Parental Differential Treatment

Social Justice Research, Vol. 18, No. 2, June 2005 ( C© 2005)DOI: 10.1007/s11211-005-7367-2

Effects of Parental Differential Treatment onRelationship Quality with Siblings and Parents:Justice Evaluations as Mediators

Thomas Boll,1,3 Dieter Ferring,2 and Sigrun-Heide Filipp1

Based on equity theory, the present study analyzes to what extent justice evalu-ations mediate the effects of perceived parental differential treatment (PDT) onrelationship quality with siblings and parents as experienced in middle adulthood.Middle-aged adult offspring (N = 709) rated how often they and a sibling receivedparental recognition, nurturance, and demands for assuming filial responsibility.In addition, they indicated their justice evaluations of PDT and completed mea-sures of relationship quality to sibling and parents. Justice evaluations emerged aseither partial or complete mediators between PDT and relationship quality. More-over, justice evaluations turned out to be more powerful predictors of relationshipquality to parents than PDT per se. Implications are discussed concerning factorscontributing to justice evaluations and the role of PDT within the context of otherjustice issues in families.

KEY WORDS: parent–child relations; justice; sibling relations; adult offspring; middle aged.

Prior research on (in)equality and (in)justice as experienced within familieshas selectively focused on particular kinds of distributive phenomena. For instance,the division of household labor in marital relationships has received the most at-tention (for an overview see Mikula, 1998). The distribution of caregiving forolder parents among siblings (e.g., Cicirelli, 1992; Ingersoll-Dayton et al., 2003)and the parental distribution of inheritance among their children (e.g., Drake andLawrence, 2000) have received little research attention. There is another kind ofinequality in the family, namely parental differential treatment (PDT) of siblings,

1Department of Psychology, University of Trier, Trier, Germany.2Department of Psychology, University of Luxembourg, Luxembourg City, Luxembourg.3All correspondence should be addressed to Thomas Boll, Department of Psychology, University ofTrier, Universitatsring 15, D-54286 Trier, Germany; e-mail: [email protected].

155

0885-7466/05/0600-0155/0 C© 2005 Springer Science+Business Media, Inc.

156 Boll, Ferring, and Filipp

that has recently become a focus of study by developmental, clinical, and familypsychologists, although surprisingly not yet from scholars of social justice. Thevast majority of this research has described and assessed PDT in purely descrip-tive terms, that is, concerning the extent to which siblings are treated (or perceivethemselves to be treated) in similar or dissimilar ways by their parents with regardto certain dimensions (e.g., affection, support, control). Most of the past researchon PDT focused on children and adolescents and has provided evidence that anotable amount of PDT—for instance, with respect to affection and control—canbe observed, and that it is related to negative outcomes for children themselves,for the quality of sibling relationships, and, to some extent, for the quality ofchildren’s relationships with their parents (for an overview see Boll et al., 2001).In addition, recent studies indicate that even in middle adulthood a considerableamount of PDT—with respect to confidentiality, emotional closeness, recogni-tion, and demands for assuming responsibility—is perceived by both parents andtheir adult children (Aldous et al., 1985; Boll et al., 2002; Ferring et al., 2003;Suitor and Pillemer, 2000). Moreover, there is evidence that adult children’s mem-ories of PDT in childhood and adolescence (Bedford, 1992; Ferring et al., 2003)as well as their current perceptions of PDT (Boll et al., 2003) are systematicallyrelated to current relationship quality with their adult siblings and older parents.In particular, adults who feel disfavored by their parents report having a worserelationship with siblings and parents than those who feel equally treated.

While most prior research on PDT has not analyzed how the links betweenPDT and relationship quality between siblings and/or between children and theirparents are mediated, a few attempts have been made recently to address thisissue. In particular, the role of justice evaluations as a possible mediator has beensuggested, and this seems to be reasonable from the perspective of social justiceresearch. Parental behavior can easily be regarded as a distribution of materialand symbolic outcomes to children of either positive (e.g., affection) or negativevalence (e.g., criticism), and it is likely to be evaluated by the children with respectto distributive justice criteria (cf. Ihinger, 1975).

A first set of studies has explicitly assessed how PDT as a purely descriptivecategory referring to (the amount of) similar or dissimilar parental treatment isevaluated with respect to justice and has provided evidence concerning the extentto which these evaluations are associated with possible outcome variables (e.g.,relationship quality). Kowal and Kramer (1997) questioned whether siblings inchildhood make distinctions between appropriate and inappropriate cases of beingdifferentially treated by their parents and hypothesized that those children whoview PDT as justified will have more positive sibling relationships than childrenwho view PDT as unjustified. The authors found that in 75% of the occasions wherechildren had reported experiencing some degree of PDT (predominantly withrespect to affection and control), PDT had been judged as fair by these children.Children who evaluated differential paternal (but not maternal) treatment as fair

Justice Evaluations as Mediators 157

also reported having a better relationship with their sibling than those childrenwho reported that differential paternal treatment was unfair. However, this studydid not consider the extent to which equal treatment was evaluated as just or unjustand, furthermore, it failed to consider the direction of PDT (i.e., whether beingfavored vs. disfavored is evaluated similarly or not).

Some of these issues were addressed in Kowal et al.’s (2002) study of chil-dren’s and adolescents’ perceptions of PDT. They found, for example, that childrenwere more likely to regard PDT as unfair when the self was disfavored only withregard to parental control (i.e., more control for self compared to sibling) andnot in the case of differential parental affection. In contrast, McHale et al. (2000)observed in a sample of children and adolescents that the direction of PDT ratedfor the warmth of the relationships, parents’ temporal involvement with the sib-lings, and allocation of household duties was not consistently related to justiceevaluations. In addition, they found that justice evaluations of parents’ differentialtemporal involvement and differential allocation of household duties were moreclosely related to the quality of sibling relationships than was PDT itself.

Even though the studies reviewed earlier demonstrate how younger childrenand adolescents evaluate PDT with regard to justice and that—at least with respectto some areas of PDT—these evaluations are powerful predictors of relationshipquality to siblings, we still need to examine to what extent justice evaluations arelinked to perceived relationship quality with parents. This avenue of research issuggested by the possibility that parents can be easily held responsible for unjustas well as for just treatment of their children and this should lead—according toattribution theory analysis—to negative relationship quality in the case of unjusttreatment and to positive relationship quality in the case of just treatment (cf.Weiner, 2001). Moreover, as these studies refer to PDT in childhood and adoles-cence, this raises the developmental issue of whether justice evaluations of PDTremain important during adulthood as well, and if so, how they are linked torelationship quality between adult siblings and between adult children and theirolder parents. Links to relationship quality among adult siblings deserve consider-ation, because even in adulthood siblings continue to be important as interactionpartners, as sources of support in times of trouble, and—most likely in middleadulthood—as partners in caring for aging parents (cf. Bedford, 1995; Cicirelli,1996). In addition, links to relationship quality with older parents deserve partic-ular attention because middle-aged adult children are among the most importantmembers in the social network of older persons (Ferring and Filipp, 1999). Con-sidering that the quality of the parent–child relationship has been found to belinked to support for older parents (Silverstein et al., 1995), one may ask whetherbeing in the role of a disfavored child reduces adult children’s willingness tocare for their aging parents, and whether being favored increases it. In our study,we focus on middle-aged adult children, because they—compared to young adultchildren—are more likely to have parents who actually need support.


In contrast to studies on PDT in childhood and adolescence, which explic-itly assess justice evaluations, adult studies have referred to justice evaluationsas hypothetical constructs to derive predictions about links between PDT andpossible outcomes (e.g., relationship quality to siblings and/or parents), whichwere then tested empirically. Bedford (1992) analyzed how middle-aged adults’global memories of being treated worse than their siblings in childhood and ado-lescence are related to current relationship quality with their parents and referredto perceived injustice as a possible mediating variable. From a social exchangeperspective, she argued that “when parents violate norms of fairness early in thechild’s life, perhaps adult children feel less duty-bound to sustain an affective bondwith their aging parents” (p. S149). In line with this reasoning she found that, forinstance, children’s global memories of being treated worse were predictive ofpoorer relationship quality with their parents. However, the study did not con-sider how memories of being treated better than one’s sibling are associated withparental relationship quality. Boll et al. (2003) focused on possible links betweenmiddle-aged adult children’s current perceptions of both maternal and paternaldifferential treatment in adulthood and their experienced relationship quality withsiblings and parents. These authors referred to perceived “equity/inequity” as apossible mediating variable and derived predictions concerning links betweenPDT and relationship quality from equity theory (Walster et al., 1973; cf. alsoSprecher and Schwartz, 1994). They argued that being disfavored in the senseof receiving less positive treatment (e.g., recognition) or more negative treatment(e.g., criticism) than one’s sibling will be perceived as inequity if the disfavoredchild, for instance, believes that both children deserve the same treatment becauseof equal inputs. This, in turn, was assumed to lead to attempts to restore “actualequity,” for example, by retaliating against the person who benefits from the in-equity or is perceived to be the harm-doer. In other words, a deterioration of thedisfavored child’s relationship quality with both the favored sibling and the parentswas predicted. On the other hand, being favored (in the sense of receiving morepositive or less negative treatment than one’s sibling) was expected to be experi-enced as inequity, too, if the favored child believes that both siblings were entitledto the same parental treatment due to equal inputs. This, in turn, was assumedto instigate attempts to restore “psychological equity,” for instance, by devaluingthe disfavored sibling’s merits and by derogating him or her. Thus, being favoredwas predicted to diminish the favored child’s relationship quality with the siblingnot favored while improving the relationship quality with the parents, because thepsychological restoration of equity is likely to lead to the favored child’s beliefthat parents basically treat their children in an equitable manner.

In testing these predictions, the authors found curvilinear relations betweenPDT and relationship quality with siblings that supported equity theory: Adultchildren’s relationship quality was best when they perceived themselves to betreated equally and deteriorated with increasing favoritism or disfavoritism. These


findings are compatible with the assumption that equal treatment is experienced asjust and that both kinds of unequal treatment (being disfavored and being favored)are experienced as unjust. In contrast, evidence concerning the links betweenPDT and relationship quality with parents only partially supported predictions.As expected, relationship quality to one’s parents deteriorated with increasingdisfavoritism and ameliorated when favoritism increased to a moderate degree.Contrary to predictions, relationship quality to parents deteriorated with extremefavoritism as well. Taken together, these findings indicate that being disfavored,as well as being extremely favored were both perceived as unjust. Yet the questionremains whether either being equally treated or being moderately favored areexperienced as most just. In addition, one should keep in mind that evaluationsof PDT as just versus injust are normative evaluations which may be based ondifferent standards (cf. Montada, 2002). Justice evaluations can be, but need not be,based on perceptions of equity versus inequity as conceptualized by Walster et al.(1973). Even though the equity principle (considering ratios of inputs and outputsof participants in a distribution) is regarded as an important justice principle,the proponents of a “multi-principle approach” argue that equity is only one ofseveral principles that may underlie evaluations of distributive justice; the justiceevaluations may also be based, for example, on the equality principle or the needprinciple (cf., Deutsch, 1975, 1985).

Beyond these ambiguities, studies that refer to justice evaluations as hypo-thetical constructs raise a more fundamental problem, even if findings are predom-inantly in line with the predictions of social exchange (Bedford, 1992) or equitytheory (Boll et al., 2003). If justice evaluations are not independently assessed onecannot rule out an alternative mediation effect involving self-interest, a variablethat is alluded to in equity theory when postulating a human desire to maximizeone’s outcomes (cf., Montada, 2002). Being disfavored can be perceived simply asa violation of self-interest, and one can even argue that being disproportionatelyfavored may lead to perceived violations of self-interest as well. With increas-ing favoritism, the favored sibling may become increasingly aware of negativeoutcomes for himself or herself, such as negative feelings and behaviors of thedisfavored sibling toward the favored one or the felt obligation to repay the ad-vantages received from parents (cf., Boll et al., 2003). Thus, in order to providemore supportive evidence for justice evaluations as mediating variables, theseevaluations should be explicitly assessed in future research.

The purpose of this study was to examine the role of justice evaluationsof PDT in middle adulthood in affecting relationship quality with adult siblingsand aging parents. In particular, we wanted to determine to what extent justiceevaluations mediate the effects of PDT on relationship quality and whether theseevaluations are more powerful predictors of relationship quality than PDT per se.To perform a proper mediation analysis, we followed the guidelines proposed byBaron and Kenny (1986). Figure 1a represents an unmediated model describing


Fig. 1. Models for the relations between PDT and relationship quality with sibling and parents:(a) unmediated model, (b) model with justice evaluations as mediating variable.

the (curvilinear) relations between PDT and relationship quality with sibling andparents (path c). Figure 1b represents a mediational model as suggested by equitytheory. In comparison to equal treatment, both being favored and being disfavoredwere expected to be evaluated as less just (path a) which in turn should give riseto negative relationship quality to sibling and parents (path b). To draw valid con-clusions, we considered several dimensions of PDT (e.g., recognition, nurturance,demands for assuming responsibility) performed by both parents across the entirespectrum from disfavoritism to favoritism and assessed both positive and negativeaspects of experienced relationship quality. Because perceived justice of PDT inadulthood is a rather neglected topic in research on (in)equality and (in)justice inadult family life, our study was expected to lead to a more complete picture ofthese intrafamilial phenomena.

METHOD

Sample

The data for the present analyses come from a longitudinal study of PDT inmiddle adulthood. Respondents first participated in 2000 (T1; see Boll et al., 2003)and a second time in 2002 (T2). The present analyses are all based on data collectedat T2 from a total of 709 German adult offspring (262 men, 447 women) betweenthe ages of 42 and 56 years (M = 46.76, SD = 3.03) who had either a motherand/or father still living at the time of the survey and at least one sibling. Whencompleting the questionnaire, respondents were instructed to refer to one specificliving sibling (focus sibling), who was designated according to the respondent’smembership in one of the following nine subsamples: sister–older sister (n =96; 13.5%), sister–younger sister (n = 121; 17.1%), sister–older brother (n = 88;12.4%), sister–younger brother (n = 135; 19.0%), brother–older sister (n = 44;6.2%), brother–younger sister (n = 85; 12.0%), brother–older brother (n = 49;


6.9%), brother–younger brother (n = 79; 11.1%); 12 respondents were twins (n =1.7%). When a respondent had several siblings within a category (e.g., older sisters,older brothers, etc.), he or she was asked to refer to the sibling that was closest tohim or her in age.

The focus siblings were between 25 and 64 years of age (M = 45.68; SD =6.08). The mean age difference between the respondents and their younger siblingwas 4.40 years (SD = 2.97, mode = 2), and between the respondents and theirolder sibling was 4.58 years (SD = 3.25; mode = 2). In addition to the focussibling, respondents had between zero and eight siblings (M = 1.38; SD = 1.57).

The majority of the respondents were either married or involved in long-termrelationships (607; 85.6%); of the remaining respondents, 38 were single (5.4%),51 were either separated or divorced (7.2%), and 12 were widowed (1.7%). Withregard to educational level and employment status, the following picture emerged:219 respondents (30.9%) reported having completed the Volkschule/Hauptschule(up to and including 9th grade), 250 respondents (35.3%) completed the MittlereReife (up to and including 10th grade), and 73 (10.3%) completed the Abitur(up to and including 13th grade, the highest school-leaving degree in Germany).Additionally, a total of 165 respondents (23.3%) completed college education.The largest portion of the sample was employed (599; 84.5%), 63 respondentswere homemakers (8.9%), and a small percentage were unemployed (29; 4.1%)or retired (14; 2.0%).

With respect to the living status of the parents, 599 respondents indicated thattheir mothers (84.5%) and 381 respondents indicated that their fathers (53.7%)were alive, and 301 respondents (43.0%) reported that both parents were living.Respondents with two living parents filled out the measures for both parents;when one of the parents was no longer living, the respondents only completed themeasures for their living parent.

Measures

Perceived Parental Differential Treatment

These variables were assessed with the Questionnaire for the Assessmentof Currently Perceived Parental Differential Treatment (see Boll et al., 2003 formore details). A respondent with two living parents rated various behaviors ofboth the mother and the father indicating recognition, nurturance, and demandsfor assuming filial responsibility (14 items for each parent). Two separate ap-praisals were required for each item on two 7-point Likert-type rating scalesranging from 1 (never) to 7 (always). First, an assessment was made about howoften the parent behaved toward the respondent in the manner described and,second, how often the parent behaved in such a manner toward the focus sibling,and a difference variable was determined for each of the 14 items from bothappraisals. This variable contained information about the extent and direction of


maternal and paternal differential treatment. A positive value indicates that therespondent receives more and a negative value indicates that he or she receivesless than the sibling in each behavioral domain; a score of zero represents equaltreatment. Based on prior factor analyses (see Boll et al., 2003), the resulting14 difference variables for each parent separately were combined into three sub-scales. High internal consistencies and high test–retest reliabilities over a timespan of 2 years (T1, T2) were found for the six scales: (1) Differential MaternalRecognition (six items, e.g., “mother shows us that she is proud of us,” α = 0.82,rT1T2 = 0.70; M = 0.10, SD = 0.88), (2) Differential Paternal Recognition (sixitems, e.g., “father shows us that he is proud of us,” α = 0.87, rT1T2 = 0.84;M = 0.14, SD = 0.96), (3) Differential Maternal Nurturance (five items, e.g.,“mother shows us her love,” α = 0.86, rT1T2 = 0.70; M = −0.13, SD = 0.85),(4) Differential Paternal Nurturance (five items, e.g., “father shows us his love,”α = 0.86, rT1T2 = 0.78; M = −0.07, SD = 0.83), (5) Differential Maternal De-mand for Assuming Filial Responsibility (three items, e.g., “mother expects us tofeel responsible for her,” α = 0.85, rT1T2 = 0.77; M = 0.22, SD = 1.41), and (6)Differential Paternal Demand for Assuming Filial Responsibility (three items, e.g.,“father expects us to feel responsible for him,” α = 0.88, rT1T2 = 0.74; M = 0.26,SD = 1.25). When considering one and the same parent, the correlations betweenparental recognition and parental nurturance were medium sized (for DifferentialMaternal Recognition and Differential Maternal Nurturance, r = 0.48, and forDifferential Paternal Recognition and Differential Paternal Nurturance, r = 0.47;ps < 0.01). Correlations between parental recognition and parental demand toassume filial responsibility were very low (for Differential Maternal Recognitionand Differential Maternal Demand for Assuming Filial Responsibility, r = 0.08,p < 0.05, and for Differential Paternal Recognition and Differential Paternal De-mand for Assuming Filial Responsibility, r = −0.08, ns). Finally, only modestcorrelations were found between parental nurturance and parental demand to as-sume filial responsibility (for Differential Maternal Nurturance and DifferentialMaternal Demand for Assuming Filial Responsibility, r = 0.39, and for Differ-ential Paternal Nurturance and Differential Paternal Demand for Assuming FilialResponsibility, r = 0.22, ps < 0.01). For the subsample of participants with twoliving parents, the correlations between the scales of perceived maternal and pa-ternal differential treatment were quite high (Differential Recognition: r = 0.68;Differential Nurturance: r = 0.71; and Differential Demands for Assuming FilialResponsibility: r = 0.85; ps < 0.01).

Justice Evaluations

Two scales—one for maternal and one for paternal differential treatment—were used to assess how respondents evaluate PDT with respect to justice. Eachscale included five items that covered the dimensions of PDT outlined earlier,


namely differential recognition, nurturance, and demand for assuming filial re-sponsibility (e.g., “Compared to my brother I regard the recognition that I receivefrom my mother/ father as . . .”). Respondents were asked rate each parent’s differ-ential behavior on a seven-point Likert-type rating scale ranging from −3 (very un-just) to +3 (very just). Internal consistencies of the scales were good (Justice Evalu-ation of Differential Maternal Treatment: α = 0.90, M = 0.84, SD = 1.12; JusticeEvaluation of Differential Paternal Treatment: α = 0.90, M = 0.75, SD = 1.10).For the subsample of participants with two living parents, the correlation betweenjustice evaluations of maternal and paternal differential treatment was quite high(r = 0.76; p < 0.01).

Experienced Relationship Quality with Sibling and Parents

Emotional and behavioral measures were included for assessing the qualityof an individual’s experienced relationship quality with his or her sibling as wellas with the mother and/or father (see Boll et al., 2003 for more details). Ratingson three emotion checklists were used as indicators of the individual’s emotionalrelationship experiences with the respective family members. Respondents wereasked to rate on seven-point Likert-type scales ranging from 1 (never) to 7 (always)how often they felt the given emotion when they thought about the person inquestion (i.e., sibling, mother, father). For the purpose of the present study, onlytwo subscales that were identical for the maternal, paternal, and the sibling versionwere employed. The Attachment/Closeness scale was composed of seven items(e.g., “deep affection”) and was characterized by high internal consistencies andhigh test–retest reliabilities over a time span of 2 years (maternal version: α = 0.95,rT1T2 = 0.85; M = 4.56, SD = 1.35; paternal version: α = 0.96, rT1T2 = 0.84;M = 4.18, SD = 1.45; and sibling version: α = 0.92, rT1T2 = 0.84; M = 3.54,SD = 1.27). The Dislike scale consisted of two items (e.g., “filled with hatred”)and had for all three target persons acceptable internal consistencies and ratherhigh test–retest reliabilities over 2 years (maternal version: α = 0.68, rT1T2 =0.72; M = 1.60, SD = 0.93; paternal version: α = 0.67, rT1T2 = 0.78; M = 1.64,SD = 1.06; and sibling version: α = 0.64, rT1T2 = 0.63; M = 1.43, SD = 0.86).The correlations between the two scales were medium sized and negative for allthree versions: for mother, r = −0.50; for father, r = −0.47; and for sibling,r = −0.38; ps < 0.01.

Three behavioral inventories were used to assess behavior-related aspects ofthe respondent’s relationship with his or her sibling, mother, and/or father. Theinventories assessing mother-related and father-related behavior were identicaland consisted of 12 items each, whereas the inventory assessing sibling-relatedbehavior contained a total of 29 items. Respondents were asked to rate on six-point Likert-type scales ranging from 1 (not at all true) to 6 (absolutely true)the extent to which a certain item described the respondent’s behavior toward the


respective family member. For the purpose of the present study, only a subset of sixitems that was identical in each version of the inventory was considered further toconstruct two subscales. The first scale Willingness to Support (the person in ques-tion) consisted of two items in each version (e.g., “I would just drop everythingin order to help my mother/father/sibling”) and had high consistencies and quitehigh test–retest reliabilities (maternal version: α = 0.85, rT1T2 = 0.76; M = 5.11,SD = 1.82; paternal version: α = 0.89, rT1T2 = 0.81; M = 4.91, SD = 1.18; andsibling version: α = 0.72, rT1T2 = 0.60; M = 4.30, SD = 1.31). The second scaledepicted Manifest Conflicts and consisted of four items in each version (e.g., “wefight a lot”), and had very high internal consistencies and quite high test–retestreliabilities (maternal version: α = 0.91, rT1T2 = 0.77; M = 2.52, SD = 1.09,paternal version: α = 0.93, rT1T2 = 0.74; M = 2.47, SD = 1.16, and sibling ver-sion: α = 0.79, rT1T2 = 0.62; M = 2.34, SD = 1.02). The correlations betweenthe two scales were moderately negative for all three versions (mother: r = −0.23;father: r = −0.23; sibling: r = −0.29; ps < 0.01).

Intercorrelations among the emotion- and behavior-related measures of re-lationship quality ranged from r = −0.50 to 0.64 (ps < 0.01) with respect tothe mother, from r = −0.47 to 0.67 (ps < 0.01) with respect to the father, andfrom r = −0.38 to 0.50 (ps < 0.01) with respect to the sibling. With respect toeach family member, two criterion variables were combined into composites de-scribing the valence of an individual’s experienced relationship quality with thatfamily member. Thus, with regard to each target (mother, father, sibling), Attach-ment/Closeness and Willingness to Support (rs = 0.64, 0.67, and 0.50; ps < 0.01)were combined into a measure called Positive Relationship Experiences. Dislikeand Manifest Conflicts (rs = 0.50, 0.49, and 0.39; ps < 0.01) were combined intoa measure called Negative Relationship Experiences. The correlations of positiveand negative measures were medium sized and negative (for mother: r = −0.48,for father: r = −0.46; and for sibling: r = −0.37; ps < 0.01).

Strategy of Data Analysis

To analyze whether justice evaluations mediate the effects of PDT on relation-ship quality with siblings and parents, we used a strategy proposed by Baron andKenny (1986); see also Kenny et al. (1998). This strategy was applied separatelyto each of the possible path models that can be constructed (1) from each of thethree scales for maternal and paternal differential treatment as initial variables, (2)from justice evaluations as possible mediating variables, and (3) from each of thetwo composite indices of positive and negative relationship quality to sibling andto the parent as outcome variables (see Fig. 1a and b for the structure of the pathmodels). In step I, we tested whether there were significant relations between PDTand relationship quality to sibling and parents (path c). Regression analyses wereconducted to predict each of the two composite measures of relationship quality to


sibling and parents from each dimension of PDT (including quadratic componentsto test for curvilinear relations). In step II, we examined whether each dimensionof PDT significantly predicted the hypothesized mediating variables (path a), thatis, justice evaluations (again including quadratic components to test for curvilinearrelations). After inspecting the relations between PDT and the possible mediatingvariables, we tested in step III whether the proposed mediators were still signif-icant predictors of relationship quality to sibling and parents while controllingfor PDT (path b) and whether the effect of PDT on relationship quality (path c′)was reduced after the mediating variables were held constant. Following Baronand Kenny (1986), four conditions must be met to establish justice evaluations asmediating variables: (1) path c must be significant, (2) path a must be significant,(3) path b must be significant, and (4) path c′ must be zero (nonsignificant) or—ifsignificant—at least smaller than path c.

RESULTS

Step I: Testing Relations Between Parental Differential Treatmentand Quality of Relationship with Sibling and Parents

A first sequence of regression analyses was performed to examine whetherthere are significant relations between PDT and relationship quality to siblingand parents which may be mediated by justice evaluations (path c in Fig. 1a).To identify possible curvilinear relations between differential maternal treatmentand measures of the respondent’s experienced relationship quality, the differentialtreatment scale under consideration was entered as the linear component in the firststep and, in the second step, as the quadratic component. To determine the formof possible curvilinear functions, we computed simple slopes at various values ofPDT and tested whether they were significantly different from zero (cf. Aiken andWest, 1991, chapter 5).

Table I reports the coefficients of the linear component of the three PDT scalesin the linear model (first step) and the coefficients of the linear and quadratic com-ponents included in the curvilinear model (second step). With respect to both thepositive and negative measures of relationship quality with the sibling (see tophalf of Table I), the quadratic component indicating curvilinear relations alwayscontributed significantly to the prediction beyond the linear component (recogni-tion: �R2 = 0.061 and 0.045; nurturance: �R2 = 0.039 and 0.026; demand forassuming filial responsibility: �R2 = 0.030 and 0.019). As the sign and the sig-nificance level of the simple slopes indicated, the relations between all differentialmaternal treatment scales and both measures of relationship quality were U shapedfor the negative experiences and invertedly U shaped for the positive experiences(see Fig. 2a and b for two prototypical examples). In other words, relationshipquality with the sibling was most positive and least negative when—according to


Table I. Summary of Hierarchical Regression Analyses to Predict an Individual’s ExperiencedRelationship Quality with Sibling and Mother from Differential Maternal Treatment

Criterion variable

Positive experiences Negative experiences

Predictor variable B SE β R2 B SE β R2

Relationship quality with siblingRecognitiona

First step: linear −0.064 0.049 −0.054 0.003 −0.039 0.037 −0.044 0.002Second step: linear −0.146 0.049 −0.123∗∗ 0.014 0.037 0.016

Quadratic −0.106 0.017 −0.257∗∗ 0.064∗∗ 0.068 0.013 0.220∗∗ 0.047∗∗Nurturancea

First step: linear 0.030 0.050 0.025 0.001 −0.139 0.037 −0.152∗∗ 0.023∗∗Second step: linear −0.088 0.055 −0.073 −0.067 0.041 −0.073

Quadratic −0.080 0.016 −0.220∗∗ 0.040∗∗ 0.049 0.012 0.178∗∗ 0.049∗∗Demandsb

First step: linear −0.010 0.032 −0.013 0.000 0.038 0.024 0.066 0.004Second step: linear 0.048 0.034 0.063 0.004 0.026 0.007

Quadratic −0.039 0.009 −0.189∗∗ 0.030∗∗ 0.023 0.007 0.147∗∗ 0.023∗∗Relationship quality with motherRecognitiona

First step: linear 0.208 0.045 0.189∗∗ 0.036∗∗ −0.320 0.037 −0.337∗∗ 0.114∗∗Second step: linear 0.135 0.045 0.123∗∗ −0.288 0.038 −0.303∗∗

Quadratic −0.094 0.016 −0.246∗∗ 0.092∗∗ 0.041 0.013 0.124∗∗ 0.128∗∗Nurturancea


Quadratic −0.068 0.015 −0.201∗∗ 0.120∗∗ 0.057 0.013 0.194∗∗ 0.097∗∗Demandsb

First step: linear 0.098 0.029 0.137∗∗ 0.019∗∗ 0.032 0.026 0.052 0.003Second step: linear 0.186 0.031 0.260∗∗ −0.026 0.027 −0.042

Quadratic −0.059 0.008 −0.308∗∗ 0.098∗∗ 0.039 0.007 0.233∗∗ 0.048∗∗

Note. B coefficients, standard errors of B, β weights, and R2 of the linear component in the linearmodel (first step) and of the linear and the quadratic component in the curvilinear model (second step).aN = 588.bN = 587.∗p < 0.05; ∗∗p < 0.01.

the participant—both siblings were treated equally by their mother. Relationshipquality became less positive and more negative with both perceived disfavoritismand perceived favoritism.

As to the relationship quality with the mother, the significant regressionweights of the quadratic component (see bottom half of Table I) also indicatedthat the curvilinear model explained significantly more of the variance of thepositive and the negative experiences than did the linear model (recognition:�R2 = 0.056 and 0.014; nurturance: �R2 = 0.032 and 0.030; demand for as-suming filial responsibility: �R2 = 0.079 and 0.045). Sign and significance levelof simple slopes of curvilinear functions revealed that the relations between thevariables only partially fit a U-shaped function, as most of the functions were not


Fig. 2. Curvilinear relations between differential maternal recognition and relationship quality withsibling and mother: (a) positive experiences with the sibling, (b) negative experiences with the sibling,(c) positive experiences with the mother, (d) negative experiences with the mother.

completely symmetric (see Fig. 2c and d for prototypical examples). In addition,the maxima of the curves referring to the Positive Relationship Experiences withthe mother and the minima of the curves referring to the Negative RelationshipExperiences with the mother were somewhat displaced: Whereas relationshipquality with the sibling was most positive and least negative when the respon-dents perceived themselves to be equally treated, the most positive and leastnegative relationship experiences with the mother emerged when the respondentsfelt moderately favored. The relationship quality decreased when the respondentsperceived that they were increasingly disfavored. Extreme cases of favoritism


deteriorated relationship quality to different degrees, depending on the area ofdifferential maternal treatment. Quite a strong deterioration of positive (but notof the negative) relationship experiences emerged when respondents perceivedmuch greater recognition in comparison to their sibling. Perceiving much morenurturance and demand for assuming filial responsibility deteriorated relationshipquality less strongly, albeit significantly.

To test for possible curvilinear relationships between differential paternaltreatment and relationship quality with the sibling and the father, we utilized thesame strategy of data analysis as discussed earlier. Significant regression weightsof the quadratic components in Table II (top half) show that the curvilinear modelexplained significantly more of the variance in both the positive and the nega-tive measures of relationship quality with the sibling than did the linear model

Table II. Summary of Hierarchical Regression Analyses to Predict an Individual’s ExperiencedRelationship Quality with Sibling and Father from Differential Paternal Treatment

Criterion variable



Relationship quality with siblingRecognition

First step: linear −0.022 0.061 −0.019 0.000 0.011 0.046 0.013 0.000Second step: linear −0.139 0.062 −0.120∗ 0.110 0.045 0.128∗

Quadratic −0.115 0.020 −0.314∗∗ 0.089∗∗ 0.098 0.014 0.358∗∗ 0.115∗∗Nurturance

First step: linear 0.181 0.068 0.138∗∗ 0.019∗∗ −0.161 0.051 −0.166∗∗ 0.027∗∗Second step: linear 0.056 0.079 0.043 −0.090 0.059 −0.093

Quadratic −0.062 0.021 −0.183∗∗ 0.043∗∗ 0.035 0.015 0.139∗ 0.042∗∗Demands

First step: linear −0.084 0.046 −0.096 0.009 0.104 0.034 0.159∗∗ 0.025∗∗Second step: linear −0.010 0.052 −0.011 0.053 0.038 0.081

Quadratic −0.044 0.014 −0.182∗∗ 0.035∗∗ 0.030 0.010 0.167∗∗ 0.047∗∗Relationship quality with fatherRecognition


Quadratic −0.082 0.020 −0.216∗∗ 0.109∗∗ 0.072 0.017 0.215∗∗ 0.140∗∗Nurturance


Quadratic −0.041 0.020 −0.116∗ 0.129∗∗ 0.037 0.018 0.120∗ 0.077∗∗Demands

First step: linear 0.128 0.048 0.141∗∗ 0.020∗∗ 0.068 0.042 0.085 0.007Second step: linear 0.211 0.053 0.232∗∗ −0.015 0.047 −0.019

Quadratic −0.049 0.015 −0.196∗∗ 0.050∗∗ 0.049 0.013 0.222∗∗ 0.046∗∗

Note. B coefficients, standard errors of B, β weights, and R2 of the linear component in the linearmodel (first step) and of the linear and the quadratic component in the curvilinear model (second step).N = 360.∗p < 0.05; ∗∗p < 0.01.


Fig. 3. Curvilinear relations between differential paternal recognition and relationship quality withsibling and father: (a) positive experiences with the sibling, (b) negative experiences with the sibling,(c) positive experiences with the father, (d) negative experiences with the father.

(recognition: �R2 = 0.089 and 0.115; nurturance: �R2 = 0.024 and 0.015; de-mand for assuming filial responsibility: �R2 = 0.026 and 0.022). Sign and signif-icance level of simple slopes of the curvilinear functions revealed that the relationsbetween differential paternal treatment and relationship quality were U shaped forthe negative experiences and invertedly U shaped for the positive experiences inthe same manner as had been observed for the differential maternal treatment (seeFig. 3a and b). Relationship quality was most positive and least negative whenrespondents perceived themselves to be about equally treated in comparison to


their sibling. Relationship quality became less positive and more negative whenrespondents perceived disfavoritism or favoritism from their fathers.

The significant regression weights of the quadratic component (see Table II,bottom half) indicate that a curvilinear model also explained significantly more ofthe variance in both the positive and the negative measure of relationship qualitywith the father than did the linear model (recognition: �R2 = 0.042 and 0.041;nurturance: �R2 = 0.010 and 0.011; demand for assuming filial responsibility:�R2 = 0.030 and 0.039). Sign and significance level of simple slopes of curvilin-ear functions clearly indicated U-shaped relations between Differential PaternalRecognition and Differential Paternal Demand for Assuming Filial Responsibilityon the one hand and at least one of the two measures of relationship quality tothe father on the other (see Fig. 3c and d for prototypical examples). Relation-ship quality was most positive and least negative when respondents perceivedslight favoritism. Relationship quality deteriorated with respondents’ perceptionsof being increasingly disfavored or favored. No clear-cut U-shaped relations werefound for Differential Paternal Nurturance. In comparison to the curvilinear func-tions found for relationship quality with the sibling, the maxima of the curvesreferring to the Positive Relationship Experiences with the father and the min-ima of the curves referring to the Negative Relationship Experiences with thefather were slightly displaced. In most of the analyses, the optimum was reachedwhen respondents perceived themselves to be somewhat favored over their sibling.Relationship quality deteriorated rapidly when respondents perceived themselvesdisfavored, whereas relationship quality decreased slightly when respondents per-ceived strong favoritism with respect to recognition and demand for assumingfilial responsibility.

Step II: Testing Relations Between Parental DifferentialTreatment and Justice Evaluations

A second sequence of analyses was conducted to examine whether thereare significant relations between PDT and the hypothesized mediator variables.Hierarchical regression analyses were performed to test for possible curvilinearrelations between PDT and justice evaluations. These analyses applied essentiallythe same strategy as the analyses of step I and were conducted to test path a in themodel outlined earlier (see Fig. 1b).

Table III shows that the quadratic component of both maternal and paternaldifferential treatment indicating curvilinear relations always contributed signifi-cantly to the prediction of justice evaluations of both maternal and paternal dif-ferential treatment beyond the linear component (recognition: �R2 = 0.043 and0.062; nurturance: �R2 = 0.049 and 0.027; demand for assuming filial respon-sibility: �R2 = 0.064 and 0.063). The relations between all dimensions of PDTand justice evaluations were invertedly U shaped as the signs and significancelevels of the simple slope of curvilinear functions revealed (see Fig. 4a and b for


Table III. Summary of Hierarchical Regression Analyses to Predict an Individual’s Justice Evaluationsof Differential Parental Treatment from Perceived Parental Differential Treatment

Criterion variable

Justice evaluation of Justice evaluation ofdifferential maternal treatmenta differential paternal treatmentb


RecognitionFirst step: linear 0.393 0.047 0.324∗∗ 0.105∗∗ 0.283 0.061 0.238∗∗ 0.057∗∗Second step: linear 0.323 0.048 0.266∗∗ 0.182 0.063 0.153∗∗

Quadratic −0.090 0.017 −0.215∗∗ 0.148∗∗ −0.100 0.020 −0.264∗∗ 0.119∗∗Nurturance

First step: linear 0.417 0.049 0.334∗∗ 0.111∗∗ 0.435 0.067 0.323∗∗ 0.104∗∗Second step: linear 0.280 0.053 0.225∗∗ 0.301 0.078 0.223∗∗

Quadratic −0.092 0.016 −0.246∗∗ 0.160∗∗ −0.067 0.020 −0.191∗∗ 0.131∗∗Demands

First step: linear 0.004 0.033 0.005 0.000 −0.007 0.048 −0.008 0.000Second step: linear 0.092 0.035 0.116∗∗ 0.113 0.052 0.125∗

Quadratic −0.058 0.009 −0.276∗∗ 0.064∗∗ −0.070 0.014 −0.283∗∗ 0.063∗∗

Note. B coefficients, standard errors of B, β weights, and R2 of the linear component in the linearmodel (first step) and of the linear and the quadratic component in the curvilinear model (second step).aN = 588.bN = 360.∗p < 0.05; ∗∗p < 0.01.

prototypical examples). The differential treatment of both mother and father wasevaluated as most just when the respondents perceived themselves to be slightlyfavored. PDT was evaluated as less just or even unjust when the respondentsperceived that they were increasingly disfavored. However, being extremely fa-vored led to an evaluation of PDT as less just or even unjust, too (see Fig. 4a andb). The total amount of variance explained was larger when justice evaluationswere predicted from differential parental recognition and from differential parentalnurturance (R2 from 0.119 to 0.160, ps < 0.001) than from differential parentaldemand for assuming filial responsibility (R2 = 0.064 and 0.063, ps < 0.01). Itshould be noticeable that the form of the curvilinear relations between PDT andjustice evaluations resembles the form of the curvilinear relations between PDTand Positive Relationship Experiences with the parents.

Step III: Testing Relations of Justice Evaluations and Parental DifferentialTreatment to Quality of Relationship with Sibling and Parents

The results that emerged for step II clearly indicated that PDT was consis-tently related to justice evaluations, thus fulfilling one necessary condition formediating the effects of PDT on relationship quality to sibling and parents. Ina third sequence of regression analyses, we tested whether justice evaluations


Fig. 4. Curvilinear relations between differential parental recognition and justice evaluations of PDT:(a) maternal recognition, (b) paternal recognition.

fulfilled additional requirements of mediating variables: (1) whether these evalu-ations are significantly related to relationship quality to sibling and parents aboveand beyond PDT, and (2) whether the curvilinear relations between PDT and re-lationship quality are reduced after justice evaluations are also entered into theregression equation. In order to test paths b and c′ of the model outlined earlier(see Fig. 1b), a series of hierarchical multiple regression analyses was performedusing Positive and Negative Relationship Experiences with sibling and parents ascriterion variables. In each analysis we entered—in addition to the linear and thequadratic component of the differential treatment scale under consideration—thejustice evaluation of PDT. Sobel tests (Goodman I version) were conducted todetermine whether there were significant decreases in effect sizes after takingthe mediator variables into account (cf. Baron and Kenny, 1986; Preacher andLeonardelli, 2003). Tables IV and V reveal that in all of the analyses, the justiceevaluations contributed significantly to the prediction of relationship quality to sib-ling and parents above and beyond the linear and the quadratic component of PDT.The incremental variance for relationship quality with parents (from �R2 = 0.108to 0.248) was much larger than the one for relationship quality with siblings (from�R2 = 0.012 to 0.040). As the β weights show, relationship quality to siblingand parents improved the more PDT was evaluated as just and decreased the morePDT was evaluated as unjust. Moreover, there was a significant reduction in the β

weights of the quadratic component of PDT in almost all of the analyses indicatinga mediation of the curvilinear relations between PDT and relationship quality byjustice evaluations (see Tables IV and V; cf. also Tables I and II).


Table IV. Summary of Hierarchical Regression Analyses to Predict an Individual’s ExperiencedRelationship Quality With Sibling and Mother from Differential Maternal Treatment and Justice

Evaluations

Criterion variable


Predictor variable B β �|β|a R2 B β �|β|a R2

Relationship quality with siblingRecognitionb

Linear −0.214 −0.181∗∗ 0.058∗∗ 0.003 0.050 0.056 0.040∗∗ 0.002Quadratic −0.087 −0.211∗∗ −0.046∗∗ 0.064∗∗ 0.058 0.188∗∗ −0.032∗∗ 0.047∗∗Justice evaluation 0.212 0.217∗∗ 0.104∗∗ −0.112 −0.152∗∗ 0.067∗∗

Nurturanceb

Linear −0.143 −0.117∗ 0.044∗∗ 0.001 −0.042 −0.046 −0.027∗ 0.023∗∗Quadratic −0.062 −0.171∗∗ −0.049∗∗ 0.040∗∗ 0.041 0.148∗∗ −0.030∗ 0.049∗∗Justice evaluation 0.194 0.199∗∗ 0.073∗∗ −0.089 −0.121∗∗ 0.061∗∗

Demandsc

Linear 0.032 0.041 −0.022∗ 0.000 0.015 0.026 0.019∗ 0.004Quadratic −0.028 −0.138∗∗ −0.051∗∗ 0.030∗∗ 0.016 0.102∗ −0.045∗∗ 0.023∗∗Justice evaluation 0.181 0.186∗∗ 0.062∗∗ −0.120 −0.164∗∗ 0.043∗∗

Relationship quality with motherRecognitionb

Linear −0.009 −0.008 −0.115∗∗ 0.036∗∗ −0.192 −0.203∗∗ −0.100∗∗ 0.114∗∗Quadratic −0.054 −0.141∗∗ −0.105∗∗ 0.092∗∗ 0.014 0.042 −0.082∗∗ 0.128∗∗Justice evaluation 0.445 0.491∗∗ 0.297∗∗ −0.297 −0.379∗∗ 0.250∗∗

Nurturanceb

Linear 0.117 0.104∗∗ −0.104∗∗ 0.088∗∗ −0.082 −0.084∗ −0.089∗∗ 0.067∗∗Quadratic −0.029 −0.086∗ −0.115∗∗ 0.120∗∗ 0.028 0.097∗ −0.097∗∗ 0.097∗∗Justice evaluation 0.421 0.464∗∗ 0.302∗∗ −0.310 −0.396∗∗ 0.229∗∗

Demandsc

Linear 0.146 0.204∗∗ −0.056∗∗ 0.019∗∗ 0.005 0.008 −0.034∗ 0.003Quadratic −0.033 −0.173∗∗ −0.135∗∗ 0.098∗∗ 0.019 0.115∗∗ −0.118∗∗ 0.048∗∗Justice evaluation 0.443 0.488∗∗ 0.321∗∗ −0.336 −0.430∗∗ 0.221∗∗

Note. B coefficients, β weights, change in β weights after entering justice evaluation, and total R2.aIn comparison to a model that includes only the linear and quadratic component of differentialmaternal treatment (see Table I).

bN = 588.cN = 587.∗p < 0.05; ∗∗p < 0.01.

However, the amount of mediation (as indicated by the amount of β weightreduction) depended on whether relationship quality with the sibling or the parentswas considered. The amount of reduction in β weights of the quadratic componentof PDT after inclusion of justice evaluations indicated that justice evaluationsmediated more strongly the effect of PDT on relationship quality to parents than theeffect of PDT on relationship quality to sibling. Another pattern of results indicatedthat justice evaluations were more important in mediating the effect of differentialpaternal treatment on relationship quality to father than in mediating the effectof differential maternal treatment on relationship quality to mother. In five out ofsix analyses, the β weights of the quadratic component of paternal differential


Table V. Summary of Hierarchical Regression Analyses to Predict an Individual’s Experienced Rela-tionship Quality with Sibling and Father from Differential Paternal Treatment and Justice Evaluations

Criterion variable


Predictor variable B β �|β|a R2 B β �|β|a R2

Relationship quality with siblingRecognition

Linear −0.170 −0.148∗∗ 0.028∗ 0.000 0.126 0.146∗∗ 0.018 0.000Quadratic −0.098 −0.267∗∗ −0.047∗∗ 0.089∗∗ 0.089 0.327∗∗ −0.031 0.115∗∗Justice evaluation 0.172 0.178∗∗ 0.116∗∗ −0.084 −0.117∗ 0.120∗∗

NurturanceLinear 0.003 0.003 −0.040∗ 0.019∗∗ −0.062 −0.064 −0.029∗ 0.027∗∗Quadratic −0.050 −0.149∗ −0.034∗ 0.043∗∗ 0.029 0.115 −0.024 0.042∗∗Justice evaluation 0.174 0.179∗∗ 0.071∗∗ −0.093 −0.129∗ 0.056∗∗

DemandsLinear −0.031 −0.035 0.024 0.009 0.066 0.100 0.019 0.025∗∗Quadratic −0.030 −0.126∗ −0.056∗∗ 0.035∗∗ 0.022 0.123∗ −0.044∗ 0.047∗∗Justice evaluation 0.191 0.197∗∗ 0.072∗∗ −0.113 −0.156∗∗ 0.070∗∗

Relationship quality with fatherRecognition

Linear 0.142 0.119∗ −0.071∗∗ 0.067∗∗ −0.202 −0.192∗∗ −0.053∗∗ 0.099∗∗Quadratic −0.035 −0.092 −0.124∗∗ 0.109∗∗ 0.041 0.123∗ −0.092∗∗ 0.140∗∗Justice evaluation 0.469 0.467∗∗ 0.301∗∗ −0.309 −0.350∗∗ 0.248∗∗

NurturanceLinear 0.247 0.183∗∗ −0.102∗∗ 0.119∗∗ −0.129 −0.108 −0.086∗∗ 0.066∗∗Quadratic −0.010 −0.029 −0.087∗∗ 0.129∗∗ 0.015 0.047 −0.073∗∗ 0.077∗∗Justice evaluation 0.458 0.456∗∗ 0.310∗∗ −0.340 −0.384∗∗ 0.205∗∗

DemandsLinear 0.153 0.168∗∗ −0.064∗ 0.020∗∗ 0.026 0.032 0.013∗ 0.007Quadratic −0.124 −0.050 −0.146∗∗ 0.050∗∗ 0.023 0.106 −0.116∗∗ 0.046∗∗Justice evaluation 0.517 0.514∗∗ 0.298∗∗ −0.362 −0.410∗∗ 0.203∗∗

Note. B coefficients, β weights, change in β weights after entering justice evaluation, and total R2.aIn comparison to a model that includes only the linear and quadratic component of differential paternaltreatment (see Table II). N = 360.

∗p < 0.05; ∗∗p < 0.01.

treatment in the prediction of relationship quality to the father (not the sibling)were reduced to a degree not significantly different from zero (see Table V); thisindicates complete mediation. In contrast, β weights of the quadratic componentof maternal differential treatment in the prediction of relationship quality to themother were only reduced below the threshold of significance in one out of sixanalyses (see Table IV); this indicates only partial mediation.

Additional Analyses for Different Types of Sibling Dyads

In a final step, similar mediational analyses as described for the total samplewere also conducted with respect to subsamples of four types of sibling dyads;


these were generated by cross-classifying gender composition of (same vs. oppo-site sex) and age gap between siblings (small: ≤3 years vs. large: >3 years). Ourfirst goal was to test the extent to which curvilinear relations between PDT on theone hand and justice evaluations and relationship quality on the other vary withthe structural features of the sibling dyad. Equity theory leads one to assume thatthe link between PDT and relationship quality as well as justice evaluations shouldbe more pronounced if the “inputs” of both siblings are perceived as similar; thisshould more likely occur if the siblings are of the same sex as compared to op-posite sex or characterized by a small as opposed to a large age gap. With regardto differential maternal treatment our findings provided partial support for ourhypotheses. As Table VI shows, the two subsamples characterized by a small agegap provided the most consistent evidence for curvilinear relations between PDTon the one hand and relationship quality and justice evaluations on the other. Incontrast, the analogous expectation for sibling dyads of the same as compared toopposite sex was not supported. Regarding paternal differential treatment siblingthe dyads of the same vs. opposite sex and of a small vs. large age gap did notdiffer remarkably in the proportion of significant curvilinear relations betweenPDT and relationship quality as well as justice evaluations. In other words, thesefindings did not support our hypotheses regarding the role of the two structuralfeatures of the sibling dyad.

A second goal of our dyad-specific analyses was to explore the extent to whichjustice evaluations of PDT are systematically related to relationship quality acrossdifferent types of sibling dyads characterized by the same vs. opposite sex and bya small vs. large age gap. Very consistent findings emerged. Justice evaluations ofPDT turned out to be significant predictors of relationship quality to the parentsin all of the analyses and in most of the analyses they were even more powerfulpredictors than perceived PDT itself. In the prediction of relationship quality tothe siblings, however, justice evaluations of differential maternal treatment werefound to be significant predictors of relationship quality in only some of theanalyses regarding each type of dyad. Justice evaluations of paternal differentialtreatment emerged as a significant predictor of relationship quality in most analysesconcerning just one type of dyad (same sex—small age gap).

DISCUSSION

Based on equity theory, the present study examined the extent to which justiceevaluations mediate the effects of currently perceived PDT in middle adulthoodon quality of relationship with adult siblings and older parents and whether jus-tice evaluations of PDT are more powerful predictors of relationship quality thanperceived PDT itself. A model was constructed including justice evaluations asmediating variables and tested in a three-step sequence of regression analysesin line with suggestions made by Baron and Kenny (1986). Step I of our data


Tabl

eV

I.Pr

opor

tion

ofSi

gnifi

cant

Eff

ects

inM

edia

tiona

lAna

lyse

sas

aFu

nctio

nof

Stru

ctur

alFe

atur

esof

the

Sibl

ing

Dya

d

Type

ofsi

blin

gdy

ad

Sam

ese

x/Sa

me

sex/

Opp

osite

sex/

Opp

osite

sex/

smal

lage

gap

larg

eag

ega

psm

alla

gega

pla

rge

age

gap

Tota

lsam

ple

Eff

ect(

path

)M

atD

Ta

PatD

Tb

Mat

DT

cPa

tDT

dM

atD

Te

PatD

Tf

Mat

DT

gPa

tDT

hM

atD

Ti

PatD

Tj

Cur

vilin

ear

effe

ctof

PDT

onre

latio

nshi

pqu

ality

(pat

hc)

8/12

7/12

5/12

1/12

11/1

25/

127/

129/

1212

/12

12/1

2C

urvi

linea

ref

fect

ofPD

Ton

just

ice

eval

uatio

ns(p

ath

a)3/

33/

31/

30/

33/

32/

33/

33/

312

/12

12/1

2E

ffec

tof

just

ice

eval

uatio

nson

rela

tions

hip

qual

ity..

.(p

ath

b)10

/12

11/1

29/

126/

128/

126/

128/

126/

1212

/12

12/1

2..

.to

sibl

ing

(4/6

)(5

/6)

(3/6

)(0

/6)

(2/6

)(0

/6)

(2/6

)(0

/6)

(6/6

)(6

/6)

...

tom

othe

r/fa

ther

(6/6

)(6

/6)

(6/6

)(6

/6)

(6/6

)(6

/6)

(6/6

)(6

/6)

(6/6

)(6

/6)

Not

e.Pr

opor

tion

ofsi

gnifi

cant

effe

cts

inre

latio

nto

the

tota

lnum

bero

freg

ress

ion

anal

yses

.Mat

DT

:mat

erna

ldif

fere

ntia

ltre

atm

ent.

PatD

T:p

ater

nald

iffe

rent

ial

trea

tmen

t.an

=14

4.bn

=87

.cn

=14

1.dn

=84

.en

=14

7/14

8.fn

=86

.gn

=13

9.hn

=92

.iN

=58

7/58

8.jN

=36

0.


analyses for the total sample provided consistent evidence for curvilinear rela-tions between PDT and relationship quality to sibling and parents. Relationshipquality with sibling was best when—according to the participants’ view—bothsiblings were treated equally, and it diminished with increasing favoritism ordisfavoritism. In addition, relationship quality with parents was best when respon-dents felt slightly favored, relationship quality deteriorated when respondents feltdisfavored, and worsened slightly when they felt extremely favored. These resultsreplicated findings from an earlier study by Boll et al. (2003) and provided evi-dence for systematic relations between perceived PDT and relationship quality inthe present data, thus justifying mediational analysis. In the regression analysesof step II conducted in the total sample justice evaluations emerged as fulfillinga first condition for being a mediator, because PDT significantly predicted justiceevaluations. Across all dimensions, both maternal and paternal differential treat-ment was evaluated as most just when the respondents regarded themselves asslightly favored. PDT was evaluated as less just (or even as unjust) when respon-dents perceived that they were either disfavored or extremely favored. Step IIIof our analyses revealed that in the total sample justice evaluations significantlypredicted relationship quality above and beyond PDT itself. After controlling forjustice evaluations, the curvilinear relations between PDT and relationship qualityto sibling and parents disappeared or decreased consistently. These findings—combined with those from step II—indicated that justice evaluations can in factbe regarded as mediating variables. Specifically, the relation between both mater-nal and paternal differential treatment and relationship quality to sibling as wellas the relation between maternal differential treatment and relationship quality tomother were partially mediated, whereas the relation between paternal differentialtreatment and relationship quality to the father was completely mediated by justiceevaluations. Similar mediational analyses conducted for subsamples of siblingdyads of the same vs. opposite sex and of a small vs. large age gap did not yieldsuch a large proportion of significant findings as did our analyses for the totalsample. This may be due in part to the smaller sizes of our subsamples (aboutone-fourth of the total sample) and thus to reduced statistical power.

Prior studies that were based on social exchange perspectives (Bedford, 1992)or on equity theory (Boll et al., 2003) and that referred to justice evaluations ashypothetical constructs could not provide definitive support for justice evaluationsas mediators between PDT and relationship quality, because an alternative me-diating path involving thwarted self-interest is conceivable (cf. Montada, 2002).As the present study explicitly assessed justice evaluations that were tested in amediation analysis, it provides stronger empirical evidence that justice evaluationsactually mediate the effects of PDT on relationship quality. However, our findingsindicating partial (as opposed to complete) mediation point to the possibility that,in addition to justice evaluations, other variables may also mediate the effects ofPDT on relationship quality to the sibling and mother. A likely hypothesis to betested in future research is that PDT can be perceived as both an injustice and a


thwarting of self-interest and that both kinds of perceptions mediate jointly theeffects of PDT on relationship quality. A fascinating issue is whether the relativecontribution of perceived injustice and of perceived thwarted self-interest dependson the direction of PDT. In particular, one may expect that both variables areabout equally important in mediating the effect of perceived disfavoritism on re-lationship quality, whereas justice evaluations are more important than thwartedself-interest in mediating the effects of perceived favoritism on relationship quality.Testing models of multiple mediation would require that, in addition to measuresof perceived injustice, separate and discriminantly valid indicators of thwartedself-interest be included in future studies.

In our mediational analysis, the amount of variance of justice evaluationsexplained by PDT was significant, but not very large (up to 16% in the total sample).Thus, the magnitude and direction of PDT should neither be equated with justiceevaluations of PDT nor be regarded as their only determinants. This leads one toinquire about boundary conditions that may provide a closer link between bothvariables, and about what other factors may contribute additively to the predictionof perceived justice. Identifying relevant predictor and moderator variables seemsto be an important issue, as perceptions of justice, in turn, have emerged asimportant predictors of relationship quality to sibling and parents (see later).

Additional analyses in subsamples of particular types of sibling dyads pro-vided only partial support for the hypothesis that relations between PDT andjustice evaluations vary with structural features of the sibling dyad. In dyadscharacterized by a small age gap between siblings, maternal (but not paternal) dif-ferential treatment was more consistently related to justice evaluations comparedto dyads with a large age gap. However, the analogous hypothesis for siblingdyads of the same as compared to opposite sex was not supported. Thus, structuralfeatures of the sibling dyads figure only to some extent as relevant inputs intosubjective calculations of equity within the family. What other similarities anddifferences between siblings might figure more strongly and moderate the linkbetween PDT and justice evaluations? Referring to the literature on justificationsof PDT (Kowal and Kramer, 1997), it seems highly likely that other perceivedsimilarities and differences between siblings (e.g., with respect to needs, merits)do moderate, because they are possible justifications for PDT. Moreover, disposi-tional sensitivity to befallen injustice (e.g., Schmitt et al., 1995) can be expectedto amplify links between magnitude of PDT and justice evaluations. Finally, per-ceived justice of PDT might not only depend on distributive justice considerationsbut also on appraisals of procedural justice (e.g., Brockner and Wiesenfeld, 1996),which to our knowledge have not been considered with respect to PDT in priorresearch. For instance, the perceived opportunities of self and sibling to influ-ence parental decisions might contribute to the perceived fairness of the resultingparental treatment.

Our analyses revealed that justice evaluations contributed significantly to theprediction of relationship quality above and beyond the direction and magnitude


of PDT itself. The incremental variance was especially large with respect torelationship quality with parents (11–25% in the total sample), but still substantialfor relationship quality with siblings as well (1–4% in the total sample). Withrespect to relationship quality to the sibling, justice evaluations of PDT contributedabout as much to the prediction as did the direction and magnitude of PDT.However, regarding the relationship quality to the parents, the justice evaluationsof PDT contributed much more than did the direction and magnitude of PDT.This finding was very robust, as it emerged in all the four types of sibling dyads(i.e., regardless of gender composition or age gap). One likely explanation whyjustice evaluations are much more powerful predictors of relationship quality toparents than of relationship quality to siblings is that adult children may holdtheir parents much more responsible for PDT than they may hold their siblings.This hypothesis should be tested in future research using explicit measures ofattributions of responsibility.

Some implications can be derived from our study for the developmental issueof whether justice evaluations of PDT remain important beyond childhood andadolescence in predicting relationship quality to family members. In childhood andadolescence, justice evaluations of PDT were systematically related to relationshipquality with sibling (cf. Kowal and Kramer, 1997), and were even stronger pre-dictors than some, though not all, dimensions of PDT (McHale et al., 2000). Thepresent study of middle-aged adults revealed that justice evaluations of PDT re-main important in later periods of life, too, and they seem to be even more importantbecause these evaluations contributed significantly to the prediction above and be-yond PDT from both the mother and the father across all three dimensions of PDT.

Our finding that middle-aged adults’ justice evaluations of PDT are morepowerful predictors of relationship quality to parents than was PDT per se leadsus to wonder whether this applies to justice evaluations of PDT in childhoodand adolescence as well. Unfortunately, prior studies on the perceived justice ofPDT in these younger age groups did not take relationship quality with parentsas a criterion variable into account (cf. Kowal and Kramer, 1997; Kowal et al.,2002; McHale et al., 2000). Thus, future studies on justice evaluations of PDT inchildhood and adolescence should also assess relationship quality to parents sothat also the developmental course of possible outcomes of PDT can be describedmore completely allowing for a further examination of the reasons for possiblesimilarities and dissimilarities across the life-span.

Taken together, the present study demonstrates that beyond the division offamily work in marital relations (cf. Mikula, 1998), there is another kind of in-equality in adult family relationships, namely PDT of middle-aged adult children,in which perceived justice matters and has been shown to be systematically linkedwith relationship quality to adult siblings and aging parents. Nevertheless, somecaveats are in order. Like most studies on PDT and its possible consequences,the present study was cross-sectional and the data were correlational. Obviously,one cannot draw conclusions regarding the direction of causality, because these


data cannot be used to determine whether perceived PDT and justice evaluationshave an effect on relationship quality, whether there is an effect of relationshipquality on perceived PDT and justice evaluations, or whether some third variablehas an effect on both. This seems to be even more important, because it has beenshown in a seemingly different though structurally similar context (i.e., disputesabout distributions of personal and business outcomes) that the relationship qualitybetween participants may strongly influence reactions toward being favored (cf.Loewenstein et al., 1989). Future research should use longitudinal designs andcross-lagged path analyses to clarify the directions of causality between the rele-vant variables. One should also notice that our study considered the perspective ofonly one sibling within each family. Future studies should consider the perspec-tive of other siblings and the parents as well to examine the extent to which thereare converging vs. diverging perceptions of PDT and justice evaluations withinfamilies.

Broadening the perspective on justice in the family and looking for otherinequalities involving aging parents and adult siblings draws attention to thedistribution of caring for older parents and to the distribution of parental inheritanceamong children. Even though both phenomena have been regarded as a likelymatter of justice (e.g., Ingersoll-Dayton et al., 2003; Drake and Lawrence, 2000)and as a possible source of sibling conflict (Kemp and Hunt, 2001; Strawbridgeand Wallhagen, 1991), yet to be analyzed are the extent to which perceptions ofjustice actually mediate such possible outcomes for relationship quality.

Considering the various inequalities and possible injustices involving adultsiblings and their older parents leads us to ask how these various phenomenamight combine to determine the relationship quality of the family members. Aninteresting possibility deserving attention is that the different inequalities interactin shaping justice evaluations which, in turn, have an effect on relationship quality.A specific hypothesis is that the link between PDT in middle adulthood and justiceevaluations is buffered if the disfavored sibling believes that he or she has providedless care for the parents than the favored sibling has. Another hypothesis a waitingempirical validation is that being disfavored with regard to parental inheritance ismore likely to be perceived as just if the disadvantaged child assumes that he or shewas favored during the lifetime of their parents. Following these lines of inquirywill provide justice and family researchers with a more complete picture of the in-equalities and injustices in sibling and parent–child relationships in adulthood andwill contribute to a more comprehensive understanding of fairness in the family.

ACKNOWLEDGMENTS

This study is part of a research project entitled “Parental Differential Treat-ment in Middle Adulthood: Dyadic and Longitudinal Analyses” which wassupported by a grant from the Deutsche Forschungsgemeinschaft (DFG; FE 502/2-3). We thank Holger Spieles for his extremely helpful assistance in this project. We


also thank Lisa Trierweiler for her very helpful support in preparing the Englishmanuscript.

REFERENCES

Aiken, L. S., and West, S. G. (1991). Multiple Regression: Testing and Interpreting Interactions, Sage,Newbury Park, CA.

Aldous, J., Klaus, E., and Klein, D. M. (1985). The understanding heart: Aging parents and theirfavorite children. Child Dev. 56: 303–316.

Baron, R. M., and Kenny, D. A. (1986).The moderator-mediator variable distinction in social psycho-logical research: Conceptual, strategic, and statistical considerations. J. Pers. Soc. Psychol. 51:1173–1182.

Bedford, V. H. (1992). Memories of parental favoritism and the quality of parent–child ties in adulthood.J. Gerontol. Soc. Sci. 47: S149–S155.

Bedford, V. H. (1995). Sibling relationships in middle and old age. In Blieszner, R., and Bedford, V. H.(eds.), Handbook of Aging and the Family, Greenwood Press, Westport, CO, pp. 201–222.

Boll, T., Ferring, D., and Filipp, S.-H. (2001). Struktur und Folgen elterlicher Ungleichbehandlungvon Geschwistern: Forschungsstand und -desiderate [Structure and consequences of parental dif-ferential treatment of siblings: State of research and desiderata]. Zeitschrift fur Entwicklungspsy-chologie und Padagogische Psychologie 33: 195–203.

Boll, T., Ferring, D., and Filipp, S.-H. (2002). Perceived Parental Differential Treatment of Siblings inMiddle Adulthood, Unpublished manuscript.

Boll, T., Ferring, D., and Filipp, S.-H. (2003). Perceived parental differential treatment in middleadulthood: Curvilinear relations with individual’s experienced relationship quality to sibling andparents. J. Fam. Psychol. 17: 472–487.

Brockner, J., and Wiesenfeld, B. M. (1996). An integrative framework for explaining reactions todecisions: Interactive effects of outcomes and procedures. Psychol. Bull. 120: 189–208.

Cicirelli, V. G. (1992). Family Caregiving: Autonomous and Paternalistic Decision Making, Sage,Newbury Park, CA.

Cicirelli, V. G. (1996). Sibling relationships in middle and old age. In Brody, G. H. (ed.), SiblingRelationships: Their Causes and Consequences. Advances in Applied Developmental Psychology,Vol. 10, Ablex, Norwood, NJ, pp. 47–73.

Deutsch, M. (1975). Equity, equality, and need: What determines which value will be used as the basisof distributive justice? J. Soc. Iss. 31: 137–149.

Deutsch, M. (1985). Distributive Justice: A Social Psychological Perspective, Yale University Press,New Heaven, CT.

Drake, D. G., and Lawrence, J. A. (2000). Equality and distributions of inheritance in families. Soc.Justice Res. 13: 271–290.

Ferring, D., Boll, T., and Filipp, S.-H. (2003). Elterliche Ungleichbehandlung in Kindheit und Jugendaus der Perspektive des mittleren Erwachsenenalters [Parental differential treatment in childhoodand adolescence from the perspective of middle adulthood]. Zeitschrift fur Entwicklungspsycholo-gie und Padagogische Psychologie 35: 83–97.

Ferring, D., and Filipp, S.-H. (1999). Soziale Netze im Alter: Selektivitat in der Netzwerkgestal-tung, wahrgenommene Qualitat der Sozialbeziehungen und Affekt [Social networks in old age:Socioemotional selectivity, perceived quality of social interactions, and affect]. Zeitschrift furEntwicklungspsychologie und Padagogische Psychologie 31: 127–137.

Ihinger, M. (1975). The referee role and norms of equity: A contribution toward a theory of siblingconflict. J. Marr. Fam. 37: 515–524.

Ingersoll-Dayton, B., Neal, M. B., Ha, J.-H., and Hammer, L. B. (2003). Redressing inequity in parentcare among siblings. J. Marr. Fam. 65: 201–212.

Kemp, S., and Hunt, F. (2001). Exploring the psychology of inheritances. Zeitschrift fur Sozialpsy-chologie 32: 171–179.

Kenny, D. A., Kashy, D. A., and Bolger, N. (1998). Data analysis in social psychology. In Gilbert,D. T., Fiske, S. T., and Linzey, G. (eds.), The Handbook of Social Psychology, McGraw-Hill,Boston, MA, Vol. 1, pp. 233–265.


Kowal, A., and Kramer, L. (1997). Children’s understanding of parental differential treatment. ChildDev. 68: 113–126.

Kowal, A., Kramer, L., Krull, J. L., and Crick, N. R. (2002). Children’s perceptions of the fairness ofparental preferential treatment and their socioemotional well-being. J. Fam. Psychol. 16: 297–306.

Loewenstein, G. F., Thompson, L., and Bazerman, M. H. (1989). Social utility and decision makingin interpersonal context. J. Pers. Soc. Psychol. 57: 426–441.

McHale, S. M., Updegraff, K. A., Jackson-Newsom, J., Tucker, C. J., and Crouter, A. C. (2000). Whendoes parents’ differential treatment have negative implications for siblings? Soc. Dev. 9: 149–172.

Mikula, G. (1998). Division of household labor and perceived justice: A growing field of research.Soc. Justice Res. 11: 215–241.

Montada, L. (2002). Justice to the justice motive. In Ross, M., and Miller, D. T. (eds.), The JusticeMotive in Everyday Life, Cambridge University Press, New York, pp. 41–62.

Preacher, K. J., and Leonardelli, G. J. (2003). Calculation for the Sobel Test. An Interactive CalculationTool for Mediation Tests, Retrieved from http://www.unc.edu/preacher/sobel/sobel.htm.

Schmitt, M. J., Neumann, R., and Montada, L. (1995). Dispositional sensitivity to befallen injustice.Soc. Justice Res. 8: 385–407.

Silverstein, M., Parrott, T. M., and Bengtson, V. L. (1995). Factors that predispose middle-aged sonsand daughters to provide social support to older parents. J. Marr. Fam. 57: 465–475.

Sprecher, S., and Schwartz, P. (1994). Equity and balance in the exchange of contributions in closerelationships. In Lerner, M. J., and Mikula, G. (eds.), Entitlement and the Affectional Bond: Justicein Close Relationships, Plenum Press, New York, pp. 11–41.

Strawbridge, W. J., and Wallhagen, M. I. (1991). Impact of family conflict on adult child caregivers.Gerontology 31: 770–777.

Suitor, J. J., and Pillemer, K. (2000). Did mom really love you best? Developmental histories, statustransitions, and parental favoritism in later life families. Motiv. Emot. 24: 105–120.

Walster, E., Berscheid, E., and Walster, G. W. (1973). New directions in equity research. J. Pers. Soc.Psychol. 25: 151–176.

Weiner, B. (2001). Responsibility for social transgressions: An attributional analysis. In Malle, B. F.,Moses, L. J., and Baldwin, D. A. (eds.), Intentions and Intentionality: Foundations of SocialCognition, MIT Press, Cambridge, MA, pp. 331–344.

Parental Differential Treatment

Documents

justice evaluations

social justice research

role of pdt

scholars of social justice

extent justice evaluations

past researchon pdt

department of psychology

university of luxembourg