mothers’ reports of events causing anger differ across family relationships

20
Mothers’ Reports of Events Causing Anger Differ Across Family Relationships Sandra Carpenter, The University of Alabama in Huntsville and Amy G. Halberstadt, North Carolina State University Abstract Mother’s reports of anger causes within the family were analyzed in terms of the family relationships of the persons experiencing and eliciting anger and three independent components of the anger-eliciting event—Type of Cause, Focus of Anger, and Tempo- ral Specificity. Mothers’ reports suggest substantial differences (a) in anger causes across relationships, (b) depending on whether parents or children were experiencing or eliciting anger events, and (c) for the three independent components of the anger events. For example, for type of cause, mothers’ reports indicated more expectancy violations for parents’ than children’s anger, but more goal blockages for children’s than parents’ anger. Also, elicitor mattered; mothers’ reports indicated more goal blockages elicited by parents than by children, but more expectancy violations elicited by children than by parents. These patterns are interpreted in terms of differential power and status within the family. Keywords: Anger; Emotion Understanding; Parent-child Relationships; Conflict ‘My husband and I were in the car alone on our way out. He unknowingly was on the wrong side of the road and I was very upset—telling him to move!! He just stopped the car (still on the wrong side of the road) and yelled “shut up.” I had a cake in my hand and I threw it at him and got out of the car and started walking.’—A woman’s anger ex- perience with her husband What makes family members angry and with whom? Psychologists have been inves- tigating the causes of anger for a century now, beginning with Hall (1899) and con- tinuing to the present day. Both early psychologists (e.g., Hall, 1899; Richardson, 1918) and more recent ones (e.g., Camras & Allison, 1989; Carpenter & Halberstadt, 1996; Fisher & Johnson, 1990; Fitness & Fletcher, 1993) have identified various specific events associated with feelings of anger in family members. Their lists are useful in comparing the specific stressors in modern-day life or across relevant groups or life experiences. It is difficult, however, to analyze these lists for the underlying fea- tures of anger-precipitating events or to make comparisons that extend beyond the surface qualities of the events. The studies that do assess underlying commonalities of anger causes tend to rely on hypothetically derived events rather than on actual events, and they typically allow anger causes to be sorted into only one category instead of several. © Blackwell Publishers Ltd. 2000. Published by Blackwell Publishers, 108 Cowley Road, Oxford OX4 1JF, UK and 350 Main Street, Malden, MA 02148, USA. Mailing Address: Dr. Sandra Carpenter, Department of Psychology, The University of Alabama in Huntsville, Huntsville, AL 35899, USA

Upload: sandra-carpenter

Post on 15-Jul-2016

222 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: Mothers’ Reports of Events Causing Anger Differ Across Family Relationships

Mothers’ Reports of Events Causing AngerDiffer Across Family RelationshipsSandra Carpenter, The University of Alabama in Huntsville and Amy G. Halberstadt, North Carolina State University

Abstract

Mother’s reports of anger causes within the family were analyzed in terms of the familyrelationships of the persons experiencing and eliciting anger and three independentcomponents of the anger-eliciting event—Type of Cause, Focus of Anger, and Tempo-ral Specificity. Mothers’ reports suggest substantial differences (a) in anger causesacross relationships, (b) depending on whether parents or children were experiencingor eliciting anger events, and (c) for the three independent components of the angerevents. For example, for type of cause, mothers’ reports indicated more expectancyviolations for parents’ than children’s anger, but more goal blockages for children’sthan parents’ anger. Also, elicitor mattered; mothers’ reports indicated more goalblockages elicited by parents than by children, but more expectancy violations elicitedby children than by parents. These patterns are interpreted in terms of differentialpower and status within the family.

Keywords: Anger; Emotion Understanding; Parent-child Relationships; Conflict

‘My husband and I were in the car alone on our way out. He unknowingly was on thewrong side of the road and I was very upset—telling him to move!! He just stopped thecar (still on the wrong side of the road) and yelled “shut up.” I had a cake in my handand I threw it at him and got out of the car and started walking.’—A woman’s anger ex-perience with her husband

What makes family members angry and with whom? Psychologists have been inves-tigating the causes of anger for a century now, beginning with Hall (1899) and con-tinuing to the present day. Both early psychologists (e.g., Hall, 1899; Richardson,1918) and more recent ones (e.g., Camras & Allison, 1989; Carpenter & Halberstadt,1996; Fisher & Johnson, 1990; Fitness & Fletcher, 1993) have identified various specific events associated with feelings of anger in family members. Their lists areuseful in comparing the specific stressors in modern-day life or across relevant groupsor life experiences. It is difficult, however, to analyze these lists for the underlying fea-tures of anger-precipitating events or to make comparisons that extend beyond thesurface qualities of the events. The studies that do assess underlying commonalitiesof anger causes tend to rely on hypothetically derived events rather than on actualevents, and they typically allow anger causes to be sorted into only one categoryinstead of several.

© Blackwell Publishers Ltd. 2000. Published by Blackwell Publishers, 108 Cowley Road, Oxford OX4 1JF, UK and 350 Main Street, Malden, MA 02148, USA.

Mailing Address: Dr. Sandra Carpenter, Department of Psychology, The University of Alabama inHuntsville, Huntsville, AL 35899, USA

Page 2: Mothers’ Reports of Events Causing Anger Differ Across Family Relationships

Anger in Families 459

Our first goal, therefore, was to create a descriptive coding scheme for three centralcharacteristics of causes for anger in families. This coding scheme was designed to(a) assess anger-precipitating events that are real events in real people’s lives, (b)describe the underlying features of the events, and (c) categorize multiple aspects ofthose anger-precipitating events. We anticipate that the dimensions we identify will beuseful in understanding both the intrapersonal aspects of emotion, such as the qualityof the emotion experienced, and the interpersonal aspects, such as individuals’responses to anger perceived in other family members. We propose a three-component framework for classifying anger-eliciting events, as a first step in investi-gating characteristics of these events that systematically influence anger experiencesand responses.

A second goal in our study of anger was to answer the second part of the questionabove, ‘. . . and with whom?’ A basic assumption of our research is that, for the mostpart, anger is a function of events within relationships, rather than located within par-ticular individuals. Our emphasis on studying anger causes across dyadic relationshipsemerges from social constructionist theory, which suggests that emotions are createdwithin social contexts and have social functions (e.g., Averill, 1979, 1980, 1982;Gergen, 1985). Social functions of anger, for example, might be to remedy a perceivedinjustice, to alter or maintain already existing power positions, or to ‘lose’ control ina socially acceptable way. Thus, in the present study we examined anger-precipitatingevents within family relationships in order to identify systematic variations within dif-ferent relationships.

We chose to study anger in the social context of the family for two reasons. First,anger experiences may be more common and intense in family settings than other set-tings due to the close, continual contact of the family members and the possibility ofcumulative transgressions (Denham & Bultemeier, 1993; Guthrie & Noller, 1987;Jones & Gallois, 1989). Second, the different dyadic relationships within a family canbe examined for differences in anger causes. Because the cognitive appraisal of anevent exerts a strong influence on how an emotion is perceived and labeled (Averill,1982; Barrett & Campos, 1987; Fitness & Fletcher, 1993; Stein & Levine, 1987),events that cause anger may be differently socially constructed across familial rela-tionships. For example, the interpretations of events and responses that individualfamily members make may vary as a function of developmental or power differencesin the relationship (Averill, 1982; Deaux & Kite, 1987), due to the different expecta-tions, beliefs, and goals for appropriate behavior of individuals. For example, childrenmay overgeneralize intentionality across situations and ages (ascribing intentionalityeven to very young siblings), whereas adults differentiate intentionality by target, per-ceiving children as less responsible for their actions than adults (Edgerton, 1985;Fincham & Bradbury, 1987). As children mature, however, their developing commu-nication skills, understanding of others’ motives and feelings, and ability to cope withfrustration and to control emotion (Dunn, 1994) may modify their anger experiencesand understanding of causality. Children are also still learning the familial and cul-tural rules for understanding and expressing anger (e.g., Averill, 1979, 1982; Burrowes& Halberstadt, 1987; Dunsmore & Halberstadt, 1997; Halberstadt, Fox, & Jones,1993; Saarni, 1985), so they may experience more or less anger and for differentreasons than parents. Finally, adults may be more accepting of certain kinds of situa-tions because they are more likely to have the perspective that change takes time,whereas children may believe that they must change the situation here and now. There-fore, there are good reasons to expect substantive differences between children’s and

© Blackwell Publishers Ltd. 2000 Social Development, 9, 4, 2000

Page 3: Mothers’ Reports of Events Causing Anger Differ Across Family Relationships

460 Sandra Carpenter and Amy G. Halberstadt

© Blackwell Publishers Ltd. 2000 Social Development, 9, 4, 2000

parents’ experiences of anger, as well as differences in the types of events that elicitanger as a function of family role.

In this research, we focus on mothers’ perceptions of the events that cause anger,because mothers have substantial impact on the family dynamics associated withanger. First, women still tend to be gatekeepers of emotion in relationships (Gottman,1990). Second, mothers’ perceptions of the events that cause family members to beangry will most likely influence their private and public labeling of situations, theirdiscussions of anger with other family members, and their monitoring and manipu-lating of family settings to avoid events that they perceive to be anger-precipitating.For example, it is likely that mothers’ perceptions about situations trigger their ownanger experiences and subsequent actions (e.g., choosing physical discipline; Peterson, Ewigman, & Vandiver, 1994). Third, mothers are more likely than any otherfamily member to stop conflicts within the family (Vuchinich, 1987), so their inter-pretations of events may be crucial to the successful resolution of problems. Fourth,mothers may ‘teach’ their causal interpretations about anger eliciting events to otherfamily members (Smetana, 1993). For example, mothers explicitly refer to social rulesduring disputes with children as young as 3 years (Dunn & Munn, 1987; Garner, Jones,Gaddy, & Rennie, 1997). Because mothers still spend more time with their childrenthan do fathers (Ishii-Kuntz, 1995; Lamb, 1987; Pleck, 1997), they are the most fre-quently available resource for modeling or purposefully communicating their causalschemas of anger. For our research purposes, we agreed with Huston and Robins(1982), who propose that perceptions of relationship events may provide betteraccounts for events than ‘actual’ events. We therefore focused on mothers’ perceptionsof anger eliciting events, because these perceptions offer a window into the causalschemas that mothers use when interacting with family members and that they teachto other family members.

Pilot Study for the Development of Our Coding Scheme

The development of our coding scheme was informed by our pilot testing of fourcoding schemes previously created for categorizing the types of events that elicit anger(Camras & Allison, 1989; Fisher & Johnson, 1990; Scherer, Wallbott, & Summerfield,1986; Shaver, Schwartz, Kirson, & O’Connor, 1987). Although these schemes wereindependently developed with different goals (e.g., to document actual emotionalexperiences versus to elicit prototypes) and different subject populations (e.g., chil-dren versus adults), they each included two types of categories: violations of expec-tations and frustration with respect to goals, help, or objects. They varied substantially,however, regarding other categories. Also, they all required that categories be mutu-ally exclusive, so that an event could be coded into only one category. We initiallycoded the present data set using all four coding schemes. This generated a wealth ofinformation, but no concise or satisfying way to clearly represent the similarities anddifferences in perceptions of anger causes across relationships.1 Therefore, to assesssimilarities and differences in causes of anger across relationships, we developed anew coding scheme for analysis of anger events. The new coding scheme included thethree components of the anger-precipitating events that we predicted would be mostapplicable to various relationships within the family. We designed our coding schemeso that we could assess underlying causes (e.g., taking away toys from children or afather not providing for school costs both represent goal blockages), and measure mul-tiple components simultaneously. Thus, this coding scheme emerges from the strengths

Page 4: Mothers’ Reports of Events Causing Anger Differ Across Family Relationships

Anger in Families 461

identified from previously created coding schemes (indicated below), yet allows forsubstantially more information to be obtained about each anger event.

Type of Cause. The first component, Type of Cause, distinguishes between the twocauses that are of central importance in theory (e.g., Averill, 1982; Fitness & Fletcher,1993; Mandler, 1975, 1980), and in all the previously created coding systems men-tioned above. Expectancy violations disrupt the normative flow of interpersonal inter-actions by breaking rules that are theoretically shared by both the person being angeredand the person eliciting the anger. Goal blockages interrupt an ongoing or plannedactivity that may be a more ‘one-sided’ desire of the angered person. Examples ofexpectancy violations in the literature are abundant and include: not feeling respectedor treated fairly (Fisher & Johnson, 1990; Scherer & Tannenbaum, 1986; Shaver etal., 1987), failure of others to conform to social norms (Scherer et al., 1986), andbroken personal or social contracts or implicit social rules (Camras & Allison, 1989;Fisher & Tannenbaum, 1990). Examples of the many previous categorizations of goalblockages include: failure to reach goals or achieve an objective, or an unexpected,unnecessary inconvenience (Scherer et al., 1986); frustration or interruptions of a goal-directed activity (Shaver et al., 1987); and desiring but not getting an object, response,or experience (Camras & Allison, 1989). The distinction between expectancy viola-tions and goal blockages is important because interpreting an event as an expectancyviolation may result in attempts to re-establish the expectancy by validating the under-lying rules or social norms, whereas interpreting an event as a goal blockage may leadto a greater focus on removing the obstacles to the goal.

In families, both types of anger are likely. However, parents may more frequentlyexperience anger due to expectancy violations than children do, as a consequence ofadults’ greater tendency toward abstraction and search for meaning in underlyingcauses. The literature on conflict in the family suggests that conflicts between chil-dren are most often due to goal blockages (see Shantz, 1987, for a review), and thatadolescents more frequently report conflict due to regulation of their activities andrelationships than do parents (Smetana, 1989). Status issues may also come into playhere—powerful others are more likely to cause anger by blocking goals than are lesspowerful others.

Thus, for Type of Cause, we predicted both experiencer and elicitor effects. (Wewill use the term ‘experiencer’ to refer to the angry person, and ‘elicitor’ to refer to the person at whom the anger is directed.) For the experiencer effect we predictedthat mothers would more frequently report expectancy violations as causes of parents’than children’s anger, and that goal blockages would be reported more frequently for children’s than for parents’ anger. For the elicitor effect, we predicted that mothers would perceive that parents elicit relatively more goal blockages than wouldchildren.

Focus of Anger. A second component of anger-precipitating events is the Focus(recipient or direction) of anger. Because the intensity of an emotion is associated withthe impact of the possible outcome (Brehm, 1999), behaviors that damage persons(mentally or physically) may elicit more extreme anger experience than behaviors thatdamage objects (Averill, 1982; Gates, 1926). In fact, interpersonal causes are the most prevalent causes in both children’s and adults’ descriptions of what would makethem or another person angry (Stein, Liwag, & Wade, 1996; Strayer, 1986). Thus,overall, we expected mothers to report more instances of person-focused than object-focused causes of anger. In the present study, all of the mothers’ reports of anger were interpersonal (due to the nature of the questionnaire requesting descriptions of

© Blackwell Publishers Ltd. 2000 Social Development, 9, 4, 2000

Page 5: Mothers’ Reports of Events Causing Anger Differ Across Family Relationships

462 Sandra Carpenter and Amy G. Halberstadt

© Blackwell Publishers Ltd. 2000 Social Development, 9, 4, 2000

anger-eliciting events caused by family members). However, anger descriptions variedas to whether an object was in dispute (e.g., food, toys, or money) or whether the causewas more clearly person-focused (e.g., an unfriendly attribution made about anotherindividual’s personality).

The coding schemes used in the pilot study reveal that person-focused causes ofanger have received the most attention, to date. A strength of the Scherer et al. codingscheme (1986) is the distinction between damage to personal property and physicalhurt. Camras and Allison (1989), however, combine damage of people and their pos-sessions into one category, and Shaver et al. (1987) refer only to damage to persons(physical or psychological). We chose to maintain the distinction on the basis of devel-opmental literature, which shows differences in focus of anger by age. Possession anduse of objects is the most prevalent cause of conflict between young children and theirpeers (Shantz, 1987) and between young children and their mothers (Dunn & Munn,1987). There also seems to be a developmental trend: play objects are the most fre-quent causes of disputes for children aged two and under (Bronson, 1975), but by agefive or so, control over the social environment (interpersonal concerns) becomes anequally likely cause of conflict (Shantz, 1987). Further, the ability to think about andunderstand interpersonal causes seems to develop substantially through middle child-hood and adolescence (Livesley & Bromley, 1973; Selman, 1980); we thereforeexpected that parents would experience more person-focused causes of anger thanwould children. Thus, a strength of several previous coding schemes is the focus oninterpersonal causes; we add and extend a focus on object-related causes.

Thus, for the component of Focus of anger, we predicted an experiencer effect, suchthat parents’ anger would be more often perceived as person-focused than would chil-dren’s anger, whereas children’s anger would be more object-focused than wouldparents’ anger. We had no reason, however, to expect an elicitor effect.

Temporal Specificity. The third component of anger, Temporal Specificity, distin-guishes between ongoing events (i.e., those that occur frequently) compared to spe-cific events (i.e., those that occur once or infrequently). Because a pattern of behaviorthat persists over time may be considered more purposeful than one ‘accident,’ anongoing, stable pattern of annoying behavior may result in more intense or enduringanger than one-time, unique events (Weiner, 1986). In addition, a stable pattern ofnegative behavior constitutes a greater threat to relationships than does a singularinstance (Fehr, Baldwin, Collins, Patterson, & Benditt, 1999), so that anger intensityand duration may be correspondingly greater for ongoing than specific events.

A strength of the Fisher & Johnson (1990) coding scheme is that they acknowledgethe importance of this differentiation by proposing two categories of expectancy vio-lations. They used ‘not fulfilling role expectations’ for generalized, ongoing patternsof behavior (such as personality traits) and ‘not fulfilling social obligations’ for moretemporally specific instances (Fisher, personal communication, 9/9/90). Adults maymore frequently experience anger due to ongoing patterns than children; adults canmore easily evaluate multiple aspects of multiple events than can children, and maythen assess the similarities in those events that suggest patterns over time (e.g., per-sonality characteristics). For example, parents are more likely than adolescents to citepersonality as a cause of conflict within the family (Smetana, 1989). Thus, for Tem-poral Specificity, we predicted an experiencer effect, such that children’s anger wouldmore often be perceived as specific than ongoing, but parents’ anger would more often be perceived as ongoing than specific. We had no reason, however, to expect anelicitor effect.

Page 6: Mothers’ Reports of Events Causing Anger Differ Across Family Relationships

Anger in Families 463

Our coding scheme thus consists of three components: Type of Cause (ExpectancyViolation versus Goal Blockage), Focus (Person-focused versus Object-related anger),and Temporal Specificity (Ongoing, trait-like actions versus Specific, circumscribedactions). This coding scheme has the advantage over previous coding schemes ofallowing any anger event to be categorized on all three components simultaneously,whereas other coding schemes have required a single choice from a set of coding categories.

Although we do not claim that these three components of the causal schema areexhaustive, we expect them to account for a significant proportion of variance inmothers’ descriptions of anger-eliciting events. In addition, we recognized that the dis-tinctions between categories for all three dimensions would not always be clear. Thesedistinctions, however, are no more ‘fuzzy’ than the distinction between positive andnegative reinforcement in operant conditioning theory. Consider the followingexample: A parent picks up a crying child, who subsequently stops crying. Does thisoutcome of the behavior constitute positive reinforcement (due to the reward of silence)or negative reinforcement (due to the cessation of crying)? As will be reported, wefound that our coders could, for the most part, agree on Type of Cause and Focus.

To obtain anger reports in the family, we asked mothers to report the causes of her anger at a target child and at the father of her child, as well as events causing anger toward herself and between the child and father. Thus, mothers reported events that frequently caused anger for themselves and events that they perceived as frequently angering other family members. To gather these event-level data, weemployed a modified version of Averill’s (1982) extensive anger questionnaire.2

Women having at least one child between the ages of five and seven years wererecruited for the study.

Children of this age were targeted for various reasons. First, parents and children tendto show consensus in their reports of anger causes by four years of age (Tergowt &Harris, 1993), so our participants’ reports are likely to correspond to reports that wouldbe provided by their children. Second, children are able to refer to social rules as justi-fications by the age of three years (Dunn & Munn, 1987); our hypotheses about Typeof Cause could thus be adequately tested for children. Third, it is at approximately fiveyears old that children experience conflict about issues related to the possession or useof objects and issues of social control in equal proportions (Shantz, 1987), allowing forsufficient variability regarding the Focus of anger. Thus, the children whose mothersparticipated in our sample should have the cognitive and social skills requisite tobecome angry in the ways anticipated by our three-component scheme.

In summary, we predicted that anger causes would vary depending on the familyrole of the person experiencing anger and eliciting anger, and that we could learn more about anger in the family by assessing multiple components of the anger cause.The three components we coded were Type of Cause, Focus of Anger, and TemporalSpecificity.

Method

Participants

Participants were 60 women who had at least one child between the ages of 5 and 7years. They had from one to nine children (median = 2 children). Their ages rangedfrom 22 to 45 yrs (M = 35.5 yrs). Their marital status was 48 married, 5 remarried, 6

© Blackwell Publishers Ltd. 2000 Social Development, 9, 4, 2000

Page 7: Mothers’ Reports of Events Causing Anger Differ Across Family Relationships

464 Sandra Carpenter and Amy G. Halberstadt

© Blackwell Publishers Ltd. 2000 Social Development, 9, 4, 2000

divorced or separated, and 1 never married. The family socioeconomic status, basedon Hollingshead’s 4-factor index (1975), spanned the entire range with the modalfamily in the fourth highest of five levels. The sample was 78% white, 17% black, and5% Asian families.

Procedure

Women filled out a questionnaire on anger-precipitating events, modified fromAverill’s anger questionnaire (1982), as part of a larger laboratory study on familyexpressiveness and emotional experience and expression in children (Halberstadt etal., 1993). Because other aspects of the study are not relevant to the present investi-gation they are not discussed further here. Participants were asked to identify anddescribe the most frequent cause of anger for each one of eight dyadic relationshipswithin the family of the woman being surveyed. The eight relationships were, in thisorder: mother to father, father to mother, mother to child, child to mother, father tochild, child to father, child to sibling(s), and siblings(s) to child. Fathers were eitherbiological, adoptive, or in common-law arrangements for at least one year. Partici-pants generally wrote phrases or a few sentences. For each most frequent cause, theythen rated the frequency, duration, intensity, and speed of onset of anger. These ex-periential data are reported elsewhere (Halberstadt, Carpenter, & Grotjohn, 2000).

The data comprised women’s free-format descriptions of their perceptions of eventsmost frequently causing anger in eight family relationships. Because we anticipatedthat coding these descriptions might be challenging, we chose to have five coders makethese judgments, and we accepted majority rule, thereby increasing stability in thefinal categorization of each anger cause. The coders (the first author, one graduatestudent in psychology, and three senior honors students majoring in psychology) inde-pendently categorized all of the participants’ responses. Coders were trained by thefirst author, using the events generated by participants in an independent sample (seeFootnote 2) as practice stimuli. Training required approximately three hours. Coderscategorized each description by assigning it to one pole of each of the three dichoto-mous categories. Overall, coders were instructed neither to read too much into descrip-tions nor to assume causes that were not explicitly stated.

First, coders determined whether a description of an anger event represented anexpectancy violation or a goal blockage. Coders were instructed that violationsreferred to implicit or explicit rules or to social contracts that have been broken. Rulesand social contracts were described as those things others ‘should’ be doing. Goalblockages were defined as obstacles to or interruptions of desired activities or desiredstates of affairs. However, noncompliance with a rule or contract was to be coded asa violation rather than as a goal blockage. A rule of thumb was helpful here: For ananger description to be coded as a goal blockage, a specific goal had to be identifi-able. For example, ‘didn’t set the table on time’ is clearly identifiable; ‘didn’t work on having a better relationship’ is too amorphous and better fits as an expectancy violation.

Second, the coders categorized anger descriptions as person-focused or as object-related. Person-focused events were defined as involving relationships, feelings, orself-esteem issues (e.g., not listening to or not spending special time with the otherperson, jealousy, unwillingness to go to marriage counseling, or being teased). Causesof anger were coded as object-related if they involved anger toward people aboutobjects. Some object-related responses were obvious (e.g., ‘she won’t share her toys’),

Page 8: Mothers’ Reports of Events Causing Anger Differ Across Family Relationships

Anger in Families 465

but we also considered these targets to be objects: work (e.g., ‘he stays too late atwork’), words (e.g., ‘talking back’), time (e.g., ‘he is always late’), and noise (e.g.,‘runs with loud footsteps in the hallway’). It was emphasized to coders that whereasanger described by the mothers in this study was always directed toward a person(because of the design of our questionnaire), it was not necessarily just about theperson. If in doubt, coders’ default was to use the object-related end of the category.This provided the most stringent test of the hypothesis that person-focused events leadto more extreme anger experiences than do object-related events.

Third, coders categorized events of anger either as ‘ongoing’ or as a ‘particular iso-lated instance.’To be coded as ongoing, the description must have depicted the patternof behavior as continuous over time. This could be indicated by the use of temporaladverbs (e.g., ‘sometimes,’ ‘often,’ or ‘occasionally’), use of gerunds (e.g., ‘my has-sling him’), or use of a trait term suggesting consistency (e.g., sloppiness, forgetful-ness). Responses were coded as particular instances if the descriptions were expressedin the past tense or contained detail that specified an isolated instance (e.g., ‘I hassledhim about wearing a clean shirt’ or ‘I forgot his birthday’).

Each response was thus categorized as associated with one alternative for each ofthree dichotomous categories: Type of Cause (expectancy violation or goal blockage),Focus (person or object), and Temporal Specificity (ongoing or particular instance).This 2 ¥ 2 ¥ 2 categorization scheme yielded eight coding categories. Abbreviatedexamples of items coded into each of the eight categories are listed in Table 1.

© Blackwell Publishers Ltd. 2000 Social Development, 9, 4, 2000

Table 1. Abbreviated Examples of Anger Events, Coded in Terms of the ThreeComponents

Focus andType of Cause

Specificity Expectancy Violations Goal Blockages

ONGOINGPerson-focused Lack of communication Won’t go to marriage counseling

Being inconsiderate Disagreements about child careNot listening Getting our divorce going

Object-related House being messy Slow to complete tasksNot doing share of work How to spend or not spend moneyDamage to toys Disagreeing about plans or activities

TEMPORALLY SPECIFICPerson-focused Didn’t call home when late No financial help for my schooling

Hurt sibling Wanted to go to her friend’s houseTeased cousin Didn’t go to bed; no time together

Object-related Didn’t make bed Took toys awayKicked door when mad Didn’t respond to request right thenDidn’t tell me about high Didn’t buy cookies

school son wants toattend

Page 9: Mothers’ Reports of Events Causing Anger Differ Across Family Relationships

466 Sandra Carpenter and Amy G. Halberstadt

© Blackwell Publishers Ltd. 2000 Social Development, 9, 4, 2000

Results

Analytic Strategy

We first assessed inter-coder reliability for our coding categories. Next, we determinedwhether the data needed to be examined separately in terms of (a) gender, (b) the dif-ferent types of relationships, (c) the three components of the anger events (type ofcause, focus, and specificity), and (d) experiencer and elicitor. These analyses indi-cated that we could collapse across gender; however, relationship type, the three com-ponents of anger events, and experiencer and elicitor effects were all necessary tosufficiently describe the variance within our data set.

We therefore tested our specific hypotheses relevant to experiencer and elicitoreffects for each of the three proposed components of anger-precipitating events. Totest these hypotheses, we conducted three maximum likelihood analyses of variance(one for each component: Cause, Focus, and Specificity), with each analysis having a2 (Experiencer: Parent vs. Child) ¥ 2 (Elicitor: Parent vs. Child) design.3

Preliminary Analyses

Inter-rater reliability in classification. As expected, judgments about anger causeswere relatively difficult, but good agreement across the five coders was obtained fortwo of the three dimensions; kappas = .63, .65, and .28 for Type of Cause, Focus, andTemporal Specificity, respectively. We believe that levels of agreement were partiallyconstrained by the brevity of the mothers’ descriptions; in retrospect, mothers in thisentirely non-university sample were much more articulate in conversation than in theirwriting.

The kappa for Specificity was very low. This was due to the lack of variability forparent causes of anger: in that these were almost entirely coded as ongoing; (see Table5), a divergence of the coders into two groups who seemed to use coherent but dif-ferent rules when coding children’s anger, and the brevity of the mothers’ descriptionsregarding the extent to which they perceived patterns in children’s anger causes. Wedecided to proceed with analyses for all three dimensions, aware that the lower reli-ability, and therefore increased noise, might suppress effects that did exist. That is,although we might be in danger of failing to identify actual relationships (Type IIerror), we were not in jeopardy of ‘finding’ effects that did not exist (Type I error). Inaddition, the use of five raters, compared to two raters, increased the stability of thefinal code for each anger cause. And, as will be seen, despite the noise generated bythe low reliability, enough of a signal was obtained in the log-linear and subsequentanalyses to identify the importance of specificity in explaining the variance in mothers’reports.

Gender similarity. We had no reason to expect gender differences (for either ex-periencer or elicitor) in the underlying structure of anger causes, as categorized usingthe three components described above. And, in the nine chi-square analyses testing forgender effects, none approached significance (X2 range = .002 to 2.05). This gendersimilarity was confirmed by a correspondence analysis. A correspondence analysissummarizes complex relationships between participants and sets of variables simulta-neously. It generates both a spatial representation indicating the structure between thevariables of interest, as well as an indication of the amount of variance accounted forby the dimensions of the representation (Weller & Romney, 1990).4 As can be seen in

Page 10: Mothers’ Reports of Events Causing Anger Differ Across Family Relationships

Anger in Families 467

Figure 1, coded responses for mothers and fathers were highly proximal, showing thatmothers’ reports of mothers’ and fathers’ anger events (in Figure: fm, mf, fc and mc)were virtually identical in how they were coded along the three components, as werethe causes of children’s anger directed at their mothers and fathers (in Figure: cf andcm). Therefore, we collapsed across gender of parents (as both experiencers and elicitors) with minimal loss of information.

Although mothers’ and fathers’ events were highly proximal in terms of their coor-dinates along the three components, mothers’ reports of fathers’ anger experiencescannot be considered mere replications of their own reasons for anger. For example,a mother might report being most frequently angry at her husband for not spendingenough time with her compared to his buddies, and that her husband is most frequentlyangry at her for nagging him about always scolding the children. Although the reasonsfor anger are very different, they end up being highly proximal in the correspondenceanalysis because both would be coded as expectancy violations that are person-focusedand ongoing. That is, the specific events differed (e.g., not enough time vs. nagging),but their underlying meanings were interpreted as the same. Descriptions of mothers’anger at fathers and fathers’ anger at mothers matched for only 7% of the descrip-tions. Mothers’ anger at their children and fathers’ anger at children matched for 42%of descriptions, and children’s anger at mothers and at fathers also matched for 42%of the descriptions. Examples of events can be found in Table 1.

Are relationships, components, and the experiencer/elicitor distinction all neces-sary? We argued in the introduction that anger is socially constructed and that angercauses will vary across different family relationships. We also argued that we need tocode at least three different components of the anger event to more fully understandanger causes. Although all previous research and theory clearly emphasized the Type of Cause distinction, just how important were Focus and especially Specificity(with its low reliability) in understanding anger causes? Our first step, then, was todetermine whether all the family relationships and all three components were actuallynecessary for a full understanding of anger in the family. This can be assessed statis-tically by conducting a series of log-linear analyses, using the method of hierarchicalbackward elimination. This method identifies main effects and interactions of vari-ables (and combinations thereof) that account for the greatest variability in frequencydata.

As shown in Table 2, results for the full model—4 Relationships ¥ 2 Causes ¥ 2Foci ¥ 2 Temporalities—indicate that the best fit to the data was generated by a com-bination of three 3-way interactions. Together, these three interactions account for96.81% of the variance, as indicated by the G2 statistic of 3.19 (which indicates thepercentage of variance not accounted for). The full model included a 3-way interac-tion between the three components (Cause, Focus, and Temporal Specificity), as wellas two 3-way interactions each containing the relationship variable (i.e., Relationship¥ Cause ¥ Specificity, as well as Relationship ¥ Focus ¥ Specificity). We then testedthe importance of including all three components in each of the four relationships separately, using a hierarchical backward elimination strategy. As can be seen in Table2, each of the three components (Cause, Focus, and Specificity) occurs in each of themodels that best account for the variability in mothers’ reports of anger in their families. This pattern of results suggests that each of the three components providesdistinct, non-repetitive, and requisite information for explaining anger-precipitatingevents. It also indicates that the patterning of the three components varies across rela-tionships, as hypothesized.

© Blackwell Publishers Ltd. 2000 Social Development, 9, 4, 2000

Page 11: Mothers’ Reports of Events Causing Anger Differ Across Family Relationships

468 Sandra Carpenter and Amy G. Halberstadt

© Blackwell Publishers Ltd. 2000 Social Development, 9, 4, 2000

Finally, we determined whether both experiencer and elicitor effects were needed to provide an adequate account of the variability of anger events in themothers’ reports. The correspondence analysis previously described revealed that thetwo components corresponding to the predicted experiencer and elicitor effectsaccounted for 30% and 27% of the variance, respectively. These results support ourplan to test for elicitor and experiencer effects separately and for each of the threecomponents of anger events; they also support the claim that anger causes vary acrossrelationships.

Tests of the Hypotheses about Type of Cause, Focus, and Specificity

An examination of the frequencies with which events were coded for each of the threecomponents indicated that the dichotomous alternatives of the components were notequally represented. Expectancy violations (74%), object-related (66%), and ongoing(70%) events predominated throughout this sample of mothers’ reports of angerevents.5 Maximum likelihood analyses of variance indicated, however, that theseevents varied significantly across these three components as a function of experiencerand elicitor. Due to the exploratory nature of our investigation and repeated measuresnature of our data, we set the alpha level at p = .01.

Frequencies for Type of Cause by Experiencer and Elicitor are shown in Table 3.As predicted, the experiencer effect indicated that mothers reported expectancy vio-lations as more frequently the cause of parents’ than children’s anger (201 versus 105expectancy violations), but goal blockages more frequently as the cause of children’sthan parents’ anger (83 versus 25 goal blockages), C2 = 35.28, p < .001. The predictedelicitor effect was also obtained. Mothers reported that parents elicited anger due to

Table 2. Models that Best Account for Variability in Mothers’ Reports of Angerin Their Families

Model Predictors G2*

All Four Relationships Cause X Focus X Specificity 3.19Relationship X Focus X SpecificityRelationship X Cause X Specificity

Separated by RelationshipParent to Parent Cause 4.56

Focus X SpecificityParent to Child Cause X Focus X Specificity 0.01Child to Parent Cause X Focus 0.12

Focus X SpecificityCause X Specificity

Child to Sibling Specificity X Cause X Focus 2.44

Note: *G2 is an index of ‘fit,’ indicating the degree of variance still to be accounted for; thus,low G2 indicates a good fit of the model. For example, for the first model on all four relation-ships, 96.81% of the variance is accounted for by the model, leaving only 3.19% of the vari-ance not accounted for.

Page 12: Mothers’ Reports of Events Causing Anger Differ Across Family Relationships

Anger in Families 469

goal blockages more frequently than did children (87 versus 21 goal blockages),whereas the children elicited expectancy violations more frequently than did parents(176 versus 130 expectancy violations), C2 = 32.32, p < .001.

Frequencies for Focus of Anger by experiencer and elicitor are shown in Table 4.As predicted, the experiencer effect indicated that mothers reported parental anger asmore frequently person-focused than children’s anger (94 versus 45), and children’sanger as more frequently object-focused than parents’ anger (143 versus 132); C2 =12.49, p < .001. There was no elicitor effect, but one other significant effect wasobtained. The interaction effect between experiencer and elicitor, C2 = 16.33, p < .001,indicated that anger causes between same-age/status pairs (parents with parents andchildren with children) were more frequently person-focused than anger causesbetween different-aged pairs (87 versus 52), whereas anger causes between different-aged pairs were more frequently object-related than anger causes between same-aged pairs (167 versus 108). Thus, an important predictor seems to be whether the

© Blackwell Publishers Ltd. 2000 Social Development, 9, 4, 2000

Table 3. Frequencies of Type of Cause as a Functionof Family Role of Experiencer and Family Role of Elicitor

Experiencer

Parent Child Totals

Elicitor EV GB EV GB EV GB

Parent 93 19 37 68 130 87Child 108 6 68 15 176 21

Totals 201 25 105 83 306 108

Note: EV = expectancy violation. GB = goal blockage.

Table 4. Frequencies of Focus of Anger as a Functionof Family Role of Experiencer and Family Role of Elicitor

Experiencer

Parent Child Totals

Elicitor PER OBJ PER OBJ PER OBJ

Parent 61 51 19 86 80 137Child 33 81 26 57 59 138

Totals 94 132 45 143 139 275

Note: PER = person-focused. OBJ = object-related.

Page 13: Mothers’ Reports of Events Causing Anger Differ Across Family Relationships

470 Sandra Carpenter and Amy G. Halberstadt

© Blackwell Publishers Ltd. 2000 Social Development, 9, 4, 2000

experiencer and elicitor are of the same or different generations (e.g., parent and parentvs. child and parent, respectively).

Frequencies for Temporal Specificity by experiencer and elicitor are shown in Table5. As predicted, the experiencer effect indicated mothers’ reports of children’s angeras more frequently due to specific rather than ongoing events (106 versus 82); however,reports of parental anger were almost exclusively due to ongoing events rather thanspecific events (209 versus 17), C2 = 91.73, p < .0001. One additional effect was sig-nificant. The elicitor effect indicated that in these mothers’ descriptions of anger,parents elicited anger due to ongoing events more frequently than children did (163versus 128), but that children elicited anger due to specific events more frequentlythan parents (69 versus 54); C2 = 6.07, p = .01.6

In summary, the perceived differences in events causing anger depended on theexperiencer and elicitor, as well as the component being considered. Mothers’ reportsof anger-precipitating events in their family can be summarized as follows: Type ofCause of anger was due to expectancy violations for most parental anger (regardlessof elicitor) and in the majority of instances when children were angered by siblings.However, goal blockage was the major cause of children’s anger at parents. Focus ofanger was relatively more person-focused for parents in comparison to children. Also,anger between same-aged pairs was more frequently person-focused than angerbetween different-aged pairs, for whom anger was most frequently object-related.Temporal Specificity of anger events was primarily ongoing if the experiencer was aparent, but was more likely to be specific if the experiencer was a child. In addition,parents elicited anger more frequently due to ongoing events, whereas children elicitedanger comparatively more often due to specific events.

Discussion

Mothers’ perceptions of events causing anger and the causal structure of these eventswithin the context of the family were investigated, using a coding scheme derived fromthe theoretical and empirical literature on anger. This coding scheme adds to recentresearch examining such emotion knowledge structures (see Fehr & Baldwin, 1996;Fitness, 1996 for reviews of this literature), by introducing three components of angerprototypes that can be coded simultaneously and compared across different relation-

Table 5. Frequencies for Temporal Specificity of Anger as a Function of theFamily Role of Experiencer and Family Role of Elicitor

Experiencer

Parent Child Totals

Elicitor Ongoing Specific Ongoing Specific Ongoing Specific

Parent 105 7 58 47 163 54Child 104 10 24 59 128 69

Totals 209 17 82 106 291 123

Page 14: Mothers’ Reports of Events Causing Anger Differ Across Family Relationships

Anger in Families 471

ships in the family. We predicted differences in the patterns of anger events acrossthese components as a function of the type of relationship of both the elicitor and theexperiencer in the anger event. Classifications of mothers’ accounts of causes of angerin their families provided support for our hypotheses. Correspondence and log-linearanalyses indicated that family role, the experiencer and elicitor distinction, and allthree underlying components are necessary to best account for mothers’ reports ofanger in their families.

In addition, our hypotheses for each of the three components of the anger-elicitingevent were supported. First, for the Type of Cause factor, mothers’ perceptions of inter-parental anger (e.g., lack of attention to family), inter-sibling anger (e.g., violation ofterritory), and parental anger at children (e.g., child’s ‘back talking’) were most oftendue to expectancy violations. Children’s anger at parents, however, was primarily dueto goal blockages (e.g., won’t let child have food between meals). It is likely that thosewith greater status or power are recognized by mothers as more effectively blockinggoals of others and are perceived as the ‘rule-makers.’ This puts them in the positionof obstructing others’ desires, and, because they have established the rules it is easierfor others to be seen as violating their expectations.

Second, with regard to Focus of anger, our stringent rules for distinguishing betweenperson-focused and object-related causes may have led to an overestimation of object-related issues in the sample as a whole. Nevertheless, differences by relationshipemerged: mothers’ reports of parents’ anger toward spouses were relatively more oftenperson-focused (e.g., my nagging), compared to the anger of children, which was moreoften object-related (e.g., not sharing toys). Also, Focus varied by same-age or different-age status, again suggesting the importance of recognizing power differencesin the home. When anger was intra-generational (intra-parental or intra-sibling),mothers’ descriptions of anger were more person-focused (e.g., father ‘spoils thedaughter’) than when it was inter-generational (child to parent or parent to child). Itmay be that when power is reciprocal, relatively more focus is on the relationship thanabout control of objects because power to control objects is equally shared. Relativelyequal levels of power therefore may be as relevant as sheer amount of power, at leastwithin the rules of emotion knowledge structures.

Third, with regard to Temporal Specificity, mothers’ descriptions of inter-parentalanger (e.g., spouse doesn’t carry his share of the work around the house) were pre-dominantly coded as ongoing, whereas children’s anger was more often perceived asspecific, perhaps because children live more in the ‘here and now’ than do adults (e.g.,child doesn’t get immediate attention when interrupting mother). We obtained theseresults in spite of the low inter-coder reliability for this component and they retainedsignificance when the coders were divided into two groups; nevertheless, these pat-terns should be considered tentative until replicated.

By combining summary statements about the experiencer and elicitor effects indi-cated above, as well as by examining the patterns obtained in the correspondenceanalysis (see Figure 1), we can make some preliminary descriptive statements abouthow components of anger-eliciting events may vary across types of relationships.Mothers perceived parents’ anger to be due to ongoing expectancy violations, andthese tended to be person-focused when another parent was the elicitor. Mothers per-ceived children’s anger, however, to be due to object-related goal blockages that wereslightly more often ongoing than specific when parents were elicitors, but specific,object-related expectancy violations when siblings were elicitors. Each of these com-ponents, then, provided additional and distinctive information about the underlying

© Blackwell Publishers Ltd. 2000 Social Development, 9, 4, 2000

Page 15: Mothers’ Reports of Events Causing Anger Differ Across Family Relationships

472 Sandra Carpenter and Amy G. Halberstadt

© Blackwell Publishers Ltd. 2000 Social Development, 9, 4, 2000

causes of anger. Now that these descriptive steps have been taken, these componentscan be used in further understanding the factors that may affect the subjective ex-perience of anger and the behavioral responses to anger-eliciting events. For example,Type of Cause may predict whether a response is to re-establish normative conditions(expectancy violations) or to remove an obstacle (goal blockages). Anger may be moreextreme when the Focus of anger is due to persons rather than to objects; the evalua-tion of the person eliciting anger is also likely to be more negative when persons areharmed. Also, the intensity and persistence of anger may be more extreme when causesof anger are more ongoing than specific.

These results also indicate the importance of studying perceptions of anger withinthe context of and across a variety of relationships. The events perceived to be causing

Figure 1. Classifications of anger-eliciting events across eight types of relationships, as depicted by a corre-spondence analysis. Capital letters refer to event classification, whereas lower case letters refer to relationships.For event classification, the first letter refers to Type of Cause (E = Expectancy Violation, G = Goal Blockages),the second letter refers to Focus of Anger (P = Person-focused, O = Object-focused), and the third letter refersto Temporal Specificity (O = Ongoing, S = Specific); thus, GPO = Goal Blockage, Person-focused, Ongoingevents. For relationships, the first letter refers to the Experiencer, the second letter to the Elicitor (c = child, f =father, m = mother, s = sibling); thus, fm = fathers’ anger at mothers.

Page 16: Mothers’ Reports of Events Causing Anger Differ Across Family Relationships

Anger in Families 473

anger varied dramatically by relationship, suggesting the importance of social roles inconstructing anger. It is likely that individuals’ goals, expectations, and needs are con-strained by the demands of their familial roles, thus influencing the kinds of angerthey experience or are perceived to experience. Or it may be that the scripts that cor-respond to these roles shape people’s understanding of the anger that is appropriatefor that familial role, thus directing their perceptions of their experiences.

In addition, maternal perceptions of anger causes may well impact the resolutionsof various family conflicts. For example, if mothers perceive that goal blockages aremore important to children, they might attempt to meet only that specific goal, ratherthan to discuss underlying power issues or to adjust children’s expectations. Andalthough mothers may not focus on children’s expectations of them, they might encour-age siblings to consider each other’s expectations in conflicts about objects. If mothersare accurate in their perceptions of anger causes, their guidance during family con-flict may well increase family harmony. But regardless of accuracy, the implicit rulesthat mothers seem to employ for labeling events are likely to be identified for chil-dren and to be used by those children in developing their own understanding of con-flict scenarios (Dunsmore & Halberstadt, 1997). Thus, mothers’ perceptions ofanger-precipitating events are especially important because children may developemotion prototypes by using their mothers as models (Dunn, Brown, & Beardsall,1991; Miller & Sperry, 1987). Future research might focus on the degree to whichchildren’s perceptions of anger-precipitating events in the family match those of theirmothers, and the degree to which family concordance in causal beliefs affects familyresolution of conflict. We suggest a ‘round-robin’ design of interview, in which eachfamily member reports their own anger-precipitating events and their perceptions ofother family members’ anger-precipitating events. We also suggest that future researchinclude asking participants to code their own responses to enhance researchers’ under-standing of the underlying structure of anger experiences.

Our next steps in investigating familial anger include examination of the experien-tial aspects of anger, for example, the parameters of intensity, frequency, and durationof anger within relationships and across the three components identified in our codingscheme (Halberstadt et al., 2000). In addition, we have compared the similarities anddifferences in the ways psychologists and laypersons understand and organize thecauses of anger (Carpenter & Halberstadt, 1996).

It is probable that patterns of anger events vary developmentally. Future researchcan expand upon our broad distinction of child versus parent by studying children ofvarying ages. We expect that (a) the experience of anger, like the experience of mostemotions, undergoes changes throughout development and (b) our rough comparisonbetween anger at parents versus anger at siblings may obscure subtle but rich differ-ences in how anger causes are perceived for younger, older, or same-aged siblings.And finally, now that we know that perceptions of anger causes vary across familyrelationships, we can begin examining how familial constructions and perceptions ofanger events influence children’s and parents’ relational outcomes, as they utilize thosebeliefs in understanding and responding to the behaviors of others in both family andnon-family settings.

References

Averill, J. R. (1979). Anger. In H. Howe (Ed.), Nebraska Symposium on Motivation (pp. 1–80).Lincoln: University of Nebraska Press.

© Blackwell Publishers Ltd. 2000 Social Development, 9, 4, 2000

Page 17: Mothers’ Reports of Events Causing Anger Differ Across Family Relationships

474 Sandra Carpenter and Amy G. Halberstadt

© Blackwell Publishers Ltd. 2000 Social Development, 9, 4, 2000

Averill, J. R. (1980). A constructivist view of emotion. In R. Plutchik & H. Kellerman (Eds.),Emotion: Theory, research, and experience (pp. 305–339). New York: Academic Press.

Averill, J. R. (1982). Anger and aggression: An essay on emotion. New York: Springer-Verlag,Inc.

Barrett, K. C. & Campos, J. J. (1987). Perspectives on emotional development II: A function-alist approach to emotions. In J. D. Osofsky (Ed.), Handbook of infant development, 2nd ed.NY: Wiley & Sons.

Brehm, J. W. (1999). The intensity of emotion. Personality and Social Psychology Review, 3,2–22.

Bronson, W. C. (1975). Developments in behavior with age mates during the second year oflife. In M. Lewis & L. A. Rosenblum (Eds.), Friendship and peer relations (pp. 131–152).New York: John Wiley & Sons.

Burrowes, B. D. & Halberstadt, A. G. (1987). Self- and family-expressiveness styles in theexperience and expression of anger. Journal of Nonverbal Behavior, 11, 254–268.

Camras, L. A. & Allison, K. (1989). Children’s and adults’ beliefs about emotion elicitation.Motivation and Emotion, 13, 53–69.

Carpenter, S. L. & Halberstadt, A. G. (1996). What makes people mad? Laypersons’ and psychologists’ categorisations of anger in the family. Cognition and Emotion, 10, 627–656.

Deaux, K. & Kite, M. E. (1987). Thinking about gender. In B. B. Hess & M. M. Ferree (Eds.),Analyzing gender: A handbook of social science research (pp. 92–177). Newbury Park, CA:Sage Publications.

Denham, G. & Bultemeier, K. (1993). Anger: Targets and triggers. In S. P. Thomas (Ed.), Womenand anger (pp. 68–90). New York: Springer Publishing Co.

Dunn, J. (1994). Experience and understanding of emotions, relationships, and membership ina particular culture. In P. Ekman & R. J. Davidson (Eds.), The nature of emotion: Funda-mental questions (pp. 352–355). New York: Oxford University Press.

Dunn, J., Brown, J., & Beardsall, L. (1991). Family talk about feeling states and children’s laterunderstanding of others’ emotions. Developmental Psychology, 27, 448–455.

Dunn, J. & Munn, P. (1987). Development of justification in disputes with mother and siblings.Developmental Psychology, 23, 791–798.

Dunsmore, J. C. & Halberstadt, A. G. (1997). How does family emotional expressiveness affectchildren’s schemas? In K. C. Barrett (Ed.), New directions for child development: The com-munication of emotion: Current research from diverse perspectives, 77, 45–68.

Edgerton, R. B. (1985). Rules, exceptions and social order. Berkeley: University of CaliforniaPress.

Fehr, B. & Baldwin, M. (1996). Prototype and script analyses of laypeople’s knowledge ofanger. In G. J. O. Fletcher & J. Fitness (Eds.), Knowledge structures in close relationships:A social psychological approach (pp. 219–245). Mahwah, NJ: Erlbaum.

Fehr, B., Baldwin, M., Collins, L., Patterson, S., & Benditt, T. (1999). Anger in close relation-ships: An interpersonal script analysis. Personality and Social Psychology Bulletin, 25,299–312.

Fincham, F. D. & Bradbury, T. N. (1987). Cognitive processes and conflict in close relation-ships: An attribution-efficacy model. Journal of Personality and Social Psychology, 53,1106–1118.

Fisher, C. B. & Johnson, B. L. (1990). Getting mad at mom and dad: Children’s changing viewsof family conflict. International Journal of Behavioral Development, 13, 31–48.

Fitness, J. (1996). Emotion knowledge structures in close relationships. In G. J. O. Fletcher &J. Fitness (Eds.), Knowledge structures in close relationships: A social psychologicalapproach (pp. 195–217). Mahwah, NJ: Erlbaum.

Fitness, J. & Fletcher, G. J. O. (1993). Love, hate, anger and jealousy in close relationships: Aprototype and cognitive appraisal analysis. Journal of Personality and Social Psychology,65, 942–958.

Garner, P. W., Jones, D. C., Gaddy, G., & Rennie, K. M. (1997). Low-income mothers’ con-versations about emotions and their children’s emotional competence. Social Development,6, 37–52

Gates, G. S. (1926). An observational study of anger. Journal of Experimental Psychology, 9,325–336.

Page 18: Mothers’ Reports of Events Causing Anger Differ Across Family Relationships

Anger in Families 475

Gergen, K. J. (1985). The social constructionist movement in modern psychology. AmericanPsychologist, 40, 266–275.

Gottman, J. M. (1990). How marriages change. In G. R. Patterson (Ed.), Depression and aggres-sion in family interaction (pp. 75–101). Hillsdale, NJ: Erlbaum.

Guthrie, D. M. & Noller, P. (1987). Spouses’ perceptions of one another in emotional situa-tions. In P. Noller & M. A. Fitzpatrick (Eds.), Perspectives on marital interaction (pp.153–181). Clevedon: Multilingual Matters Ltd.

Halberstadt, A. G., Carpenter, S. L., & Grotjohn, D. K. (2000). How mad are you? The fre-quency, intensity, duration, and onset latency of anger in family relationships. Manuscriptin progress.

Halberstadt, A. G., Fox, N. A., & Jones, N. A. (1993). Do expressive mothers have expressivechildren? The role of socialization in children’s affect expression. Social Development, 2,48–65.

Hall, G. S. (1898–1899). A study of anger. American Journal of Psychology, 19, 516–591.Hastings, P. & Grusec, J. E. (1997). Conflict outcome as a function of parental accuracy in per-

ceiving child cognitions and affect. Social Development, 6, 76–90.Hollingshead, A. B. (1975). Four factor index of social status. Unpublished manuscript, Depart-

ment of Sociology, Yale University, New Haven, Conn.Huston, T. L. & Robins, E. (1982). Conceptual and methodological issues in studying close

relationships. Journal of Marriage and the Family, 44, 901–925.Ishii-Kuntz, M. (1995). Paternal involvement and perception toward father’s roles: A compar-

ison between Japan and the United States. In W. Marsiglio (Ed.), Fatherhood: Contempo-rary theory, research, and social policy (pp. 102–118). Thousand Oaks, CA: Sage.

Jones, E. & Gallois, C. (1989). Spouses’ impressions of rules for communication in public andprivate marital conflicts. Journal of Marriage and the Family, 51, 957–967.

Lamb, M. E. (1987). The father’s role: Cross-cultural perspectives. Hillsdale, NJ: Erlbaum.Livesley, W. J. & Bromley, D. B. (1973). Person perception in childhood and adolescence.

London: Wiley.Mandler, G. (1975). Mind and emotion. New York: Wiley.Mandler, G. (1980). The generation of emotion: A psychological theory. In R. Plutchik & H.

Kellerman (Eds.), Emotion: Theory, research, and experience (pp. 219–243). New York: Academic Press.

Miller, P. & Sperry, L. L. (1987). The socialization of anger and aggression. Merrill-PalmerQuarterly, 33, 1–31.

Peterson, L., Ewigman, B., & Vandiver, T. (1994). Role of parental anger in low-income women:Discipline strategy, perceptions of behavior problems, and the need to control. Journal ofClinical Child Psychology, 23, 435–443.

Pleck, J. H. (1997). Paternal involvement: Levels, sources, and consequences. In M. E. Lamb(Ed.), The role of the father in child development, 3rd Edition (pp. 66–103). New York, NY:Wiley.

Richardson, R. F. (1918). The psychology and pedagogy of anger. Baltimore: Warwick & York.Saarni, C. (1985). Indirect processes in affect socialization. In M. Lewis & C. Saarni (Eds.),

The socialization of emotions (pp. 187–209). New York: Plenum Press.Scherer, K. R., Wallbott, H. J., & Summerfield, A. B. (1986). Experiencing emotion: A cross-

cultural study. Cambridge: Cambridge University Press.Selman, R. L. (1980). The growth of interpersonal understanding: Developmental and clinical

analyses. New York: Academic Press.Shantz, C. U. (1987). Conflicts between children. Child Development, 58, 283–305.Shaver, P., Schwartz, J., Kirson, D., & O’Connor, C. (1987). Emotion knowledge: Further explo-

ration of a prototype approach. Journal of Personality and Social Psychology, 52,1061–1082.

Smetana, J. G. (1989). Adolescents’ and parents’ reasoning about actual family conflict. ChildDevelopment, 60, 1052–1067.

Smetana, J. G. (1993). Understanding of social roles. In M. Bennett (Ed.), The development ofsocial cognition: The child as psychologist (pp. 111–141). New York, NY: Guilford Press.

Stein, N. L. & Levine, L. J. (1987). Thinking about feelings: The development and organiza-tion of emotional knowledge. In R. E. Snow & M. Farr (Eds.), Aptitude, learning and instruc-tion: Cognitions, conation, and affect (vol. 3, pp. 165–197). Hillsdale, NJ: Erlbaum.

© Blackwell Publishers Ltd. 2000 Social Development, 9, 4, 2000

Page 19: Mothers’ Reports of Events Causing Anger Differ Across Family Relationships

476 Sandra Carpenter and Amy G. Halberstadt

© Blackwell Publishers Ltd. 2000 Social Development, 9, 4, 2000

Stein, N. L., Liwag, M. D., & Wade, E. (1996). A goal-based approach to memory for emo-tional events: Implications for theories of understanding and socialization. In R. D.Kavanaugh, B. Zimmerman, & S. Fein (Eds.), Emotion: Interdisciplinary perspectives (pp. 91–118). Mahwah, NJ: Erlbaum.

Strayer, J. (1986). Children’s attributions regarding the situational determinants of emotion inself and others. Developmental Psychology, 22, 649–654.

Terwogt, M. M. & Harris, P. L. (1993). Understanding emotion. In M. Bennett (Ed.), The devel-opment of social cognition: The child as psychologist (pp. 62–86). New York: The GuilfordPress.

Vuchinich, S. (1987). Starting and stopping family conflicts. Journal of Marriage and theFamily, 49, 591–601.

Weiner, B. (1985). An attributional theory of achievement motivation and emotion. Psycho-logical Review, 92, 548–573.

Weller, S. C. & Romney, A. K. (1990). Metric scaling: Correspondence analysis. NewburyPark: Sage.

Notes

1. Our comparison of these four coding schemes across family relationships, based on five coders apply-ing the coding schemes to the same data set described in this paper, is available from the first author.

2. The temporal reliability and consensual validity of this modified version were assessed in an inde-pendent sample. Participants were women with 4- to 8-year-old children enrolled in a daycare center orafter school program. Across a 2-week time period, women (n = 52) consistently reported exactly the sameanger event 60% of the time, and reported the same type of anger event (receiving the same code in theeight-choice coding scheme developed for this article) 88% of the time. Female and male adults from the same family (n = 31 parent pairs) separately reported the frequent events of anger within their fami-lies; 56% of their reports about anger events in the family matched exactly, and 91% of their reportsreceived the same code in the eight-choice coding scheme. These data suggest good reliability and a sur-prising degree of agreement between mothers and their family members in their reporting about angerevents.

3. We initially tested these hypotheses using a series of simple chi-square analyses, but these were cum-bersome and repetitive when describing the interactions. For the sake of a clear and succinct descriptionof our results, we re-tested our hypotheses with maximum likelihood ANOVAs. Although an underlyingassumption of such analyses is that observations are independent, maximum likelihood ANOVAs are highlyrobust, and in this case conformed very closely to both the patterns and strength of the effects found inthe simple X2 analyses.

4. Interestingly, the combination of Specific Person-focused Goal Blockages (GPS) had frequencies of zero for each of the eight relationships. This combination of causes is therefore excluded from the analysis.

5. Throughout our description of these results, we are referring to coders’ classifications of mothers’reports.

6. An inspection of percent agreement between pairs of coders showed the presence of two groups ofcoders with reasonable agreement within the group, but divergence across groups. The percent agreementof pairs in one group was .79, .77, and .80; the percent agreement in the other group was .85. We there-fore re-tested the two significant effects for Specificity for the results from each coder group. That is, wetested whether the same findings would emerge when we examined the two sets of coders’ data separately.The main effect for experiencer was quite robust, regardless of which coder set was used. For the 3-coderteam, X2 = 209.09, p < .001, and for the two-coder team, X2 = 9.07, p < .01. The main effect for elicitorwas replicated by the 2-coder team, X2 = 5.58, p < .05, but did not achieve significance or near signifi-cance with the 3-coder team. These analyses suggest confidence in the predicted experiencer effect, butonly very tentative support for the elicitor effect.

Author Note

Sandra Carpenter, Department of Psychology, The University of Alabama, Huntsville; Amy G. Halber-stadt. Department of Psychology, North Carolina State University.

Funding for this project was provided by The University of Alabama in Huntsville mini-grant to thefirst author, and NIMH and NICHHD grants to the second author. We thank Walter Sullins of The Uni-versity of Alabama in Huntsville, David Dickey, Joy Smith, and Kim Eaton of North Carolina State University, and John Nesselroade of University of Virginia for their statistical advice and expertise. We also thank Sheri Clark for her help in gathering data at the day care center (independent sample in

Page 20: Mothers’ Reports of Events Causing Anger Differ Across Family Relationships

Anger in Families 477

Footnote 2), and the anonymous reviewers, Kim Eaton, and Nancy McElwain for their comments on pre-vious versions of the paper. The two authors contributed equally to this research; first authorship wasawarded on the basis of coin toss.

Correspondence concerning this article may be sent to Sandra Carpenter, Department of Psychology,The University of Alabama in Huntsville, Huntsville, AL 35899 ([email protected]) or to Amy Halberstadt, Department of Psychology, North Carolina State University, Raleigh, NC 27695-7801([email protected]).

© Blackwell Publishers Ltd. 2000 Social Development, 9, 4, 2000