation coefficient of 0.996, 95%CI:0.995, 0.996, F(290,580) = 223.07, p < .001.

Need satisfaction

Participants completed the need satisfaction questionnaire regarding how they felt during the writing task (12 items total, three items for each of the four needs, e.g., ‘I felt disconnected’), rated on a 5-point scale from 1 (not at all) to 5 (extremely; Williams, 2009). Higher numbers indicate more need satisfaction (α=.86).

Mood

Participants were presented with a list of 12 adjectives (e.g., bad) rated from 1 (not at all) to 5 (extremely; Williams, 2009). Instructions were identical to those of the need satisfaction questionnaire. Higher scores indicate more negative mood (α=.91). Both the psychological need satisfaction measure and mood measures are commonly used to assess the effects of ostracism.

Worldview defense

Participants were presented with the police report of a woman who had been arrested by the local police department for prostitution. Participants were instructed to set her bail ranging from $0–$999, with higher numbers indicating more worldview defense (e.g., Rosenblatt et al., 1989). The prostitution bail measure is a commonly used measure of worldview defense within the Terror Management literature.

Procedure

This study was conducted online through Qualtrics. Following consent, participants were randomly selected to write a brief response to one of three writing prompts. In the mortality salience condition, participants wrote about the emotions they would attribute to their own death and what the experience of dying would be like. In the ostracism condition, participants wrote about a time that they felt socially excluded or ignored, what occurred, and how they felt. Participants in the control condition wrote about what they experience on a typical Wednesday (e.g., where they normally go, what they do, and how they typically feel) as well as their routine.

After responding to the writing prompt, participants completed the word-fragment task, rated their need satisfaction and mood, engaged in a filler task used as a delay to ensure the nonconscious awareness of death, and completed the measure of worldview defense. Then, participants completed a second worldview defense measure on bail for shoplifting.³ Additionally, participants completed measures assessing attitudes towards prostitution, reflective need satisfaction and mood (participants rated how they felt at the moment), personality, and trait self-esteem. Finally, participants provided demographic information, and they completed a manipulation check.

Results

Manipulation check

We asked participants to select the prompt they wrote about during the writing task: either ostracism, death, or typical Wednesday. A Chi-Squared Test (3×3) demonstrated a significant relationship between condition and the writing prompt variable χ²(4,291) = 502.82, p < .001, indicating our manipulations were successful. Within the mortality salience condition, 97% selected the death prompt, within the ostracism condition 89.7% selected the ostracism prompt, and within the control condition 98.9% selected the typical Wednesday prompt. Additionally, two independent coders who were blind to condition analyzed the content of participant essays to ensure that participants were writing about what they were instructed to write about. They rated each essay on the degree to which the essay was death-related, related to ostracism, or related to what people typically do in a day. All intra-class correlation coefficients were above .9, indicating a high degree of inter-rater reliability for each rating type. Condition was also significantly associated with each essay rating, ps < .001, indicating that participants wrote about what they were instructed to write about.⁴

Primary Analyses⁵

Death-thought accessibility

A significant One-Way ANOVA suggested that differences in death-thought accessibility were dependent upon condition, F(2,288) = 7.22, p < .001, η_p² = 0.05. Post hoc analyses with Bonferroni correction revealed that participants primed with mortality (M = 3.04, SD = 1.54) produced more death words compared to those who were ostracized (M = 2.35, SD  = 1.31), t(288) = 3.57, p =.001, 95% CI[0.22,1.15], d = 0.48, and compared to those in the control condition (M = 2.48, SD = 1.19), t(288) = 2.88, p=.01, 95%CI[0.09,1.03], d = 0.41. Those who were ostracized did not differ significantly in death-thought accessibility from those in the control condition, t(288)=-0.66, p = 1.00, 95%CI[−0.60,0.34], d=- 0.10. See Figure 1.

Figure 1. Effect of condition on death-thought accessibility for Study 1.

Note. Mortality salience increases death-thought accessibility relative to ostracism and control. Error bars represent standard errors.

Worldview defense

We conducted a One-Way ANOVA to examine the effect of condition on worldview defense. The overall model did not reach significance, F(2,284) = 1.06, p = 0.35, η_p² = .01.

Need satisfaction

A significant One-Way ANOVA, F(2,287) = 22.51, p < .001, η_p² = 0.14, suggests that differences in need satisfaction were dependent upon condition. Post hoc analyses revealed that ostracized participants (M = 2.84, SD = 0.86) reported lower need satisfaction compared to those in the mortality salience condition (M = 3.33, SD = 0.54), t(287) = -5.15, p < .001, 95%CI[−0.73,-0.26],d = -0.71, and compared to those in the control condition (M = 3.46, SD = 0.58), t(287) = -6.31, p < .001, 95%CI[−0.85,-0.38],d=- 0.86. There was no difference in need satisfaction between those in the control and mortality salience conditions, t(287) = 1.25, p = 0.64,95%CI[−0.11,0.35], d = 0.22. See Figure 2.

Figure 2. Mean need satisfaction by condition for Study 1.

Note. Ostracized participants reported decreased need satisfaction compared to control and mortality salience participants. Error bars represent standard errors.

Mood

A One-Way ANOVA revealed a significant effect of experimental condition on mood, F(2,288) = 18.99, p < .001, η_p² = 0.12. Post hoc analyses showed that ostracized participants reported more negative mood (M = 2.84, SD = 0.94) compared to those in the mortality salience condition (M = 2.47, SD=0.74), t(288) = 3.21, p=.004, 95%CI[0.09,0.65], d =  0.44, and compared to those in the control condition (M = 2.12, SD=0.75), t(288) = 6.16, p < .001, 95%CI[0.44,1.01], d = 0.86. Those in the mortality salience condition reported worse mood than those in the neutral control condition, t(288) = 3.02, p=.008, 95%CI[0.07,0.63], d = 0.48.

Discussion

We examined the effects of ostracism and mortality salience on variables associated with each construct. Specifically, we wondered whether ostracism and mortality salience would 1) increase death-thought accessibility, 2) decrease need satisfaction, 3) worsen mood, and 4) increase worldview defense compared to a control condition. In the first study we found that mortality salience increased death-thought accessibility compared to both ostracism and the control condition. We also found that ostracism reduced need satisfaction and increased negative mood compared to both mortality salience and the control condition. Although, participants in the mortality salience condition reported more negative mood than those in the control condition, our results suggest that ostracism has the most negative effect on mood.

Some of our results contradict existing theory and previous findings. Mortality salience did not increase worldview defense compared to the control condition, as it should have according to Terror Management Theory. It is possible that our measure of worldview defense, the prostitution bail stimulus, was not believable or was difficult to comprehend. However, we included the same instructions used successfully in previous research (Boyd et al., 2017; Rosenblatt et al., 1989). Thus, it is possible that the contradictory findings of the first study, were spurious findings. A follow-up study was needed to further elucidate this finding.

Previous research by Steele et al. (2015) found that ostracism produced higher death-thought accessibility compared to inclusion. We wondered if our results differed because of the manipulation we used. To examine this, we designed the second study to employ the Cyberball paradigm (Williams et al., 2000), an online ball-tossing game commonly used to manipulate ostracism, as in the study by Steele et al. (2015).

Study 2

In the second study, we used Cyberball to manipulate ostracism and inclusion, and we designed a mortality salience condition to resemble the two Cyberball conditions. We called our new manipulation of mortality salience, Cybergrave. When creating Cybergrave, we were initially inspired by Pyszczynski et al. (1996), who had participants complete their study while standing in front of a funeral home, which represented a naturally occurring prime for mortality salience. We wondered if we could use tombstones within the context of Cyberball to manipulate mortality salience. In addition, we drew inspiration for the instructions and appearance of Cybergrave from Cybertree – a control condition for Cyberball designed by Dvir et al. (2019) which involves the mental visualization of trees in place of the Cyberball players.

Additionally, we replaced the prostitution bail measure with a measure gauging the support for killing animals, which is an indicator of worldview defense because the human body can elicit death anxiety due to its creatureliness and ultimate surrender to death (Goldenberg et al., 2000). For example, animals and their susceptibility to death serve as reminders of our own mortality because of what we share in common with them; we are also animals, and we also ultimately succumb to death. Mortality salience is then hypothesized to increase the support for killing animals as an attempt to separate the self from other non-human animals. In support of this notion, Lifshin et al. (2017) showed that those primed with mortality supported the killing of animals more regardless of gender, whether the killing was justified, or their prior attitudes toward killing animals. Our research questions were the same as in Study 1. The purpose of Study 2 was to examine whether the initial pattern of results would replicate using different methods or whether our results would instead indicate similarities between ostracism and mortality salience.

Method

Participants

Introduction to psychology students (64.8% female; 66.7% White; M_age=18.69, SD_age = 0.97) were recruited from a large Midwestern university. We conducted a priori power as we did for Study 1 and determined at least 69 participants would need to be recruited for 80% power. We originally recruited 448 participants. However, eight participants did not consent to the use of their data and 14 participants did not provide their age or were underage, and three took the study more than once. Our final sample consisted of 423 participants who were randomly assigned to be ostracized (n=175), included (n=175), or to mentally visualize tombstones (n=73). They received course credit following participation.

Measures⁶

Worldview defense. Participants rated their support for killing animals (16 items; e.g., ‘It is often necessary to control for animal overpopulation through different means, such as hunting, or euthanasia’; Lifshin et al., 2017) on a 7-point scale ranging from 1 (strongly disagree) to 7 (strongly agree). Higher scores indicate increased support for killing animals (α=.90).

Procedure

This study was conducted online through Qualtrics. Following consent, participants were randomly assigned to one of three conditions. In the ostracism and inclusion conditions, participants were told to mentally visualize the scene while playing an online ball tossing game. Ostracized participants received the ball a couple of times and were excluded for the remainder of the game, whereas those who were included received the ball equally. Participants in the Cybergrave condition were instructed to mentally visualize the scene and click on the parts of the scene that attracted their attention. They were then presented with a white background and three tombstones arranged in the positions that the three players would be arranged in for the Cyberball conditions. Additionally, in Cybergrave a dot would temporarily appear on the screen wherever a participant would click. See Figures 3 and 4 for images of the Cybergrave condition. All conditions lasted around one minute and 50 seconds. The procedure order was nearly identical to Study 1, with exception of a second filler task following the first, included to ensure non-conscious activation of mortality salience during the worldview defense measure. We also did not include any exploratory worldview defense measure or measure of attitudes toward prostitution.

Figure 3. Instructions for Cybergrave.

Note. The image above depicts the instructions for Cybergrave. This is the first screen participants are shown when they begin the task.

Figure 4. Cybergrave mortality salience manipulation.

Note. The imagine above depicts the Cybergrave mortality salience manipulation. When participants click around the screen a dot appears on the location they click on.

Results

Manipulation check

Participants were asked to select which mental visualization task they participated in: (1) a ball toss game in which they were included, (2) a ball toss game in which they were excluded, (3) a set of tombstones. A Chi-Squared Test (3×3) demonstrated a significant association between condition and the manipulation check variable χ²(4,422) = 611.96, p < .001. In the mortality salience condition, 97.3% selected the tombstones option. In the ostracism condition, 87.4% selected the excluded option. In the inclusion condition, 85.1% selected the included option.

Primary analyses

Death-thought accessibility

A One-Way ANOVA revealed a significant effect of condition on death-thought accessibility, F(2,420) = 18.74, p < .001, η_p² = 0.08. Post hoc analyses showed that those primed with mortality produced more death words (M=3.53, SD=1.71) than those who were ostracized (M=2.42, SD=1.32), t(420) = 5.75, p < .001, 95%CI[0.65,1.58], d = 0.74, and compared to those who were included (M=2.46, SD=1.32), t(420) = 5.55, p < .001, 95%CI[0.61,1.54], d = 0.71. Those who were ostracized and those who were included did not differ significantly in death-thought accessibility, t(420)=-0.27, p = 1.00, 95%CI[−0.40,0.32], d=- 0.03. See Figure 5.

Figure 5. Mean death-thought accessibility by condition for Study 2.

Note. Mortality salience increases death-thought accessibility relative to ostracism and inclusion. Error bars represent standard errors.

Worldview defense

A One-Way ANOVA revealed a significant effect of condition on worldview defense, F(2,420) = 4.82, p=.009, η_p² = 0.02. When all data were included post hoc analyses showed that ostracized participants (M=3.54, SD=1.02) supported killing animals less than included participants (M=3.82, SD=1.01), t(420)=- 2.58, p=0.03, 95%CI[−0.54, −0.02], d =-0.28, and less than those primed with mortality (M=3.90, SD=1.03), t(420)=-2.58, p=.03, 95%CI[−0.71,- 0.03], d=-0.36. Additionally, participants primed with mortality and those who were included did not differ significantly in the support for killing animals, t(420) = 0.60, p = 1.00, 95%CI[−0.25,0.43], d = 0.08. However, when we filtered out those who failed the manipulation check, the difference in support for killing animals between ostracized and included participants was no longer statistically significant, t(369)=-2.35, p=.06, 95%CI[−0.57, 0.01], d=-0.27. See Figure 6.

Figure 6. Mean support for killing animals by condition for Study 2.

Note. Mortality salience increases support for killing animals compared to ostracism, but not compared to inclusion. The means depicted were calculated with the manipulation check failures filtered out. Error bars represent standard errors.

Need satisfaction

The One-Way ANOVA was significant, F(2,420) = 42.24, p < .001, η_p² = 0.17. Post hoc analyses revealed that participants in the ostracism condition (M=2.42, SD=0.93) reported lower need satisfaction compared to those in the mortality salience condition (M=3.02, SD=0.69), t(420)=-5.28, p < .001, 95%CI[−0.87,-0.32], d =-0.74, and compared to those in the inclusion condition (M=3.20, SD=0.73), t(420) =-8.97, p < .001,95%CI[−0.99,-0.57], d=-0.94. When all data were included, there were no significant differences between those in the inclusion and mortality salience conditions, t(420) = 1.61, p = 0.33, 95%CI[−0.09, 0.45],d = 0.26. However, when we filtered out participants who failed the manipulation check, those in the inclusion condition (M=3.29, SD=0.70) reported more need satisfaction than those in the mortality salience condition (M=3.00, SD= 0.69), t(369) = 2.54, p = 0.32, 95%CI[0.02,0.56], d = 0.41. See Figure 7.

Figure 7. Mean need satisfaction by condition for Study 2.

Note. The means depicted here were calculated with manipulation check failures filtered out. Error bars represent standard errors.

Mood

The One-Way ANOVA was significant F(2,420) = 23.05, p < .001, η_p² = 0.10. Post hoc analyses revealed that participants in the ostracism condition (M = 2.82, SD = 0.90) reported more negative mood compared to those in the mortality salience condition (M = 2.52, SD = 0.77), t(420) = 2.69, p=.02, 95%CI[0.03,0.57], d = 0.36, and compared to those in the inclusion condition (M = 2.24, SD = 0.71), t(420) = 6.79, p < .001, 95%CI[0.38,0.80], d = 0.72. Those in the mortality salience condition also reported more negative mood than those in the inclusion condition, t(420) = 2.52, p = 0.04,95%CI[0.01,0.55], d = 0.38.

Discussion

Using different methods across two studies, a consistent pattern of results began to emerge. Our results mirrored those of Study 1. We found that participants in the mortality salience condition showed increased death-thought accessibility compared to those who were ostracized and included, indicating that ostracism does not elicit effects commonly associated with mortality salience primes. We also found that ostracized participants showed decreased need satisfaction and worse mood compared to those in the mortality salience condition and compared to those who were included. Additionally, we found that mortality salience decreased need satisfaction and increased negative mood compared to inclusion.

One difference between the two studies is that the effect of condition on worldview defense was significant in Study 2. Although mortality salience increased support for killing animals compared to ostracism, it did not do so compared to inclusion. It is possible that ostracism is driving the effect by decreasing the support for killing animals compared to mortality salience. If this is the case, then ostracized individuals may have shown decreased support for killing animals either because they felt dehumanized (Bastian & Haslam, 2010) and may have been able to relate more to non-human animals or due to a desire to engage in anthropomorphism following ostracism (Brown et al., 2016). However, although ostracized participants on average supported killing animals the least, the difference in support for killing animals between ostracized and included participants was no longer significant after filtering out those who failed the manipulation check.

Moreover, the inclusion condition in Cyberball tends to have neutral effects on participants (Dvir et al., 2019), which would provide us with a neutral comparison group for the mortality salience and ostracism manipulations. With this in mind, we are not sure why our results did not support existing theory that mortality salience leads to increased worldview defense through endorsement for killing animals (Goldenberg et al., 2000). Given the inconsistency in worldview defense results from studies 1 and 2, we used the same measure of support for killing animals in follow-up studies to determine whether this inconsistency was due to a difference in the measure used.

There were fewer participants in the mortality salience condition than in the ostracism and inclusion conditions and we felt that participants assigned to the mortality salience condition were more apt to not complete the study. Moreover, the instructions for Cybergrave mistakenly had participants mentally visualize trees (from Cybertree), though there were no trees on the screen. To address these issues, we conducted similar studies with corrected instructions.⁷

Studies 3a and 3b

The purpose of studies 3a and 3b was to clarify the remaining uncertainty involving effects on worldview defense. Studies 3a and 3b were identical. Study 3a was conducted in person prior to COVID-19quarantine⁸ and study 3b was conducted online during quarantine. Both studies were pre-registered on the Open Science Framework (https://osf.io/3tajq), and we have reported all pre-registered hypotheses in the body of the paper.⁹ Pre-registration occurred prior to data analysis. Given our consistent pattern of findings for the effects of ostracism and mortality salience and support from existing theory (Solomon et al., 1991; Williams, 2009), we had two main hypotheses for studies 3a and 3b: (1) We hypothesized that mortality salience would increase death-thought accessibility compared to ostracism and inclusion, but that (2) ostracism would decrease need satisfaction and increase negative mood compared to mortality salience and inclusion.

Method

Participants

For studies 3a (50% female; 52% White, M_age = 19.44, SD_age = 1.55) and 3b (63.3% male; 66.8% White, M_age = 19.64, SD_age = 1.85), we recruited introduction to psychology students from a large Midwestern university in the United States. Using η_p² = 0.13¹⁰ for the effect of condition on death-thought accessibility, we determined at least 85 participants would be needed to achieve a power of .80. Some exploratory analyses involved interactions, so we recruited participants until the end of the semester.

In Study 3a, we originally recruited 101 participants prior to being quarantined, however one participant was removed from the dataset because they did not consent to the use of their data. In Study 3b, we originally recruited 371 participants but 21 were removed from the data due to being underage, not providing age, or not consenting to the use of their data. One participant was removed entirely from the data because they took the survey twice, leaving us with a final sample of 349 participants for Study 3b. Participants were randomly assigned to be ostracized (Study 3a:n = 36; Study 3b:n = 108), included (Study 3a:n = 32; Study 3b:n = 103) or to mentally visualize tombstones (Study 3a:n = 32; Study 3b:n = 138). All participants received course credit following participation.

Measures

Manipulation checks

All measures were identical to Study 2, although we added six new manipulation check items. We asked participants the extent to which they felt ignored and excluded rated on a 5-point scale from 1 (not at all) to 5 (extremely). Participants who played Cyberball were asked to estimate the percentage of ball throws they received. Finally, all participants were asked three questions probing what they were asked to do during the mental imagery task, and they indicated either ‘yes’ or ‘no’ to each: 1) play a ball game, 2) look at graves, and 3) imagine a cow pasture.

Procedure

Both studies were taken on Qualtrics. The procedure for studies 3a and 3b was identical to Study 2, excluding the new manipulation check questions we added to studies 3a and 3b. For Study 3b, we over-sampled participants in the Cybergrave condition to address the condition-specific participant attrition that occurred in our previous online studies.

Results

Manipulation Checks¹¹

For studies 3a and 3b, the 3 × 2 Chi Square Tests revealed significant associations between condition and ‘Play a ball game,’ χ²(2,99)_3a = 99, p < .001; χ²(2, 344)_3b = 299.18, p < .001. In Study 3a, for example, 100% of participants in the mortality salience condition answered ‘no’, within the ostracism and inclusion conditions 100% of participants answered ‘yes.’ In both studies, there were also significant associations between ‘Look at graves’ and condition, χ²(2,98)_3a = 98, p < .001; χ²(2,340)_3b = 283.99, p < .001. In Study 3b, for example, 99.3% of participants in the mortality salience condition answered ‘yes,’ within the ostracism condition 96.10% answered ‘no’ and within the inclusion condition 89.9% answered ‘no.’ All Study 3a and most Study 3b participants selected ‘no’ for ‘Imagine a cow pasture.’

Overall, results of these analyses suggest our manipulations of ostracism, mortality salience and inclusion were successful.

Primary analyses

Death-thought accessibility

One Way ANOVAs revealed that the effect of condition on death-thought accessibility was significant for Study 3a, F(2,97) = 5.58, p=.005, η_p² = 0.10, and for Study 3b, F(2,346) = 20.98, p < .001, η_p² = 0.11. For both studies, mortality salience increased death-thought accessibility compared to ostracism and compared to inclusion. There were no significant differences in death-thought accessibility between ostracized and included participants in either study. See Table 1.

Table 1. Post hoc comparisons of death-thought accessibility by condition for studies 3a and 3b.

Study	Comparison		Mean Difference	t	df	SE	p	95% CI	d
3a	Mortality	Ostracism	1.06	2.90	97	0.36	0.014	0.17, 1.95	0.67
		Inclusion	1.09	2.91	97	0.38	0.013	0.18, 2.0	0.69
	Ostracism	Inclusion	0.03	0.10	97	0.36	1.00	−0.85,0.92	0.03
3b	Mortality	Ostracism	0.96	5.19	346	0.19	<0.001	0.5, 1.41	0.67
		Inclusion	1.08	5.74	346	0.19	<0.001	0.63, 1.53	0.73
	Ostracism	Inclusion	0.12	0.59	346	0.20	1.00	−0.36, 0.59	0.09

Note. Post hoc comparisons with Bonferroni correction of all conditions are shown for studies 3a and 3b, where 95% CI represents the 95% confidence interval (lower bound, upper bound), and d represents Cohen’s d.

Worldview defense

The One-Way ANOVAs used to examine the effect of condition on worldview defense, were both non-significant for studies 3a, F(2,97) = 0.38, p=.682, η_p² = 0.01, and 3b, F(2,346) = 0.18, p=.837, η_p² = 0.001.

Need satisfaction

For studies 3a and 3b, the One-Way ANOVAs were both significant, indicating an effect of condition on need satisfaction, F(2,97)_3a = 17.48, p < .001, η_p² = 0.27; F(2,346)_3b = 46.06, p < .001, η_p² = 0.21. In both studies, post hoc analyzes revealed that ostracized participants reported lower need satisfaction compared to those in the mortality salience condition and compared to included participants. There were no differences between the inclusion and mortality salience conditions in either study. See Table 2.

Table 2. Post hoc comparisons of need satisfaction by condition for studies 3a and 3b.

Study	Comparison		Mean Difference	t	df	SE	p	95% CI	d
3a	Ostracism	Mortality	−0.85	−5.41	97	0.16	<0.001	−1.23, −0.46	−1.38
		Inclusion	−0.73	−4.66	97	0.16	<0.001	−1.11, −0.35	−1.06
	Inclusion	Mortality	−0.12	−0.73	97	0.16	1.00	−0.51,0.28	−0.19
3b	Ostracism	Mortality	−0.83	−8.95	346	0.09	<0.001	−1.05, −0.60	−1.14
		Inclusion	−0.75	−7.64	346	0.10	<0.001	−0.99, −0.52	−0.96
	Inclusion	Mortality	−0.07	−0.75	346	0.09	1.00	−0.29, 0.15	−0.11

Note. Post hoc comparisons with Bonferroni correction of all conditions are shown for studies 3a and 3b, where 95% CI represents the 95% confidence interval (lower bound, upper bound), and d represents Cohen’s d.

Mood

In studies 3a and 3b both One-Way ANOVAs indicated a significant effect of condition on mood, F(2,97)_3a = 4.58, p < 0.05, η_p² = 0.09; F(2,346)_3b = 14.78, p < 0.001, η_p² = 0.08. Post hoc analyses revealed ostracized participants in both studies reported more negative mood compared to mortality salience and included participants. In both studies, mortality salience did not differ from inclusion. See Table 3.

Table 3. Post hoc comparisons of negative mood by condition for studies 3a and 3b.

Study	Comparison		Mean Difference	t	df	SE	p	95% CI	d
3a	Ostracism	Mortality	0.47	2.57	97	0.18	0.035	0.02, 0.91	0.62
		Inclusion	0.48	2.62	97	0.18	0.03	0.03, 0.92	0.62
	Inclusion	Mortality	−0.01	−0.06	97	0.19	1.00	−0.47, 0.48	−0.01
3b	Ostracism	Mortality	0.50	5.05	346	0.10	<0.001	0.26, 0.74	0.66
		Inclusion	0.46	4.36	346	0.11	<0.001	0.21, 0.72	0.57
	Inclusion	Mortality	0.04	0.38	346	0.10	1.00	−0.20, 0.28	−0.05

Note. Post hoc comparisons with Bonferroni correction of all conditions are shown for studies 3a and 3b, where 95% CI represents the 95% confidence interval (lower bound, upper bound), and d represents Cohen’s d.

Discussion

Studies 3a and 3b continue to support the notion that mortality salience and ostracism produce distinct effects. In studies 3a and 3b we expected that mortality salience would increase death-thought accessibility compared to ostracism and inclusion and that ostracism would decrease need satisfaction and increase negative mood compared to mortality salience and inclusion. As in the previous studies, mortality salience produced higher death-thought accessibility compared to both ostracism and inclusion. Ostracism lowered need satisfaction and worsened mood compared to both mortality salience and inclusion. The studies conducted thus far compared momentary ostracism to mortality salience. However, uncertainty remained as to whether chronic ostracism’s effects would resemble those of mortality salience, given its severity.

Study 4

The fourth study examined the relationship between chronic ostracism and death-thought accessibility, with the goal of determining whether the effects of chronic ostracism resemble those of mortality salience. Experimental research on chronic ostracism is difficult to conduct due to ethical concerns, so we chose a correlational design. Research links chronic ostracism to severe effects like depression, alienation, detachment, suicidal ideation, and suicide attempts (Riva et al., 2017; Williams, 2001). Although we repeatedly demonstrated differences between temporary ostracism and mortality salience, we considered that chronic ostracism might be perceived as a permanent state and its effects might resemble those of mortality salience. To address this question, we examined the correlation between a measure of chronic ostracism and death-thought accessibility. We also examined correlations between chronic ostracism and general need satisfaction and negative mood. We wondered whether 1) chronic ostracism would be positively correlated with death-thought accessibility, whether 2) chronic ostracism would be negatively correlated with general need satisfaction, and whether 3) chronic ostracism would be positively correlated with negative mood. We pre-registered this study on the Open Science Framework (https://osf.io/5adxu). Pre-registration occurred prior to data analysis.

Method

Participants

Participants were recruited from a large Midwestern university. An a priori power analysis using a small effect, r = 0.2, suggested we would need 193 participants to achieve 80% statistical power. To account for possible missing data, we oversampled by about 25%. We recruited 250 participants. After excluding one 16-year-old and eight participants who did not consent to the use of their data, the total sample consisted of 241 participants (51.9% female; 63.9% White; M_age = 19.34, SD_age = 1.43).

Measures

Chronic ostracism

Participants rated how often they generally experienced what was described by each item (8 items; e.g., ‘In general, others leave me out of their group’; Carter-Sowell, 2010) on a 7-point scale from 1 (hardly ever) to 7 (almost always). Higher scores indicate more chronic ostracism (α=.94).

Death-thought accessibility

We used the same word completion measure used in previous studies.

General need satisfaction

Participants rated 24 items (e.g., ‘I feel invisible’) according to how they feel in general on a 5-point scale from 1 (not at all) to 5 (extremely). We used a longer version of the Need Satisfaction Scale that also tapped into the need for certainty. Higher scores indicate more general need satisfaction (α=.94).

General mood

We used the same measure of mood as in previous studies. Higher scores indicate more negative mood (α=.86).

Depression

We measured depression using the Center for Epidemiologic Studies Depression Scale Revised (CESDR; Eaton et al., 2004). Participants were shown 20 items (e.g., ‘My sleep was restless’) and were asked to rate each item on using the following 5-point scale: 1) Not at all or less than 1 day, 2) 1–2 days, 3), 3–4 days, 4) 5–7 days, 5) Nearly every day for 2 weeks. Scores across the 20 items were summed to form a composite of depression with higher numbers indicating more severe depression (α = 0.94).

Emotional stability

We used the emotional stability subscale of the Ten-Item Personality Inventory (Gosling et al., 2003) consisting of two items (e.g., ‘Anxious, easily upset’). Participants rated the extent to which each item applied to them from 1 (disagree strongly) to 7 (agree strongly). After reverse-scoring one item, the two items were averaged to form a composite of emotional stability with higher numbers indicating more emotional stability (α = 0.63). A Spearman-Brown coefficient used to assess reliability suggested moderate to high internal consistency between the items, r=.64, N = 241.

Procedure

Following consent, participants completed measures of chronic ostracism, death-thought accessibility, general need satisfaction and mood, depression, personality, and demographics. They were debriefed online. Measures and items within each measure were randomized apart from demographic items, which participants completed last.

Results

Those scoring higher in chronic ostracism showed increased death-thought accessibility, r(239) = 0.14, p=.025, 95%CI[0.02,0.27]. Moreover, those higher in chronic ostracism were less likely to generally feel that their needs are satisfied, r(239)=-0.51, p < .001, 95%CI[−0.60,-0.41]. Finally, those higher in chronic ostracism were more likely to feel negative mood, r(239) = 0.45, p < .001, 95%CI[0.34,0.55]. See Table 4 for the correlation matrix.

Table 4. Correlation matrix for study 4.

Variable	M	SD	1	2	3	4	5
1. Chronic Ostracism	2.10	1.15
2. General Need Satisfaction	3.32	0.71	−.51**
			[−.60, −.41]
3. General Negative Mood	2.29	0.68	.45**	−.78**
			[.34, .55]	[−.82, −.72]
4. Death-Thought Accessibility	2.90	1.33	.14*	−.23**	.15*
			[.02, .27]	[−.35, −.11]	[.02, .27]
5. Depression	41.30	15.44	.50**	−.69**	.72**	.20**
			[.40, .59]	[−.75, −.62]	[.65, .78]	[.07, .32]
6. Emotional Stability	4.26	1.38	−.19**	.45**	−.39**	−.47**
			[−.31, −.07]	[.34, .54]	[−.50, −.28]	[−.56, −.36]

Note. M and SD are used to represent means and standard deviations. Values in square brackets indicate the 95% confidence interval for each correlation. * indicates p < .05. **indicates p < .01.

Though the correlation between chronic ostracism and death-thought accessibility was positive, it was small. Considering that the Temporal Need Threat Model of Ostracism (Williams, 2009) proposes that depression is an outcome of chronically depleted psychological needs and considering previous research supporting chronic ostracism’s positive association with depression (Riva et al., 2017), we conducted an exploratory analysis controlling for the effect of depression. We wanted to determine whether chronic ostracism would still be associated with higher death-thought accessibility above and beyond depression. We also added emotional stability to this exploratory model because lower emotional stability is characterized by more negative emotionality (Gosling et al., 2003) and may, thus, be correlated with threats like chronic ostracism and mortality salience. Emotional stability was also negatively correlated with chronic ostracism (r=−0.19) and depression (r = -0.47) in the data. Similarly, we wanted to determine whether chronic ostracism would still be associated with death-thought accessibility above and beyond the effect of emotional stability. When controlling for the effect of depression, the association between chronic ostracism was no longer significant and shrunk in size from a standardized beta of ß = 0.14, t = 2.26, p = 0.025, to a standardized beta of ß = 0.06, t = 0.814, p = 0.416. Moreover, depression was positively associated with death-thought accessibility, but the effect was small, ß = 0.17, t = 2.29, p = 0.023. When the effect of emotional stability was included in the model, none of the predictors were significantly associated with death-thought accessibility.

Discussion

Although we initially found a small positive association between chronic ostracism and death-thought accessibility, this effect disappeared when controlling for depression and emotional stability. Although this result aligns with our previous findings and suggests ostracism and mortality salience produce unique effects, correlational research is limited, and other methods should be used to examine this relationship in future research. Nevertheless, the moderate to large associations we found between chronic ostracism and general need satisfaction and mood support the resignation stage of the Temporal Need Threat Model of Ostracism, by demonstrating that those who are chronically ostracized may be depleted of the resources necessary to restore their threatened needs and improve their mood (Williams, 2009).

General discussion

Inspired by death-related themes within the ostracism literature, we aimed to determine the accuracy of using death-themed language to describe ostracism by examining whether ostracism and mortality salience produce similar effects. Comparing ostracism to mortality salience is important because the language that is used can influence how ostracism is conceptualized and can have downstream consequences for the questions that are addressed within the literature. Specifically, death-related themes help to illustrate ostracism’s severity by portraying the isolation and the sometimes-perceived permanence of the situation. Still, negative consequences may be associated with using death-themed language, such as focusing more on ostracism’s harmful effects rather than on recovery. Though our research only reliably addresses short-term ostracism, we argue that it would be generally beneficial for ostracism research to emphasize recovery by focusing on the differences between these constructs. One very tangible difference between ostracism and mortality salience is that ostracism is reversible, whereas mortality salience often involves an awareness of the permanence of death.

In Studies 1, 2, 3a, and 3b, we focused on whether ostracism and mortality salience would increase death-thought accessibility, worldview defense, negative mood, and decrease need satisfaction. We consistently found that mortality salience increased death-thought accessibility compared to ostracism and a control condition, and that ostracism decreased need satisfaction and increased negative mood compared to mortality salience and a control condition. Our results may be due to differences inherent to both ostracism and mortality salience. Although mortality salience represents a source of existential threat (Solomon et al., 1991) and individuals may ponder losing their loved ones, an important theoretical and methodological distinction between ostracism and mortality salience is that when someone is ostracized, they are directly experiencing the event. Mortality salience, instead, involves the conscious and unconscious awareness of death, with individuals sometimes imagining what the experience would be like. Directly experiencing ostracism, even when it involves reflecting on a past experience, may, thus, worsen mood and deplete need satisfaction to a greater degree than just contemplating what it would be like to have death separate us from those we love.

We conducted Bayesian analyzes (Dienes, 2014) on the death-thought accessibility and need satisfaction dependent variables across all four experimental studies to assess support for the null and alternative hypotheses. For death-thought accessibility, we found consistent support for the finding that mortality salience increased death-thought accessibility relative to ostracism and the control condition. We also consistently found support for the null difference between ostracism and the control condition. For need satisfaction, we found consistent support for the finding that ostracism decreases need satisfaction compared to inclusion and mortality salience. For the difference between inclusion and mortality salience, we found support across studies 1, 3a, and 3b for the null hypothesis, but found that the data were insensitive to detect this difference for Study 2. These analyzes provide further support that ostracism and mortality salience produce distinct effects.¹²

In Study 4, the positive relationship we found between chronic ostracism and death-thought accessibility was small and was no longer significant when we controlled for emotional stability and depression in exploratory analyses. Correlational designs, however, are limited in many ways including their inability to determine causality. Chronic experiences of ostracism were also less frequent in our sample of undergraduate students, with the data being right-skewed and the mean falling toward the lower end of the scale (M = 2.10, SD = 1.15). Thus, our research does not address whether the ‘social death’ metaphor accurately captures experiences of chronic ostracism. Future research could sample from populations struggling with chronic ostracism (e.g., online support groups), or may employ the use of an experimental method like the life-alone paradigm (Bernstein & Claypool, 2012; Twenge et al., 2001) to better simulate experiences of chronic ostracism in a controlled setting. The life-alone paradigm, which involves participants being given bogus feedback that their personality will likely lead them to experience a lonely life characterized by a string of unsuccessful relationships, is to our knowledge the only experimental manipulation of ostracism that most closely resembles chronic ostracism.

Nevertheless, we propose that it is still important to highlight differences between the two constructs, given the role that language may play in shaping how we think of phenomena and what we focus on (e.g., Gibbs, 2011). As mentioned previously, one important difference is that ostracism is reversible and not subject to the permanence of death. A quote from author Philip K. Dick’s novel Do Androids Dream of Electric Sheep? illustrates this point: ‘Mors certa, vita incerta’ (1968, p. 19) means ‘death is certain, life is not certain.’ Although interpreted out of context here, the quote carries with it an important point of emphasis for those who are ostracized: all aspects of life including one’s social or inclusionary status are not set in stone and there is opportunity for change within the uncertainty. Emphasizing this point may engender hope among those who are chronically ostracized and feel they are stuck in a hopeless, permanent state. Additionally, research should focus on examining interventions that will help the chronically ostracized restore their threatened needs and mood.

Limitations and future directions

Although previous research and theory support the notion that mortality salience increases worldview defense (e.g., Florian & Mikulincer, 1997; Solomon et al., 1991), we did not find consistent evidence for this. The effect of mortality salience on worldview defense was null in all studies except Study 2. In Study 1, we considered the possibility that our measure of worldview defense was not believable. In Study 2, although mortality salience increased worldview defense compared to ostracism, we did not replicate this effect in studies 3a and 3b using the same measure. Moreover, in Study 2, mortality salience participants did not differ significantly from included participants in the support for killing animals, suggesting that the ostracism condition was driving our pattern of findings. While this finding is inconsistent with previous findings and theory, it mirrors other studies that have failed to replicate mortality salience effects. For example, Schindler et al. (2021) conducted three pre-registered and carefully crafted studies (e.g., knowing the worldviews of the sample, using delays) aimed at validating mortality salience effects on worldview defense and found non-significant effects of mortality salience across all studies.

Relatedly, using the same measure of chronic ostracism we employed in Study 4, Chen et al. (2020) found that both temporary and chronic ostracism were positively associated with suicidal ideation, and that this effect was mediated by perceptions of less meaning in life. Our results may have differed because of the outcome variables employed. Across most of their studies Chen et al. (2020) measured thoughts of suicide using a self-report questionnaire. One distinction between mortality salience and suicidal ideation is that with suicide, death is a method chosen to escape reality. Participants in Chen et al. (2020) were not offered other methods of escape, and it is not clear whether other methods would have been preferred.

It is worth noting a few limitations associated with our experimental manipulations. First, we experienced participant attrition in the online studies involving our mortality salience manipulation, Cybergrave. Some participants randomly assigned to the Cybergrave condition clicked out of the study after seeing the mental visualization task. Given that these participants did not complete any measures, they were removed entirely from the final sample. Participant attrition may indicate an issue related to the manipulation and may have introduced bias into our final samples for the online Cybergrave studies. For example, it is possible that participants assigned to the Cybergrave condition who completed the study may have not found the manipulation as threatening as those who clicked out of the study, which then may have contributed to the observed differences between mortality salience and ostracism (e.g., mortality salience only affecting death-thought accessibility). We argue, however, that this is not likely, because we found the same pattern of results in Study 3a, which was conducted in person, and in Study 1, which employed different manipulations of ostracism and mortality salience. We believe this speaks to the robustness of our findings.

Another limitation of our experimental manipulations involves possible confounds. In Study 1, the mortality salience manipulation focused on anticipating death and was forward-thinking in nature, whereas the ostracism manipulation involved recalling being ostracized. In studies 2, 3a, and 3b, the mortality salience manipulation involved mentally visualizing tombstones, whereas those who were ostracized were actively excluded by the ‘other players.’ These confounds may have contributed to the differences we found between mortality salience and ostracism, given that something recalled or actively experienced may be especially detrimental to psychological need satisfaction. Future research should work to enhance experimental control by removing these confounds. For example, participants can be randomly assigned to imagine they will be ostracized or to imagine what death might be like.

Another limitation of the current research is that it was conducted with college students who are predominantly White, and our effects may not generally apply to all individuals. Future research should employ more diverse samples to examine whether ostracism and mortality salience behave similarly among people with different social identities. It is possible that their effects may be more similar among members of stigmatized groups who generally experience invisibility because of the marginalization of their group.

Conclusion

Is ostracism a metaphor for death? The current set of studies suggests that for short-term ostracism it may not be an accurate one. Rather, short-term ostracism produces distinct effects to those produced by priming mortality. Moreover, while use of death-related language within ostracism research may help to depict the severity of ostracism’s effects, we suggest that ostracism researchers remain mindful of the ways language can influence how ostracism is conceptualized and how it is studied. This suggests taking a closer look at areas of focus that require more research, such as ways to address ostracism’s harmful effects, especially for cases of chronic ostracism. Our results suggest that consequences of both ostracism and mortality salience should be addressed distinctly, with the former focused on recuperating threatened needs and emphasizing the impermanence of ostracism.

Data not available- participant consent

The institutional review board at the university where the studies were conducted did not allow us to share the data with reviewers, because they are now requiring that we include in our post-session consent form exactly how and where data will be shared. Due to the sensitive nature of the research, supporting data is not available. We have shared, instead, the study materials, analysis code, and output files on the Open Science Framework.

Disclosure statement

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

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/15534510.2024.2342308

Notes

1. We incorrectly powered for a large effect for Studies 1 and 2 as opposed to a medium effect (η² = 0.06). An a priori power analysis with a medium effect yields a suggested sample size of 156 participants. In both Studies 1 and 2 we exceeded that number.

2. All study materials are uploaded to a project OSF page, https://osf.io/r9ntb

3. All participants completed this measure after the prostitution bail measure. It was included for exploratory purposes. Similarly, however, no analyses with shoplifting bail were significant. Details concerning this measure are included in the supplemental materials.

4. These analyses are in the supplemental materials.

5. For all studies, results were analyzed with those who failed the manipulation check included and excluded. If the pattern and significance of results did not change, we reported the analyses with the full sample. Degrees of freedom vary slightly across studies due to missing data. All 95% confidence intervals are of the mean difference.

6. Only measures that did not appear in the previous study will appear in this section. All other measures and their reliabilities can be found in the supplemental materials.

7. We directly replicated study 2 in a follow-up study, but still experienced participant attrition in the mortality salience condition. We report study 2a in the supplemental materials.

8. Due to COVID-19 quarantine we had to stop in-person recruitment of participants.

9. Reflective need satisfaction (measured several minutes after the manipulation) analyses are listed as primary analyses in the pre-registration but were not included in the manuscript because we did not have relevant hypotheses about the effects of the manipulations after time had passed. The analyses involving self-esteem as a moderator of the independent variable are listed as primary analyses in the pre-registration but are included in the supplemental materials as exploratory. These analyses are reported as exploratory because self-esteem was considered as a possible moderator only because Terror Management Theory suggests that it can moderate the effects of mortality salience. However, its interaction with ostracism on relevant dependent variables is not theoretically based.

10. We chose this effect size from Study 2a, our replication of Study 2, included in the supplemental materials.

11. Given that each study had six manipulation checks, the complete set of analyses can be found in the supplemental materials. Only participants who answered all three categorical questions correctly passed the manipulation check. Primary analyses were conducted with these participants included and excluded. Any deviations in pattern and significance are reported.

12. Please see the supplemental materials for the analyses.