Anger expression toward parents and depressive symptoms among undergraduates in Taiwan

http://tcp.sagepub.com/The Counseling Psychologist

http://tcp.sagepub.com/content/33/1/72The online version of this article can be found at:

DOI: 10.1177/0011000004270343

2005 33: 72The Counseling PsychologistHsiu-Lan Cheng, Brent Mallinckrodt and Li-Chuan Wu

Anger Expression Toward Parents and Depressive Symptoms Among Undergraduates in Taiwan

Published by:

http://www.sagepublications.com

On behalf of:

Division of Counseling Psychology of the American Psychological Association

can be found at:The Counseling PsychologistAdditional services and information for

http://tcp.sagepub.com/cgi/alertsEmail Alerts:

http://tcp.sagepub.com/subscriptionsSubscriptions:

http://www.sagepub.com/journalsReprints.navReprints:

http://www.sagepub.com/journalsPermissions.navPermissions:

http://tcp.sagepub.com/content/33/1/72.refs.htmlCitations:

What is This?

- Dec 1, 2004Version of Record >>

at UNIV OF TENNESSEE on March 30, 2014tcp.sagepub.comDownloaded from at UNIV OF TENNESSEE on March 30, 2014tcp.sagepub.comDownloaded from

http://tcp.sagepub.com/


http://tcp.sagepub.com/content/33/1/72

http://tcp.sagepub.com/content/33/1/72



http://www.div17.org

http://www.div17.org

http://tcp.sagepub.com/cgi/alerts

http://tcp.sagepub.com/cgi/alerts

http://tcp.sagepub.com/subscriptions

http://tcp.sagepub.com/subscriptions

http://www.sagepub.com/journalsReprints.nav

http://www.sagepub.com/journalsReprints.nav

http://www.sagepub.com/journalsPermissions.nav

http://www.sagepub.com/journalsPermissions.nav

http://tcp.sagepub.com/content/33/1/72.refs.html

http://tcp.sagepub.com/content/33/1/72.refs.html

http://tcp.sagepub.com/content/33/1/72.full.pdf

http://tcp.sagepub.com/content/33/1/72.full.pdf

http://online.sagepub.com/site/sphelp/vorhelp.xhtml

http://online.sagepub.com/site/sphelp/vorhelp.xhtml





10.1177/0011000004270343ARTICLETHE COUNSELING PSYCHOLOGIST / January 2005Cheng et al. / ANGER EXPRESSION AND DEPRESSIVE SYMPTOMSAnger Expression Toward Parents and DepressiveSymptoms Among Undergraduates in Taiwan

Hsiu-Lan ChengBrent Mallinckrodt

University of Missouri, Columbia

Li-Chuan WuNational Taiwan Normal University

This study surveyed 559 Taiwanese undergraduates regarding anger expression towardparents and depressive symptoms. Five modes of anger expression were explored. Allmodes of expression were either neutral or positively associated with depressive symp-toms. This result is contrary to studies of Western culture suggesting that some modes ofanger expression may be beneficial. However, a significant buffering interaction effectwas detected for women. Higher relative preference for outward verbal expression ofanger (Verbal Out) was associated with fewer depressive symptoms at increasing levelsof anger antecedents. For men, in contrast, greater preference for physical demonstra-tions of anger (Action Out) was associated with more depressive symptoms at increasinglevels of anger antecedents. Implications of these findings for U.S. counselors servingclients with Taiwanese cultural values are discussed.

Problems stemming from the maladaptive expression of anger are amongthe most common sources of emotional distress (Deffenbacher & Stark,1992; Greenberg, 2002). However, what qualifies as maladaptive expressionand how an individual experiences anger are affected by one’s cultural con-text (Lee, 1993; Sharkin, 1996). A failure to appreciate cultural and contex-tual influences on a client’s emotional expression and experience of angermay limit a counselor’s effectiveness. Developing a more sensitive apprecia-tion for cultural context is also in keeping with a recent call for international-izing counseling psychology (Leong & Ponterotto, 2003). Counseling pro-fessionals are urged to move beyond an appreciation of multiculturalism onlywithin the borders of the United States to instead develop a broader interna-tional perspective in their work (Leong & Blustein, 2000). Counselorsshould expose themselves to different worldviews and multiple research per-spectives to be prepared for work with an increasingly diversified and globalclientele (Leong & Ponterotto, 2003).

72

Correspondence concerning this article should be addressed to Brent Mallinckrodt, Departmentof Educational, School, and Counseling Psychology, 16 Hill Hall, University of Missouri,Columbia, Missouri, 65211. E-mail: [email protected].

THE COUNSELING PSYCHOLOGIST, Vol. 33 No. 1, January 2005 72-97DOI: 10.1177/0011000004270343© 2005 by the Society of Counseling Psychology.

at UNIV OF TENNESSEE on March 30, 2014tcp.sagepub.comDownloaded from



The current article responds to this call by acquainting counseling psy-chologists with processes of anger expression and depressive symptomsamong young adults in Taiwan. We hope the findings will be useful for U.S.readers in at least two ways: (a) Learning more about the psychology of per-sons in another culture is intrinsically beneficial for equipping counselingpsychologists to work in a global community, in keeping with the call tointernationalize our field; (b) although this study was not a direct multicul-tural comparison, learning about how young adults in Taiwanese culturecope with anger may provide a contrast that leads to a greater understandingfor U.S. readers about the process of coping with anger in their own culture.

Anger can be especially problematic in traditional Chinese culturebecause outward expressions of anger are strongly discouraged, particularlyanger directed toward parents by their children (Lin & Wang, 1995; O-Yang& Wu, 1998; Yu, 1987). A well-known saying in Taiwan can be translated as“Under heaven, no parent is ever wrong.” Children are expected to weigheach decision they make and each action they take with regard to the impact itwill have on their entire family. Filial piety is a strong value in traditional Chi-nese culture, requiring children to be at all times obedient and accepting oftheir parents’ decisions. Parenting practices in Taiwan are strongly influ-enced by these cultural values. Consequently, many parents behave in waysthat might be considered quite authoritarian by Western standards. For exam-ple, most Taiwanese parents expect unquestioning obedience in return for thesupport and love they give to their children. Parental control and authority areconsidered justified and are not open to challenge. Children are expected tofeel gratitude and to express only affection and admiration, never anger, totheir parents (Lin & Wang, 1995; O-Yang & Wu, 1998; Yu, 1987). However,in these circumstances, it cannot be doubted that children in Taiwan at leastoccasionally experience anger toward their parents, even if the emotion isconsidered quite unacceptable and must never be openly expressed.

Many studies conducted in Western cultures have found that repressedanger is associated with depressive symptoms (Biaggio & Godwin, 1987;Clay, Anderson, & Dixon, 1993; Moore & Paolillo, 1984; Riley, Treiber, &Woods, 1989). Other studies suggest that both repressed anger and outwardanger expression are associated with depressive symptoms (Deffenbacher etal., 1996). Some investigators have attempted to identify different styles ofsuppressing or expressing anger, together with different consequences thateach style may have for the development of depressive symptoms. For exam-ple, Kopper and Epperson (1996) conducted an exploratory factor analysisthat discovered three dimensions of anger expression in a sample of U.S.undergraduates: (a) aggressive acting out, (b) uncontrolled anger expression,and (c) anger suppression. Correlations with Beck Depression Inventory(BDI) scores were r = .20, r = .26, and r = .51, respectively, for the three

Cheng et al. / ANGER EXPRESSION AND DEPRESSIVE SYMPTOMS 73




dimensions. These findings suggest that anger suppression has the strongestassociation with depressive symptoms but that uncontrolled or aggressivestyles of expressing anger are also implicated. In this study, gender wasrelated to aggressive acting out and uncontrolled verbal and indirect expres-sion of anger, although gender was not associated with anger suppression.Men were more likely than women to express anger in aggressive oruncontrolled ways (Kopper & Epperson, 1996).

A comparison of Mexican American and White adolescents that focusedonly on outward expression of anger (Deffenbacher & Swaim, 1999) discov-ered three types of externalized anger expression: (a) verbal aggression (e.g.,cursing, being sarcastic), (b) physical aggression directed at other people(e.g., hitting, shoving, fighting), and (c) physical aggression directed towardthings in the environment (e.g., throwing things, slamming doors, stompingaround). Significant sex differences were observed. Compared with girls,boys reported significantly more aggression against people. Mexican Ameri-can adolescents of both sexes were less likely than White adolescents to useverbal aggression. All three types of expression were significantly associatedwith depressive symptoms for boys and girls in both ethnic groups.

One of the most widely used models of anger expression/suppression incontemporary U.S. studies was developed by Spielberger (1988) and wasoperationalized with the three Anger Expression scales of the State/TraitAnger Expression Inventory: (a) Anger Out, expressing anger toward otherpeople or objects in the environment; (b) Anger In, experiencing anger butattempting to suppress its outward expression; and (c) Anger Control,involving a struggle against both the subjective experience of anger (e.g.,efforts to calm oneself down) and the efforts to suppress any form of outwardanger expression. Using this formulation, Deffenbacher et al. (1996) foundthat scores on the BDI were significantly positively correlated with Anger Inand Anger Out but significantly negatively correlated with Anger Control.

Thomas and Williams (1990) urged a broader approach to anger researchthat moves beyond simply comparing Anger In versus Anger Out to include amore complete range of the diverse forms that anger expression may take.Practitioners of emotion-focused approaches to counseling (e.g., Greenberg,2002) and anger-reduction interventions (e.g., Deffenbacher, Thwaites,Wallace, & Oetting, 1994) emphasize that for a given client, some expres-sions of anger may be maladaptive, whereas other forms of expression areadaptive, depending on the nature of the client’s anger and the functional rolethat anger serves in a particular client’s context. Both individual differences(e.g., gender difference, gender-role socialization) and cultural differencesshould be taken into account when considering the adaptive nature of an indi-vidual’s anger and anger expression (Sharkin, 1996).

74 THE COUNSELING PSYCHOLOGIST / January 2005




It has long been argued that as a result of socialization patterns, men andwomen differ in their experience and expression of anger. For women, out-ward expression of anger may be seen as a defect in femininity and a threat tointerpersonal relationships, which therefore is strongly discouraged and seenas inconsistent with a feminine gender role (Collier, 1982). In the socializa-tion of men, in contrast, anger expression is perceived as more positive andacceptable, and anger tends to be seen as more congruent with the masculinegender role (Shields, 1987). Thus, anger suppression is an important part offemale gender identity and self-esteem development and may be responsiblefor the higher incidence of depressive symptoms in women than men(Kaplan, 1986). Empirical studies in Western cultures have found that menreport significantly more Anger Out expression than women, whereaswomen report more Anger In expression than men (Greenglass & Julkunen,1989; Iqbal & Ahmad, 1993). Cox, Stabb, and Hulgus (2000) found thatpreadolescent girls reported more anger suppression than boys, althoughanger suppression was not associated with depressive symptoms for eithersex. In contrast, Newman, Gray, and Fuqua (1999) found that although menand women did not differ in mean levels of Anger In expression, the correla-tion between Anger In and depressive symptoms was significantly higher forwomen than for men. Some studies report that gender-role characteristics,but not biological sex per se, contribute to different forms of anger expression(Kopper, 1993; Kopper & Epperson, 1991, 1996), although other researchhas failed to confirm this finding (Adams, 1997). Sharkin (1996) argued thatmen and women may experience different secondary feelings after a similarexpression of anger. In fact, Deffenbacher et al. (1996) found that womenwere more likely than men to report negative secondary emotions, such asshame and embarrassment, associated with expression of anger.

In addition to gender, cultural and contextual factors are of great impor-tance in shaping a person’s anger experience and expression (Sharkin, 1996).Lee (1993) argued that context is especially important when studying formsof anger expression among persons with traditional Chinese cultural values.For example, the cultural value of showing obedience and respect to one’sparents may prompt undergraduates in Taiwan to experience and expressanger toward parents very differently than U.S. undergraduates. Moreover,for persons in Taiwan, outward anger expression toward friends may be moreacceptable than toward one’s parents. Thus, Lee (1993) contended thatmodes of anger expression (i.e., Anger In or Anger Out) must be studied onlyin connection with the antecedents that give rise to the anger.

To more appropriately account for the Taiwanese cultural context ofanger, Lee (1993) developed an “anger process model” that incorporates theconcepts of anger antecedents and modes of anger expression into the assess-ment of anger. In Lee’s model, an anger antecedent is a specific class of inci-





dent experienced in a particular relationship domain (e.g., with friends, withparents, with teachers) that a respondent reports is especially likely to giverise to the experience of anger (e.g., “parents blamed me without figuring outthe facts”). After reporting the frequency of antecedents of a given type,respondents next report which combinations of the following five modes ofanger expression they are most likely to use following the specific anteced-ent: (a) Verbal Out, in which verbal expressions are used to directly conveyanger to the target person; (b) Action Out, in which physical actions or behav-iors are used to express anger; (c) Indirect Out, in which anger is communi-cated indirectly through sarcasm, cursing, or complaints expressed outsidethe presence of the target person; (d) Anger In, in which respondents allowthemselves to experience anger inwardly but suppress all forms of outwardexpression; and (e) Anger Control, in which in addition to suppressing alloutward expression of anger, respondents struggle to repress the experienceof anger itself. Note that even though Anger In and Anger Control are consid-ered modes of anger expression, strictly speaking, neither mode involves theoutward expression of anger.

Some readers may be struck by the fact that none of the five modes inLee’s model encompass what by Western standards would be considered“assertive,” “direct,” or “positive-adaptive” expression. However, Western-style assertiveness can be expected to occur with very low frequency when aTaiwanese young adult expresses anger to a parent because any direct out-ward expression of anger toward a parent is strongly prohibited and wouldmost likely have very negative consequences (Lin & Wang, 1995; O-Yang &Wu, 1998; Yu, 1987). The same forms of assertive expression that producepositive results for members of the dominant U.S. culture can be expected tohave far-reaching negative consequences when practiced in Asian cultures(Wood & Mallinckrodt, 1990). Offspring are likely to be considered selfish,inconsiderate, and disrespectful if they demonstrate any direct expression ofanger toward their parents. Thus, forms of direct expression of anger, whichmight be considered adaptive by U.S. standards, are not considered positiveor adaptive in Taiwanese culture and occur with very low frequency (O-Yang& Wu, 1998; Yu, 1987). Consequently, these positive or adaptive modes (byU.S. standards) were not included in Lee’s model.

Among the very few Taiwanese empirical studies using Spielberger’smeasurement model of anger, Zhong (1990) found that self-esteem was posi-tively associated with trait anger, Anger In and Anger Control, in Taiwanesecollege students. Xu (1996) found that in personal interactions, Taiwaneseundergraduates reported using Anger In and Anger Control to deal with theiranger, regardless of their self-esteem level; however, when it came to societaland political issues, undergraduates with high self-esteem reported moreAnger Out than students with low self-esteem. Important sex differences





have also been reported in Taiwanese samples. Zhong (1990) found that col-lege males reported more Anger Out expression, whereas females reportedsignificantly more Anger Control. The two sexes did not differ in their use ofthe Anger In expression. Lee (1993) found that college men and women didnot differ in the frequency that they became angry, but men reported signifi-cantly more Anger Indirect Out, Anger Control, and Anger Verbal Out thanwomen, whereas women reported more Anger In than men.

Although there are obvious similarities between Lee’s model and theWestern classification schemes for anger expression described previously,there are two important unique features. First, all of Lee’s five anger-expression modes are not captured entirely in any single Western model (e.g.,Deffenbacher & Swaim, 1999; Kopper & Epperson, 1996; Spielberger,1988, 1999). Second, because a respondent reports the frequency of angerantecedents in the context of a specific type of relationship, together with thefrequency of particular anger-expression modes for a given antecedent, a rel-atively complete picture of how anger is typically expressed and experiencedin different relationship domains is possible with Lee’s measurement model.For example, the mode of Indirect Out, which involves cursing and sarcasm,may be considered relatively more permissible to express anger felt toward afriend rather than toward a teacher or parent. However, some of the virtues ofLee’s instrument also introduce limitations. For example, using only context-dependent assessment means that if a respondent reports having experiencednone of the specific anger antecedents presented, then no useful data aboutanger expression can be collected. In addition, as with each of the other mod-els discussed, Lee’s model relies on self-report data and retrospectiveaccounts of behavior. The advantage in this approach is that much angerexperienced by Taiwanese subjects would not be visible to an externalobserver, but of course, the disadvantages include all the biases inherent inself-report data.

Lee’s (1993) full model also incorporates components of anger durationand intensity that were thought to be important for investigating associationsbetween chronic anger arousal and negative health outcomes such as highblood pressure. Because in the present study we were not interested in nega-tive health outcomes, we used a simplified version of Lee’s model thatexcluded the anger duration and intensity components. We were especiallyinterested in exploring whether, given the experience of high levels of angerantecedents involving parents, particular modes of anger expression tend toexacerbate or ameliorate depressive symptoms. Although a considerablenumber of Western studies have suggested a linkage between repressed angerand depressive symptoms (Biaggio & Godwin, 1987; Clay et al., 1993;Deffenbacher et al., 1996; Kopper & Epperson, 1996; Moore & Paolillo,1984; Riley et al., 1989), we questioned whether this linkage would be found





in Taiwanese students and, if so, whether there are important differencesbetween Anger In, in which affect is experienced—at least to some degree—but outward expression is suppressed, versus Anger Control, in which onestruggles to deny and repress the basic affective experience of anger itself.Similarly, we wondered if there were any forms of outward expression ofanger that might be beneficial for Taiwanese students.

Using the research literature reviewed above and our anecdotal clinicalexperience, we developed the model shown in Figure 1. Proposed causallinks in the figure are labeled with lowercase letters: (a) Anger antecedentsare positively associated with depressive symptoms; (b) particular angermodes are associated with depressive symptoms, irrespective of level ofanger antecedents toward parents, as long as at least some anger is reported;and (c) particular modes of anger expression serve as moderators of the rela-tionship between anger antecedent and depressive symptoms. Exploring thislast link involved examination of statistical interactions. Specifically, weinvestigated whether the strength of association between depressive symp-toms and parental anger antecedents varied as a function of a student’s prefer-ence for particular anger modes of expression. We explored paths b and c asexploratory analyses without specific hypotheses because there are not yetsufficient studies of Taiwanese subjects to allow predictions about specificmodes of anger expression. Finally, because of the sex difference reported inboth U.S. and Taiwanese studies, we investigated each of the proposed pathsin Figure 1 separately for men and women.

METHOD

Participants

Undergraduates were recruited from a variety of departments in nine uni-versities and colleges in Taipei Metropolis (Taipei City and Taipei County).Deliberate steps were taken to sample a wide variety of departments acrossthese institutions. The sample was drawn from classes in Chinese literature,education, mathematics, political science, physics, social work, electricalengineering, medicine, fine arts, and accounting, among others. Divorce isrelatively rare among the parents of Taiwanese college students. As the pack-ets were distributed, students were instructed that if their parents weredivorced or if one or more parents were deceased, they should return a blanksurvey in their envelope. Data were not solicited from students whose parentswere divorced or deceased to eliminate the following potential confounds:(a) The introduction of a step-parent adds a third, or perhaps fourth, source ofanger antecedents, whereas other students have at most only two sources; (b)






a deceased parent, or a divorced parent who abandons the family, eliminates atarget for expressing anger. Thus, a packet could be returned without a com-pleted survey inside because a student declined to participate or because hisor her parents were divorced or deceased.

A total of 883 survey packets were distributed, with 763 (86%) completedsurveys returned. For reasons explained in greater detail in the Results sec-tion of this article, data from the 204 (27%) of the students who reported vir-tually no angry interactions with their parents were excluded from furtheranalyses. The remaining sample contained 220 (39%) males and 339 (61%)females. Data for age ranges rather than specific age were solicited. Thebreakdown for age was as follows: younger than 18 (2%), 18 to 20 years old(50%), 20 to 22 years old (40%), 22 to 24 years old (6%), and older than 24(2%). Regarding class level, 25% were freshman, 36% were sophomores,38% were juniors, and one student (0.2%) was a senior. The culture in Tai-wanese colleges and universities is virtually monoethnic. Thus, questionsabout ethnic identification typical of U.S. surveys would not have beenappropriate for this context.

Instruments

BDI. The BDI (Beck, Ward, Mendelson, Mock, & Erbaugh, 1961) is oneof the most frequently used self-report measures of depressive symptoms. Itconsists of 21 items, each containing four self-report statements representing

Anger Antecedents

Depression(a)

(b)

(c)

Modes of Anger Expression

-- verbal out -- action out -- indirect out -- anger in -- anger control

Figure 1. Anger process causal model.





a cluster of symptoms. The statements reflect different cognitive-behaviorallevels of depressive symptoms and are scored on a 0 to 3 scale, with higherscores indicating more depressive symptoms. The Chinese translation of theBDI used in this study was developed by Ko (1989) using back-translationmethods and a team of bilingual experts who refined the scale item by item.Ko (1989) presented evidence of validity in that the translated BDI success-fully differentiated groups of hospitalized patients based on levels of depres-sive symptoms assessed independently through personal interviews. Split-half reliability, after applying the Spearman-Brown correction, was reportedto be .94 in a Taiwanese sample. Huang (1992) reported that internal consis-tency reliability (coefficient alpha) was .83, and retest reliability (3-weekinterval) was .83 in a sample of Taiwanese respondents. In the current study,the coefficient alpha was .86. To our knowledge, Taiwanese researchers havenot developed an indigenous Chinese-language measure of depressive symp-toms but instead rely on translations of English-language measures. The BDIis by far the most frequently used Chinese-language measure of depressivesymptoms (Chan, 1991). It was selected for this study because it captures thecognitive as well as the affective components of depressive symptoms thatmay be especially associated with anger.

Two items of the BDI assess interpersonal irritation (Item 11) and thebelief that one is being punished (Item 6). To examine the possibility that pos-itive correlations between depressive symptoms and anger toward parentswere due to conceptual overlap, all of the correlation and multiple regressionanalyses reported in this study were run again after removing Items 11 and 6and replacing the values with the mean of the remaining 19 items. In mostcases, the regression coefficients were higher in these parallel analyses thanfor the full BDI scale. In no case was a statistically significant result obtainedwith the full BDI that was not statistically significant when the reduced scalewas used.

Lee Anger Process Questionnaire (LAPQ). The LAPQ (Lee, 1993) pres-ents respondents with anger-antecedent scenarios determined by previousresearch to be common situations likely to produce experiences of anger inTaiwanese young adults. Lee conducted a pilot qualitative study in which 120Taiwanese college students were asked to describe the most frequentlyoccurring situations in daily interactions likely to produce feelings of anger.Although separate sets of anger scenarios were developed from these qualita-tive responses for the relationship domains of friends, teachers, and otherrelationships, only the four scenarios developed from anger-producing inter-actions with parents were used in this study. The specific anger antecedentspresented as stimulus items in the LAPQ were as follows: (a) “My parents




misunderstood me and blamed me without figuring out the facts”; (b) “Myparents did not agree with my opinions and kept nagging me or forced me todo things their way”; (c) “I thought I did a good job, but my parents did notrecognize and praise me and instead continued to criticize me”; and (d) “Myparents corrected my behavior, gave me lessons, criticized, or punished me infront of other family members or friends.” Table 1 provides an example itembased on the first scenario from the portion of the LAPQ used in this study.For each scenario, students indicated how often this general class of situationoccurred in the past year, using a 4-point scale (0 = almost never happened;1 = sometimes happened; 2 = often happened; 3 = always happened). A totalparental anger-antecedent score was calculated from values summed acrossthe four scenario frequency ratings (Item 1 shown in Table 1) and thereforecould range from 0 to 12. Lee (1993) reported that retest reliability (2-weekinterval) for the total parent anger-antecedent scale was .83.

After the antecedent frequency Question 1, a second question was pre-sented in each scenario asking, “When you felt angry in the above situations,how did you deal with your anger?” Labels for each of the five anger modeswere then presented. To provide an operational definition of each angermode, a list was attached to provide concrete examples of each mode. Defini-tions for the modes and sample-stimulus examples presented to respondentsfollow: (a) Verbal Out, “Use verbal expressions to convey one’s anger to theperson whom one is angry with”; (b) Action Out, “Use physical actions/behaviors to deal with one’s anger” (e.g., “bang the door,” “throwingobjects”); (c) Indirect Out, “Use sarcasm, curses, or covert objections toexpress one’s anger outside the presence of the person with whom one isangry” (e.g., “complain to others about the faults of the person I am angrywith”); (d) Anger In, “Put your anger inside your heart without expressingit”; and (e) Anger Control, “Try to stay calm or take it easy to deal with one’sanger” (e.g., “calm down very quickly,” “pretend there is nothing happeningand get away from the situation”). After each of the five anger modes, stu-dents rated the frequency that they used this mode (in this scenario) on a 4-point scale (0 = almost never happened; 1 = sometimes happened; 2 = oftenhappened; 3 = always happened). The five anger modes were developed byLee from open-ended questions that asked students (n = 135) how they usu-ally respond to their anger. Their responses were then converted to items thatwere presented to a second sample (n = 650), factor analyzed, and condensedto form the five response modes and attached list of examples.

It should be emphasized that the LAPQ does not use a forced-choice for-mat. Rather, it asks respondents to report their frequency of using each modein each anger-antecedent scenario. Thus, the sum of the frequency scores fora given mode could range from 0 to 12. A score of 12 would be assigned in the





case of a student who reported that this particular mode of anger expressionwas “always” used in every one of the four scenarios. Because the LAPQdoes not force respondents to choose only one mode for a given anger ante-cedent, hypothetically, scores for two or more modes could equal 12, pro-vided that two or more modes are reported to be used “always” when each ofthe four scenario occurred. In the present study, internal consistency reliabil-ity (coefficient alpha) for the four items summed to give a specific angermode score ranged from .72 (Verbal Out) to .83 (Anger Control). Theserelatively large coefficients indicate that if a mode is used for one anger-antecedent scenario, a student tended to use it for other scenarios as well.Lee’s model has been used in a number of studies in Taiwan. Evidence ofvalidity is provided by associations of particular anger modes in the theoreti-cally predicted directions with self-esteem (Wang, 2002), locus of control(Xu, 1996), and heart disease (Gao, 1997) in adults, as well as with depres-sive symptoms and anxiety in middle school students (Shieh, 2001).


TABLE 1: Sample Item From the Portion of the Lee Anger Process Questionnaire Usedin This Study

(A) My parents misunderstood me and blamed me without figuring out the facts.1. In the past year, how often did the situation mentioned above occur?

(0) almost never happened(1) sometimes happened(2) often happened(3) always happened

2. If you felt angry in the above situations, how did you deal with your anger? (please seethe attachment to find examples for each way a person can express anger)

(a) Used verbal expressions:(0) almost never (1) sometimes (2) often (3) always

(b) Used physical actions to express your anger(0) almost never (1) sometimes (2) often (3) always

(c) Use indirect outward expressions(0) almost never (1) sometimes (2) often (3) always

(d) Put your anger inside your heart without expressing it (AI):(0) almost never (1) sometimes (2) often (3) always

(e) Control your anger (CO):(0) almost never (1) sometimes (2) often (3) always

NOTE: The sequence of Questions 1 and 2 above, with all the subsections, was repeated for threeother anger antecedents in addition to the “A” scenario listed above: (B) “My parents did notagree with my opinions, and kept nagging me or forced me to do things their way.” (C) “I thoughtI did a good job, but my parents did not recognize and praise me, and instead continued to criti-cize me.” (D) “My parents corrected my behavior, gave me a lesson, criticized, or punished me infront of other family members or friends.”




Procedure

Instructors of courses known to have large enrollments of undergraduateswere contacted at colleges and universities in the Taipei metropolitan area.All but a few of those contacted agreed to cooperate and to make time avail-able for students to complete the survey materials in class. No special incen-tive was offered to participants. (It is common for instructors in Taiwan toagree to such requests and for students to agree to complete surveys withoutspecial incentives.) On their own initiative, some instructors offered extracredit for participation, and a few other instructors asked for a summary ofthe results to be provided to their students when the study was completed. Allstudents in the classes of cooperating instructors received a packet containingthe survey, an answer sheet that could be optically scanned, and a cover sheetassuring them that their responses would remain completely anonymous. Nopersonally identifying information was solicited on the survey answer sheet.Thus, although we later became concerned about some of the very high BDIscores in the returned surveys (18% of the sample scored higher than 16, and6% scored higher than 25), there was no way to contact these students indi-vidually to refer them for counseling services. Surveys were distributed andcollected by researchers during the same class period. Students who wishednot to participate or whose parents were no longer married to each other(deceased or divorced) returned surveys uncompleted.

RESULTS

A shortcoming of the LAPQ is that the 0 point for all the rating scales isanchored by almost never. Thus, a student who absolutely never experiencedone of the scenarios, and never used a particular anger mode, is neverthelesscompelled to answer “almost never” in each circumstance. Despite the care-ful steps taken in developing the LAPQ to include scenarios depicting themost frequent types of conflicts with parents, in our sample, 204 students(27%) answered “almost never” to every one of the four anger scenarios,resulting in a score of 0 for this variable. This group included 88 men (43%)and 116 women (57%). Most of these students also gave an answer of“almost never” in response to each question about the frequency of angermodes. Although it is likely that an appreciable portion of these studentsactually experienced one or more of the four anger scenarios and eitherunconsciously repressed the experience or consciously denied it whenresponding to our survey, it is also possible that a significant proportion ofthese students never actually experienced any of these four situations withtheir parents. In Taiwanese culture, this result is quite expectable, given the





strong cultural prohibition against anger toward one’s parents. Because wehad no way of discriminating between these two possibilities and becausestudents who chose the 0 point for every question have no within-subjectvariability, we decided to exclude these students for every anger scenario.Subsequent analyses were based on the remaining 559 students.

We offer an analogy to further support our decision. Consider a surveystudy that presented participants with four possible property crimes, fol-lowed by “How often has this type of crime happened to you?” This is thenfollowed by five possible responses to being the victim of a property crime—not hypothetically how might you respond but rather “how did you respondwhen the crime happened to you?” For a person who had never been the vic-tim of any of the four crimes, the most accurate response would be to leavethe items blank or, if directed to answer every item, to choose the 0 point ofthe response scale for each crime scenario and each response mode. How-ever, if 27% of respondents have never experienced any of the four crimes,choose the 0 point for each possible crime response, and are still included inthe analyses, the results of a study that claimed to report “the ways peopletypically respond to property crime” would be skewed. We contend that inthis example, and for our study, the 0 point of the scenario scale does not rep-resent the lowest point on a continuum but rather indicates respondents whoare members of a different population than the remainder of the sample andtherefore should be excluded from the analyses.

Before exploring the proposed relationships shown in Figure 1, sex differ-ences were investigated using a one-way multivariate analysis of variance(MANOVA) to compare men and women with regard to anger antecedentlevel, mode of anger expression (five variables), and depressive symptoms.Results indicated that there were significant differences, F(7, 551) = 5.39; p <.001. Table 2 shows the results of univariate follow-up analyses. Comparedwith women, men reported a much higher preference for the Anger Controlmode of expression. There were no sex differences in depressive symptoms.

Because the LAPQ relies on sets of four single items to assess anger-expression modes and anger antecedents, structural equation model analysesare not appropriate for these data (Ullman, 2001). Instead, we began ourexploration of the relationships shown in Figure 1 by calculating Pearsoncorrelations between anger antecedents, anger modes of expression, anddepressive symptoms. Results of these analyses are shown in Table 3. Coeffi-cients for women are shown above the diagonal, and coefficients formen appear below the diagonal. As expected, anger antecedents weresignificantly positively associated with depressive symptoms for women (r =.26, p < .001) and for men (r = .39, p < .001). Table 3 shows associationsbetween use of particular anger modes and depressive symptoms, with-out regard to the level of anger antecedents. Thus, these analyses explore path





b of the model shown in Figure 1. Table 3 shows that the mode of Anger Inhad the numerically highest positive association with depressive symptomsof any mode for both women and men, (r = .24 and r = .23, respectively).Consistent with Taiwanese cultural values, every statistically significantassociation between a mode of anger expression and depressive symptomswas a positive correlation. Table 3 also shows correlations between angermodes for men and women. As might be expected, subscales measuring themost direct outward forms of anger expression, Verbal Out and Action Out,were generally negatively correlated with subscales designed to measuresuppression (Anger In) and repression (Anger Control). In fact, the Anger Inand Anger Control subscales were only moderately correlated with eachother (r = .44 for men, r = .32 for women), indicating that these subscalesmeasured relatively distinct aspects of students’ attempts to refrain fromexpressing anger directly. All correlations were less than .44, findings thatsupport the conclusion that the LAPQ measures five differentiated modes ofexpressing anger.

The final set of analyses was used to explore the relationships representedby path c in Figure 1. Whereas the previous bivariate correlations were calcu-lated irrespective of the level of anger antecedents, testing path c involvedexploring statistical interactions between anger antecedents and a particularmode of anger expression in connection with depressive symptoms. Beforethe analyses could be conducted, a difficulty needed to be addressed in thatraw scores for each of the five anger modes, taken individually, do not pro-vide the most useful information about relative preference for a particularmode. For example, Student X, whose score for Indirect Out is 4 (on a 0 to 12scale), may appear identical to Student Y, whose score for Indirect Out is also


TABLE 2: Sex Differences in Anger Process and Depression

Men Women(n = 220) (n = 339)

Variable M SD M SD F(1, 761)

Anger antecedents 2.56 1.90 2.55 1.66 0.02Anger-expression mode

Verbal Out 4.39 2.92 4.61 2.73 0.76Action Out 1.76 2.27 1.85 2.17 0.25Indirect Out 3.32 2.86 3.61 2.63 1.52Anger In 4.39 3.11 4.83 2.90 2.82Anger Control 4.92 3.39 3.66 2.91 21.73***

Beck Depression Inventory 11.22 8.77 10.70 7.40 0.56

NOTE: MANOVA for this analysis, F(7, 551) = 5.39, p < .001.***p < .001.




86

TA

BL

E 3

:C

orre

lati

ons

Am

ong

Ang

er A

ntec

eden

ts, M

odes

of

Ang

er E

xpre

ssio

n, a

nd D

epre

ssio

n

Ang

er E

xpre

ssio

n

Bec

kD

epre

ssio

nA

nger

Ant

eced

ents

Verb

al O

utA

ctio

n O

utIn

dire

ct O

utA

nger

In

Ang

er C

ontr

olIn

vent

ory

Ang

er a

ntec

eden

ts—

.09

.15*

.24*

**.2

1***

.05

.26*

**A

nger

exp

ress

ion

Ver

bal O

ut.2

7***

—.1

0.0

1–.

18**

*–.

23**

*–.

12*

Act

ion

Out

.18*

*.3

2***

—.1

8***

–.06

–.16

**.1

2*In

dire

ct O

ut.1

9**

.17*

.39*

**—

.21*

**–.

07.1

4**

Ang

er I

n.1

6*–.

24**

*–.

04.0

9—

.32*

**.2

4***

Ang

er C

ontr

ol.0

7–.

16*

–.16

*–.

05.4

4***

—.0

6D

epre

ssio

n.3

9***

.13

.20*

**.0

7.2

3***

–.02

—

NO

TE

: Cor

rela

tions

for

wom

en (

n=

339

) ar

e sh

own

abov

e th

e di

agon

al, a

nd m

en (n

= 2

20)

are

show

n be

low

the

diag

onal

.*p

< .0

5. *

*p<

.01.

***

p<

.001




4. However, if Student X’s scores for the other four modes are all greater than7, whereas Student Y’s scores for the other four modes are all less than 2, therelative preference of these two students is obviously quite different. Toaddress this problem, anger-expression scores were converted to proportionsprior to analyzing statistical interactions. For a given student, the frequencyof a particular anger mode served as the numerator of this ratio, with the totalfrequency of all anger modes used as the denominator. Thus, higher scoresrepresented a greater preference for a given anger-expression mode relativeto any other mode. Among women, the mean proportions of anger-expression preferences were as follows: Verbal Out = .27; Anger In = .26;Anger Control = .20; Indirect Out = .19; and Action Out = .09. Among men,the relative preferences were as follows: Anger Control = .26; Verbal Out =.25; Anger In = .23; Indirect Out = .17; and Action Out = .09.

Previous research on sex- and gender-role differences in anger expres-sion, plus the mean sex differences shown in Table 2 and the pattern of differ-ent correlations for men and women shown in Table 3, suggests the possibil-ity that interactions between anger antecedents and expression modes mightbe significantly different for men and women. Therefore, we began ourinvestigation of path c in Figure 1 by conducting a series of hierarchical mul-tiple regression analyses to examine the three-way interactions between sex,level of anger antecedents, and mode of anger expression in predictingdepressive symptoms. Following the recommendations of Cohen, Cohen,West, and Aiken (2003), we used “effects” dummy coding for the variable ofsex, with men coded as –1 and women as +1. To avoid problems withmulticolinearity, we used recommended procedures to “center” the other twoindependent variables by subtracting the respective group means of angerantecedents and modes of anger expression from each student’s individualscore prior to analysis. Next, in the first step of each hierarchical multipleregression, the zero-order variables were entered, sex (dummy coded),anger-antecedent level, and mode of anger expression. In the second step, 3two-way interactions were entered consisting of the product of Sex × AngerAntecedents, Sex × Anger Mode, and Anger Antecedents × Anger Mode. Inthe final step, the three-way interaction of Sex × Anger Antecedents × AngerMode was entered. If the increment in variance for depressive symptomsassociated with this third step is statistically significant, an inference of sig-nificant three-way interactions among the independent variables is warranted(Aiken & West, 1991; Cohen et al., 2003).

Of the 5 three-way interactions conducted (one for each mode of angerexpression), results suggested significant Sex × Antecedent × Mode Interac-tions for two modes, Verbal Out, change in R2 = .01, F(1, 551) = 6.70, p < .01;and Action Out, R2 = .01, F(1, 551) = 3.94, p < .05. Cohen et al. (2003) recom-mend exploring three-way interactions by examining two-way interactions





at fixed values of the third independent variable. Thus, the logical next step inthis study was to examine two-way interactions separately for women andmen. One set of follow-up analyses was conducted for the Verbal Out mode,with a second set conducted for the Action Out mode. In each hierarchicalregression, depressive symptoms served as the criterion variable, and anger-antecedent level (centered as before) was entered in the first step. In the sec-ond step (which differentiated each analysis), preference for the centeredVerbal Out or Action Out expression mode was entered. Finally, in the thirdstep, the product of anger antecedents and preference for the particular angermode was entered to test for interaction effects. After accounting for simpledirect effects in the first two steps, a statistically significant increment in R2

for the third step indicates a significant interaction between anger anteced-ents and relative preference for a particular anger mode. Of the four hierar-chical regression analyses conducted (Verbal Out and Action Out for menand for women), two met this test for significant interaction effects. Table 4shows results for the significant interaction for women involving a prefer-ence for the Verbal Out mode and for men involving the Action Out mode.These statistically significant interactions accounted for an increment ofabout 2% in the explained variance in depressive symptoms. Taken together,the entire regression model accounted for 12% of the variance in depressivesymptoms for women and 18% of the variance in depressive symptoms formen.

Using procedures recommended by Frazier, Tix, and Barron (2004) andby Jaccard, Turrisi, and Wan (1990), we constructed graphs of the two signif-icant interactions to examine the precise nature of the moderating effects.Separate regression equations were constructed for high (one standard devia-tion above the mean) and low (one standard deviation below the mean) levelsof anger mode preference. Anger antecedent levels of 1 and 10 (from the 0- to12-point scale) were then entered into these equations to plot the values ofdepressive symptoms shown in the panels of Figure 2. Panel A shows that forwomen, higher use of the Verbal Out mode of anger expression was associ-ated with a lower rate of increase in depressive symptoms for each unitincrease in anger antecedents. In contrast, women who had a relatively lowerrelative preference for the Verbal Out mode (dashed line) experienced greaterincrements in depressive symptoms for each unit increase in parental angerantecedents. This type of moderator interaction is consistent with a “buffer-ing effect” for the Verbal Out mode among women because as anger anteced-ents increase, a high level of using this mode appears to protect women fromthe otherwise greater increments in depressive symptoms for each unitincrease in anger antecedents reported by women who do not prefer to use theVerbal Out mode.





89

TA

BL

E 4

:H

iera

rchi

calM

ulti

ple

Reg

ress

ions

toE

xam

ine

Inte

ract

ions

ofA

nger

Ant

eced

ents

and

Mod

esof

Exp

ress

ion,

Wit

hD

epre

ssio

nas

Cri

teri

on

Fin

al M

odel

Coe

ffici

ents

Step

/Var

iabl

eR

2C

hang

e in

R2

FC

hang

eB

SE B

Ana

lysi

s 1:

For

wom

en (

n=

339

), V

erba

l Out

mod

eSt

ep 1

. Ang

er a

ntec

eden

ts.0

7.0

724

.20*

**0.

980.

23.2

19**

*St

ep 2

. Ver

bal O

ut m

ode

.10

.04

13.5

4***

–9.2

12.

21–.

219*

**St

ep 3

. Ant

eced

ents

×M

ode

inte

ract

ion

.12

.02

7.20

**–3

.62

1.35

–.14

2**

Ana

lysi

s 2:

For

men

(n

=22

0), A

ctio

n O

ut m

ode

Step

1. A

nger

ant

eced

ents

.16

.16

39.8

7***

1.74

0.29

.377

***

Step

2. A

ctio

n O

ut m

ode

.16

.00

1.77

8.94

5.50

.102

Step

3. A

ntec

eden

ts×

Mod

e in

tera

ctio

n.1

8.0

24.

47*

6.70

3.17

.132

*

NO

TE

: The

se a

naly

ses

wer

e pe

rfor

med

on

subs

ets

of m

en a

nd w

omen

who

rep

orte

d at

leas

t one

ang

er a

ntec

eden

t.*p

< .0

5. *

*p<

.01.

***

p<

.001

.




In contrast, the remaining significant interaction shown in Panel B of Fig-ure 2 suggests that for men, a higher preference for the Action Out anger-expression mode was associated with greater increments in depressive symp-toms as anger antecedents increased. Men who had a low relative preferencefor the Action Out mode experienced lesser increments in depressive symp-toms for each unit of increase in anger antecedents. Note that especially formen, the slopes of these regression lines are so steep that high combinationsof anger antecedents and Action Out modes correspond to levels of BDIscores far above the cutoff of 16, representing serious depressive symptomsfor Western samples.

DISCUSSION

This study was intended to test the model of anger process and depressivesymptoms for Taiwanese young adults shown in Figure 1. In general, resultsprovided relatively strong support for the proposed links a and b, with moremarginal support for the proposed moderating path c. Specifically, parentalanger antecedents were positively and strongly associated with depressivesymptoms for both men and women. This finding is in accordance with the


A. Women, Ve rba l Out Mode

High Ange r Mode

Low Ange r Mode

B. Men, Anger Out Mode

Ange r Antecedents

De

pre

ssi

on

10

20

30

2 4 6 8 10 12

Ange r AntecedentsD

ep

res

sio

n

10

20

30

2 4 6 8 10 12

Figure 2. Significant interactions of anger modes with anger antecedent level in predictingdepression.




clinical experience of the first and third authors, who have worked with manyTaiwanese adolescents and young adults who experienced considerabledepressive symptoms together with the anger and frustration they felt towardtheir parents. In light of these findings, we turned our attention with height-ened interest to examining whether any particular mode of anger expressionmight be beneficial, that is, path c of our model. Path c involves interactionsbetween anger antecedents and five modes of anger expression in predictingdepressive symptoms. The five modes of anger expression were (a) VerbalOut, (b) Action Out, (c) Indirect Out, (d) Anger In, and (e) Anger Control.Many theoretical systems of Western psychotherapy—for example,supportive-expressive (Greenberg, 2002), cognitive-behavioral (Deffen-bacher & Stark, 1992), or psychoanalytic (Alexander & French, 1948)—hold that the adaptive expression of anger is beneficial and can be expected tolead to lower levels of depressive symptoms.

However, results of bivariate correlations in our study (shown in Table 3)suggest that none of the five anger-expression modes measured by the LAPQwere negatively associated with depressive symptoms. All the significantcorrelations between anger modes and depressive symptoms for men andwomen shown in Table 3 were positive. Thus, as Taiwanese cultural normspredict, both suppression of anger (Anger In) and forms of outward expres-sions of anger (Action Out) were associated with depressive symptoms. Thefinding that Anger In was significantly associated with depressive symptomsis congruent with studies of Western samples (Biaggio & Godwin, 1987;Clay et al., 1993; Moore & Paolillo, 1984). Recall that Anger In involves apersisting subjective experience of the emotion and a simultaneous struggleto suppress any outward manifestation of anger. Similarly, Deffenbacher etal. (1996), using a different measure of anger expression, found that bothAnger Out and Anger In were positively associated with depressive symp-toms. In contrast, Anger Control in Lee’s model involves efforts to quell thesubjective experience of anger itself, as well as inhibit any expression ofthese feelings. Our findings suggested that Anger Control was not signifi-cantly positively or negatively associated with depressive symptoms. In oneU.S. study, Anger Control was found to be negatively associated withdepressive symptoms (Deffenbacher et al. 1996).

We were especially interested in the interaction relationships representedby path c in our model because we hoped to discover modes of anger expres-sion that interacted with high levels of antecedents, such that use of a particu-lar anger mode might buffer the otherwise high levels of depressive symp-toms associated with high anger antecedents. Unfortunately, of the twostatistically significant interactions, only one exhibited this type of positivebuffering pattern. Specifically, our findings suggest that—comparing acrosswomen—as anger antecedents increase, so do depressive symptoms, but the





increase in depressive symptoms associated with a given increase in angerantecedents is much less for women who have a strong preference for directverbal expression of anger to their parents (Verbal Out). In other words, ourresults suggest that for any pair of women with the same level of anger ante-cedents, the one who has a higher preference for the Verbal Out mode willtend to experience fewer depressive symptoms than her counterpart.

In contrast to this buffering effect, for men, the only significant interactionsuggested that a relative preference for the Action Out mode was likely tomake their situation worse. That is, for any pair of men with the same level ofanger antecedents, the one who has a higher relative preference for the ActionOut mode tends to report much higher levels of depressive symptoms than hiscounterpart. It must be emphasized that these explorations of moderatoreffects are based on relative preference for a particular expression mode.Thus, the significant interactions are observed for women who report a highpreference for the Verbal Out mode and for men who report a high preferencefor the Action Out mode, in both cases to the relative exclusion of the otherfour modes.

Findings suggesting that a relative preference for Verbal Out mode rela-tive to all other modes may be beneficial for women but not for men might bedue to different expectations of males and females in traditional Chinese orTaiwanese culture with regard to the experience and expression of emotions.In Confucian teaching, men earn respect through self-discipline and controlof emotions, especially negative emotions. Women are not expected to exertan equally high degree of emotional self-control in Taiwanese culture, andsomewhat more emotional expression is considered allowable for them. Tra-ditionally, Confucius’ teachings applied only to men. For example,

The superior man . . . in regard to his countenance, he is anxious that it shouldbe benign. In regard to his demeanor, he is anxious that it should be respectful.When he is angry, he thinks of the difficulties his anger may involve him in.

Thus, for men, verbally expressing anger is considered a lack of self-discipline, whereas the same behavior may not be equally shameful forwomen. A parallel in Western culture may be the observation that many cli-ents experience two other secondary emotions in quick succession after anexperience of intense anger, namely, sadness and shame (Greenberg, 2002;Teyber, 2000). Perhaps a similar process in Taiwanese young adults results indepressive symptoms and feelings of shame following an expression ofanger, especially the Action Out type of expression for men.

Results shown in Table 4 indicate that the two significant interactionsaccounted for about 2% of the variance in depressive symptoms, from a totalof 12% of the variance in depressive symptoms accounted for in women and





18% in men. At first blush, identifying variables that account for 2%, or even12%, of the variance in depressive symptoms may not seem very practicallysignificant (even if statistically significant). However, a perspective on therelative importance of these effects may be seen by considering that for twoyoung women experiencing anger antecedents at about two standard devia-tions above the mean (i.e., antecedents = 6.0), Panel A of Figure 2 shows thatthe difference between a high and low preference for the verbal out mode(i.e., one standard deviation above the mean vs. one standard deviation belowthe mean) corresponds to a difference in BDI scores of 10.94 versus 22.02.Considering the recommended cutoff score of 16 for Western samples,results comparable to ours for a U.S. sample of undergraduate women wouldmean that high versus low preference for the Verbal Out mode corresponds tothe difference between relatively functional versus fairly serious levels ofdepressive symptoms.

With 12% to 18% of the variance in depressive symptoms accounted forby the variables we measured in this study, it may be interesting to speculateabout what unmeasured factors might account for the remaining variance.First, it should be emphasized that we assessed only four specific antecedentsfor anger in the parent-child relationship. Other antecedents with parents, orperhaps anger in other relationships (e.g., romantic partners, teachers,friends, coworkers), may account for additional variance in depressive symp-toms. Life in Taipei City, one of the largest and most crowded cities in theworld, can be difficult and demanding. Undergraduate students throughoutthe world, although perhaps especially in the Taiwanese higher educationalsystem, face considerable pressures that can lead to depressive symptomswhen achievement fails to match expectation. It must remain for futurestudies to empirically investigate these possibilities.

There were a number of important methodological limitations in thisstudy. By definition, some forms of anger experience may not be consciouslyavailable to an individual, for example, when efforts at repression have beenso successful as to keep the anger outside conscious awareness. This phe-nomenon may render findings regarding the Anger Control mode particu-larly suspect because this mode most closely taps repression of affect. Per-haps an important proportion of students who reported no parental angerantecedents at all had actually repressed anger incidents completely outsideof conscious awareness. If so, because students who reported zero angerantecedents were excluded from exploration of interactions, findingsconcerning Anger Control may have been compromised.

Second, although the surveys were completely anonymous, shame andreluctance to admit anger-producing incidents with parents may have influ-enced students’ responses. Although the LAPQ has demonstrated adequatereliability and validity, we could locate only four researchers other than its





original developer who have used the scale. More research is needed to con-firm its psychometric properties. Perhaps some of the most adaptive forms ofanger expression in Taiwanese culture are not included among the five modesof expression—which, after all, were developed primarily to study linksbetween anger expression and physical health problems. For example, seek-ing emotional support through confiding one’s anger in a trusted friend orsibling is not among the ways of dealing with anger assessed by the LAPQ. Interms of generalizability, the study was limited to the experience of angertoward parents. More research is needed to explore depressive symptoms andanger in the context of other types of close relationships.

Third, an additional limitation was introduced because the BDI assesseddepressive symptoms only over the past 2 weeks, whereas the LAPQ assessesanger process over the past year. The study was also limited to college stu-dents, although a very wide range of disciplines was sampled, with an excel-lent rate of research cooperation. Exploration of the moderator relationshipsinvolved 5 three-way analyses conducted with an alpha level of .05. Becausethese were exploratory analyses, a correction to alpha was not applied, butthe consequent inflation of experiment-wise error is another important limi-tation of this study. Finally, the correlational nature of this study does not per-mit conclusions about causality. For example, because the data are not longi-tudinal, it is possible that depressive symptoms develop first for reasonslargely unrelated to anger. Perhaps subsequently depressive symptoms leadto interpersonal irritability, anger, and fixed patterns of communication(Joiner & Coyne, 1999). The particular direction and nature of the causalrelationships we propose in Figure 1 can be confirmed only by experimentalor longitudinal studies.

Implications of these findings for counseling must remain tentative. It isespecially tenuous to expect that our findings will apply to persons with Tai-wanese cultural values who now live in the United States. The influence ofacculturation may have a considerable impact on the norms for interactionsbetween parents and children who move to the United States from Taiwan.We had hoped that positive buffering interactions, if replicated in other stud-ies, might become the basis for interventions with students in Taiwan whocannot escape from family situations that expose them to high levels of angerantecedents. However, it would be premature (and perhaps quite harmful) tomake a blanket suggestion that young women in Taiwan who experience highlevels of anger antecedents should be encouraged to use the Verbal Out mode(the one positive buffering interaction we found) as a way to reduce the levelsof depressive symptoms they might otherwise experience. Whether verbalexpression of anger in a particular client’s family may be adaptive no doubtdepends on many contextual factors that a counselor must be careful toassess. However, our findings suggest that a high versus low preference for





the Verbal Out mode of expressing anger may be associated with a clinicallymeaningful difference in BDI scores.

Our findings suggest that students who do report high levels of anger ante-cedents to a counselor should be screened for depressive symptoms. Coun-selors should expect that some young men who report outward expressionsof anger to their parents (i.e., Action Out) may feel a great deal of secondaryshame and depressive symptoms as a result. Furthermore, our findings sug-gest that methods routinely used by Western counselors to encourage expres-sion of anger openly in session need to be modified for clients with Taiwan-ese cultural values. Perhaps the most important conclusion to be drawn fromour findings is that the context of a client with traditional Taiwanese culturalvalues must be carefully assessed before developing a treatment plan thataddresses anger and depressive symptoms. One of the most helpful direc-tions for future research would be a longitudinal study that begins to disen-tangle the causal sequence of anger antecedents, modes of expression, anddepressive symptoms for adolescents and young adults.

REFERENCES

Adams, L. S. (1997). Sex differences in reports of emotional experiences. Dissertation AbstractsInternational, 58(03), 1582B.

Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions.Newbury Park, CA: Sage.

Alexander, F., & French, T. M. (1948). Studies in psychosomatic medicine: An approach to thecause and treatment of vegetative disturbances. New York: Ronald Press Co.

Beck, A. T., Ward, C. H., Mendelson, M., Mock, J., & Erbaugh, J. (1961). An inventory for mea-suring depression. Archives of General Psychiatry, 5, 462-467.

Biaggio, M. K., & Godwin, W. H. (1987). Relation of depression to anger and hostility construct.Psychological Reports, 61, 87-90.

Chan, D. W. (1991). The Beck Depression Inventory: What difference does the Chinese versionMake? Psychological Assessment, 3, 616-622.

Clay, D. L., Anderson, W. P., & Dixon, W. A. (1993). Relationship between anger expression andstress in predicting depression. Journal of Counseling and Development, 72, 91-94.

Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2003). Applied multiple regression/correlationanalysis for the behavioral sciences (3rd ed.). Mahwah, NJ: Lawrence Erlbaum.

Collier, H. V. (1982). Counseling women: A guide for therapists. New York: The Free Press.Cox, D., Stabb, S. D., & Hulgus, J. F. (2000). Anger and depression in girls and boys: A study of

gender differences. Psychology of Women Quarterly, 24(1), 110-112.Deffenbacher, J. L., Oetting, E. R., Thwaites, G. A., Lynch, R. S., Baker, D. A., Stark, et al.

(1996). State-Trait anger theory and the utility of the Trait Anger Scale. Journal of Counsel-ing Psychology, 43, 131-148.

Deffenbacher, J. L., & Stark, R. S. (1992). Relaxation and cognitive-relaxation treatments ofgeneral anger. Journal of Counseling Psychology, 39, 158-167.

Deffenbacher, J. L., & Swaim, R. C. (1999). Anger expression in Mexican American and Whitenon-Hispanic adolescents. Journal of Counseling Psychology, 46, 61-69.





Deffenbacher, J. L., Thwaites, G. A., Wallace, T. L., & Oetting, E. R. (1994). Social skills andcognitive-relaxation approaches to general anger reduction. Journal of Counseling Psychol-ogy, 41, 386-396.

Frazier, P. A., Tix, A. P., & Barron, K. E. (2004). Testing moderator and mediator effects in coun-seling psychology research. Journal of Counseling Psychology, 51, 115-134.

Gao, M.-B. (1997). The Coronary-Prone Behavior in Coronary Heart Disease Patients: Hostil-ity and Anger Process. Unpublished master’s thesis, Chung-Yuan Christian University, Tai-wan, ROC.

Greenberg, L. S. (2002). Emotion-focused therapy: Coaching clients to work through their feel-ings. Washington, DC: American Psychological Association.

Greenglass, E. R., & Julkunen, J. (1989). Construct validity and sex differences in Cook-MedlryHostility. Personality and Individual Differences, 10, 209-218.

Huang, T.-E. (1992). Comparison on the negative beliefs and transitory strategies of depressedand nondepressed college students. Unpublished master’s thesis, National Taiwan NormalUniversity, Taiwan, ROC.

Iqbal, N., & Ahmad, H. (1993). Gender differences and birth order in expressed and suppressedanger. Journal of Personality & Clinical Studies, 9, 73-77.

Jaccard, J., Turrisi, R., & Wan, C. K. (1990). Interaction effects in multiple regression. NewburyPark, CA: Sage.

Joiner, T., & Coyne, J. C. (Eds.). (1999). The interactional nature of depression. Washington,DC: American Psychological Association.

Kaplan, A. (1986). The self-in-relation: Implications for depression in women. Psychotherapy,23, 234-242.

Ko, H.-C. (1989). Comparison on cortisol suppression, cognitive bias, and daily stress of endog-enous and nonendogenous depression. Unpublished doctoral dissertation, National TaiwanUniversity, Taiwan, ROC.

Kopper, B. A. (1993). Role of gender, sex role identity, and Type A behavior in anger expressionand mental health functioning. Journal of Counseling Psychology, 40(2), 232-237.

Kopper, B. A., & Epperson, D. L. (1991). Women and anger: Sex and sex role comparisons in theexpression of anger. Psychology of Woman Quarterly, 15, 7-14.

Kopper, B. A., & Epperson, D. L. (1996). The experience and expression of anger: Relationshipswith gender, gender role socialization, depression, and mental health function. Journal ofCounseling Psychology, 43, 158-165.

Lee, R-L. (1993). A biological-psychological-social model to investigate the influences of angerprocess on the psychophysiological mechanism of blood pressure. Unpublished doctoral dis-sertation, National Taiwan University, Taiwan, ROC.

Leong, F. T. L., Blustein, D. L.(2000). Toward a global vision of counseling psychology. TheCounseling Psychologist, 28, 5-9.

Leong, F. T. L., Ponterotto, J. G. (2003). A proposal for internationalizing counseling psychologyin the United States: Rationale, recommendations and challenges. The Counseling Psycholo-gist. 31, 381-395.

Lin, W.-E., & Wang, C.-W. (1995). Chinese parenting: Discipline or punishment? Journal ofResearch on Indigenous Psychology (Taipei, Taiwan), 3, 2-92.

Moore, T. W., & Paolillo, J. G. (1984). Depression: Influence of hopelessness, locus of control,hostility, and length of treatment. Psychological Reports, 54, 875-881.

Newman, J., Gray, E. A., & Fuqua, D. R. (1999). Sex differences in the relationship of anger anddepression: An empirical study. Journal of Counseling & Development, 77(2), 198-203

O-Yang, Y., & Wu, L.-C. (1998). The intergenerational transmission model of child-rearing styleand attachment for adolescence and their mothers in Taiwan. Bulletin of Educational Psy-chology (National Taiwan Normal University, Taipei, Taiwan), 30 (2), 33-58.





Riley, W. T., Treiber, A., & Woods, M. G. (1989). Anger and hostility in depression. Journal ofNervous and Mental Disease, 177, 668-674.

Sharkin, B. S. (1996). Understanding anger: Comment on Deffenbacher, Oetting, et al. (1996),Deffenbacher, Lynch, et al. (1996), and Kopper and Epperson (1996). Journal of CounselingPsychology, 43(2), 166-169.

Shieh, C.-J. (2001). Anger antecedents and reactions in conflict between junior high studentsand father/mother. Unpublished master’s thesis. Chung Yuan Christian University, Taiwan,ROC.

Shields, S. A. (1987). Women, men and the dilemma of emotion. Review of Personality andSocial Psychology, 7, 229-250.

Spielberger, C. D. (1988). Manual for the State-Trait Anger Expression Inventory. Odessa, FL:Psychological Assessment Resources.

Spielberger, C. D. (1999). Manual for the State-Trait Anger Expression Inventory. Odessa, FL:Psychological Assessment Resources.

Teyber, E. (2000). Interpersonal process in psychotherapy. (4th ed.). Pacific Grove, CA: Brooks/Cole.

Thomas, S. P., & Williams, R. L. (1990). Relationships among perceived stress, trait anger,modes of anger expression and health status of college men and women. Paper presented atthe Annual Society of Behavioral Medicine Scientific Sessions, Chicago. (ED 326 808)

Ullman, J. B. (2001). Structural equation modeling. In B. G. Tabachnick & L. S. Fidell, Usingmultivariate statistics (4th ed.) (pp. 653-771). Needham Heights, MA: Allyn & Bacon.

Wang, A.-W. (2002). The influence of self-esteem on anger process and poster-anger satisfac-tion. Unpublished master’s thesis, Chung-Yuan Christian University, Taiwan, ROC.

Wood, P. S., & Mallinckrodt, B. (1990). Culturally sensitive assertiveness training for minorityclients. Professional Psychology Research and Practice, 21, 5-11.

Xu, J.-Y. (1996). The relationships among self-esteem, locus control, control process and angerexpression. Unpublished master’s thesis, National Chengchi University, Taiwan, ROC.

Yu, T.-H. (1987). Chinese parenting—The relatedness and parent’s grace. Taipei, Taiwan, ROC:Teacher-Chuang.

Zhong, M.-J. (1990). Sex differences, self-esteem, arbitrary style, empathy, and associationswith anger expression. Unpublished master’s thesis, National Chengchi University, Taiwan,ROC.





Anger expression toward parents and depressive symptoms among undergraduates in Taiwan

Documents