Focusing on the Negative: Cultural Differences in Expressions of Sympathy

Birgit Koopmann-Holm; Jeanne L Tsai

doi:10.1037/a0037684

. Author manuscript; available in PMC: 2015 Dec 1.

Published in final edited form as: J Pers Soc Psychol. 2014 Sep 22;107(6):1092–1115. doi: 10.1037/a0037684

Focusing on the Negative: Cultural Differences in Expressions of Sympathy

Birgit Koopmann-Holm ¹, Jeanne L Tsai ¹

PMCID: PMC4340666 NIHMSID: NIHMS638231 PMID: 25243416

Abstract

Feeling concern about the suffering of others is considered a basic human response, and yet, we know surprisingly little about the cultural factors that shape how people respond to the suffering of another person. To this end, we conducted four studies that tested the hypothesis that American expressions of sympathy focus on the negative less and positive more than German expressions of sympathy, in part because Americans want to avoid negative states more than Germans do. In Study 1, we demonstrate that American sympathy cards contained less negative and more positive content than German sympathy cards. In Study 2, we show that European Americans want to avoid negative states more than Germans do. In Study 3, we demonstrate that these cultural differences in “avoided negative affect” mediate cultural differences in how comfortable Americans and Germans felt focusing on the negative (vs. positive) when expressing sympathy for the hypothetical death of an acquaintance's father. To examine whether greater avoided negative affect results in lesser focus on the negative and greater focus on the positive when responding to another person's suffering, in Study 4, American and German participants were randomly assigned to one of two conditions: (1) to “push negative images away” (i.e., increasing desire to avoid negative affect), or (2) to “pull negative images closer” to themselves (i.e., decreasing desire to avoid negative affect). Participants were then asked to pick a card to send to an acquaintance whose father had hypothetically just died. Across cultures, participants in the “push negative away” condition were less likely to choose sympathy cards with negative (vs. positive) content than were those in the “pull negative closer” condition. Together, these studies suggest that cultures differ in their desire to avoid negative affect, and that these differences influence the degree to which expressions of sympathy focus on the negative (vs. positive). We discuss the implications of these findings for current models of sympathy, compassion, and helping.

Keywords: Culture, Emotion, Avoided Negative Affect, American, German, Sympathy, Compassion

Imagine that the father of one of your acquaintances just died, and you know she is extremely sad. You want to express your sympathy, so you decide to send her a card. But what type of card do you choose? Do you choose a black and white card that contains phrases such as “In deep sadness,” and “Words will not lighten a heavy heart,” or a colorful card that contains phrases such as “Love lives on,” and “Memories will bring comfort”?

In this paper, we test the hypothesis that cultural factors shape how people express concern for another person's suffering. Specifically, we predict that because American culture encourages people to avoid negative emotion more than German culture does, American expressions of sympathy focus on the negative less and the positive more than German expressions of sympathy. We conducted four studies to test this hypothesis. Prior to describing the studies, we briefly review previous work on sympathy, compassion, and other feelings of concern about another person's suffering, and then propose how culture may influence expressions of sympathy based on our theoretical framework, Affect Valuation Theory (Tsai, 2007).

Previous Research

For decades, social scientists have been interested in the functions of sympathy and compassion in everyday life¹ (e.g., Clark, 1997) as well as the factors that influence these emotions (e.g., Batson et al., 1989; Batson et al., 1991; Cialdini, Darby, & Vincent, 1973; Davis, 1983; Eisenberg et al., 1989; Eisenberg et al., 1998; Eisenberg & Miller, 1987; Lennon & Eisenberg, 1987; Toi & Batson, 1982). For instance, the empathy-altruism hypothesis originally proposed by Batson and colleagues (Batson, 1981, 1983; Batson et al., 1991) argues that the more people feel distressed (vs. empathetic) when they see another person suffering, the less likely they are to help that person. In contrast, the negative state relief model proposed by Cialdini and colleagues (1973; 1987) argues that the more empathetic people feel, the sadder they feel, and people's desire to reduce this sadness leads them to help others. In recent years, there has been a resurgence of interest in sympathy, compassion, and other feelings of concern for another's suffering (e.g., Condon & DeSteno, 2011; Condon & Feldman Barrett, 2013; Goetz, Keltner, & Simon-Thomas, 2010; Keltner, 2009; Nussbaum, 1996; Oveis, Horberg, & Keltner, 2010; Valdesolo & DeSteno, 2011; Van Kleef et al., 2008; Zaki & Ochsner, 2012). However, because the vast majority of these studies have focused primarily on European American samples, we still know little about the role culture plays in shaping people's responses to another's suffering (Wuthnow, 2012, p. 306), despite arguments that the rules and logic surrounding when to give and receive sympathy are culturally shaped (Clark, 1997).

The Role of Culture

Of the studies that have examined the role of culture, both cultural similarities and differences have been observed. Across different cultures, sympathy and compassion are viewed as emotions (Shaver, Murdaya, & Fraley, 2001) that can be differentiated from other emotions via touch (Hertenstein, Keltner, App, Bulleit, & Jaskolka, 2006), have similar triggers (e.g., person who is not responsible for his/her suffering) (Zhang, Xia, & Li, 2007), and are highly valued (Miller & Bersoff, 1994). Furthermore, cultural similarities have been observed in the links between experiencing sympathy/compassion and helping others (e.g., helping someone who has just lost a valued toy, or giving money to a needy stranger) (Eisenberg, Zhou, & Koller, 2001; Kitayama & Markus, 2000; Trommsdorff, Friedlmeier, & Mayer, 2007) as well as in the specific patterns of brain activity that occur during empathy for another person's pain (e.g., de Greck et al., 2012; Jiang, Varnum, Hou, & Han, 2014).

Differences have been observed as well. For instance, on average, people from countries such as Brazil and Costa Rica that value “simpatia” (being concerned about another's well-being) and poorer nations such as Malawi and India are more likely to help a blind person, a person with a hurt leg, or someone who has dropped a pen (Levine, Norenzayan, & Philbrick, 2001) compared to people from countries that do not have a tradition of “simpatia” such as Singapore and Malaysia and wealthier nations such as the Netherlands and the U.S. Indeed, these findings are consistent with work by Stellar, Manzo, Kraus, and Keltner (2012) demonstrating that within the United States, individuals from lower socioeconomic contexts show more compassion towards a peer undergoing a stressful job interview than do those from higher socioeconomic contexts. Other studies suggest that although there are cultural similarities in the patterns of brain activity associated with empathizing with another person's pain, there are also differences (e.g., de Greck et al., 2012; Jiang et al., 2014). For example, whereas Chinese participants showed a pattern of brain activity suggesting that they were regulating their emotions when empathizing with a familiar angry target, German participants showed a pattern of brain activity suggesting that they were assuming the perspective of the angry target (de Greck et al., 2012).

No studies, however, have examined how culture shapes the different ways in which sympathy, compassion, or other feelings of concern for another's suffering might be expressed. Previous research suggests that sympathy and compassion are mixed (i.e., positive and negative) feeling states (e.g., Condon & Feldman Barrett, 2013; Eisenberg et al., 1994); however, cultures might differ in whether expressions of sympathy focus on the positive (e.g., the “silver lining”) or the negative (e.g., the pain and suffering of another). For instance, while most people want to avoid feeling negative, cultures and individuals may differ in the degree to which they want to avoid negative states, which may influence how people express their sympathy. In cultures that encourage people to avoid negative states more, people may focus more on the positive and less on the negative when expressing their sympathy, whereas in cultures that encourage people to avoid negative states less, people may focus on the negative more and positive less. We test this prediction in this paper. First, we define “avoided affect” and then describe the theoretical framework motivating this work, Affect Valuation Theory (Tsai, 2007).

The Role of Avoided Negative Affect

“Verweinen lasst die Nächte mich, Solang ich weinen mag” [Let me pass the nights in tears, As long as I want to cry] (Johann Wolfgang von Goethe; 1749-1832)

“Be still, sad heart! And cease repining; Behind the clouds is the sun still shining” (Henry Wadsworth Longfellow; 1807-1882)

Feeling negative--- sad, angry, afraid, sluggish----is a part of daily human life. We are passed up for a job; we argue with a friend; we lose a loved one. However, as suggested by the above two quotes, people differ in their responses to feeling bad. Whereas some people appear to accept and embrace their negative feelings, as reflected in the quote by German poet Goethe, others try to stop their negative emotions and focus on the positive, as reflected in the quote by American poet Longfellow. Although much research has focused on the experience of negative emotions (e.g., Brainerd, Stein, Silveira, Rohenkohl, & Reyna, 2008; Chorpita, Albano, & Barlow, 1998; Forgas, 2008; Frasure-Smith, Lesperance, & Talajic, 1995; Gross et al., 1997; Gross & Levenson, 1997; Kiefer, 2005; Kuppens, Realo, & Diener, 2008), surprisingly few studies have examined variation in the degree to which people want to avoid negative emotions, and the factors that account for such variation.

Instead, studies have demonstrated individual, situational, and cultural differences in people's attitudes toward negative emotion (e.g., Andrade & Cohen, 2007; Chambless, Caputo, Bright, & Gallagher, 1984; Grossmann & Kross, 2010; Harmon-Jones, Harmon-Jones, Amodio, & Gable, 2011; Izard, 1971; Sommers, 1984; Tamir & Ford, 2009; Tamir, Mitchell, & Gross, 2008; Taylor & Rachman, 1991; Västfjäll & Gärling, 2006; Williams, Chambless, & Ahrens, 1997). However, these studies primarily focus on how much people dislike negative emotions rather than how much people want to avoid negative emotions. For example, Harmon-Jones et al. (2011) examined how much people like or dislike negative emotions (e.g., “I like thinking about sad things”, “I dislike how it feels when I am angry”). Similarly, Izard (1971) and Sommers (1984) asked their participants to report how much they dreaded different negative emotions (e.g., ”which emotion do you dread the most”; “which emotions do you dread having”), which could reflect both how much people dislike feeling different negative emotions as well as how much they want to avoid feeling negative emotions. Previous research illustrates the differences between liking and wanting (e.g., Dai, Brendl, & Ariely, 2010), differences that extend to disliking and avoiding as well. For example, while someone might not like feeling sad, she might still want to feel sad (i.e., not want to avoid it) because she believes that feeling sad will allow her to come to a deeper understanding of herself. Although disliking negative emotions and wanting to avoid negative emotions are both important topics of study, we were interested in the latter because we thought that it would influence how people respond to another's suffering. Although some scholars have examined the avoidance of emotions in general (Hayes et al., 2004; Maio & Esses, 2001), none have examined the desire to avoid negative emotions specifically.

Affect Valuation Theory

In Affect Valuation Theory (AVT; Tsai, 2007; Tsai, Knutson, & Fung, 2006), we proposed that: (1) the affective states that people actually feel (their “actual affect”) is different from the affective states that people ideally want to feel (their “ideal affect”), (2) cultural factors shape ideal affect more than actual affect, and temperamental factors shape actual affect more than ideal affect, and (3) ideal affect shapes what people do to feel good, as well as their conceptions of health and happiness. Because most people want to feel good, research on ideal affect has primarily focused on positive states (Tsai, 2007; Tsai et al., 2006; Tsai, Louie, Chen, & Uchida, 2007; Tsai, Miao, & Seppala, 2007; Tsai, Miao, Seppala, Fung, & Yeung, 2007). However, here we were interested in the negative states that people want to avoid, and therefore, we briefly discuss how AVT may apply to “avoided affect.”²

The affective states that people want to avoid feeling (their “avoided affect”) should differ from their actual and ideal affect. Whereas ideal affect is a desired state, one that people consciously or unconsciously actively work to achieve, avoided affect is an undesired state, one that people consciously or unconsciously actively work to stay away from or elude. Both are different from actual affect, which is a response to an event or an outcome. In the same way that on average, people do not feel how they want to feel, reflecting a mismatch between actual and ideal affect (e.g., having less fun at a party than one wished), we predict that people may often feel how they do not want to feel, reflecting a mismatch between actual and avoided affect (e.g., getting angry at a friend or loved one).

We also predict that cultural factors shape avoided affect more than actual affect. Rozin (2003) and Shweder (2003) argue that cultural factors shape what people view as good, moral, and virtuous. In AVT, we applied this idea to affective states; i.e., cultural factors shape what affective states people view as good, moral, and virtuous. Similarly, culture should shape the affective states that people view as bad, immoral, and sinful. Indeed, previous research has documented cultural differences in undesirable states (Eid & Diener, 2001; Izard, 1971; Sommers, 1984). For example, Eid and Diener (2001) found that people in individualistic cultures viewed “guilt” as more undesirable than people in collectivistic cultures. Although cultural factors also shape what affective states people actually feel (Kitayama, Markus, & Kurokawa, 2000; Mesquita & Markus, 2004), decades of empirical research suggest that across cultures, actual affect is also shaped by people's temperament (Costa & McCrae, 1980; David, Green, Martin, & Suls, 1997; Diener & Lucas, 1999; Gross, Sutton, & Ketelaar, 1998; Lykken & Tellegen, 1996; McCrae, Costa, & Yik, 1996; Rusting & Larsen, 1997; Schimmack, Radhakrishnan, Oishi, Dzokoto, & Ahadi, 2002) as well as their immediate circumstances and regulatory abilities (e.g., Gross, 1998). Thus, we predict that cultural factors should shape avoided affect more than actual affect.

Finally, AVT predicts that avoided affect should influence how people respond to various negative events, including another person's suffering, above and beyond actual affect. If people want to avoid negative states to a great extent, they likely try to avoid people, places, and events that elicit negative emotion (e.g., by trying not to run into an irritable colleague, by not walking home late at night, or by deciding not to see a movie with disturbing content). These attempts, however, may be unsuccessful (e.g., people cannot avoid the irritable colleague, late nights in the office, or unexpected content of films). Under those circumstances, avoided affect should predict how people respond to and cope with the experience of negative emotion. For instance, in response to the death of a loved one, the more individuals want to avoid negative affect, the less they may focus on the negative and the more they may focus on the “silver lining,” or the positive aspects of a negative event (e.g., by thinking of the beautiful memories they have of their loved ones) and try to “cease repining” (e.g., by engaging in pleasant activities and encouraging others to do the same). Conversely, the less people want to avoid feeling negative, the less they may focus on the positive and the more they might focus on the negative (e.g., by acknowledging how painful it is to lose a loved one) and try to “pass the night in tears” (e.g., by allowing oneself to grieve and encouraging others to do the same). We predict that these different responses to and expressions of concern for others’ suffering are driven by how much people want to avoid negative affect above and beyond how much they actually experience negative affect (as suggested by some models of helping). The present study is the first to test these predictions, and in doing so, to expand AVT to negative states.

American vs. German Cultural Contexts

Most research on emotion in cultural and cross-cultural psychology, including our own previous work, has focused on differences between East Asian and Western cultures, and has attributed observed differences to individualism-collectivism. Much less work has focused on variation across different individualistic Western cultures (e.g., American vs. German). The few studies that have do find important variation; for instance, American and German cultures differ in their value priorities: while Americans give higher priority to “mastery” values such as achievement and controlling the environment than Western Europeans (including Germans), Western Europeans give higher priority to “harmony” values such as fitting into the environment than Americans (Schwartz & Ros, 1995). Furthermore, American and German cultures differ in their display rules of negative emotions (Koopmann-Holm & Matsumoto, 2011). For instance, the expression of anger and sadness is more acceptable in German than in American contexts. We were particularly interested in comparing Americans and Germans because ethnographic, historical, and personal accounts suggest that Americans want to avoid negative emotions more than Germans do. For instance, German scholar Hedderich (1999) conducted semi-structured interviews with American and German employees (from four American and four German companies), who had spent at least six months in the other country. He asked them about differences between the cultures and concluded that Americans resist talking about their failures, indirectly referring to them as ‘items for improvement’ (Hedderich, 1999, p. 161) and instead, praise each other for their achievements. Consistent with this observation, in “American Cool,” historian Peter Stearns (1994) states that in American society, “fear and anger had no positive function [...]; rather than being directed, they were to be avoided as fully as possible” (p. 96). Indeed, American culture considers “negativity, complaining, pessimism” sinful (Held & Bohart, 2002, p. 961).

Similarly, McAdams (2004) describes the key feature of contemporary American identity as “the transformation of personal suffering into positive-affective life scenes that serve to redeem and justify one's life” (p. 96), as reflected in American stories, which characteristically have positive endings. Positive endings are also found in American personal accounts of trial and tribulation. For instance, “Chicken Soup for the Soul: The Cancer Book” contains “101 Stories of Courage, Support and Love” (Canfield, Hansen, & Tabatsky, 2009), implying that even in the face of a life-threatening illness such as cancer, Americans can and should focus on the positive (Ehrenreich, 2009). This may be why some cancer survivors, including cyclist Lance Armstrong, refer to cancer as “the best thing that ever happened” to them (Henneberger, 2013). Indeed, as described in her book, “Bright-Sided: How positivity thinking is undermining America,” when American Ehrenreich (2009) publically expressed her anger and anxiety about having breast cancer, other breast cancer patients admonished her and suggested she seek psychological help.

In contrast, Germans openly and explicitly state when something is bad; for instance, many of the respondents in Hedderich's study (1999) commented that Germans are more likely to use the phrase, “das war Mist,” or “that was rubbish,” when something bad happened, compared to Americans. Unlike American personal accounts of illness, German accounts primarily focus on the negative. For instance, in his book about his battle with cancer entitled “Ein Jahr Hölle [One year of hell],” German actor Michael Lesch primarily described the anxiety and horror of having cancer, and even years later, talked about cancer as mainly being a horrible experience (Lesch, 2008). Indeed, German culture is often described as being melancholic and pessimistic, as the terms “Weltschmerz” and “Angst” suggest (Clair, 2005; Gelfert, 2005) and as reflected by the “Sturm und Drang” [“Storm and Drive”] movement in German literature and music in the 18^th century, in which negative emotions were not just accepted, but glorified.

Why might American and German contexts differ in their desire to avoid negative affect? One possible explanation is that these differences stem from the historical origins of the United States. Although ancestors from both countries originated in Europe, the founding fathers of the United States left Europe to escape religious persecution, famine, and financial hardship (Brooks, 2004; Wilson, 2008). In essence, these individuals responded to their negative life circumstances by dreaming of a “brighter future” in the New World, and by leaving their European homelands to realize that dream (Brooks, 2004; Wilson, 2008). According to the “voluntary settlement hypothesis” (e.g., Kitayama, Ishii, Imada, Takemura, & Ramaswamy, 2006; Kitayama, Park, Sevincer, Karasawa, & Uskul, 2009), the traits and values of these early settlers became the basis of American independence and individualism (Lipset, 1997; Wilson, 2008). Several scholars have referred to these traits as the “frontier spirit” (“American Frontier” Spirit; Turner, 1921), which included not only a desire to escape one's negative circumstances to achieve one's dreams but also a willingness to tame and master the wilderness of the New World in order to do so (Conway, Houck, & Gornick, 2014). In his book, “The Frontier Spirit and Progress,” historian Tucker (1980) defined frontiers as “those areas in which men are most dynamically in motion to master the environment” (p. 4).

Thus, by leaving Europe and moving to the United States, the early settlers of the United States may have created a culture in which individuals want to avoid the negative and seek the positive. In contrast, the ancestors of modern Germany (and other parts of Europe) chose to stay in their European homelands. Although they may have stayed for many reasons, one possibility is that instead of dreaming of a new life, they responded to economic and religious hardship by accepting and adjusting to their negative life circumstances. Moreover, they may have preferred their current---albeit imperfect---life circumstances to the idea of crossing a vast ocean and then taming the wilderness of the New World. Thus, these individuals may have been part of or may have created a culture in which individuals are more accepting of the negative, and therefore, have less of a desire to avoid negative emotion. Indeed, a frontier spirit might explain why there are few differences between the US and Germany on the broad dimensions of individualism-collectivism but many differences in specific “individualistic” values such as self-enhancement (e.g., achievement) and specific “collectivistic” values such as self-transcendence (e.g., fitting into nature) (e.g., Koopmann-Holm & Matsumoto, 2011; Schwartz & Ros, 1995).

In sum, we predicted that contemporary American culture wants to avoid negative states more than contemporary German culture in part because American culture endorses a frontier spirit (i.e., achieving one's goals, influencing one's circumstances, overcoming nature) more than German culture does. Although not the main focus of this paper, we begin to test this hypothesis in Study 3. Many sociologists and historians have written about the frontier spirit and its links to independence and individualism (e.g., Lipset, 1997; Tucker, 1980; Turner, 1921; Wilson, 2008). Additionally, empirical studies have tested these ideas (e.g., Kitayama et al., 2006; Kitayama et al., 2009; Varnum, 2013, 2014; Varnum & Kitayama, 2011). For instance, one empirical study found that positive emotions were more independent and less interdependent among Japanese living in the frontier of Hokkaido compared to those living in mainland Japan (Kitayama et al., 2006). In the present study, we examine whether differences in the endorsement of specific values associated with the frontier spirit (achievement and fitting into nature) are related to differences in views of negative emotion between the United States and Germany. By comparing views of negative emotion in American and German contexts, this work significantly broadens the current literature on emotion across cultures, which has primarily compared North American and East Asian contexts.

Overview of Studies

The present studies examined whether cultural differences in expressions of sympathy exist, and if they do, whether cultural differences in the desire to avoid negative affect can explain these differences. Study 1 examined whether American sympathy cards focus less on the negative and more on the positive compared to German sympathy cards. Study 2 examined whether avoided affect is distinct from actual affect and ideal affect, and then examined whether Americans want to avoid negative affect more than do Germans. As described by circumplex/dimensional models of emotion, negative states vary in terms of arousal (Russell, 1980; Russell, Lewicka, & Niit, 1989). Therefore, in Study 2, we sampled high arousal negative (e.g., fearful, hostile, nervous), low arousal negative (e.g., dull, sleepy, sluggish), and moderate arousal negative (e.g., sad, unhappy, lonely) states to examine whether differences in actual, ideal, and avoided affect held for different types of negative states. Study 3 then examined whether these cultural differences in avoided negative affect mediated cultural differences in expressions of sympathy among Americans and Germans in a hypothetical scenario. In Study 3, we also examined whether cultural differences in avoided negative affect were due to cultural differences in specific frontier values. Finally, in Study 4, we experimentally manipulated avoided negative affect in American and German samples to assess whether we could alter the degree to which people focused on the negative (vs. positive) when expressing sympathy for another's suffering in a hypothetical scenario.

Study 1: Cultural Differences in Emotional Content of Popular Sympathy Cards

To begin to examine whether Americans and Germans differ in how they respond to the suffering of another person, we compared the content of American and German sympathy cards. Various studies have examined the content of “cultural products” (e.g., newspapers, books, television, advertisements, architecture, laws, websites), or widely-distributed objects that both reflect and reinforce cultural ideas (e.g., Han & Shavitt, 1994; Lamoreaux & Morling, 2012; Markus, Uchida, Omoregie, Townsend, & Kitayama, 2006; Morling & Lamoreaux, 2008; Tsai, Louie, et al., 2007; Tsai, Miao, & Seppala, 2007). Indeed, Morling and Lamoreaux (2008) argue that in some cases, examining the content of widely distributed products is an even better way of illustrating dominant ideas and practices because these products are typically marketed to a general public, and therefore, target culturally shared ideas. This is in contrast to responses to questionnaire measures, for example, which typically reflect a combination of shared (e.g., culture) and unique (e.g., temperament) influences.

Moreover, because these cultural products are widely distributed, members of a culture learn how to think and feel by being exposed to them. For example, in previous work examining cultural differences in ideal affect, our lab compared the emotional content of best-selling American and Taiwanese children's storybooks and observed that American storybooks contained characters with more excited and fewer calm smiles. We then showed that being exposed to storybooks that promoted either excitement or calm altered children's affective preferences assessed immediately after (Tsai, Louie, et al., 2007).

In the present study, we focused on sympathy cards because in American and German contexts, sympathy cards are designed to show one's sympathy and compassion toward someone who has lost a loved one, and to help the receiver cope with his or her pain. In both cultures, the cards are widely distributed and therefore readily available. Moreover, Americans and Germans are regularly exposed to these cards because they are sold in a variety of stores. Finally, because cards are a consumer product like storybooks and magazines, we assumed that when people purchase the cards, they are expressing a preference for the type of emotions conveyed in the cards, which we predict is shaped at least partially by one's cultural values.

In addition to examining American and German sympathy cards, as a control comparison, we compared the emotional content of American and German “baby cards,” or cards sent when a friend or acquaintance has just had a baby. We chose baby cards as a control comparison because like sympathy cards, baby cards are sent in response to a significant emotional event in someone's life. However, compared to the death of a loved one, the birth of a baby is typically viewed as a positive event, and therefore, we predicted that cards sent in response to the birth of a baby should not be as influenced by the degree to which a culture encourages people to avoid feeling negative as cards sent in response to the death of a loved one.

Hypotheses

We predicted that: (1) American sympathy cards would contain less negative (i.e., fewer words related to sadness and death, lesser acknowledgment of suffering and grief, less dying imagery) and more positive (i.e., more words related to positive feelings and optimism, more encouragement, and more living imagery) content than would German sympathy cards, and (2) cultural differences in the emotional content of sympathy cards would be greater than that of baby cards.

Method

Card Selection

In order to select a representative sample of sympathy and baby cards that are widely distributed in American and German contexts, we first asked 28 European American (67.86% female; mean age = 19.00, SD = 1.12) and 22 German undergraduate students (81.82% female; mean age = 21.95, SD = 1.53) from large top-tier universities in the US and Germany to list the top three places they would go to purchase a sympathy card if they wanted to send one to a friend or acquaintance who had just lost a loved one. We determined how often each store was mentioned as well as whether the store was mentioned first, second, or third (for more information, please contact the first author). European American participants most often mentioned Hallmark, CVS, Target, and Walmart (Target and Walmart were tied). German participants most often mentioned Kaufhof, “Buchhandlung” (German for “bookstore”), and “Schreibwarenladen” (German for “stationary store”).

Most of the American cards that the American stores carried were sold by Hallmark and American Greetings (including its greeting card brands, Carlton Cards and Papyrus), and most of the German cards that the German stores carried were sold by bsb, Taurus, Hallmark, and Hanra. We first contacted these companies for a complete catalogue of their cards. In Germany, bsb and Taurus provided their entire catalogues; Hallmark provided a random selection of their cards (cards in their on-line catalogue were not sold in stores), and Hanra provided some cards and suggested that we obtained the others from their website (cards in their on-line catalogue were the same as those sold in stores). This process produced a total of 376 German (246 sympathy, 130 baby) cards.

In the US, Hallmark and American Greetings said that they did not have their entire catalogues and suggested that we purchase the cards at various stores. Therefore, we went to a total of five stores (2 Hallmark, 2 Target, 2 CVS Pharmacy, and 1 Walmart) in California and purchased all the sympathy and baby cards they carried. This process produced 338 American (198 sympathy, 140 baby) cards. ³

Card Coding

Frequency of negative and positive word use (Percentage of total words that were positive or negative)

We used the English version of the Linguistic Inquiry and Word Count Program (LIWC; Pennebaker, Francis, & Booth, 2001) to code the text of the American and German sympathy and baby cards.⁴ A research assistant fluent in English and German translated the content of the German cards into English, and another research assistant also fluent in English and German back-translated these English translations into German to ensure equivalence. Both translators were blind to the study hypotheses. The few discrepancies that arose were discussed, and modifications were made to the original translation to arrive at the best possible English translation of the German text. Then we ran the text of the American and German cards through the English version of LIWC. LIWC calculates the total number of words, and then calculates the percentage of total words that fall into a given category. To assess the frequency of negative words, we added the percentages of total words that fell into the LIWC categories of “Sadness or depression” (e.g., grief, cry, sad) and “Death and dying” (e.g., dead, burial, coffin). To assess the frequency of positive words, we added the percentages of total words that fell into the categories “Positive feelings” (e.g., happy, joy, love) and “Optimism and energy” (certainty, pride, win).

Context of word use

While the types of words that occur in the cards are important, so is the context or way in which each word is used. For instance, the experience of sadness can be encouraged and acknowledged (“take time to grieve”) or discouraged (“we hope time will take your grief away”). Therefore, in addition to the LIWC analyses, two trained research assistants (different individuals than those who did the translations-back translations described above), one European American female (who was fluent in English and German and who had lived in the US most of her life) and one German female (who was fluent in German and English and who had lived in Germany most of her life) coded the sympathy cards in their original languages to determine whether the text on the card encouraged/acknowledged grief (0 = no, 1 = yes) and whether the text on the card encouraged/wished something positive (0 = no, 1 = yes). For instance, the phrases “In deep sadness,” “Sharing your sorrow,” “We mourn the loss” were coded as encouraging/acknowledging grief, and “May you find comfort,” “The memories are yours,” and “Hold on to hope” were coded as encouraging/wishing something positive. Both coders were blind to study hypotheses. To assess reliability, coders overlapped on 18% of the cards (79 sympathy cards). Inter-rater reliability was high according to Landis and Koch (1977) (Cohen's kappa for encouraged/acknowledged grief = .73, Cohen's kappa for encouraged/wished something positive = .74).

Card Images

The coders (who were the same as the ones who coded the context of word use) also coded the images on the sympathy cards in terms of whether they were living or dying. If there was only one image on the card, that image was coded. If there was more than one image on the card, the coders coded the most prominent image. Images on the cards were coded as: (1) dying (e.g., shriveled leaves), (2) living (e.g., living flower), or (3) neutral (e.g., rocks, water). Again, inter-rater reliability was high according to Landis and Koch (1977) (Cohen's kappa = .69).

Data Analyses and Results

We first examined whether the sympathy cards differed in number of words they contained. We conducted a univariate analysis of variance (ANOVA) by Culture (US, Germany) on word count, which we obtained from LIWC. The American sympathy cards contained significantly more words than the German sympathy cards (American mean = 39.10, SE = 1.39; German mean = 10.32, SE = 1.25; F[1, 442] = 237.38, p < .001, partial eta squared = .35). For analyses of frequencies of negative and positive words, we did not control for number of words since these measures were percentage scores, and therefore already took number of words into account. However, for the analyses of context codes (which are based on occurrences), we controlled for the number of words in the cards (our findings did not differ when we did not control for number of words in the cards).

With one exception (for German cards, the correlation between context code of encouraged/acknowledged grief and LIWC coded sadness/death was .86), the correlations among the variables were generally weak to moderate across cultures, with a mean of .03 (SD = .17) for European Americans and with a mean of .14 (SD = .38) for Germans, suggesting that the codes were non-overlapping. Thus, we concluded that most of the codes referred to different aspects of the cards and therefore should be analyzed separately.

Do American Sympathy Cards Contain Less Negative and More Positive Content Than German Sympathy Cards?

Frequency of Negative and Positive Words (LIWC) (Percentage of total words that were positive or negative)

To examine whether the American and German cards differ in the frequency of negative and positive words (based on LIWC) described above, we conducted a 2 (Culture: American, German) × 2 (Card Type: Sympathy, Baby) × 2 (Valence: negative LIWC category, positive LIWC category) repeated measures ANOVA with Culture and Card Type as between-group factors and Valence as within-group factor. We found a significant main effect for Valence (F[1, 710] = 11.14, p = .001, partial eta squared = .02), which was qualified by significant Valence by Culture (F[1, 710] = 10.87, p = .001, partial eta squared = .02), Valence by Card Type (F[1, 710] = 81.32, p < .001, partial eta squared = .10), and Valence by Culture by Card Type (F[1, 710] = 13.42, p < .001, partial eta squared = .02) interactions.

As predicted and depicted in Figure 1, simple effects analyses revealed that American sympathy cards contained fewer negative words (American mean = 2.90, SE = .79; German mean = 7.30, SE = .71), F(1, 710) = 17.35, p < .001, partial eta squared = .02) and more positive words (American mean = 3.50, SE = .33; German mean = 1.35, SE = .30), F(1, 710) = 23.51, p < .001, partial eta squared = .03, than did German sympathy cards. However, American and German baby cards did not differ in the frequency of negative (American mean = .04, SE = .94; German mean = .08, SE = .97; F[1, 710] = .001, p = .98, partial eta squared < .001) or positive (American mean = 5.68, SE = .39; German mean = 6.07, SE = .41; F[1, 710] = .47, p = .50, partial eta squared = .001) words.⁵

Frequency of negative words and positive words (percentage of total words that are negative or positive) in American and German sympathy and baby cards in Study 1.

Context of Word Use

Based on the manual coding, we examined whether there were cultural differences in (1) encouraged/acknowledged grief, and (2) encouraged/wished something positive in the sympathy cards. We conducted two logistic regression analyses in which these two variables were regressed onto culture and number of words. As predicted, the American sympathy cards were less likely to encourage/acknowledge grief, B = −.76, SE = .38, p = .046, Nagelkerke R square = .06, and were more likely to encourage/wish something positive, B = 2.56, SE = .30, p < .001, Nagelkerke R square = .57, compared to the German cards.

Sympathy Card Images

A chi-square test of independence of the relationship between culture and images revealed that American and German sympathy cards differed in the type of images on the cards, χ² [2, 444] = 50.44, p < .001, Cramer's V = .34. Even though a large percentage of the American (31.31%) and German sympathy cards (48.78%) were decorated with images other than dying or living ones, as predicted, significantly fewer American than German cards contained dying images (2.53% of American cards vs. 16.26% of German cards). In contrast, more American cards (66.16%) than German cards (34.96%) contained living images, χ² [1, 262] = 36.23, p < .001, Cramer's V = .37.

Study 1 Summary and Discussion

As predicted, American sympathy cards contained less negative and more positive content than German sympathy cards. These cultural differences emerged when we examined the text (both the frequency of negative and positive word use and context in which negative and positive words were used) as well as the images on the cards. Cultural differences in the frequency of positive and negative words (based on LIWC) did not generalize to baby cards, suggesting that they may be specific to primarily negative events, such as the death of a loved one.⁶ While the frequency of positive and negative emotion words in the sympathy cards may appear low, they are well within the range of frequencies reported in the 43 studies listed in the LIWC manual (Pennebaker et al., 2001).

How can these differences in sympathy card content be explained? We hypothesized that these differences are due at least in part to cultural differences in the desire to avoid negative affect. Therefore, in Study 2, we examined whether Americans and Germans indeed differ in the degree to which they want to avoid negative emotion. However, before examining cultural differences in avoided negative affect, we assessed whether avoided negative affect is a separate construct from actual negative affect and ideal negative affect (i.e., we wanted to make sure that avoided negative affect is not just the opposite of ideal negative affect), and whether people want to avoid negative states more than they ideally want to or actually feel them, as predicted by Affect Valuation Theory (Tsai, 2007). Because most of our previous work has focused on global actual and ideal affect (i.e., affective states experienced or desired on average or over the course of a typical week), we focused on global avoided affect in this study. We also sampled negative states that varied in terms of high and low arousal, in order to assess whether our findings differed as a function of the type of negative affect.

Study 2: Cultural Differences in Avoided Negative Affect Hypotheses

We predicted that: (1) avoided negative affect, ideal negative affect, and actual negative affect are distinct factors, and that this distinction holds in American and German contexts, (2) across groups, people want to avoid negative states more than they ideally want to or actually feel them, and (3) European Americans want to avoid feeling negative more than Germans do, controlling for how much they actually experience negative states.⁷

Method

Participants

One hundred nineteen European American (73.95% female) students and 104 German students (81.73% female) participated in this study. There were no group differences in gender (χ² [1, 223] = 1.93, p = .17, Cramer's V = .09). However, Germans were significantly older (mean age = 23.71, SD = 4.61) than European Americans (mean age = 20.22, SD = 4.47), F (1, 221) = 32.89, p < .001, partial eta squared = .13. Therefore, we included age in the analyses described below when it emerged as significant covariate. American participants received a $10 Amazon gift certificate, and German participants received a Euro 7 Amazon gift certificate as compensation for their participation.

Instruments

Assessment of Actual, Ideal, and Avoided Affect

To assess actual and ideal affect, participants completed the Affect Valuation Index (AVI; Tsai & Knutson, 2006). Participants used a five-point scale ranging from 1 (“never”) to 5 (“all the time”) to rate how often they actually felt and how often they ideally wanted to feel 37 different affective states that sampled each octant of the affective circumplex (high-arousal positive [HAP; elated, enthusiastic, euphoric], positive [P; content, happy, satisfied], low-arousal positive [LAP; peaceful, calm, relaxed], high-arousal negative [HAN; fearful, hostile, nervous], negative [N; sad, unhappy, lonely], low-arousal negative [LAN; dull, sleepy, sluggish], high-arousal [HA; astonished, surprised], and low-arousal [LA; idle, passive, inactive]) (Russell, 1980; Russell et al., 1989) over the course of a typical week. The other affective states sampled were excited, strong, fatigued, angry, activated, rested, quiet, still, contemptuous, guilty, stressed, disgusted, ashamed, and serene⁸. To assess avoided affect, participants used the same rating scale to indicate how often they wanted to avoid feeling the same states over the course of a typical week. Reliabilities are reported below.

Demographic Questionnaire

Participants completed a demographics questionnaire, which assessed participants’ gender, age, place of birth, and cultural upbringing.

Procedure

Participants were recruited via email announcements and flyers distributed at large top-tier universities in the US and Germany. European American participants completed all measures in English, and German participants completed all measures in German. Standard translation-back-translation procedures were used. A research assistant fluent in both German and English translated all scales into German. B. K.-H. (also fluent in German and English) then back-translated these German translations into English. The few discrepancies that emerged were discussed and modifications were made to the original translations to arrive at the best possible translation. All participants completed the measures online. The order of the ideal, actual, and avoided versions of the AVI was counterbalanced. Because the order of these measures did not have an effect on the results, we will not discuss it further. Other measures were included in the study as fillers.

Data Analyses and Results

Given group differences in age, we included age as a covariate. For parsimony, we retained age in the model when it was a significant covariate and removed it when it was not.

Hypothesis 1: Is Avoided Negative Affect Different From Actual and Ideal Negative Affect?⁹

We were first interested in whether among the negative states, avoided affect was distinct from actual and ideal affect, and if so, whether we would see this differentiation among high arousal negative (HAN), negative (N), and low arousal negative (LAN) states (see above). To test Hypothesis 1, we conducted confirmatory factor analyses using AMOS 21. We included three items from each of the negative octants (HAN: fearful, hostile, nervous; N: sad, unhappy, lonely; and LAN: dull, sleepy, sluggish) for avoided, ideal, and actual affect, and treated each as indicators for the three latent variables of ideal, actual, and avoided negative affect.

We hypothesized a model (see Figure 2) that included nine separate latent variables (avoided HAN, N, and LAN; ideal HAN, N, and LAN; and actual HAN, N, and LAN) and three second-order factors (the first combining avoided HAN, N, and LAN into an overall avoided negative factor, the second combining ideal HAN, N, and LAN into an overall ideal negative factor, and the third combining actual HAN, N, and LAN into an overall actual negative factor). In the model, we allowed these three second-order factors to covary with each other. Thus, our model included 3 covariance links. Across both groups, the confirmatory factor analysis testing this model revealed that the data fit the model well, χ²(312, N = 223) = 573.34, p < .001; the root mean square of error approximation (RMSEA) = .06, with confidence interval .05 to .07; and the Akaike Information Criterion (AIC) = 759.34. The covariance links between the second-order avoided negative and ideal negative factors (−.03, p = .049), between the second order avoided negative and actual negative factors (−.02, p = .44), and between the second order ideal negative and actual negative factors (.04, p = .001) were small to moderate.

Factor structure of ideal, actual, and avoided negative affect in Study 2. Latent variables are in ovals; their indicators are in rectangles; covariances are along the curved arrows; and error terms are e1 through g9.

To test whether the constructs of ideal and avoided negative affect overlap, we constrained the covariance link between ideal and avoided negative affect to be −1. Across both groups, this constrained model had a significantly worse fit than the unconstrained model, χ²(313, N = 223) = 775.98, p < .001; the RMSEA = .08, with confidence interval .07 to .09; and the AIC = 959.98, Δχ² = 202.64, Δdf = 1, p < .001. We also tested whether the constructs of actual and avoided negative affect overlap. To do this, we constrained the covariance link between actual and avoided negative affect to be −1. Across both groups, this constrained model had a significantly worse fit than the unconstrained model, χ²(313, N = 223) = 725.48, p < .001; the RMSEA = .08, with confidence interval .07 to .08; and the AIC = 909.48, Δχ² = 152.14, Δdf = 1, p < .001.

In sum, these findings suggest that consistent with Hypothesis 1, avoided negative affect is weakly if at all correlated with actual negative affect and ideal negative affect, and all three types of affect are distinct constructs at both the second-order level (i.e., across all negative states) and at the first-order level (i.e., for each type of negative affect).

Does This Structure Hold Across Groups?

Before we can compare psychological constructs across different groups, we have to demonstrate measurement equivalence (Little, 1997). In order to examine whether the model described above held for European Americans and Germans (i.e., all measurement weights are the same across both groups), we performed a multiple-group confirmatory factor analysis using AMOS 21 (e.g., Cheung & Rensvold, 1999). All path coefficients were free to vary across the two groups. This unconstrained model had a good fit, χ²(624, N = 223) = 1020.08, p < .001; the RMSEA = .05, with confidence interval .05 to .06; and the AIC = 1392.08. Then all measurement weights were fixed to be the same across both groups. This fully constrained model provided a significantly worse fit than the unconstrained model, χ²(642, N = 223) = 1056.19, p < .001; the RMSEA = .05, with confidence interval .05 to .06; and the AIC = 1392.19, Δχ² = 36.11, Δdf = 18, p = .01. Further analysis revealed that the path between the indicator variable ideal hostile and the latent variable ideal HAN varied across cultural groups. When we unconstrained this weight, and fixed all the other measurement weights to be the same across both groups, we found that this constrained model did not provide a significantly worse fit than the unconstrained model, χ²(641, N = 223) = 1046.43, p < .001; the RMSEA = .05, with confidence interval .05 to .06; and the AIC = 1384.43, Δχ² = 26.35, Δdf = 17, p = .07. This suggests that consistent with Hypothesis 1, our proposed model holds for European Americans and Germans, except for the link between ideal hostile and ideal HAN affect. Ideal hostile was related to ideal HAN and overall ideal negative affect in the European American (estimate = 1.92, p = .003) but not in the German sample (estimate = .33, p = .06). However, the two constructs that we primarily focused on in the present studies, avoided negative and actual negative affect, had the same structure for both European Americans and Germans. As we did not find measurement equivalence for the ideal HAN composite, in the subsequent analyses, we only use this composite when we examine the cultural groups separately. Together, these findings suggest that the structure of avoided and actual negative affect is empirically valid and equivalent across American and German contexts.

Mean Aggregates: Correlations and Internal Consistency Estimates

Mean aggregate scores were computed for items that sampled each negative octant of the affective circumplex. For European Americans, Cronbach's alpha's ranged from .64 to .80 for actual negative affect, from .60 to .66 for ideal negative affect, and from .76 to .92 for avoided negative affect. For Germans, it ranged from .69 to .84 for actual negative affect, from .43 to .69 for ideal negative affect, and from .76 to .88 for avoided negative affect.

Calculating mean deviated scores

European Americans had a lower overall mean response to all actual affect items (American mean = 2.47, SE = .02; German mean = 2.81, SE = .03; F(1, 221) = 94.70, p < .001, partial eta squared = .30) and to all ideal affect items (European American mean = 2.44, SE = .02; German mean = 2.62, SE = .02; F(1, 220) = 42.56, p < .001, partial eta squared = .16) than did Germans. However, European Americans had a higher overall mean response to all avoided affect items (European American mean = 2.91, SE = .05; German mean = 2.54, SE = .06; F(1, 219) = 22.53, p < .001, partial eta squared = .09) than did Germans. This is in line with previous observations of American-German differences in response-style (e.g., Koopmann-Holm & Matsumoto, 2011).

Because of these cultural differences in response style and previous findings showing that people differ in how much they want to avoid feeling emotions in general (e.g., Maio & Esses, 2001), we mean deviated participants’ responses by subtracting each individual's overall mean response to all the 37 avoided affect items from the raw score for each avoided affect item (e.g., sad), and then calculated the mean aggregate for each type of avoided negative affect (HAN, N, and LAN) (e.g., for avoided HAN, we calculated the mean of mean deviated avoided fearful, avoided hostile, avoided nervous). We followed the same procedure for actual and ideal negative affect. We used these scores for avoided, actual, and ideal negative affect in all subsequent analyses.¹⁰ Correlations among actual, ideal, and avoided negative affect ranged from −.35 (p < .001) to .64 (p < .001) for European Americans and from −.32 (p = .001) to .42 (p < .001) for Germans.¹¹

Hypothesis 2: Do People Want To Avoid Negative States More Than They Ideally Want to and Actually Feel Them?

To test Hypothesis 2 that people want to avoid negative states more than they ideally want to or actually feel them, we conducted pairwise t-tests on avoided HAN, N, and LAN with ideal HAN, N, and LAN as well as with actual HAN, N, and LAN affect for Americans and Germans separately. We also compared the means of ideal HAN, N, and LAN with actual HAN, N, and LAN. All means statistically differed from each other, p < .001. As Figure 3 shows, European Americans (top graph) and Germans (bottom graph) wanted to avoid HAN, N, and LAN states more than they ideally wanted to feel or actually felt them. These findings support the above findings that actual negative affect, ideal negative affect, and avoided negative affect are distinct constructs. For the rest of the paper, we focus on avoided and actual negative affect; however, we ran all of the analyses controlling for ideal negative affect to ensure that the results were not due to ideal negative affect. The pattern of our findings did not change, and therefore, we do not discuss ideal negative affect further.

Mean-deviated ratings of actual, ideal, and avoided affect in Study 2 for (a) European Americans and (b) Germans. HAN = high-arousal negative; N = negative; LAN = low-arousal negative.

Hypothesis 3: Do Americans Want to Avoid Feeling Negative More Than Do Germans?

We initially ran our analyses with Gender as a between subjects factor; however, because there were no significant main effects or interactions involving Gender, we dropped Gender from our final analyses. We controlled for actual affect when examining group differences in avoided affect, and we controlled for avoided affect when examining group differences in actual affect to determine the independent effects of group on each type of affect. The findings did not change when we excluded these covariates.

To test Hypothesis 3 that European Americans want to avoid feeling negative states more than Germans do, we conducted a 2 × 2 × 3 (Group: European Americans, Germans × Affect Type: Avoided Affect, Actual Affect × Octant: HAN, N, LAN) repeated measures analysis of covariance (ANCOVA) with Group as between-subjects factor and Affect Type and Octant as within-subjects factors, controlling for age. The Group by Affect Type interaction was significant, F(1, 218) = 14.54, p < .001, partial eta squared = .06 as well as the Group by Affect Type by Octant interaction, F(1.82, 397.65) = 5.55, p = .01, partial eta squared = .03. Consequently, for avoided and actual affect, we conducted separate ANCOVAs for avoided and actual negative affect for each of the three different octants. More specifically, we conducted three separate ANCOVAs on avoided HAN, N, and LAN, with Group (European Americans, Germans) as the independent variable, controlling for actual HAN, N, and LAN, respectively, and age when it emerged as a significant covariate. As predicted and depicted in Figure 4 (left), European Americans wanted to avoid HAN, N, and LAN states significantly more than did Germans (HAN: European American mean = 1.31, SE = .06; German mean = .97, SE = .07, F(1, 217) = 13.62, p < .001, partial eta squared = .06; N: European American mean = 1.27, SE = .07, German mean = .91, SE = .07, F(1, 218) = 12.10, p = .001, partial eta squared = .05; LAN: European American mean = 1.15, SE = .05, German mean = .95, SE = .06, F(1, 219) = 5.68, p = .02, partial eta squared = .03).

Group differences in avoided and actual negative affect in Study 2 (HAN = high-arousal negative; N = negative; LAN = low-arousal negative), controlling for actual and avoided negative affect, respectively. * p < .05; ** p < .01; *** p < .001.

We also examined group differences in actual negative affect. We conducted ANCOVAs on actual HAN, N, and LAN separately with group as the independent variable, controlling for avoided HAN, N, and LAN, respectively and age when it emerged as a significant covariate. As depicted in Figure 4 (right), European Americans and Germans did not differ in how much they actually felt HAN (European American mean = −.46, SE = .05; German mean = −.34, SE = .05; F[1, 217] = 2.45, p = .12, partial eta squared = .01) or LAN (European American mean = .12, SE = .06; German mean = −.01, SE = .06; F[1, 219] = 2.35, p = .13, partial eta squared = .01). Although European Americans (marginal mean = −.23, SE = .06) actually felt N states significantly less than did Germans (marginal mean = −.05, SE = .06), F (1, 219) = 4.07, p = .045, partial eta squared = .02, the effect size of the cultural difference in actual N was smaller than that of the cultural differences in avoided N, HAN, and LAN.

Study 2 Summary

As predicted, our results suggest that (1) avoided negative affect, ideal negative affect, and actual negative affect are distinct constructs, and that this holds for European Americans and Germans, (2) across groups, people want to avoid negative states more than they ideally want to or actually feel them, and (3) European Americans want to avoid feeling negative more than do Germans, even after controlling for actual negative affect. Furthermore, while we found cultural differences for all three negative octants of avoided affect, we only found cultural differences for one out of the three negative octants of actual affect. This suggests that consistent with AVT and our predictions, culture shapes avoided affect more than actual affect.

In Study 1, we established that there are cultural differences in expressions of sympathy, and in Study 2, we demonstrated that there are cultural differences in the desire to avoid negative affect. In Study 3, we examined whether there was a direct link between how much people want to avoid negative affect and the degree to which their expressions of sympathy focused on the negative (vs. positive). In addition, we also examined whether cultural differences in avoided negative affect were due to differences in the endorsement of specific frontier values.

Study 3: Do Cultural Differences in Avoided Negative Affect Mediate Cultural Differences in Expressions of Sympathy?

In Study 3, participants completed measures of avoided affect, and then were asked to rate and choose between sympathy cards that either focused more on the negative or more on the positive. Because Study 2 revealed that European Americans wanted to avoid HAN, LAN, and N states more than did Germans, for parsimony, in Study 3, we collapsed across the three types of negative affect and created a general avoided negative affect score.¹²