Relational Dynamics and the Expression of Aggression and Comforting Between Siblings

Sex variables as antecedent

Sex differences between siblings influence communication and caretaking (facets of sibling closeness) and conflict (Maccoby, 1988). Female sex of participant and sibling should influence sibling bond and assertion of power. Female siblings tend to be more emotionally involved in comparison with male siblings (Tillman, 2008), so female participants and female siblings should have a positive effect on sibling bond, as shown in Figure 2. Female dyads have also been found to have less conflict than male dyads (Cole & Kerns, 2001; Graham-Berman, 1994; Matthews & Conger, 2004; Vespo, Pedersen, & Hay, 1995). Hence, the model in Figure 2 predicts a negative effect of female sex of participant and sibling on assertion of power.

The interaction between the sex of participant and sibling also appears to be important in that siblings tend to prefer same-sex dyads. Sex-segregation begins in school at around 3 to 4 years of age. In school, both boys and girls show a preference for same-sex playmates in addition to avoiding mixed-sex play (Maccoby, 1988). This preference for same-sex play compared with mixed-sex play is also apparent in sibling relationships. As with peers, the sex of the sibling dyad influences the siblings’ relationship. Maccoby (1998) argues that compatibility is increased when the dyad is of the same-sex. Hence, sex-matched dyads should show a stronger sibling bond than mixed-sex dyads. The literature is unclear, however, on whether conflict occurs more often in mixed-sex sibling dyads (Deater-Deckard et al., 2002) or same-sex sibling dyads (Vespo et al., 1995). In fact, the disparity in findings might be due to sampling error. Hence, the model in Figure 2 did not predict a sex-match effect on assertion of power.

Research has found that men are higher in verbal aggressiveness than women (Hamilton et al., 2008). This effect occurs even after controlling for age and argumentativeness. Such effects have been observed between siblings (Maccoby, 1998) but not always (Vespo et al., 1995). The present study explored whether the negative effect of female sex on sibling verbal aggression would be direct or indirect.

Communicating affection tends to be more appropriate between siblings than acquaintances (Deater-Deckard et al., 2002; Floyd, 1997; Maccoby, 1998). Women tend to express affection more than men; this effect may be due to the fact that they engage in less verbal aggression, given that criticism generally decreases affection. It is unclear whether the positive effect of female sex on comforting is indirect (mediated by reduced verbal aggression which in turn decreases comforting) or direct. In the interest of parsimony, the model in Figure 2 proposed that the positive effect of female sex on verbal comforting would be mediated by verbal aggression.

Age as an antecedent

Siblings’ relationships change substantially as they progress through various life stages. The nature and quality of the sibling relationship changes as individuals transition from childhood to adolescence to emerging and later adulthood (Noller, 2005; Riggio, 2006). The sibling relationship transforms with age, becoming more egalitarian, symmetrical, and voluntary (Scharf, Shulman, & Avigad-Spitz, 2005). The sibling bond, reflected in part by amount of contact, substantially decreases from adolescence to emerging adulthood (Segrin & Flora, 2005).

Most adult siblings, despite decreased contact, are very close and report very positive feelings (Shortt & Gottman, 1997). Older aged adult siblings report greater closeness and solidarity, compared with siblings in emerging adulthood (Goetting, 1986). For many older adults, the bond with a sibling is almost as close as the parent–child bond (White & Riedmann, 1992), such that most adult siblings share a strong bond and many rate their sibling as one of their strongest intimate relationship (White & Riedmann, 1992). Thus, the high level of support and companionship that siblings shared during childhood returns during older adulthood; this is especially true during times of crisis, as when dealing with ailing parents (Riggio, 2000).

As children reach late childhood, the strength of their sibling bond is supposed to weaken (Pike, Coldwell, & Dunn, 2005). Children are thought to begin seeking companionship outside the home as they move from childhood to adolescence. That is, from mid-childhood to late adolescence, interest in relationships with peers should increase. As this interest outside of the family increases, the level of sibling companionship should decrease (Cole & Kerns, 2001; Kramer & Kowal, 2005). Peer relationships become more salient and influential than family relationships during this developmental stage (Scharf et al., 2005). Additionally, adolescence marks a progressive decline in sibling intimacy, amount of contact, and intensity (Kramer & Kowal, 2005, p. 503). Hence, sibling bond is thought to peak during childhood and pick up later in life after the siblings have raised their respective families (Leigh, 1982). Figure 2 therefore proposes that age will have a quadratic effect on sibling bond, forming a U-shaped function.

Relative power tends to increase with age, as siblings begin to challenge their parents’ authority. Previous research (Hamilton et al., 2004) suggests that age has a small inhibitory effect on verbal aggressiveness. This inhibitory effect should be partly mediated by assertion of power. That is, age should decrease the assertion of power, with assertion of power increasing verbal aggression. The present study explored whether the negative effect of sibling age on verbal aggression is indeed indirect, mediated by the assertion of power.

From early childhood into later adult years, siblings continue to provide affection and social support to one another (Mikkelson, 2005, 2006). The behavior of young siblings has been found to be generally cooperative (Graham-Bermann, 1994), where the trend is for age to decrease comforting into young adulthood and beyond (Salt, 1991). The model in Figure 2 predicts that this negative effect of age on comforting would be very weak because it is mediated by three personality variables.

First-order factor structure

Two separate CFAs were conducted. The first analysis considered the 11 personality factors. The sibling bond items formed four factors rather than a single sibling closeness factor. As Table 2 indicates, the 11 factors were all internally consistent with alpha values ranging from .79 to .97. Because of internal consistency concerns, one item each was discarded from the parent–child Verbal Aggression Scale, the Verbal Comforting Scale, and the Family Function Scale. The parallelism tests indicated that the parent–child Verbal Aggression Scale, the Self- and Other-Assertion of Power Scales, and the Talk Frequency Scale were all flat. That is, the items on these scales were approximately equal in quality. By contrast, the parallelism tests indicated a strong–weak gradient for the items on the parent–child Closeness Scale, Sibling Empathy, Hostility, Closeness, Self-disclosure, and Shared Experiences Scales.

OPEN IN VIEWER

Table 2.

Tests of First-Order Factor Structure

Variable	No. of Items	Coefficient α	Parallelism RMSEA
			Flatness	Gradient
PCC Mother	9	.95	.02	.01
PCC Father	9	.97	.06	.05
PCCTS Verbal aggression, parent–child	3	.80	.00	.02
STQ Self-assertion of power	2	.79	.07	.07
STQ Other-assertion of power	2	.80	.11 a	.11
LSR/STQ Empathy	16	.97	.06	.04
STQ/LSR Hostility	3	.82	.08	.06
STQ/LSR Closeness	6	.97	.05	.04
LSR/ACI/RM Self-disclosure	6	.90	.07	.05
STQ/LSR Shared experience	5	.89	.10	.05
LSR/RM Talk frequency	3	.88	.05	.09
STQ/RM Verbal aggression, siblings	3	.80	.00	.00
RM/LSR/ACI Verbal comforting	16	.97	.11	.08
ACI Comforting acts	8	.93	.01	.04
RM family function	2	.86	.08	.08

Note: RMSEA = root mean squared error of approximation; PCC = parent–child closeness; PCCTS = Parent–Child Conflict Tactics Scale; STQ = Sibling Type Questionnaire; LSR = lifelong sibling relationship; ACI = Affective Communication Index; RM = relational maintenance.

a.

Exceeds acceptable levels of error.

Only the Other-Assertion of Power Scale exceeded the standard RMSEA level of acceptable error (.08). Inspection of the RMSEA values for each personality factor for two other-assertion items indicated that the RMSEA were noticeably high for hostility (.17), closeness (.17), self-disclosure (.17), shared experiences (.15), and talk frequency (.13). These findings suggest that the two other-assertion items relate somewhat differently to the temperament variables shown in Figure 2.

The second first-order analysis considered the four derived factors presented in Figure 2. As Table 2 shows, the four factors were internally consistent with α values ranging from .80 to .97. One item was discarded from both the verbal comforting and family function scales because those items reduced internal consistency. The parallelism tests found that the sibling Verbal Aggression Scale, the Comforting Acts Scale, and the Family Function Scales were all flat, indicating that the items on these scales were of approximately equal quality. The parallelism tests found a strong–weak gradient for the verbal comforting items.

Second-order factor structure

EFA of the first-order factors suggested three patterns. First, the two parent–child closeness factors formed a second-order factor of parent–child closeness. Second, the self-assertion and other-assertion factors formed a second-order factor of general assertion of power; the structure of this second-order factor was closely inspected given the parallelism concerns of the other-assertion factor. Third, the closeness, self-disclosure, shared experience, and talk frequency factors formed a second-order factor labeled strength of sibling bond. The strength of adult sibling relationships is influenced by shared experiences (Myers, 2001) as well as the amount of contact and communication siblings share. The parallelism tests across the other first- and second-order factors indicated that the parent–child closeness factor was flat (RMSEA = 0), as was the assertion of power second-order factors (RMSEA = 0). The parallelism tests found a strong–weak gradient for the sibling bond second-order factor (RMSEA = .08).

Items on the first-order factors were summed to generate scale scores. For parent–child closeness, assertion of power, and sibling bond, the first-order factor scores were summed to generate the three second-order factors.

Siblings’ cooperative affect system

On three of the five variables in Table 3, female participants interacted with biological relatedness. These interactions were with linear biological relatedness rather than quadratic (half-sibling) biological relatedness. Specifically, the effect of biological relatedness on sibling bond, verbal comforting, and family function depended on whether the participant was male or female. The interaction was ordinal, so Table 3 shows a main effect for biological relatedness on all three measures. Among women, biological relatedness had a moderately large positive effect on sibling bond (r = .37), verbal comforting (r = .40), and family function (r = .22). By contrast, among men, biological relatedness had a moderately small positive effect on sibling bond (r = .20), verbal comforting (r = .21), and family function (r = .05). Thus, full-siblings were more likely to experience prosocial emotions than half-siblings, and half-siblings were more likely to experience prosocial emotions than step-siblings, a trend that was stronger for female participants than male participants.

Quadratic age decreased empathy (β= −.18) and strength of sibling bond (β= −.20), indicating that these prosocial emotions between siblings peak at middle age. Female siblings increased affection displays (β= .12). Women are generally regarded as more affectionate than men, but in the present study this effect could not be attributed to women reporting greater empathy, closer attachment (sibling bond), or verbal comforting. The interaction between female participant and female sibling (same sex) increased sibling bond (β = .16) and verbal comforting (β = .14). These findings indicate a stronger bond between same-sex siblings than opposite-sex siblings and as a consequence more supportive verbal exchanges.

Siblings’ individualistic affect system

On two of the three variables in Table 4, female participants moderated the impact of biological relatedness. These interactions were with linear biological relatedness rather than quadratic (half-sibling) biological relatedness. Specifically, the effect of biological relatedness on assertion of power and verbal aggression depended on whether the participant was male or female. The interaction was disordinal, so the main effect for biological relatedness was nullified for the two dependent measures. For women, biological relatedness had a modest positive effect on assertion of power (r = .21) and verbal aggression (r = .13) whereas for men, biological relatedness had a small negative effect on assertion of power (r = −.07) and verbal aggression (r = −.11). Thus, full-siblings were more likely to experience prosocial emotions than half-siblings, and half-siblings were more likely to experience selfish emotions than step-siblings, a trend that held only for women. Among male participants, this trend was either reversed or nonexistent.

Linear age decreased assertion of power (β= −.26), hostility (β = −.20), and verbal aggression (β = −.33), suggesting that the individualistic affect system becomes less active with age. Quadratic age increased assertion of power. This indicates middle age is characterized by less assertion of power, which complements the increase in empathy and strength of sibling bond at that age.

The parent–child relationship

As anticipated, age decreased both parent–child closeness (ρ = −.20) and parent–child verbal aggression (ρ = −.20). That is, the most emotionally intense communication occurs at younger ages. Parent–child closeness decreased parent–child verbal aggression (ρ = −.44). The model in Figure 2 implies that had parental hostility been measured, it would decrease closeness to the child and increase verbal aggression toward the child, producing a spurious correlation between closeness and verbal aggression of −.13. Yet the observed effect of −.44 is substantially larger than a value of −.13. This discrepancy suggests that there is a direct negative effect of closeness on verbal aggression, as shown in Figure 3.

Prosocial emotions within the cooperative affect system

Parent–child closeness increased empathy (ρ = .18), as predicted in Figure 2. However, parent–child verbal aggression had a direct negative effect (ρ = −.24) on empathy. The model in Figure 2 proposed that the negative correlation between parent–child verbal aggression and empathy would be spurious due to the common causal antecedent parent–child closeness. Instead, the model in Figure 3 shows that parents’ verbal aggression against a child decreases the child’s (siblings’) empathy. Finally, biological relatedness increased empathy (ρ = .30). Having a full-sibling produces more sibling empathy than having a half-sibling, and having a half-sibling produces more sibling empathy than having a step-sibling.

Empathy had a massive positive effect on strength of sibling bond (ρ = .85), an effect that was much larger than that observed in acquaintance studies (see Figure 2). The biological relatedness by female participant interaction slightly increased strength of sibling bond (ρ = .08); the regression analysis had indicated that biological relatedness had a larger positive effect on sibling bond for women than men. Finally, same-sex siblings had a stronger sibling bond than opposite-sex siblings (ρ = .14). Two effects failed to materialize, neither hostility nor age weakened the sibling bond. The apparent independence between hostility and sibling bond distinguishes the sibling relationship from acquaintance relationships (see Figure 2). The indirect effect of age on sibling bond, mediated by both parent–child closeness and sibling empathy was near zero (.03).

Research on acquaintances, represented in Figure 2, holds that the positive effect of empathy on verbal comforting is partially mediated by strength of attachment. Verbal comforting did indeed increase with strength of sibling bond (ρ = .30) and empathy (ρ = .64). However, the direct positive effect of sibling empathy on verbal comforting was more than twice the size of the indirect effect mediated by sibling bond. Just as it had for sibling bond, the biological relatedness by female participant interaction had a small positive effect on verbal comforting (ρ = .07); the regression analysis had indicated that the positive effect of biological relatedness on verbal comforting was larger for women than men.

Verbal comforting increased attendance at family functions (ρ = .65) and affection displays (ρ = .78). In addition, affection displays were somewhat more likely among half-siblings (ρ = .78), female participants (ρ = .10), and female siblings (ρ = .11). Half-siblings did not influence verbal comforting. These findings suggest that female siblings are more likely to express prosocial emotion nonverbally than male siblings. Finally, half-siblings were more likely to express prosocial emotion nonverbally than full-siblings or step-siblings.

Selfish emotions within the individualistic affect system

The biological relatedness by female participant interaction increased assertion of power (ρ = .15). The regression analysis revealed that biological relatedness had a positive effect on assertion of power for women but not men. Female siblings increased assertion of power (ρ = .15) but female participants did not. As the model in Figure 2 proposed, assertion of power increased hostility (ρ = .72). The size of this effect was quite large. The model in Figure 2 also predicted that empathy would have a negative effect on hostility but that this effect would be mediated by assertion of power. Instead, empathy had a direct negative effect on hostility (ρ = −.36).

The model in Figure 2, drawing on acquaintance research, predicted that the positive effect of assertion of power on verbal aggression would be partially mediated by strength of hostility. Verbal aggression did indeed increase with hostility (ρ = .64) and assertion of power (ρ = .27). The indirect positive effect of assertion of power on verbal aggression mediated by hostility was nearly twice the size of the direct effect from assertion of power. Parent–child verbal aggression had an unexpected direct effect on sibling verbal aggression (ρ = .07). Age directly decreased verbal aggression (ρ = −.14), an effect that has been observed between acquaintances as well as siblings. Finally, female participant decreased verbal aggression (ρ = −.21), an effect that can be reframed as male participants showing higher verbal aggression.

The cooperative affect system that underlies the “light side” of sibling relationships

Consistent with prediction, parent–child closeness increased empathy. Contrary to prediction, parent–child verbal aggression did not influence egocentric assertion of power. Instead, parent–child verbal aggression decreased empathy. This verbal aggression finding for siblings is at odds with that observed for acquaintances.

Biological relatedness strengthened sibling bond and increased verbal comforting in two ways. Indirectly, biological relatedness increased empathy, with empathy strengthening sibling bond and increasing verbal comforting. Half-siblings were found to display the most molar affection, suggesting that they may be compensating for their sibling bond deficit by demonstrating greater affection. This pattern is much more pronounced for female than male participants in the present study, where the positive effect of biological relatedness on sibling bond was nearly twice as large. Closeness between parents and child strengthened the sibling bond by increasing empathy. This finding supports the results of previous studies that indicate positive, emotional, parent–child relationships increase harmony and closeness in siblings’ relationships (Yu & Gamble, 2008). Directly, biological relatedness strengthened sibling bond and increased verbal comforting, although these positive effects were greater for women than for men. As predicted in Figure 2, sibling bond was stronger for same-sex siblings than opposite-sex siblings. Although the regression results suggested a reversal of the predicted age effect on sibling bond, the causal analysis shown in Figure 3 determined that this finding was spurious.

As predicted in Figure 2, empathy had a positive effect on verbal comforting that was partially mediated by strength of sibling bond. The direct positive effect of sibling empathy on verbal comforting was more than twice the size of the indirect effect mediated by sibling bond. In turn, verbal comforting increased attendance at family functions and molar affection displays. Women and half-siblings were more prone to molar affection displays than men or full- and step-siblings.

The individualistic affect system that underlies the “dark side” of sibling relationships

Biological relatedness was found to have a moderately small effect on assertion of power for women in sibling relationships, although this effect may not hold for men. Jenkins, Dunn, O’Connor, Rasbash, and Behnke (2005) found a similar effect, with full-siblings showing more negativity than half-siblings, yet they did not find a significant difference between half-siblings and step-siblings. Female siblings were rated as more likely to assert power than male siblings. However, a parallel sex effect was not present when siblings rated themselves on propensity to assert power: Both men and women are more inclined to blame female siblings than male siblings for their competition and strife. This sibling effect stands in stark contrast to acquaintances, where men are more likely to assert power and be argumentative. As predicted in Figure 2, egocentric assertion of power had a positive effect on verbal aggression that was partially mediated by hostility. That is, asserting power increased hostility, with hostility fueling verbal aggression; yet assertion of power also directly increased verbal aggression. This “direct” effect of power assertion on verbal aggression may actually be mediated by an as yet unidentified variable.

Parent–child verbal aggression “directly” increased sibling verbal aggression in that this effect was not mediated by either assertion of power or hostility. Two direct identity effects held for siblings as well as acquaintances, indicating that familiarity does not moderate these effects. The indirect positive effect of assertion of power on verbal aggression mediated by hostility was nearly twice the size of the direct effect. Age decreased verbal aggression and male participants increased verbal aggression. Contrary to prediction, the current study did not find that siblings were closer at younger ages and older ages and less close at middle age. The results indicate that age influenced assertion of power but not sibling bond.