Love to Help: The Roles of Compassion and Empathy in Regards to Altruism
David Lindsten Minelius and Felix Nilsson Örebro University
VT 2020
Abstract
Unresolved global problems, such as extreme poverty, ask for a better understanding of what predicts altruism and what does not. The aim of this thesis project was to address this topical and timely research question by studying the predictive role of compassion and empathy in
understanding altruism. In past research on the relationship between altruism and empathy, distinct empathic processes (Perspective taking, Empathic concern, Personal distress, Emotional contagion, and Behavioral contagion) have been often lumped together and the context dependency of the relationship has been insufficiently taken into account, resulting in confusion and contradictory findings. Compassion overcomes these issues. The present web-based survey with previous or current university students (age 18-45; N=240) aimed to clarify relationships between components of empathy, compassion, and altruism. It was hypothesized that (1) compassion would predict altruism beyond all components of empathy; (2) Empathic concern would mediate the relationship between Perspective taking and altruism; (3) compassion would mediate the relationship between Empathic concern and altruism, and (4) higher levels of compassion would result in a reduced negative
relationship between Personal distress and altruism. The results supported all hypotheses except for the final one. These findings are discussed in context of previous research and theory, considering the current study limitations and with focus on theoretical and practical implications. In sum, the findings suggest that efforts to motivate altruism should focus on invoking positive emotions of warmth, concern, and relatability. Care should be taken to avoid unnecessary Personal distress when invoking altruism, as this reduces its likelihood.
Key words: Altruism, Directed Altruism, Compassion, Empathy, Perspective taking, Empathic concern, Personal distress, Emotional contagion, Behavioral contagion, Self-report Survey, Cross-Sectional design.
Human beings are social in nature, relying on each other for help, support, and cooperation (Tomasello, 2014). Certain circumstances ask for prioritizing beyond the immediate needs of ourselves, families, and friends to benefit others. Take, for instance, the problem of global poverty. According to the most recent estimates, 736 million people live in extreme poverty around the globe, surviving on less than 1.90 dollar per person, a day (World Bank, 2018). This is in sharp contrast with the richest billion people who dispose of 32 dollars or more per day (Rosling et al., 2018). The United Nations (2019) estimate that the most opportune time to end the predicament of those in extreme poverty is during the coming decades, as population increase falls alongside decreases in child mortality. To achieve this, we need to prioritize beyond the immediate needs of ourselves, our families, and friends by acting. An interesting and relevant question that remains largely unanswered is whether we can predict people’s tendency to help, even when this comes with costs. The focus of this thesis is on comparing psychological predictors of such helping behavior, as well as investigating their relationships to one another.
Altruism
Altruism is defined as behavior intended to benefit another, even when this action risks possible sacrifice to the welfare of the actor (Monroe, 1996; Post, 2005). The term altruism can refer anything from watering a neighbor’s plants, to giving a fellow concentration camp prisoner your last piece of bread. When defining altruism, it is not the magnitude of the gesture, but rather the contextual setting and/or underlying motivation which matters. Indeed, if context and motivation is left unspecified, altruism could just as well refer to giving Elon Musk another expensive car, as saving a child from drowning. While donating a car to a wealthy business owner may constitute prosocial behavior (a general term used to describe any behavior assessed to benefit society), whether or not it constitutes altruism is
de Waal (2008) distinguishes between several distinct forms of altruism based on the merits of their context and/or underlying motivation. In learned altruism, an individual acts altruistically because this behavior has produced benefits in the past. In intentionally selfish altruism, on the other hand, helping is motivated by a belief that it will lead to personal gains. Finally, in directed altruism, an individual is motivated to act altruistically by making efforts to aid another by attempting to alleviate the other’s pain or distress, such as helping people living in poverty. The type of altruism investigated in this thesis is the directed form, and any references henceforth will be referred to simply as “altruism”.
In previous research, altruism has usually been operationalized as giving something up for another in need or offering assistance. In a laboratory setting, the stakes tend to be smaller than in real life, exemplified by having participants making decisions with very low sums of money. A caveat in previous research on what motivates altruism is the limited number of studies measuring helping behavior outside of the lab, which limits generalizability to real-world scenarios.
Additionally, it is noteworthy how few studies rely on self-report compared with experimental designs. While surveys carry the problem of social desirability and non-credible responding, they allow us to investigate altruism outside the controlled environment of a laboratory. Although a carefully controlled set-up allows for testing the relationship of this behavior to an independent variable, the survey format opens up the possibility to investigate trends of behavior across any variety of contexts. Furthermore, scenarios such as those including danger or high levels of distress cannot be recreated in a laboratory for ethical or practical reasons. Using self-report allows investigation into these areas as well.
Previous research has been devoted to understanding psychological predictors of altruism (Batson et al., 1981; Cialdini et al., 1987; Eisenberg & Miller, 1987). For instance, can we predict which people have a stronger tendency to act altruistically as compared to
others based on their traits? Traits refer to stable inner qualities, such as personality, that are assumed to be internalized and not dependent of the situation. Previous research has focused on trait predictors for altruism. Focus has primarily been in the context of empathy, where findings have shown contradictory results (Eisenberg & Miller, 1987). More recently, research into compassion as a predictor of altruism has peaked the interest of many
researchers (Böckler et al., 2018; McCall et al., 2014; Weng et al., 2013; Weng et al., 2015). The aim of this thesis was to contribute to the understanding of how dispositional empathy and compassion relate to altruism.
Empathy
Research on what predicts altruism has so far been almost exclusively focused on empathy. Although a widespread and prominent psychological construct as well as a term commonly used by the general population, definitions of empathy vary significantly. According to a recent review by Cuff, Brown, Taylor, and Howat (2016) an astounding 43 discrete definitions of empathy were identified. While each definition emphasizes different aspects of the phenomenon, a widely accepted view is that in order for a response to be considered empathic, there needs to be an emotional response in relation to the state of another.
Descriptions of empathy range between viewing empathy as a bottom-up and a top-down phenomenon. The bottom-up aspects of empathy that have been commonly described as contagion and Personal distress. While contagion is defined as the contagiousness of
affective states and behaviors between individuals, Personal distress represents an unpleasant reaction in the context of an interpersonal situation. In contrast, top-down descriptions of empathy emphasize intellectually grasping the situation of another, usually referred to as Perspective taking. A final component of empathy, which is not characterized by bottom-up nor top-down processes, is Empathic concern (or sympathy). Empathic concern is defined as
having warm, concerned feelings for others. Note that Empathic concern is distinct from the sharing of emotions occurring in contagion.
Clearly, empathy is a very broad term, encompassing several distinct phenomena. In fact, although frequently co-occurring, it is questionable whether these distinct empathic processes are part of the same psychological phenomenon or not (Cuff et al., 2014; Decety & Cowell, 2014). By examining the different components separately, the linguistic clarity and precision needed for a more nuanced and complex investigation and understanding of empathy becomes possible. This is of crucial importance for the present thesis work. Let’s therefore have a closer look at the different components of empathy mentioned before: contagion, Personal distress, Perspective taking, and Empathic concern.
Contagion. Contagion, commonly referred to as Emotional contagion in the literature, is the process through which emotional states and behaviors resonate between individuals. As a bottom-up process, contagion represents the more primitive, non-cognitive, and often unconscious process of empathy which can also be found in non-human animals (Panksepp & Panksepp, 2013). When another individual exhibits a behavior or experiences affect, be it happiness, sadness, joy, or distress, we are influenced by our perception of their experience and tend to mirror it in some way ourselves. This can manifest as the irresistible impulse to return the smile of a baby. A less pleasant example would be infants who tend to cry together
in maternity wards, a neutral one being the impulse to yawn when observing another person yawning (Franzen et al., 2018; Panksepp & Panksepp, 2013).
Personal distress. Although being affected by the state of others may be beneficial, it can also lead to unpleasant experiences and trigger avoidance. This unpleasant arousal in response to the experience of another person defines Personal distress (Davis, 1983). Shared fear, for example, may well incentivize us to leave the other behind, rather than giving the other aid, especially when severely aversive. Take, for instance, a parent who feels extremely unsettled and distressed when unable to soothe their baby. Leaving the crying child is often not an option due to several reasons. The powerful instinct to protect and respond directly to the needs of your child, shame at your own distressing impulse to leave, along with the awareness that the situation will not be resolved by leaving, can present a parent with an ostensibly unsolvable problem. In utter desperation and severe distress, a parent may resort to shaking the baby as a last resort, seeing no other recourse to regulate their own Personal distress. Shockingly, this is a common occurrence and may be a leading cause of infant mortality and morbidity (Runyan, 2008).
While animals and humans alike benefit from an impulse to get out of harm way, it can become a hindrance to adaptive behavior. Most animals are at the mercy of these impulses. However, more advanced, social species such as elephants, cetaceans, great apes, and humans are able to think about the situation and make decisions that conflict with the experienced need to escape (de Waal, 2008).
Perspective taking. Perspective taking - or in other words being able to intellectually place oneself in another's position - allows for a more nuanced understanding of the other’s experience. Perspective taking opens one up to the differing experiences and views of others, and allows for an understanding beyond the immediate self. This, in turn, allows for the possibility of complex behaviors such as anticipating the actions of others, teaching, or
requesting help with a specific problem only the other can solve (Panksepp & Panksepp, 2013; de Waal, 2008). So, Perspective taking allows one to go beyond merely reacting to contagion of emotions, or avoid the situation due to the Personal distress associated with it. Rather, one is able to adapt one’s response based on the actual needs of the other, which may differ significantly from one’s own. Take for example the crying baby. By having an
understanding of the child's wants and needs intellectually, you may be able to figure out that the crying is due to the missing teddy-bear, rather than an immediate physical need like hunger.
Empathic concern (a.k.a. sympathy). While even primitive animals can generate feelings of warmth and concern for their offspring, it has been suggested that it was through Perspective taking that these feelings could be extended to others than own children.
Processes originally evolved in the context of child-rearing thus expanded to occur in
potentially any relationship, giving rise to Empathic concern (Panksepp & Panksepp, 2013; de Waal, 2008).
Sometimes referred to as sympathy, Empathic concern is not characterized by bottom-up nor top-down processes, but has its own affective profile (Cuff et al., 2014; Decety & Cowell, 2014) characterized by concern, and other affective states such as sorrow and wishing the other well (Eisenberg, 2000). This is in contrast to contagion, where one “catches”
emotions of the other person. Rather, Empathic concern is defined on the merit of feeling for, rather than with, the other person (Cuff et al., 2014; Decety & Cowell, 2014). Returning to the example with the crying baby, you may “catch” some distress by way of contagion, however, your love and parental care triumphs over these emotional obstacles, allowing you to engage in caring behaviors that hopefully soothe the child.
Taken together. Different empathic processes elicit different responses and therefore
it has been recommended to break apart the construct of empathy when studying its behavioral consequences (Decety & Cowell, 2014). However, it is important to note that although components of empathy are in fact discrete, they can interact and influence each other, altering their impact on altruism. For instance, when Perspective taking is added to contagion, genuine concern for the other’s well-being becomes possible. This is argued to be the main motivation behind altruism in advanced species (Panksepp & Panksepp, 2013; de Waal, 2008).
Empathy & altruism
There is a large body of empirical evidence on the relationship between empathy and altruism, providing mixed results. In the eighties and nineties, empathy in relation to altruism has been tested extensively experimentally (Batson et al., 1983; Eckel & Grossman, 1996; Cialdini et al., 1987; Smith, et al., 1989). Meta-analyses have found either largely non-significant relationships between the two (Underwood & Moore, 1982), or small to moderate effects with large variability in effect size (Eisenberg & Miller, 1987). More recent studies have found that the effect of empathy on altruism depends on contextual factors. For instance, if there is distance in regards to social relationships or out-group membership relative to us, we are far less likely to provide assistance. If the other person is perceived as a competitor, we may even feel pleasure at their misfortune (Cikara et al., 2011). In situations where one holds negative stereotypes about the out-group member, induced empathy may in fact result in less positive attitudes and reduced inclinations to help (Vorauer & Sasaki, 2009). On the more extreme end, the disidentification process reflected by the us/them categorization seen in the examples above has historically turned to dehumanization. This is reflected by completely dichotomous thinking about in-group/out-group, where individuals belonging to a certain
outgroup are not even regarded as human. This has tragically resulted in persecution and genocide (Moshman, 2007).
A possible explanation for the mixed results regarding the effects of empathy on altruism is, as previously mentioned, that research on the relationship between empathy and altruism has generally been performed on the level of the umbrella-term empathy, rather than on a component level (Decety & Cowell, 2014). As noted, the concept of empathy
encompasses several distinct processes (Cuff et al., 2014), which prompts the critical question as to whether all empathic components encourage altruism, and whether they do so across different conditions and situations.
What does this mean for altruism? It is indeed a reasonable assumption that each empathic component influences likelihood of altruism. There are two predominant views in the literature as to how this occurs. In one view, referred to as the Aversive-Arousal
Reduction Hypothesis (AARH), contagion of emotions and Personal distress is thought to be the key components which compels one to act altruistically (Cialdini et al., 1987). The opposing view, the Empathy-Altruism Hypothesis (EAH), holds that Personal distress promotes egotistical behavior rather than altruism, whereas it is in fact Empathic concern which promotes altruism (Batson et al., 1983).
Going back to the example with the crying infant, the AARH holds that the parent is influenced by the state of the baby. As the baby cries inconsolably, the parent’s distress increases. The need to alleviate the Personal distress evoked by the crying child triggers the impulse is to escape the unpleasant state. However, as previously mentioned, the impulse to escape can be in conflict with other imperative needs, situational demands, and instincts. Soothing the crying baby is thus the only recourse. The AARH holds that it is this
predicament which compels altruism, as relieving the distress of another, relieves the distress of the actor as well (Cialdini et al., 1987). An attentive reader would of course feel concerned
that, should the parent be unable to soothe the child, the distress could lead to desperation, and there would be a risk that attempts at soothing could turn to shaking.
The EAH on the other hand conceptualizes Empathic concern as the driving force behind altruism. In this paradigm, situations are compared on high contra low-cost of escape, when a person is faced with someone in need of help. Where escape comes with a high cost, everyone helps. If escape is easy, only those motivated by Empathic concern help, thereby illustrating Empathic concern as the driving force behind altruism (Batson et al, 1981).
While this offers a theoretical framework for testing empathy in regards to altruism, it does not address why the feelings associated with Empathic concern, developed in the context of child-rearing (Panksepp & Panksepp, 2013; de Waal, 2008), arise in relation to strangers or animals.
Imagine a scenario in which a single parent is doing the weekly grocery shopping. Finding no available babysitter, the parent brings the baby along. As the parent finally is just about to pay at check-out after waiting for a long time, with an increasing line build-up, the infant starts to cry inconsolably. The parent is distressed, overwhelmed, and panics.
Meanwhile a fellow customer in a different line, also a single parent, observes the situation. It would be easy to ignore what is going on (low-cost of escape), but the fellow customer
identifies with the parent through Perspective taking, and is moved by their plight through Empathic concern and decides to abandon their own place in queue and steps in to help.
With these opposing views of the AARH and the EAH in mind, it becomes important to investigate and compare outcomes associated with the distinct components of the umbrella-concept empathy in the context of predicting altruism. The following heading gives an
overview of what is known empirically so far about the different components of empathy mentioned before (contagion, Personal distress, Perspective taking and Empathic concern), and how they are related to altruism and related outcomes.
Predicting altruism on the component level of empathy
While people have been shown to act altruistically due to experienced Personal distress in scenarios where escape is difficult, the effect is smaller and less reliable than that of Empathic concern. (Stocks et al., 2009). This indicates that the views of the EAH and AARH may not be completely incompatible with each other, although there is substantially more empirical evidence for the EAH (Batson & et al, 1981; Batson et. al, 1991; Stocks et al., 2009).
Personal distress has been positively associated with negative outcomes such as secondary traumatic stress (Thomas, 2013), whereas the opposite relationship has been found for other components of empathy (Hayuni et al., 2019). Moreover, people are prone to avoiding situations where they will feel empathetic if they expect the cost of helping the person in need will be high (Shaw et al., 1994, Cameron et al., 2019). Furthermore, Personal distress is highly correlated with the Big-5 personality dimension of neuroticism, unlike the other components of empathy (Kim & Han, 2018), and activates separate neural networks compared to Empathic concern (Decety & Michalska, 2009).
Unsurprisingly, when people experience Personal distress, the manner in which they deal with it appears to be what dictates their resulting behavior (Lebowitz & Dovidio, 2015). Suppression-based emotion regulation, unlike reappraisal of Personal distress, is associated with stigmatizing and distancing behaviors. Additionally, it negatively correlates with Empathic concern as well as altruism (Lebowitz & Dovidio, 2015). Taken together, previous findings do not favor the AARH as Personal distress triggers a classical avoidance behavior, lessening the likelihood of altruism.
On the other hand, the effect of Empathic concern, as well as Perspective taking, does seem to motivate helping, which has been well established experimentally (Batson & et al, 1981; Batson et. al, 1991; Fultz et al., 1986; Stocks, Lishner & Decker, 2009). Moreover,
survey studies have found a positive relationship between self-reported empathetic concern and charitable giving (Bekkers, 2006; Kim & Kou, 2014). For these reasons, recent studies on what motivates altruism have increasingly focused largely on Perspective taking and
Empathic concern (Burkard & Know, 2004; Gini et al., 2007; Lim & DeSteno, 2016), which are predictive of prosocial behavior, as proposed the by the EAH. This picture is further nuanced by taking into account the question of whether the positive effects of Perspective taking on altruism depend on the emotional response associated with Empathic concern. Indeed, there is evidence that Perspective taking only works so far as it induces Empathic concern, demonstrating that the relationship between Perspective taking and altruism is mediated by Empathic concern (Lim & DeSteno, 2016; Jordan et al., 2016). Empathic concern, in turn, is a potent motivator of altruism, yet appears to contingent actually being confronted by a person in need (Einolf, 2008; Small et al., 2007).
In sum, the available empirical findings suggest that Perspective taking and Empathic concern motivate helping behavior, while the Personal distress invoked by seeing someone in dire need does the opposite by inducing the need to escape. Largely, the literature is in support of the EAH as compared with the AARH as an explanation for altruism. Empathic concern and Perspective taking are successful predictors of altruism. That said, they are insufficient in explaining altruism, as they do not account for the discomfort associated with Personal distress. Contagion is most commonly mentioned as the process through which states and behaviors are transmitted and does not in and of itself appear to have an inherent directional effect on altruism. Rather, the resulting response is dependent on the affective state transmitted and its effects of the receiver.
Conditions for caring & prediction in practice.
In a real-world scenario, there are many factors at play. All of which can influence our behavior, but some specifically do so through our empathic responses. An intriguing question
is: what is actually needed in order for empathy to invoke someone to act and help someone in need? Consider the following example:
You see a person begging for food as you exit your local supermarket. Perspective taking enables you to imagine being hungry and homeless yourself. You identify with the other and is more likely to offer that person help. Likewise, if you have a close family member in a similar situation, it hits closer to home and Empathic concern is more readily accessible. However, being exposed to a person in obvious misery, triggers Personal distress by way of contagion. As Personal distress is aversive by definition, you are naturally inclined to avoid what causes it, in this case the begging person. This needs not become an issue as the familiarity of her situation, as well as the Empathic concern for her well-being carries more weight than your discomfort, and compels you to help.
Now imagine the same scenario, but the begging individual is far more foreign to you. She clearly belongs to a different culture. You do not share a language and she is somewhat erratic in her behavior. It is more difficult to put yourself into her position through
Perspective taking, as you have far less in common. Accessing positive emotions involved in Empathic concern proves more difficult, as she is so unrelatable. Thus, motivation to help is far harder to access and Perspective taking and Empathic concern hold less sway over your actions.
Now add the fact that being exposed to this person is more aversive than before. She is moving in an unpredictable way. You are unsure of her body language and she is physically deformed in a way you find disturbing. Your level of Personal distress is far higher, and with less motivation to help due to a lack of Empathic concern and Perspective taking, you very are likely to leave as quickly as possible.
Most people in need of aid to meet their basic needs are not situated outside of our local supermarket, but live in distant countries with customs foreign to us. Because we often
have no obvious relationship to them, there are many obstacles in the way of relating and caring. By extension, we are often not incentivized to offer help. Adding to this, in less developed countries the degree of hardship is far higher, amounting to deadly disease and starvation. This evokes greater levels of Personal distress in us, making the positive impulses rooted in Empathic concern and Perspective taking less likely to actualize in altruistic
behavior.
The example presented above illustrates some of the more pressing issues with empathy as a predictor of altruism. While able to functionally predict altruism in some scenarios, empathy is unreliable and heavily dependent on external conditions. The Personal distress invoked through contagion triggers avoidance, which makes helping less likely. Empathic concern and Perspective taking encourages helping, but are unreliable, and require specific conditions to be met in order to emerge.
In sum, all components of empathy can play a role in predicting altruism. However, they do so in different ways. Their influence depends on context and which components are in play at a given time. Perspective taking is contingent on putting oneself in the position of another and subsequently identifying with them. Likely, this prompts the warm feelings and a wish for the other’s well-being are central to Empathic concern. Whether or not aversion becomes overpowering is central to Personal distress. Likely, noticing the distress of another is facilitated by some degree of contagion, but may also add to the experienced Personal distress. Ideally, these predictive components would be represented in a construct which takes into account the (contextual) issues of social distance, relatability and tolerance for distress. This would then be superior as a predictor of altruism. Fortunately, such a construct is available: Compassion.
Compassion as an alternative to empathy
Compassion is a several thousand-year old concept represented in all major world religions. Through the popularization of Buddhist ideas, it made its way into the realm of Western science in the early nineties (Strauss et al., 2016). Today, it is represented in several areas of the psychological field, including psychotherapy, psychophysiology, and
neuroscience (Kirby et al., 2017; Luberto et al., 2017; Stellar et al., 2015). Moreover, several compassion-based treatments have gathered initial empirical support as a treatment for depression, anxiety, and improving general well-being, although small sample-sizes underscore the need for further study (Kirby et al., 2017).
In spite of increasing interest, there has been little consensus in the field of psychology about what constitutes compassion. A recent review of existing definitions and measurements of compassion (Strauss et al., 2016) resulted in the following summarizing definition
highlighting five characteristics of compassion: Table 2.
1) Recognition: Recognizing suffering
2) Understanding: Understanding the universality of suffering in human experience 3) Feeling with: Feeling empathy for the person suffering and connecting with the
distress (emotional resonance)
4) Tolerance: Tolerating uncomfortable feelings aroused in response to the suffering
person (e.g. distress, anger, fear) so remaining open to and accepting of the person suffering.
5) Motivation: Motivation to act/acting to alleviate suffering.
When comparing these five key characteristics of compassion with the distinct components of empathy it becomes clear that there are significant overlaps between the two constructs (Strauss et al., 2016). While the components of compassion in one way or another
touch upon and rely on each component of the empathy construct, they do so in a very specific and deliberate fashion. Recognition and understanding share a reliance on
Perspective taking, but are worded so that their definition presupposes a lack of interference with identification with the person suffering, argued by de Waal (2008) to be crucial in
generating concern for another. Feeling with makes use of contagion and Personal distress but is formulated in a way that necessitates the absence of suppression and avoidance. Acceptance permeates the wording of the entire compassion construct which in and of itself should help mitigate the predictive issues that come with empathy. Compassion takes this one step further by including a discrete component solely dedicated to dealing with issues brought about by distress. The tolerance component seemingly represents emotion regulation based on acceptance, which again presupposes a lack of suppression or avoidance (Lebowitz & Dovidio, 2015). This concept lacks representation in the empathy construct. As suppression and avoidance (Cameron et al., 2019; Shaw et al., 1994) caused by Personal distress appears central to the issue of predicting altruism, the tolerating uncomfortable feelings dimension, along with the acceptance-based wording of compassion should mitigate this issue. A crucial difference though between compassion and empathy is that unlike empathy, which can arise in response to any state expressed by another person (e.g. anger, joy, suffering, curiosity and so on), compassion specifically occurs in response to suffering in another person. In the context of compassion, the term suffering refers to a wide range of unpleasant states and emotions, including but not limited to feeling upset, fearful, and frustrated, or to struggle.
Compassion training was originally developed in the context of Buddhist practice where it was referred to as “Metta”. Today, Metta is used interchangeably with the term Loving-Kindness Meditation (Salzberg, 1995). It involves thinking of yourself or someone you care deeply for and wishing them well-being, relief from suffering, as well as feeling the warm positive emotions associated with these wishes. These wishes is then expanded towards
increasingly challenging individuals in one’s life. This gradual expansion trains the
practitioner to see the universality of suffering and emotionally relate to everyone, and their wish to be happy (Klimecki et al., 2012).
Compassion training has been found to result in changed neural activity in response to suffering, as well as to make others’ suffering more salient (Desbordes et al, 2012; Weng et al., 2013). Adding to this, stimuli of others’ suffering usually result in experienced negative affect. After compassion training, the suffering instead elicits positive affect (Klimecki et al., 2012). Individuals instructed in compassion show activity in neural circuits associated with parental love, whereas those instructed to empathize with others show brain activation associated with their own suffering (Klimecki et al., 2013).
Moreover, and especially relevant for this thesis, compassion-induced changes have been found to predict altruism (Weng et al., 2013), and has been tied to altruistic outcomes such as redistributing money to help a perceived victim in a common goods game (Böckler et al., 2018; McCall et al., 2014; Weng et al., 2013; Weng et al., 2015;). Training in compassion also translates into being more helpful when playing a game designed to measure prosocial behavior (Böckler et al., 2018; Leiberg et al., 2011).
We propose that compassion as a construct is a better fit than empathy for explaining why people choose to help others in need even if there is a cost to themselves. Compassion involves tolerance for the unpleasantness that the suffering of others evokes in us, making us less likely to turn away (Strauss et al., 2016). Furthermore, the warm concerned feelings of Empathic concern can be found represented in compassion, and experimental research has found that compassion actually mediates the relationship between Empathic concern and altruism. However, the strength of predictors has not been compared (Lim & DeSteno, 2016), thus it is not possible to determine whether compassion, Empathic concern, or Perspective taking explains more of the variance in altruism. While several components of empathy can
be valuable when predicting altruism, it appears that compassion encapsulates the predictive aspects of these while adding other benefits, such as the tolerance of distress and universality of suffering components. The goal of the present study is to investigate if level of
self-reported trait compassion indeed predicts altruism. The present study Summary
Empathy influences altruism, but the different components of empathy have different effects. Whereas Empathic concern and Perspective taking both predict altruism, Personal distress evoked by contagion predicts avoiding those in need if possible. Compassion on the other hand includes a component of tolerating distress specifically in response to suffering in another person. Compassion encapsulates both Empathic concern and Perspective taking which motivate helping, while addressing their conditional reliance. Furthermore, it addresses the avoidance issues associated with Personal distress through contagion, which demotivates altruism.
Rationale and justification
This study builds upon and extends previous research on empathy and altruism. It offers a novel addition to the understanding of trait predictors of altruism in several ways. Firstly, empathy is a concept plagued by problems related to confusing terminology and varying definitions. By testing how Personal distress on one hand, and Empathic concern and Perspective taking on the other, relate to altruism, we aim to clarify what actually motivates altruism.
Secondly, the nature of the relationship between empathy and compassion is largely unexplored beyond concluding that there is significant overlap between the two (Strauss et al., 2016). As previously mentioned, past research has found that Empathic concern and
the predictive qualities of Empathic concern and compassion have not been compared, and such a comparison is needed in order to establish which construct is the better fit for predicting altruism. Establishing the superior construct is beneficial for directing future research as well as for practical applications, such as facilitating solicitation of charitable donations.
We compared compassion and empathy on a component level in regards to altruism prediction, as well as replicated and expanded on the previous findings by Lim & DeSteno (2016) in two ways: (1) by examining in a self-report survey outside of the controlled lab environment, the positive relationship between Perspective taking and altruism, the mediating influence of Empathic concern on this relationship, as well as the mediating influence of compassion on the positive relationship between Empathic concern and altruism. (2) by investigating, to our knowledge for the first time, whether compassion also moderates the relationship between Personal distress and altruism. If this relationship were to be confirmed, one could conclude that there would be little benefit in measuring empathy rather than
compassion when predicting altruism. Compassion encapsulates the predictive value of Perspective taking and Empathic concern. Additionally, the negative association of Personal distress with altruism would then be mitigated by the tolerance aspect of compassion.
Thirdly, a self-report approach to predicting altruism comes with several advantages. In the laboratory, altruism is most commonly operationalized by having participants make decisions with extremely small sums of money (pennies and cents), which hardly translates to real-world scenarios. In fact, this questionable operationalization sheds doubt the construct validity of these designs, as giving up something substantial for the benefit of someone else is central to altruism as a concept.
The advantages of laboratory research, such as operationalization of altruism through narrow, predefined behaviors, as well as the ability to design a highly controlled situation,
come with a natural trade-off regarding external validity. There are also practical and ethical constraints inherent to such a design, as it is difficult and necessitates deception on part of the researches. As a person can vary in altruism both in terms situational tendencies and
behavioral frequencies, this prompts the need for an approach with emphasis on width in order to establish whether the findings from experimental research hold outside of the laboratory.
Measuring altruism by way of self-report enables inquiry into the contexts and situations which does not fit the experimental approach. The survey format addresses the issue of behavioral frequency by measuring trends in behavior, as compared to single
behavioral outcomes. Additionally, self-report measures enquire about behavior using a wide range of concrete real-world examples, thereby addressing the lack of external validity and construct validity that has characterized experimental designs to date. Only a few survey studies have been published on the relationships between altruism and empathy (Bekkers, 2006; Einolf, 2008; Kim & Kou, 2014), and none with regards to altruism and compassion. This could be due to legitimate skepticism regarding the credibility of responses in such a design. In the current study, this was mitigated by a completely anonymous design and a control for social desirability. These benefits, together with the control for social desirability, as well as complete anonymity, make for a much-needed addition to what is clearly lacking in previous research.
Fourth, we use valid and reliable questionnaires for all constructs of interest. Investigating how a psychometrically sound scale of compassion relates to altruism adds another dimension of validity to the scale. It does so by testing whether this measure of compassion, characterized by care and concern for others, actually translate into behavior.
Finally, the current research question is highly relevant and timely. Pending real-world issues, such as the dire need of hundreds of millions of people living under extreme poverty,
stresses the importance of establishing a clear understanding of what motivates altruism and what does not. If it works as suggested, a measure of compassion would be highly useful in order to identify individuals predisposed to helping. This knowledge could be used to nudge them towards important work or donating behavior. In addition, since it is a trainable skill, it could be applied to make people more likely to engage in altruism. Hopefully, this would contribute to the end of extreme poverty and preventable suffering, making for a kinder more compassionate world.
Purpose and research questions
The aim of this survey study was to examine and compare the predictive value of compassion and empathy on self-reported altruism.
Research questions:
1. Does compassion predict altruism above and beyond empathy?
2. Does Empathic concern mediate the relationship between the Perspective taking and altruism?
3. Does compassion mediate the relationship between Empathic concern and altruism? 4. Does compassion dimensionally moderate the relationship between Personal distress
and altruism where higher levels of compassion result in lower predictive value of Personal distress?
For these reasons, we believe compassion to have higher predictive value of altruism than empathy in the classical sense, including contagion, Personal distress, Empathic concern, and Perspective taking.
Our hypotheses were as follows:
1. Compassion will account for the largest variance in altruism.
2. Empathic concern fully mediates the relationship between Perspective taking and altruism.
3. Compassion fully mediates the relationship between Empathic concern and altruism.
4. A negative relationship between Personal distress and altruism is
dimensionally moderated by compassion, as compassion increases, the effects of Personal distress on altruism decreases.
Method Design, procedure and participants
Participants were recruited via social media platforms. A short recruitment text was posted in two Facebook-groups, where one group was dedicated to all students of Örebro University, and the other to psychology students from all over Sweden. Furthermore, it was published on the official Swedish subreddit website. The recruitment posts all used similar wording while avoiding terminology related to “helping”. Finally, instructions to encourage sharing of posts were included. The recruitment text included a link to the online survey.
Exclusion criteria were good comprehension of English and Swedish in written form as well as being at least 18 years of age. The requirement of good comprehension in Swedish served the purpose of maximizing homogeneity of the sample. This, in order to better
represent a Swedish population. The requirement of good comprehension of English allowed for use of the scales of interest in their original, validated form.
556 persons followed the link to the online survey. Out of those, 235 did not complete the questionnaires. This left 321 participants who completed all study measures. For reasons of comparison with previous research (cf. Gu et al., 2019), inclusion criteria were set to those who reported to either currently attending university or had attended university at some point in the past but were no longer students and 18 - 45 years of age. University attendance was included in order to allow for comparisons with the sample in which the measure of
compassion was developed. In order to satisfy inclusion and exclusion criteria, demographic data consisting of age as well as educational status was collected.
The resulting final sample consisted of 241 participants. Out of the included 241 participants, 161 (66.8%) identified as female, 75 (31.1%) as male, 3 (1.2%) as non-binary gender identity and 2 (0.8%) as uncertain about gender identity. The mean age of participants was 27.22 (SD = 4.86) years, with a mean of 26.96 (SD = 4.80) for females and 27.92 (SD = 4.91) for males.
Ethics
During recruitment participants were informed that participation was completely voluntary, that it was conducted as part of a master thesis in psychology at Örebro University, that participation was anonymous and that no sensitive identifying data was to be collected and that they could cancel their participation in the study at any time by exiting the browser. Noteworthy is also the inclusion of a final question at the end of the survey, asking
participants to consent to have their data analyzed as part of the study.
The data collection did not include a physical intervention or methods with the purpose of affecting a research person, nor did it include any apparent risks of physiological or psychological injury. We followed the GDPR guidelines as set by Örebro University throughout the whole research process. The study was approved by the ethics advisory board, psychology program, Örebro University, but as no sensitive or identifying personal data was collected, participants were not asked for GDPR consent. Due to the fact that the current study was carried out in the context of a university master’s thesis, and will not be used in a
research project or used for publication, no external ethical review was requested, in accordance with Swedish law ("Lag (2003:460) om etikprövning av forskning som avser människor Svensk författningssamling 2003:2003:460 t.o.m. SFS 2019:1144 - Riksdagen", 2020).
During recruitment, participants were informed that the aim of the study was to investigate individual differences in interpersonal relationships. This rationale was worded in an unspecified manner in order to avoid the word “helping” and related constructs in
recruitment, as to not skew the sample. The rationale was deliberately formulated to do so while avoiding deception by encompassing and in no way contradicting a more specific description of the purpose.
Participants were given the opportunity to contact the researchers if they had any questions or wanted to take part of the result of the study. No such questions were submitted, but several participants expressed interest in being informed of the study’s results, which they were briefed on via email alongside an offer to receive a copy of the complete thesis.
Measurements
A survey consisting of the following self-report scales (original English versions) was used to collect the data: Self-Report Altruism Scale (SRA), Moral Identity Measure,
Interpersonal Reactivity Index (IRI) intermixed with supplemental Empathy Index, Sussex-Oxford Compassion for Others Scale (SOCS-O) and the Social Desirability Scale (SDS). They were presented in the order as listed here. Our primary outcome measure for behavioral altruism (SRA) was given first as to not be influenced by order effects. This was followed by our secondary outcome measure, which focuses on altruistic values (MIM). Thereafter came the measure for empathy (IRI), consisting of items that measure Perspective taking, Empathic concern and Personal distress, followed by the measure for compassion (SOCS-O). Finally, as to not affect any of the other measures, a control for social desirability was given last (SDS).
Self-Report Altruism Scale (SRA). The Self-Report Altruism Scale (Rushton et al., 1981) consists of 20 items. Each item is worded as a statement describing some form of altruistic behavior. Items represent specific behaviors, such as “I have given directions to a stranger” or “I have donated blood”. On each item, respondents are asked to rate how often
they have engaged in the behavior by picking the option most representative of their self-assessment on a 5-point scale: “Never”, “Once”, “More than once”, “Often” and “Very often”. Each item is scored from 1-5, with total scores ranging from 20 to 100. Internal consistency tests in the scale development showed acceptable to good Cronbach’s alpha values, ranging from .78 to .87 (Rushton et al., 1981). In the current sample Cronbach’s alpha was .82 suggesting good internal consistency.
Internalization subscale of the Moral Identity Measure (MIM). The Moral Identity Measure (Aquino & Reed, 2002) consists of 10 items and two subscales, of which only
Internalization was relevant to capture how important certain qualities are for the respondent. The scale measures altruistic values instead of behaviors. The respondent is asked to imagine a person that is “caring, compassionate, fair, friendly, helpful, hardworking, honest, kind”. Each item is worded as a statement. On each item, respondents are asked to rate how important it is to them to possess the presented qualities, as well as how they portray themselves as having these qualities. They do so by picking the option on a 7-point scale ranging from 1 (“Strongly Disagree”) to 7 (“Strongly agree”). An item example is: “Being someone who has these characteristics is an important part of who I am”. Out of all items, 2 were reversed. Total scores range from 1 to 35 for each subscale respectively.
Cronbach’s alpha for the Internalization subscale found in the context of scale development was .84, suggesting a good internal consistency (Aquino and Reed, 2002). Internal consistency for the current sample was good as well, with a Cronbach’s alpha of .80.
Interpersonal Reactivity Index (IRI). The Interpersonal Reactivity Index (Davis, 1980) consists of 28 items. It contains four subscales, each designed to tap a distinct
dimension of empathy. Each subscale consists of 7 items. Each item is worded as a statement describing some form of empathic response. The subscales are Perspective taking (“Before criticizing somebody, I try to imagine how I would feel if I were in their place”), Fantasy (“I
get really involved with the feelings of the characters in a novel”), Empathic concern (“I am often quite touched by things I see happen”) and Personal distress (“I tend to lose control during emergencies”). On each item, respondents are asked to rate the empathic response by picking the option most representative of their self-assessment on a 5-point scale ranging from A (“Does not describe me very well”) to E (“Describes me very well”). Three items are
reversed on the Empathic concern scale, as well as two respectively for Perspective taking, Fantasy and Personal distress. Scoring on items from all scales range from 0-4, with the lowest total score for each subscale being 0 and the highest being 28. Higher scores indicate being higher in the relevant dimension. The subscales are interpreted separately, and no total score is calculated for the IRI. As there is no indication of Fantasy being associated with altruism in past research, this scale was not included in the survey.
Cronbach’s alpha for the subscales of the IRI, found in the context of scale development was .71 to .77, suggesting an acceptable internal consistency (Davis, 1980). Cronbach’s alpha values for the current sample ranged from acceptable to good. They were .81 for Empathic concern, .77 for Perspective taking and .81 for Personal distress, indicating sufficient internal consistency.
Empathy Index. The Empathy Index (Jordan et al., 2016) was developed as
supplement to the IRI and contains 14 items divided into two subscales, “Empathy”, (“If I see someone who is excited, I will feed excited myself”) and Behavioral contagion (“If I see someone else yaw, I am also likely to yawn”). Each subscale consists of 7 items. Each item is answered with five options ranging from A (“Does not describe me very well”) to E
(“Describes me very well”). The scale separates contagion into two distinct kinds, Emotional contagion, represented by the “Empathy” subscale and Behavioral contagion, represented by the “Behavioral contagion” subscale. Unlike the IRI, no items are reversed. Scoring on items from all scales range from 0-4, with the lowest total score for each subscale being 0 and the
highest being 28. Higher scores indicate being higher in the relevant dimension. The subscale of “Empathy” will, in this thesis be referred to as “Emotional contagion” to facilitate
differentiation from the other variables of interest as well as to avoid confusion. The two subscales are interpreted separately from one another, no total score is calculated for the Empathy Index.
Cronbach’s alpha for both subscales of the Empathy Index found in the context of scale development was .71, suggesting an acceptable internal consistency (Jordan et al., 2016). Cronbach’s alpha in the current sample was .72 for Emotional contagion and .65 for Behavioral contagion, suggesting an acceptable internal consistency of Emotional contagion and an inadequate one for Behavioral contagion.
Sussex-Oxford Compassion for Others Scale (SOCS-O). The Sussex-Oxford Compassion for Others Scale (Gu et al., 2019) was used to measure compassion as defined by Strauss et al. (2016). The scale consists of 20 items which represent the components
“Recognizing suffering”, “Understanding the universality of suffering”, “Feeling for the person suffering”, “Tolerating uncomfortable feelings” and “Acting or motivation to act to alleviate suffering". Examples being “I notice when others are feeling distressed” or “I understand that everyone experiences suffering at some point in their lives”. Each dimension is represented by 4 items formulated as statements. Respondents are asked to assess “how they might relate to other people” on a 5-point Likert scale ranging from 1 (“Not at all true”) to 5 (“Always true”), with total scores ranging from 20 to 100. A higher score on the scale indicates a higher level of compassion. The scale is interpreted based on the total score of all items.
Cronbach’s alpha values found the context of scale development ranged from .90 to .94 (Gu et al., 2019), suggesting excellent internal consistency. In the current sample, Cronbach’s alpha was recorded at .88, suggesting a good internal consistency.
Social Desirability Scale (SDS). The most recent version of the scale (Stöber, 2001) consists of 16 items designed to capture the tendency of the responder to describe themselves in a socially desirable fashion. The questions are binary and make statements about the respondent, such as “I sometimes litter” or “I always eat healthy” with the option to answer either “true” or “false”. Six of the items are reversed, with the false option indicating higher levels of social desirability. After reversion, each item is worth 1 or 0 score, with the highest possible score being 16 and the lowest being 0.
Cronbach’s alpha found in the context of scale development was .80 (Stöber, 2001), suggesting good internal consistency. In the current sample Cronbach’s alpha was .72, suggesting an adequate internal consistency.
Research design & statistical Analysis The study has a cross-sectional design
IBM SPSS 26 was used for statistical analysis.
Descriptive analyses were conducted to get an overview of the data (means, standard deviations, zero-order correlations and percentiles). In order to test the null hypotheses, a multiple hierarchical regression was run, as well as two mediation analyses and a moderation analysis which was performed using PROCESS, version 3.4 by Andrew F. Hayes.
Upon inspection of the data of the final sample, no issues were detected for skewness or kurtosis for any of the measures, all values being within the stricter recommended cutoffs of +/-1. Beyond this, the assumptions of normality for the outcome-measure was met in regard to Kolmogorov-Smirnov (p = .07), but not for Shapiro-Wilks (p = .04). Based on the information presented in this paragraph, along with visual inspection of Q-Q plot, the outcome variable Altruism was assessed to meet assumptions of normality, allowing for parametric tests.
Outliers (IQR criterion) were identified in SRA (3), Personal distress (1) and Empathic concern (2) SOCS-O (8). However, their deletion from the data-set did not alter the outcome of the main analyses. Therefore, no univariate outliers were excluded from the further analyses.
The Moral Identity Measure, which was only included as a secondary outcome measure, did not meet assumptions of normality and was severely skewed. For this reason, and in order to allow for use of parametric tests, the Moral identity measure was excluded from all further analyses.
Multiple hierarchical regression was used to estimate which portions of the variance could be attributed to the predictors of interest. These were entered in several steps,
systematically building up the regression model with altruism as the outcome variable. In the first step, SDS was entered into the equation as a control for social desirability. In the second step, compassion was entered into the equation as measured by the SOCS-O. In the third step, the subscales Empathic concern and Perspective taking of the IRI were added together. In the fourth step, the Personal distress subscale of the IRI was added. Finally, Emotional and Behavioral contagion as measured by the Empathy Index were added in the fifth step of the equation.
As we expect compassion to account for the largest portion of the variance, it was entered as the first predictor after the control measure. As Empathic concern and Perspective taking have been shown to be positively associated with altruism, and seem to be interrelated, these were added together in the same step. Personal distress was entered separately as it is distinct and a negative association with altruism was expected. Finally, the Empathy Index consisting of Emotional and Behavioral contagion were entered last. These measures have seen the least use in past research and are less validated compared with the other measures of empathy.
Mediation analysis was then used to further explore the relationship between Perspective taking and Empathic concern, as well as Empathic concern and compassion, in relation to altruism. Finally, a moderation analysis was used to see whether the effect of Personal distress on altruism was moderated by compassion. Initially, the moderation did not detect the main effect of Personal distress on altruism seen in the multiple hierarchical regression, possibly due to collinearity issues. A transformation was performed which centered the variables by subtracting the means from each score for Personal distress and compassion respectively, after which the expected associations were found. Note that this transformation did not, and was not intended to, alter the findings regarding an interaction effect between Personal distress and compassion, it merely allowed interpretability of the interaction effect.
Results Data screening
Linearity between the outcome measure of altruism and all predictor variables were confirmed by means of visual inspection of scatter plots. Visual inspection of scatter plot with standardized residuals and predicted values showed a spread with no pattern for all predictor variables. This indicates adequate homoscedasticity for the data on the multiple regression analysis as well as mediation and moderation.
Independent errors were within recommended cutoffs (Durbin-Watson = 2.20), indicating that residuals are not correlated. Collinearity statistics showed no indication of issues with multicollinearity, with VIF and tolerance within acceptable range of the stricter criterion (VIF < 2.5) for all variables at all stages of the model.
Multiple hierarchical regression analyses revealed one multivariate outlier on the variables Empathic concern (2.80) and compassion (1.43), (Mahalanobis distance = 26.62, p < .001). This outlier was excluded from all further analyses. Deletion of the outlier did not alter
the general findings, but did improve the predictive value multiple regression model from 15.9% to 16.5%. This left 240 participants for the remaining analyses.
Sample characteristics and correlations
Pearson correlations were computed between all measures as can be seen in Table 3. Effect sizes are estimated using recommended values with .10 indicating a small effect, .30 a medium effect, and .50 and above, a large effect. Alfa criterion for rejecting null hypotheses was set to < .05.
To allow for comparison between measures, means and standard deviations will be reported based on scale scores rather than the total sum of scores. This, since the measures used in analyses, were all answered on a scale from 1-5, while differing in maximum and minimum total scores for several measures. Basing means and standard deviations on scale scores facilitates visual comparison of results.
Correlations with altruism were as follows: As anticipated, Perspective taking, Empathic concern, and Compassion, were positively correlated. Perspective taking and Empathic concern both showed a small positive association, while compassion showed a medium positive association. Also, in line with expectations, Personal distress showed a small negative association. Emotional contagion showed no significant association. Unexpectedly, Behavioral contagion showed a small positive association. Beyond this, compassion showed a medium size association with Perspective taking, and a large association with Empathic concern, which was also in line with expectations.
Table 3.
Social desirability (SDS) (M = 7.30, SD = 3.31), did not show a significant
relationship with altruism. It would be unhelpful to report statistics based on scale scores for a binary variable. These are instead reported based on the total sum of scores (where the highest level of social desirability is reflected by the maximum 16), in contrast to all other variables. Multiple hierarchical regression
The results of the multiple hierarchical regression analysis are presented in Table 4. Taken together, social desirability did not significantly explain any of the variance in altruism. Compassion explained 10% of the variance. Specifically, people high in compassion were more likely to have high scores on altruism. Perspective taking and Empathic concern did not significantly explain any additional the variance. Personal distress on the other hand explained another 4% of the variance. Specifically, persons with high scores on Personal distress were less likely to engage in altruism. Intriguingly, adding Personal distress made Empathic concern significantly impact the model. Finally, when Emotional contagion and Behavioral contagion where added, this did not significantly help predict altruism.
Mediation
A mediation analysis was conducted to examine the role of Empathic concern in the relationship between Perspective taking and altruism. See Figure 1. Results revealed that Perspective taking positively predicted Empathic concern, b = .43, t(237) = 6.75, p < .001, BootCI[.30, .56]. Moreover, Empathic concern positively predicted altruism after controlling for the effect of Perspective taking, b = .16, t(236) = 3.42, p < .001, BootCI[.06, .25].
Sobel test indicated a significant indirect effect of Perspective taking on altruism through Empathic concern, bind = .07, z = 3.06, p = .002, BootCI[.03, .12]. This effect was then divided by the total direct effect of Perspective taking on altruism, b = .10, t(236) = 2.25, p = .025, 95% CI[.01, .19], in order to attain the proportion of the total effect operating
through Empathic concern. The effect of Empathic concern on altruism that operated indirectly was 65%, indicating a medium indirect effect.
Perspective taking predicts individuals being more emphatically concerned, which in turn predicts altruism. In other words, having controlled for Empathic concern, there is no predictive value of Perspective taking on altruism.
Figure 1.
Mediation analysis of Empathic concern on the relationship between Perspective taking and altruism.
Next, to establish what the active component of Empathic concern was in predicting altruism we conducted a mediation analysis with compassion as mediator, see Figure 2.
Results revealed that Empathic concern positively predicted compassion, b = .36, t(237) = 11.66, p < .001, 95% BootCI[.30, .42]. Moreover, compassion positively predicted altruism after controlling for the effect of Empathic concern, b = .30, t(236) = 3.43, p < .001, 95% BootCI[.13, .47]
Sobel test indicated a significant full mediation effect of Empathic concern on altruism through compassion, bind = .11, z = 3.29, p < .001, 95% BootCI[.03, .18]. This was then divided with the total effect of Empathic concern on Altruism, b = .17, t(236) = 4.07, p < .001, 95% CI[.09, .26], in order to attain the proportion of the total effect operating through compassion. The effect of Empathic concern on Altruism that operated indirectly was 62%, indicating a medium indirect effect.
Empathic concern predicts individuals being more compassionate, which in turn predicts altruism. In other words, having controlled for compassion, there is no predictive value of Empathic concern on altruism.
Figure 2.
Mediation analysis of compassion on the relationship between Empathic concern and altruism.
Moderation
Finally, it was examined whether the negative association between Personal distress and altruism was moderated by compassion. A regression analyses with centered predictors revealed the following result.
The main effect of Personal distress on altruism was significant, b = .11, t(235) = -2.86, p = .005, 95% BootCI[-.19, -.03]. Furthermore, the main effect of compassion on Personal distress was significant, b = .35, t(235) = 5.08, p = .109, 95% BootCI[.19, .49]. However, compassion did not significantly moderate the relationship between Personal distress and altruism, as reflected in a non-significant interaction between Personal distress and compassion on altruism, b = -.01, t(235) = -.44, p = .883, 95% BootCI[-.19, .21].
There was no indication that the level of compassion influenced the effect of Personal distress on altruism. That is, the degree of compassion did not mitigate the negative predictive effect of Personal distress on altruism.
Discussion
The present survey aimed to investigate empathy and compassion as trait predictors of
altruism. Self-report measures of empathy included measures of Perspective taking, Empathic concern, Personal distress, Emotional contagion, and Behavioral contagion. We hypothesized that (1) compassion would explain more variance in altruism than empathy variables, (2) that any positive association between Perspective taking and altruism would be mediated by Empathic concern and, (3) that any positive association between Empathic concern and altruism would be mediated by compassion. Finally, we hypothesized (4) that the expected negative association between Personal distress and altruism would be moderated by
compassion. That is, individuals high in Personal distress would be less inclined towards altruism, but only if they were low in compassion. In contrast, individuals with high compassion would not be deterred from acting altruistically by Personal distress.
The results can be readily summarized and support our first three hypothesis. Compassion accounted for the largest portion of the variance in self-reported altruism, the relation between Perspective taking and altruism was mediated by Empathic concern, and the relation between Empathic concern and altruism was mediated by compassion. The relation between Personal distress and altruism was not moderated by level of compassion, meaning our fourth hypothesis was not supported. These findings are further discussed in what follows.
Empathy
In the present study, several components of empathy showed significant correlations with altruism. Perspective taking and Empathic concern showed a positive association, while Personal distress showed a negative association. This is in line with previous research, which has found Perspective taking and Empathic concern to positively predict altruism (Lim & DeSteno, 2016) and Personal distress to negatively predict it (Kim & Han, 2018).
Surprisingly, Behavioral contagion was positively associated with altruism. While the effect was small in size, its presence still raises the question as to why this occurred. Contagion is closely theoretically associated with Personal distress (de Waal, 2008), and was expected to show a negative relationship as it has in previous research (Jordan et al., 2016). Emotional contagion has shown a negative relationship with altruism in the past, but had no significant relationship in the present study.
One possible explanation for the positive association between Behavioral contagion and altruism could be that the relationship between contagion and altruism is moderated by compassion. The high levels of compassion in the sample may have altered the impact of contagion, which would be an interesting find in itself. Helping someone necessitates noticing the need for help, and contagion can arguably facilitate noticing. People can be distressed without expressing it verbally. As such, Emotional contagion may alert a person to the
distress of another, allowing for the noticing of a need that would otherwise have gone
unnoticed. If the person is high in compassion, and notices this need, they would theoretically be more inclined to help. Someone low in compassion may by contrast be overwhelmed by Personal distress, and avoid the source of the Emotional contagion. Future research is needed to clarify contextual conditions for the effect of contagion on altruism, as well as its
relationship with the other components of empathy and compassion.
As expected, the positive association between Perspective taking and altruism was mediated by Empathic concern. It seems that the warm, invested feelings associated with Empathic concern is the driver behind altruism. Being able to put oneself in the position of another in need of help, only motivates behavior to the degree it generates this affective state, which confirms findings from previous research (Lim & DeSteno, 2016; Jordan et al., 2016).
Unlike most previous studies which focus exclusively on the predictive value of Empathic concern and Perspective taking in relation to altruism, the current study also takes into account the possible roles of Emotional contagion, Behavioral contagion and Personal distress. This provides a more complete picture of how empathy relates to altruism and
compassion. This is further explored below under the headings Compassion and Tolerance for distress.
Compassion
Similar to the findings made in earlier psychometric studies of the SOCS-O compassion scale, the SOCS-O and IRI empathy scale showed a significant overlap. Perspective taking showed a moderate relationship (.45) and Empathic concern showed a strong correlation at (.60), Personal distress a small correlation at (-.10). There were no associations above .80 (Gu et al., 2019). The only deviation from previous research was on the overlap of Personal distress and compassion, where no significant correlation was found in the present study, although the correlation was similar in size. Thus, indicating that the
