Social foundations of emotions in family consumption decision making

Abstract

Although emotions are believed to be socially constructed and important features of family life they are little understood in the context of families making consumption decisions. Our research focuses on understanding individual and social aspects of emotions, including whether parents mirror each other's emotions during family consumption decisions. Our Social Relations Model analyses provide evidence consistent with hypotheses that anger and frustration in family consumption decision making are interdependent, and incorporate individual and relationship levels, bidirectionality, and reciprocity. Additionally we find that the mirroring of these emotions between the parents is pervasive, occurring across spending, model/make, and final decisions involved in buying a new car. Our theory and findings provide insights into how families share emotions during consumption decision making and provide a foundation for future investigations on discrete emotions, emotion mirroring and contagion, and family decision making.

Keywords:

• Emotion
• Interdependence
• Mirroring
• Family decision making
• Social relations model

Emotions are believed to be “among the most prominent features of our ongoing relationships with family members” (Lazarus, 2006, p. 15). While emotions are experienced as individual psychological phenomena (Bagozzi, Gopinath, & Nyer, 1999; Oatley, 1992), they are also believed to have social origins because they are socially constructed, serve social functions, and usually involve two people in relationships of significance like family members (Lazarus, 1991; Mikulincer & Shaver, 2005). Family consumer decision making is likely to be emotion-laden, as family members together recognize needs and wants, gather information, evaluate alternatives, and ultimately make decisions on purchases for the family (Commuri & Gentry, 2000, 2005; Cotte & Wood, 2004; Palan & Wilkes, 1997; Su, Zhou, Zhou, & Li, 2008).

However, little is known about how emotions are experienced and expressed within families during consumption decision making. Investigating this topic requires seeing family members as operating together rather than simply as a collection of individual emoters. Our contribution lies in bringing a social perspective to theory building, coupled with appropriate methods and measures to reflect family members’ relationships, and to testing long-standing theoretical beliefs about emotions as social phenomena. We thereby contribute to the emotion and family decision-making literatures by addressing two research questions:

	RQ#1: What is the pattern of individual and social aspects of emotions experienced within families during consumption decision making?
	RQ#2: Do parents mirror each other's emotions during consumption decision making?

Past research has examined social aspects of affectivity, a broadly construed positive or negative experience (Rasbash, Jenkins, O’Connor, Tackett, & Reiss, 2011). We focus on discrete emotions because they are readily recognized and coped with (Ekman et al., 1987; Lazarus 1991), and they influence judgments, coping responses, and behavioral intentions and actions (Frijda, Kuipers, & ter Schure, 1989; Roseman, Wiest, & Swartz, 1994). Anger and frustration are emotions that individuals sometimes experience during decision making and consumption (Bougie, Pieters, & Zeelenberg, 2003; Richins, 1997), particularly when decision making is blocked. These emotions may be important as families encounter obstacles when jointly making important and risky decisions such as purchasing a car. By comparison other emotions such as delight, disappointment, or satisfaction may be more associated with post-decision assessments. Accordingly, we examine two specific discrete emotions—anger and frustration—that may be experienced by family members in the process of buying a car. By addressing these two research questions we add to knowledge of social factors affecting what consumers feel within an important social unit—the family—during consumer decision making.

Literature Review

Psychological perspectives describe emotions as mental states of readiness arising from appraisals of situations and involving a felt component, physiological processes such as heightened arousal, and verbal and non-verbal communication such as facial expressions that allow emotions to spread to others (Bagozzi et al., 1999; Hennig-Thurau, Groth, Paul, & Gremler, 2006; Howard & Gengler, 2001; Oatley, 1992). Emotions may result in actions to affirm or cope with the emotion and its meaning (Lazarus, 1991).

While emotions can be individually experienced psychological phenomena, “many emotions are social events” because they are recognized by, shared with, and affect other people (Frijda & Mesquita, 1994, p. 73). Some social aspects of emotions have been studied in consumer decision-making contexts (Weber & Johnson, 2009). For example, emotion contagion research shows that a “sender” consumer's emotion can transfer to a “receiver” consumer and affect the receiver's product evaluations (Howard & Gengler, 2001). Brief interactions with unacquainted others can elicit emotions during decision making, such as consumers who experience guilt when asked to buy a product from an unacquainted salesperson (Dahl, Honea, & Manchanda, 2005). Even thinking about social factors, such as social consequences of product use, can affect consumers’ emotions during decision making (Lau-Gesk & Drolet, 2008). While showing effects of some social factors, these studies likely do not map fully to family life, where intimates rather than strangers have extensive interactions with one another and members play particular roles within the family (Cox & Paley, 1997; Milardo & Duck, 2000).

We build on these studies to investigate social aspects of emotions arising in family decision making. The foundation of social constructs is that they are inherently interdependent (Ickes & Gonzalez, 1996). Social constructs neither arise solely within, nor are they completely attributable to, an individual but are instead dependent on the person with whom one is interacting. Interdependent constructs incorporate both the individual level of analysis, representing each person's individual dispositions in their behavior towards all others, and the relationship level of analysis, representing the jointly constructed portion of behavior they create together over and above their separate dispositions. Interdependence also can manifest as bidirectionality, where each person is both a sender and receiver who affects, and is affected by, the other. Additionally interdependence can manifest as reciprocity where what a person sends out is reflected back by the partner in either similar or different ways. Studying interdependent social constructs requires including the perspectives of all participants and using an analysis method appropriate for interdependent data (Ickes & Gonzalez, 1996).

If emotions are social they should be interdependent and reflect these foundational qualities. Interdependence underlies the notion that emotions “are not just [the] feelings of isolated individuals, but part of the substance of social interactions” (Johnson-Laird & Oatley, 2000, p. 472). Accordingly the emotions of one person should affect the emotions of another, and vice versa. Emotions should incorporate both the individual and relationship levels because they arise within an individual and yet are also attributable to the other. Emotions should be bidirectional because each partner sends and receives emotion; that is, each partner both expresses and evokes emotion (Keltner & Kring, 1998). Emotions should be reciprocal, where reciprocity may help regulate and coordinate behaviors of partners in marketing contexts (Bagozzi, 1995).

We expect these social qualities of emotions to be present within family consumption decision making. Emotions are evoked at significant junctures of a plan (Oatley, 1992), and family decision making often involves recognizing a problem and enacting a plan to address wants and needs in the family (Commuri & Gentry, 2000). Family car buying is a joint and highly interactive process involving well-acquainted individuals in ongoing relationships (Burns & Granbois, 1980). Because families are social and the decision-making context is expected to be emotion-laden (Richins, 1997), we investigate the social foundations of two emotions—anger and frustration—that mothers, fathers, and children might experience and express during consumption decision making. Anger and frustration are similar emotions associated with unpleasant situations, perceived goal-obstacles, and moderate attention and certainty. Yet the two emotions differ in attributions of responsibility and control for the situation. Anger involves attributions that another person or entity is responsible for the negative situation, whereas frustration is associated with situations that are beyond anyone's control (Roseman et al., 1994; Smith & Ellsworth, 1985). How individuals cope varies somewhat by emotion. For example, in service failure settings angry customers exhibit more confrontative coping compared to frustrated customers who exhibit more support-seeking coping (Gelbrich, 2010; Yi & Baumgartner, 2004).

Therefore while there may be some differences in overall patterns of effects, we expect that family members’ anger and frustration during consumption decision making will exhibit the social foundation of interdependence, manifesting in at least one of the social aspects described herein. Thus we propose family members’ anger will:

-	H1a: be interdependent rather than solely independent;
-	H1b: incorporate the individual and relationship levels of analysis;
-	H1c: be bidirectional;
-	H1d: be reciprocal.

Similarly, we propose that family members’ frustration will:

-	H2a: be interdependent rather than solely independent;
-	H2b: incorporate the individual and relationship levels of analysis;
-	H2c: be bidirectional;
-	H2d: be reciprocal.

Additionally, because research suggests that mirroring is another form of interdependence we expect to observe mirroring during family decision making. Social functioning requires not only being able to “read” emotional states of others but also to feel what they feel. Mirroring is accomplished in part by mirror neurons that fire during the execution and observation of a specific action by others (Iacoboni, 2009; Keysers & Gazzola, 2010; Mukamel, Ekstrom, Kaplan, Iacoboni, & Fried, 2010). Mirroring occurs as a function of the strength of mirror neuron activation, although here we only scrutinize the outward manifestations of such internal, neural excitation.

Emotions involve changes in the brain, the body, and the environment, and motivate behaviors to adapt to those changes (Adolphs, 2010). While mirroring is only one part of how emotions are perceived and processed, it fosters an understanding of what social adaptation is needed (Adolphs, 2009). Appraising another person's emotions and feeling the same emotion as another helps predict that person's intentions (Anderson, Keltner, & John, 2003; Parkinson & Simons, 2009), increases liking and evaluations (Guéguen, 2009), and enhances helping behaviors (Stel, van Baaren, & Vonk, 2008).

Emotion mirroring is likely in pairings with high emotional engagement and mutual interest (Grammer, Kruck, & Magnussen, 1998; Iacoboni, 2009). Accordingly we expect mirroring of anger and frustration to occur between parents because of the special generational role among equal partners who are pursuing common goals for the family and not across generations as with a child and a parent (Cox & Paley, 1997; Parkinson & Simons, 2009). We do not expect mirroring to occur in dyads involving children because the ability to mirror emotions develops with age (Darling & Clarke, 2009). Children learn to understand and deal with emotions and emotional contexts through parental coaching (Gottman, Katz & Hooven, 1997). Thus we propose:

-	H1e: Parents’ anger will be mirrored.
-	H2e: Parents’ frustration will be mirrored.

Reinforcing these hypothesized effects are the social communicative and facilitative effects of anger and frustration. When people in relationships, especially close ones, express anger or frustration, they are signaling displeasure and might be aiming to resolve conflict, repair hard feelings, and sustain their relationships (Fischer & Roseman, 2007; Hutcherson & Gross, 2011). Thus anger and frustration can strengthen relationships in the long-run and be functional when family members feel they can openly express these emotions.

Method

We adopt a theoretical and statistical model of interpersonal relationships that appropriately assesses interdependent dyadic social data: the Social Relations Model (SRM; Kenny, 1994; Kenny, Kashy, & Cook, 2006). The SRM decomposes variance in interdependent dyadic responses into: variance attributable to each individual's experience that is independent of others and interdependent variance emerging as partners interact. SRM studies typically employ round-robin designs where each person rates all dyadic relationships in the group, producing reciprocal ratings where each person is rated as sender and receiver. Variations in perceptions are preserved because they are theoretically and empirically meaningful (e.g., fathers, mothers, and children may all rate the child's frustration with the mother at different levels). The SRM has been applied in studies of forgiveness in families (Hoyt, Fincham, McCullough, Maio, & Davila, 2005), parenting style (Cook, Kenny, & Goldstein, 1991), and relationship assessments (Cook & Kenny, 2004). To the best of our knowledge the SRM has not been applied to social qualities of discrete emotions. We adopt the family SRM, rather than the basic SRM, because family members are distinguishable due to their unique roles in the family (Kenny et al., 2006).

The study

We studied family car purchases within 336 three-person families from youth groups in a small eastern city. Eligible families had an adolescent in the ninth through eleventh grades, and both a mother and a father at home (including step-parents). Families received a cover letter, three identical questionnaires, and three pre-paid envelopes for the mother, father, and child. Family members completed questionnaires individually. Questionnaires contained three sections corresponding to three decisions in car choice: spending, make/model, and final decisions (Hopper, Burns, & Sherrell, 1989). Each decision section asked respondents to rate specific emotions for each of the six dyadic relationships existing among the three family members. A total of 110 families provided complete data, meeting the rule of thumb of 100 families thought satisfactory for model estimation (Bagozzi & Hsiung, 1994).

Measures

As part of a larger study on family consumption, anger and frustration were assessed for all three decisions, yielding a total of six emotion conditions. Recognizing that emotions are experienced and expressed within relationships and toward a target person (e.g., mother expresses frustration to the father; Gottman, 1993) means that socially shared emotions should not be assessed with measures of an individual's emotions. Therefore we developed measures of emotions experienced by an actor and expressed toward a specific partner so that each measure is indexed to each of six possible actor–partner dyads in the family. Emotion ratings were prefaced with the question “To what extent did each person express each of the following feelings towards the other?” A label appeared above each emotion scale identifying the relationship and direction of the rating (e.g., “mother's frustration with father”, “father's frustration with mother”, etc.). A 6-point scale appeared beneath the label ranging from 0 (“no frustration”) to 5 (“much frustration”). Mothers, fathers, and children rated the emotions expressed in all six relationships. Following Cook et al. (1991) these were averaged to produce the six manifest variables (x₁–x₆) shown in Figure 1.

Figure 1. Family social relations model of emotions with mirroring during family consumption decision making.Adapted from Hsiung and Bagozzi (2003). M or m = mother, F or f = father, C or c = child, A = actor, P = partner. For example with respect to the emotion frustration, mf x₁ refers to all three raters’ averaged rating of the mother's (actor) frustration with the father (partner). Factor loadings are set to 1.

Model components and estimation

Figure 1 depicts Hsiung and Bagozzi's (2003) family SRM that we adapted in two ways: through adding mirroring correlations between parents’ actor and partner effects, and by retaining only the mother-father dyadic reciprocity correlation. Table 1 describes the model components assessed by decomposing variance and examining correlations.

Table 1. Proposed social qualities of emotions in family decision making: Social Relations Model effects and correlations

	Model components	Description	Level of analysis	Social characteristic demonstrated for this component	Elements in Figure 1
Variance	Actor effect	A person consistently exhibits an emotion toward all partners	Individual	None: A given person's emotions are independent of others’ emotions	ξ 1–3
	Partner effect	A person consistently elicits an emotion from all partners	Individual	Interdependence	ξ 4–6
	Relationship effect	A person departs from a general tendency to emote consistently to/from all, and uniquely expresses the emotion to a particular relationship partner	Relationship	Interdependence	δ 1–6
Correlations	Individual reciprocity	How a person exhibits an emotion toward all partners is associated with how the person elicits that emotion from all partners	Individual	Interdependence	φ^irs between one person's actor and partner effects
	Dyadic reciprocity	How a person uniquely expresses emotions to a particular partner is associated with how the particular partner expresses the emotion to the person	Relationship	Interdependence	φ^drs between one dyad's two relationship effects (e.g., mother–father and father–mother)
	Mirroring (actor)	How a parent expresses an emotion to all partners is associated with how the other parent expresses the emotion to all partners	Individual	Interdependence	φ A^mir between the parents’ actor effects
	Mirroring (partner)	How a parent elicits an emotion from all partners is associated with how another parent elicits the emotion from all partners	Individual	Interdependence	φ P^mir between the parents’ partner effects

Variance is decomposed into actor, partner, and relationship effects. Significant actor effects indicate that an individual consistently exhibits that emotion toward all partners, an individual-level phenomenon because it is independent of any partner. Significant individual-level partner effects indicate that an individual consistently elicits an emotion inwardly from all partners. This is a form of interdependence since both partners contribute to the individual's emotion. Relationship effects are estimated while controlling for individual-level actor and partner effects and, if significant, indicate that an individual uniquely expresses his or her emotion toward a particular partner. Such effects are interdependent and relationship-specific.

Significant correlations between an individual's actor and partner effects indicate individual reciprocity; the emotion the individual consistently directs toward all family members is associated with the emotion consistently elicited from them. Dyadic reciprocity, a correlation between relationship effects within one dyad, indicates that the emotion one parent uniquely expresses toward the other parent is associated with how that particular partner uniquely expresses the emotion in return. There are two mirroring correlations. The actor mirroring correlation between parents’ actor effects indicates that mothers’ and fathers’ expressions of the emotion to family members parallel each other. A partner mirroring correlation between the parents’ partner effects indicates that their elicitations of the emotion from family members parallel each other.

Our first model adaptation involved adding mirroring correlations only for the parents. Mirroring correlations are related to similarity correlations (Ackerman, Kashy, Donnellan, & Conger, 2011; Cook, 2001). “Mirroring” is a more process-oriented term implicated in the decoding of others’ emotions (Pfeiffer, Iacoboni, Mazziotta, & Dapretto, 2007) and prosocial behavioral adaptations (Stel, Mastrop, & Strick, 2008). Social appraisals and behavioral adaptations are important in processing and regulating emotions in social contexts (Parkinson & Simons, 2009). Incorporating mirroring into the SRM emphasizes the role of outward manifestations of physiological processes. We add mirroring correlations to the model a priori and only anticipate positive parental mirroring correlations (although they may occur in other pairings for different constructs).

Our second model adaptation involves restricting dyadic reciprocity to just the mother–father dyad. The parental dyad is where affectivity is highly reciprocal (Eichelsheim, Dekovic, Buist, & Cook, 2009; Rasbash et al., 2011), and spouses may particularly reciprocate negative emotions (Gottman, 1998) such as anger and frustration.

To conclude, if only actor effects are found, the emotion is experienced solely as a private psychological phenomenon. If, in addition to actor effects, interdependent partner and/or relationship effects are found, the emotion has a social foundation and involves a combination of individual and shared mechanisms. If actor and/or partner effects are found alongside of relationship effects, the emotion incorporates both the individual and relationship levels. Bidirectionality is indicated if actor effects are found along with partner effects (outward and inward) or by multiple relationship effects (directed toward different family members). Reciprocity is indicated if reciprocity correlations are found (whether at the individual or dyadic level). Finally, mirroring is indicated if mirroring correlations are found (whether between actor or partner effects).

Analyses and Results

Parameter estimation and model fit assessment occurred via LISREL8.8 (Jöreskog & Sörbom, 2007). Following Bagozzi and Yi (2012), for a comprehensive assessment of model fit we used the χ ² test and four indexes that complement each other's weaknesses: RMSEA, CFI, NNFI, and SRMR. Because RMSEAs penalize model complexity, and the SRM is complex by design, the offsetting indexes may be emphasized.

Tables 2 and 3 present measures of model fit as well as estimates for statistically significant model components for anger and frustration, respectively. Five of six models fit the data well (the SRM for frustration in the final decision did not). Three of the models fit on all measures. The model for anger in the final decision was accepted because it fit according to the χ² test alone, and fit well according to the CFI and NNFI. Although the χ ² and p value for frustration in the model decision were not strictly within guidelines, it was accepted because three indexes indicated the model fit well.

Table 2. Estimates for anger during the spending, model, and final decisions*

Decisions	Spending	Model	Final**
Actor effects
Mothers	.28	.31	.22
Fathers	.20	.23	.16
Children	.06
Partner effects
Mothers	.23	.24	.18
Fathers	.24	.32	.23
Children
Relationship effects
Mother–father
Mother–child			.05
Father–mother
Father–child	.04	.03	.05
Child–mother	.06	.08	.05
Child–father
Individual reciprocity
Mothers		.02
Fathers
Children
Dyadic reciprocity
Mother–father
Mirroring (Mother–Father)
Actor	.24	.26	.19
Partner	.22	.28	.21
Fit Indices	χ ^2(3) = 1.47, p = .69 RMSEA = .00 CFI = 1.00 NNFI = 1.00 SRMR = .03	χ ^2(3) = 3.86, p = .28 RMSEA = .05 CFI = 1.00 NNFI = 1.00 SRMR = .05	χ ^2(4) = 9.22, p = .06 RMSEA = .11 CFI = .99 NNFI = .98 SRMR = .11
*Statistically significant effects are shown. **Improper solutions can occur in SRMs. Following Kenny, Kashy, & Cook (2006), the model for anger in the final decision was re-estimated with the child's small negative partner variance constrained to a small positive value (.001).

Table 3. Estimates for frustration during the spending and model decisions*

Decisions	Spending	Model
Actor effects
Mothers	.30	.33
Fathers	.27	.23
Children	.12	.12
Partner effects
Mothers	.27	.22
Fathers	.30	.31
Children
Relationship effects
Mother–father	.40
Mother–child
Father–mother	.45	.24
Father–child	.05	.06
Child–mother	.04	.06
Child–father
Individual reciprocity
Mothers	.03
Fathers	.05
Children
Dyadic reciprocity
Mother–father	.34
Mirroring (Mother–Father)
Actor	.26	.24
Partner	.23	.23
Fit Indices	χ ^2(3) = 2.95, p = .40 RMSEA = .00 CFI = 1.00 NNFI = 1.00 SRMR = .03	χ ^2(3) = 8.76, p = .03 RMSEA = .13 CFI = .99 NNFI = .96 SRMR = .07
*Statistically significant effects are shown.

Results for anger

For the spending decision three actor effects were found. That is, all family members consistently exhibited anger to all partners. The sizes of actor effects indicate greater variability among mothers (.28) in the anger displayed toward others, compared to the anger displayed to others by fathers (.20) or children (.06). Two partner effects were found, indicating that mothers (.23) and fathers (.24) consistently elicited anger from all family members, but children did not. The two significant relationship effects indicate that fathers adapted their expressions of anger to children (.04) in a unique manner not shown toward others, and children adapted their expressions of anger to mothers (.06) in a unique manner. Neither individual reciprocity nor dyadic reciprocity was observed.

Both forms of mirroring were evident in the spending decision. The actor form of mirroring (.24) indicates a linkage, or matching, between the mother expressing anger toward all others and the father expressing anger toward all others. Partner mirroring (.22) indicates a linkage between the mother eliciting anger from all others and the father eliciting anger from all others.

For the model decision, actor effects were found for the mother and father, who consistently exhibited their anger to all partners. Partner effects indicate that mothers and fathers consistently elicited anger from all family members. Relationship effects indicate that fathers adapted their expressions of anger to children in a unique manner not shown toward others, and children adapted their expressions of anger to mothers in a unique manner. Mothers exhibited individual reciprocity during this decision. No dyadic reciprocity was observed. Two forms of mirroring indicate that parents’ expressions of anger moved in concert with one another, as did their elicitations of anger.

For the final decision, mothers and fathers consistently exhibited their anger to all partners (actor effects), and mothers and fathers consistently elicited anger from all others (partner effects). Mothers adapted expressions of anger to children in a unique manner, as did fathers to children and children to mothers (relationship effects). Neither individual reciprocity nor dyadic reciprocity was observed. The two forms of mirroring in the final decision indicate that parents’ expressions of anger moved in concert with one another, as did their elicitations of anger.

In sum, across the three decisions, anger was characterized by both independent (actor effects) and interdependent effects (partner and relationship effects), demonstrating the emotion is indeed social. Family members’ anger incorporates the individual and relationship levels (actor and partner effects at the individual level alongside of relationship effects), is bidirectional (co-occurring actor and partner effects, and multiple directional relationship effects), and is reciprocated (individual reciprocity), consistent with H1a–d, respectively. Parents also exhibited mirroring of anger during these decisions, consistent with H1e.

Results for frustration

For the spending decision, all family members consistently exhibited their frustration to all others (actor effects). Mothers and fathers consistently elicited expressions of frustration from all family members but children did not (partner effects). Mothers adapted expressions of frustration to fathers in a unique manner not shown toward others, as did fathers to mothers, fathers to children, and children to mothers (relationship effects). Also the more mothers expressed frustration to others, the more mothers elicited frustration from others; the same held for fathers’ expressions and elicitation of frustration to and from others (individual reciprocity). Further, parents adapted frustration to each other in the same manner (dyadic reciprocity). Finally, parents’ expressions of frustration were linked (actor mirroring), as were parents’ elicitations of frustration (partner mirroring).

For the model decision, all family members consistently exhibited their frustration to all others (actor effects). Mothers and fathers consistently elicited expressions of frustration from all family members but children did not (partner effects). Fathers adapted their expressions of frustration to mothers in a unique manner not shown toward others, as did fathers to children and children to mothers (relationship effects). Unlike the spending decision, no individual reciprocity or dyadic reciprocity was observed. However, both forms of mirroring were observed, indicating that parents’ expressions of frustration were linked, as were parents’ elicitations of frustration.

In sum, across these two decisions, frustration was characterized by both independent (actor effects) and interdependent effects (partner and relationship effects), demonstrating the emotion is social. Family members’ frustration incorporates the individual and relationship levels (actor and partner effects at the individual level alongside of relationship effects), is bidirectional (co-occurring actor and partner effects, and multiple directional relationship effects), and is reciprocated (individual and dyadic reciprocity), consistent with H2a–d, respectively. Parents also exhibited mirroring of frustration during these decisions, consistent with H2e.

Discussion

Because much of consumption life is social, and emotions are an important part of consumption decision making, we highlighted the need to shift from a primarily individualistic psychological perspective to a more social view of emotions. Accordingly we adopted a theoretical, methodological, and analytical approach to investigate both interdependent and independent aspects of family members’ anger and frustration during consumption decision making.

The pervasiveness of the interdependent components in the families’ experience of anger and frustration in consumption decisions demonstrates that these emotions are social. Our results also show what form that interdependence took, incorporating the individual and relationship levels, bidirectionality, and reciprocity. We further expanded the theoretical discussion of emotions by theorizing how the innately social process of mirroring could be manifested and within which family relationships. Our adaptation of the SRM shows that both forms of mirroring are remarkably stable, evident in both anger and frustration in all decisions. Our findings have implications for theory and future research regarding discrete emotions, mirroring and emotion contagion, and family decision making.

Discrete emotions

Tracing the pattern of SRM results for each emotion across decisions illustrates the explanatory power gained by examining related but distinct discrete emotions. Whereas anger shows consistency across decisions, frustration's patterns change more noticeably. Particularly for the spending decision, frustration appears to be densely relational, with more relationship effects than observed for anger and also more than observed during the model decision.

Frustration was even more reciprocal than anger in the spending decision. Positive individual reciprocity reflects amplification and positive dyadic reciprocity represents synchronization (Hsiung & Bagozzi, 2003). The individual reciprocity correlations for mothers and fathers show that they amplify frustration; that is, the more that each expresses frustration to others the more each elicits frustration from others. The dyadic reciprocity demonstrates synchronization where the parents adapt their frustration to each other in the same way. Consistent with reciprocity of affectivity (Ackerman et al., 2011; Rasbash et al., 2011), frustration's dyadic reciprocity occurred within the parents’ equal power relationship. However, dyadic reciprocity was neither found for frustration in the model decision, nor for anger in any decision. Thus dyadic reciprocity is sensitive to discrete emotions and changes in the decision context.

Within families, parents may feel comfortable in both expressing and reciprocating frustration to one another because this emotion reflects a situational blockage to the goal, such as realizing the family budget is insufficient for the preferred car. Expressing frustration, as opposed to anger, may represent a softer communication strategy to get family members to change their decisions or simply to accept less preferred outcomes. Because expressing and reciprocating anger could represent a relationship threat (Roseman et al., 1994), the attribution of blame—to either uncontrollable conditions (e.g., new cars are expensive) or a relationship partner (e.g., one's spouse)—likely changes the nature of the emotion and how relationship partners express their emotions. Future research should consider the dynamics of these processes.

Because anger and frustration were shown to be interdependent emotions, future research on discrete emotions in families should take a social rather than individual perspective. Only after testing for the joint aspects and then ruling them out should research proceed to focus on emotions in family decision making as simply individually and privately experienced feelings.

Emotion mirroring and contagion

Mirroring of emotion within parental dyads was pervasive, occurring in all five decisions. While the mirroring of anger may be unintentional or automatic, mothers and fathers are consistent in their emotional expressions and they match the emotional experience of the other. Mirroring is not likely restricted to these two particular emotions or emotions generally. We examined mirroring using a methodology requiring the participation of at least three people. But mirroring itself only requires two people, making it applicable to many social consumer contexts such as consumer and sales force interactions and even consumers’ responses to facial expressions in ads (see for example Stel et al., 2011).

Our results point toward new pathways of emotion transmission, suggesting that emotional contagion could be expanded beyond unidirectional transfers of emotion from one sender to one receiver. For example, family members express emotions at the same time they elicit emotions (co-occurring actor and partner effects). Amplification of anger and frustration demonstrates that the emotion one receives can be partly due to, or reinforced by, one's own emotional expressions towards others. Synchronization of frustration demonstrates that partners uniquely express emotions in the same way within a special relationship. Mirroring of emotions shows that sending emotions between people can be linked (actor form), as can receiving emotions (partner form). Therefore emotional contagion is not monolithic: it is not always of one form, those forms can co-occur, and it is sensitive to interaction partners and the relationships between them.

Family decision making

Family decision making encompasses the nature of the decision, the varied participation of family members, and the dynamics of the process. Clearly the type of decision matters because frustration was highly reciprocal in the spending decision but not the model decision. Perhaps families’ interactions about spending involved price and quality trade-offs, which are emotion-provoking for individuals (Luce, Payne, & Bettman, 1999). Perhaps the emotion focused attention on a goal (Bagozzi, Baumgartner, & Pieters, 1998) such as spending, which limited choices and smoothed decision making about the model and final choices.

Clearly children's and parents’ roles in sharing these emotions are not interchangeable, and the emotion and the decision matter. Children consistently express anger only during the spending decisions; yet they consistently express frustration in both the spending and model decisions. Unlike their parents, children do not consistently elicit either anger or frustration. Children adapt anger and frustration to the mother, and have anger and frustration adapted to them by the father. When children participate it is largely independent and outward, although some bidirectionality was evidenced by emotions being specifically targeted to them. Bidirectionality was more prevalent for the parents, occurring in all five models. Only the parents’ emotions reflected reciprocity.

From a family dynamics perspective, although anger was found in all decisions it did not keep the family from making a decision. When events involve anger or frustration, participants may seek a “resolution of the episode [that] often includes … a better understanding between the two parties [and] more realistic expectations about their mutual behaviors” (Johnson-Laird & Oatley, 2000, p. 469). Perhaps more self-control is exhibited when anger is experienced so as not to hurt other family members, whereas family members might more freely express frustration. Emotions can serve prosocial functions (Parkinson & Simons, 2009; Fischer & Roseman, 2007; Hutcherson & Gross, 2011), and maybe expressing anger led to more cooperative behavior (Su et al., 2008).

Our social perspective and method may be especially appropriate for future investigations of socially rich processes involving decision making, family consumption, or other discrete emotions. For example, influence has been conceptualized as largely rational, steeped in authority, and pre-planned, where the influential decision maker is seen as successfully persuading others to choose the decision maker's preferred alternative (Commuri & Gentry, 2000). Because our results show anger and frustration are interdependent these emotions necessarily also involve influence (Huston, 1983), even if this influence is not necessarily pre-planned because emotions are experienced dynamically in light of emerging situations. Although family studies have not generally investigated influence and emotions together, our results suggest that they should be. Maybe the sharing of emotions is what makes family members influential, or the sharing prompts give and take in decision making.

We expect our approach would generalize to larger families where social qualities of emotions and mirroring are expected between partners of near-equal stature (e.g., adolescent siblings). Our theory and method should be applicable to family decisions for other product categories such as housing that involve feelings family members have about proximity to friends, school, and work. However, our approach may not apply to less socially risky choices or those that are individual or habitual in nature, where emotions are subdued (Wood, Quinn, & Kashy, 2002). Nevertheless even for habitual decisions, after choices have been made and outcomes later experienced, emotions may emerge as family members become bored, have bad outcomes, or encounter new choices, and the family reconsiders heretofore automatic choices.