Professional Documents
Culture Documents
Downey AngerTransmissionMother 1999
Downey AngerTransmissionMother 1999
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
National Council on Family Relations is collaborating with JSTOR to digitize, preserve and
extend access to Journal of Marriage and Family
A Comparison of Mothers in
Chronic Pain and Well Mothers
Figure
Key Words: anger, daily diary, emotion transmission, pain, 1 summarizes this basic model of distress
parenting. transmission. We acknowledge that the transmis-
Path A
sion process is undoubtedly bidirectional. Chil- test of whether the operation of the model de-
dren's distress influences parents, and parents' dis- pends on the context or the source of the distress.
tress influences children. (For reviews, see Dix, The context or source of distress may possibly lead
1991; Repetti & Wood, 1997a.) However, in this to either the accentuation or the dampening of the
article we focus on transmission from parent to transmission process outlined in Figure 1. We
child. Consistent with the model in Figure 1, have termed these moderations of the basic model
researchers have shown that parents' distress helps of distress transmission the accentuated distress-
explain the association between chronic and acutetransmission model and the distress-containment
parental stressors and disruptions in parentingmodel.
(Conger et al., 1995; Elder, Caspi, & Downey,
1986; McLoyd & Wilson, 1990). Parenting qual- Accentuated distress-transmission model. Under
ity, in turn, helps account for the link between severe chronic stress, such as that experienced by
parental distress and child difficulties. Althoughfamilies in which the mother has chronic pain, the
tests of the model generally have yielded evidence relation between the mother's distress and the
consistent with the model, a number of important child's distress may be stronger than in more typi-
questions remain. This study addresses three of cal families. Perhaps living continuously with
these questions. pain or with other stressors taxes mothers' abili-
ties to inhibit translating their distress into hostile
Does the Model Operate Within Families interactions or complete withdrawal, which are
as Well as Between Families? distressing to children. Living with a parent who
is chronically distressed because of illness also
Prior studies of the model of distress transmission may sensitize children to be hyperreactive to daily
typically have treated parental distress as a stable, changes in the ill parent's behavior.
individual difference variable (e.g., Conger et al.,
1995). Considerably less attention has been paid Distress-containment model. Even when parents
to determining whether the emotions of parents are highly distressed, however, the source of the
and their children covary over time within families distress may determine how much it influences
and whether, if such an association exists, it is me- parenting and child distress. Experimental social-
diated by concomitant fluctuations in parenting psychological research has shown that people use
quality. (For exceptions, see Repetti & Wood, their experimentally induced current mood state
1997b; Snyder, 1991.) Our study uses daily diary as a basis for evaluating other people's behavior
data from mothers and their adolescent children to unless they are explicitly reminded of the source
examine the links among maternal distress, parent- of their mood (Cervone, 1994; Schwartz & Clore,
ing, and child distress in typical families and in 1983, 1988). When reminded, they correct for the
chronically stressed families in which the mother externally induced mood and make their evalua-
experiences heightened distress because of chronic tion of the person's behavior. These experimental
pain. (See Larson & Almeida, 1999, for a discus- findings suggest that when parents have a com-
sion of the advantages of the diary approach.) pelling noninterpersonal explanation for their dis-
tress, such as physical pain, they may adjust for
Does the Source or Context of Distress Matter? the contribution of pain to their distress when
they evaluate the negativity of the child's behav-
Both typical families and families in which the ior, and thus they may be less likely to engage in
mother has chronic pain are included to allow a harsh parenting. For mothers with chronic pain,
knowing the connection between their pain and found that when mothers were induced to feel
their distress may make them aware that, when angry, they had more negative expectations of their
they feel distressed, they are vulnerable to over- children's behaviors in problem situations, made
perceiving and overreacting to negative child be- more negative attributions for child behavior, and
havior. In contrast, typical mothers may be less recommended more negative parental responses
aware of the effect of their distress on their stan-than when they were in a neutral emotional state.
dards for acceptable child behavior, and they may In a daily diary study of families with preschool
be more likely to use their distress as a barometer children, Downey et al. found that a mother's daily
of the negativity of their child's behavior. Conse- anger had an effect on her parenting that was both
quently, when they feel more distressed than stronger than and independent of the effect of either
usual, typical mothers may be more likely than daily depressed mood or anxiety, neither of which
mothers with chronic pain to withdraw positivity had a significant independent effect on parenting.
or to engage in harsh, coercive parenting. Thus, prior research provides some evidence
Similarly, equivalent levels of harsh or distant that maternal anger is particularly likely to prompt
parenting may evoke less anger in children when negative, unsupportive interactions between
they can attribute parents' harshness or disinterest mother and child, interactions that, in turn, induce
to something (i.e., uncontrollable pain) other than anger in the child. We test whether there is
parental malevolence. This proposition has a basis stronger evidence of transmission of emotion in
in social-psychological studies showing that people the case of daily anger than in the case of daily
tend to attribute other people's negative behavior depressed mood or anxiety. If so, we will test
(e.g., a parent ordering the child to clean up his or whether this is because of the disruptive impact of
her room) to dispositional negativity on the part of anger on parenting. If the hypothesized process of
the other person (e.g., the parent is a jerk) in the anger transmission operates at a daily level, we
absence of a more obvious situational explanation will test whether it does so irrespective of the
(e.g., the parent is in pain; Nisbett & Ross, 1980). mother's concomitant levels of depressed mood
and anxiety.
Which Parental Emotion Has the Most Current Study
Disruptive Impact on Children?
We tested the three alternative models of distress
Despite mounting evidence that parents' emotions transmission with daily diary data from mothers
and their adolescent children in control families
mediate the impact of parental stressors on parent-
ing and children, relatively little is known aboutand in families in which the mother had RSDS, a
rare, unpredictable complication of soft-tissue
the distinctive influence of specific emotions, such
as anger, anxiety, and depression (Dix, 1991). trauma characterized by persistent burning pain,
Prior research typically has used either a global lowered pain threshold, and pain induced by non-
measure of distress or has focused on a single as-noxious stimuli (e.g., minor friction; Chard, 1991).
pect of distress, usually depression (Downey & This disorder is particularly well suited for testing
Coyne, 1990). However, the aspect of distress withalternative models of the anger-transmission
the most disruptive impact on parenting may be hypothesis using daily diaries. Because people
anger, rather than depression or anxiety. To ex- with RSDS tend to experience clinically signifi-
plain why children of clinically and subclinicallycant levels of pain-related distress (Feldman,
depressed parents show externalizing difficulties Downey, & Schaffer-Neitz, 1998), daily fluctua-
as well as internalizing difficulties, Downey and tions in distress occur around a higher mean level
Coyne posited that the anger and irritability thatof distress than is typical. Thus, the disorder allows
often accompany parental depression explain why us to examine whether the process of distress
transmission is accentuated in families in which
depressed parents tend to engage in the harsh,
the mother has chronically high levels of distress.
coercive parenting that contributes to externalizing
difficulties. Furthermore, pain provides a compelling noninter-
In support of this position, researchers havepersonal explanation for RSDS patients' daily dis-
shown that induced and naturally occurring anger
tress and for any harsh and unsupportive parenting
promotes harsh, unsupportive parenting (Dix,in which they might engage. Because the condi-
Reinhold, & Zambarano, 1990; Downey, Osatin- tion is not life threatening, any negative effects on
ski, & Pettit, 1993; Elder et al., 1986; Patterson,
family members do not reflect realistic fears about
the prognosis, as might be the case for people with
1982). In a particularly illuminating study, Dix et al.
cancer (Compas et al., 1994; Compas, Worsham, study is limited to families in which the mother had
Ey, & Howell, 1996). Finally, although pain expe- RSDS. One family in which both parents had
rienced by people with RSDS is more severe, on RSDS was included. On average, mothers had been
average, than other forms of chronic pain (De- diagnosed 3 years before participating in the study
Good, Cundiff, Adams, & Shutty, 1993; Melzack (SD = 2.1 years). Many RSDS mothers experienced
& Katz, 1992), there are large day-to-day fluctua- substantial disability: 39% reported loss of the use
tions in pain and mood (Feldman et al., 1998). of at least one arm, and 40% reported either being
Throughout this article, we refer to families in unable to walk or needing crutches.
which the mother has RSDS as RSDS families. RSDS families and control families did not
To demonstrate support for the unmoderated differ significantly in the age, grade, or sex of the
distress-transmission model depicted in Figureparticipating
1 children. The mean age of adoles-
we must show: (a) a positive link between the cents in the study was 14 years (SD = 1.9), and the
mother's distress and the child's distress, total Path
median grade was eighth grade. Forty-six percent
A; (b) a positive link between the mother's distresswere girls, and 54% were boys. The number of
and her unsupportive and negative parenting, Path children in the families ranged from 1 to 6, with a
B; (c) a positive link between the mother's nega- mean of 2.4 (SD = 1.1). The mean number of
tive, unsupportive parenting and the child's dis- children was significantly lower in RSDS families
tress, when controlling for the mother's distress, than in control families (RSDS families: M= 2.2,
Path C; and (d) a substantially reduced link be- SD = .83; control families: M = 2.67, SD = .13;
tween the mother's and the child's distress when t(79) = 2.09, p < .05).
statistically controlling for negative, unsupportive Participating families were Caucasian, with two
parenting, unmediated Path A. Finding a stronger exceptions (one Native American, one Hispanic).
association between the relevant variables depicted Both non-Caucasian families were in the RSDS
by Paths A, B, and C in RSDS families than in group. The median family income was in the
control families would support the accentuated $35,000-$45,000 range and did not differ signifi-
distress-transmission model. If the association be- cantly between RSDS families and control fami-
tween the relevant variables were significantly lies. Ninety percent of fathers were currently em-
weaker in RSDS families than in control families, ployed, and the rate of paternal employment did
then the distress-containment model would be not differ significantly between RSDS families
supported. Based on prior research, we expect and control families. Sixty-eight percent of control
that anger is the component of distress thatmothers will were employed outside the home, but
show the strongest evidence of transmission from only 20% of women with RSDS were employed,
mother to child. X2(1, n = 82) = 19, p < .001. The relatively low
employment rate of RSDS mothers is not surpris-
METHOD ing, given the level of physical impairment expe-
rienced by them.
Both RSDS mothers and control mothers had
Sample
completed an average of 14 years of education
Data were from mothers and their adolescent chil- (SD = .42). Fathers in RSDS families had com-
dren in a community sample of 82 two-parent fam- pleted significantly fewer years of education than
ilies that included a child who was 10-18 yearscontrol fathers (RSDS families: M = 14, SD = 1.7;
old. In 40 of these families, the mother had RSDS.control families: M = 15, SD = 2.3; t(76) = 2.75, p <
The remaining 42 families made up the control .01). Parents in RSDS families were significantly
group. In households with more than one childyounger than the control parents-mothers in
between 10 and 18 years, the mother was asked toRSDS families: M = 37 years, SD = 4.1; mothers in
select one child as the focal child of the study. control families: M = 41 years, SD = 3.9; t(79) =
Control families were recruited through adver-3.9, p < .001; fathers in RSDS families: M = 40
tisements placed in a newspaper serving suburban,years, SD = 6, fathers in control families: M = 44
semi-rural, and rural areas, paralleling the residen-years, SD = 6.1; t(75) = 3.19,p < .01.
tial distribution of the RSDS families. Families in
which one parent had RSDS were recruited nation- Procedure
wide through advertisements posted in the RSDS
Association newsletter and through announcements We asked the mother and the participating child to
made at meetings of RSDS support groups. The separately complete a brief structured question-
naire (diary) before retiring each night on 28 con- pushed, shoved, or hit my child," "I tried to make
secutive days. Where relevant in the questionnaire, my child feel guilty," "I yelled or screamed at my
the mother's responses were focused on the child child," "I told my child that I was disappointed in
in the study. If a participant forgot to complete a him/her" (a = .62). The five positive parenting
particular day's dairy, we asked them to complete it items were: "I showed my child how happy she/he
as soon as possible and to note the date and time makes me," "I considered my child's needs when
when the diary was completed. A packet of seven making my plans," "I negotiated or reasoned with
questionnaires was mailed to participants at the be- my child," "I made time to listen to my child," and
ginning of each week, and participants returned the "I helped my child with things that were important
packets by mail at the end of the week. Each fam- to him/her" (a = .76). Measures for negative and
ily member returned the completed week's worth positive parenting were obtained by averaging the
of diaries in a separate sealed envelope. The first ratings of the respective items.
packet also contained a background questionnaire.
After completing the study, families received $35. Compliance and Attrition
RSDS families were given the option of having the
money donated to the RSDS Association. Ninety percent of mothers and 85% of children
completed at least 3 weeks of the diaries. Mothers
Measures completed an average of 26 diary days. Children
completed an average of 25 days. Most diary en-
Background questionnaire. The mother completed tries were completed on the appropriate day (moth-
the background questionnaire at the beginningers: of 97%; children: 99%). Most of the remaining
the diary study. It included questions on the fam-diaries were completed 1 day late (mother: 2%;
child: 1%). RSDS family members and control
ily's composition, the family's income, and parents'
education levels and occupations. Questionnaires family members did not differ significantly in the
completed by mothers with RSDS included items number of diaries completed or in the number of
about the onset and severity of their disability. days when the diary was completed late. We used
analyses of variance (ANOVAs) and chi-square
Diary questionnaire. The daily diary questionnairetests to determine if there were systematic differ-
included items about distress and mother-child in- ences in completion rates on demographic variables
teraction. The distress items were adapted from the (gender, religion, parents' education, employment
Affect Balance Scale (Derogatis, 1975) and as- status, age of child, and family income) and, for
sessed anger, anxiety, and depressed mood. Partici- RSDS families, on medical variables (length of ill-
pants rated the intensity of each of these feelings ness, RSDS stage, and level of physical disability).
during the past 24 hours using a 3-point scale, We found no systematic differences. Moreover, the
ranging from no to very much. The four items as- demographic variables did not interact with RSDS
sessing anger were angry, irritable, enraged, frus- versus control status to predict completion rates for
trated (mother's a = .80, children's ca = .75). The either mothers or children.
eight items assessing depressed mood were guilty,
hopeless, sad, depressed, worthless, unhappy, blue,
Analyses
miserable (mother's a = .89, children's ( =.86).
The five items assessing anxiety were nervous, This study yielded a data set with two levels of
afraid, agitated, tense, anxious (mother's ax = .81, analysis. The within-family level reflects daily
children's c = .70). variation over time within a family or within a
We assessed mother-child interaction by hav- focal family member (e.g., daily variation in child
ing mothers indicate the extent to which they had anger). The between-family level reflects differ-
engaged in various indicators of positive and nega-ences between families or focal members of fami-
tive parenting each day on a 3-point scale, ranging lies (e.g., differences in mean anger between RSDS
from no to a lot. The items used to index parenting mothers and control mothers). The within-family
were drawn from existing measures of positive level of analysis can be used, for example, to esti-
and negative parenting and were tested in a prior mate how strongly the mother's anger and the
daily diary study (Downey et al., 1993). Factor child's anger are linked within a family, as well as
analyses of data from our sample showed that the the average level of anger for members of a family.
positive and negative items loaded on two distinct The between-family level of analysis can be used
factors. The four negative parenting items were: "I to examine whether the processes in families that
include an RSDS mother (RSDS families) differ are over the control children. We assume that ei is
from the processes in control families (e.g., a residual component of the dependent variable
whether RSDS mothers and control mothers dif- specific to each family and that ei is a normally
fer in average daily levels of anger, whether
distributed random variable with a mean of 0 and
RSDS families differ from control families in how a constant variance.
strongly the anger of mother and child is linked).
We conducted analyses using a multilevel orDifferences in relationships among variables. A
hierarchical linear model approach, which permitsmultilevel approach to assessing the relation
the simultaneous analysis of within-family andamong variables measured at the daily level (e.g.,
between-family variation (Bolger & Zuckerman,the mother anger-child anger relation) and to es-
1995; Bryk & Raudenbush, 1992; Kenny, Kashy,tablishing whether the relation differed for RSDS
& Bolger, 1998). In contrast, conventional linearfamilies and control families also involves speci-
models either aggregate across within-family fying two equations. The within-family equation
data, resulting in information loss, or conflate specifies that the value of the dependent variable
within-family and between-family variation, re-(e.g., child's anger) for a given family on a given
sulting in incorrect tests of significance. (See day, CAngt, is predicted by the level of the inde-
Kenny et al., 1998.) We wanted to examine twopendent variable (e.g., mother's anger) on the
types of differences between RSDS families andsame day, MAngt, and by a residual component of
control families. First, we looked at mean differ-the dependent variable, rt, which is specific to
ences in daily levels of distress and parenting each day and is assumed to have a mean of 0 and
quality. We asked, "What is the main effect ofa constant variance across families and days. The
RSDS on the dependent variables of interest?" equation is:
Then we looked at differences in the relationship
among variables measured at the daily level. We CAngt = ao + alMAngt + rt (3)
asked, "Does the relationship between mother's
and child's distress differ in RSDS and control
Estimates of a0 and a, are obtained for all fami-
families? That is, is there an interaction effect?"
lies in the sample. The estimation of a0 is given in
Equations 1 and 2. The estimation of a, follows.
Mean differences. A multilevel analysis approach
The between-family equation specifies that for
to assessing whether focal members of RSDS
each family, the effect (a1i) on the dependent vari-
families differed from focal members of control
able of the independent variables is a function of
families in anger, for example, on the average day
whether family i is an RSDS family:
during the diary period requires that we estimate a
within-family and a between-family equation.
ali = Co + ClRSDSi +fi (4)
The within-family equation specifies that a family
member's (e.g., the child's) anger on a given day,
CAngt, is a function of their mean level across all Assuming that RSDSi is coded 0 for the control
days, a0, plus a residual component specific to group and 1 for the RSDS group, co is the mean
each day, q,:
association between maternal anger and child
anger in the control group, and cl is the number
of units higher that the association between ma-
CAng = ao +qt (1)
ternal anger and child anger is in RSDS families
than in control families. Thus the coefficient, cl,
The between-family equation specifies that
can be thought of as mother's anger x RSDS inter-
members of RSDS families (e.g., children) differ
action effect and will be reported as such in the
in their mean level of anger acrossresults all days from
section. We assume that f is a residual
members of control families:
component of the dependent variable specific to
each family and that f is a normally distributed
aoi = bo + blRSDSi + ei (2) random variable with a mean of 0 and a constant
variance.
Assuming that RSDSi is coded 1 for children of If we substitute bo + blRSDSi + ei for aoi and co
RSDS mothers and 0 for control children, then+ bo clRSDSi + fi for ali in Equation 3, we get the
is the mean anger of the control children, and b1following
is equation:
how many units higher in anger the RSDS children
CAngt = bo + b1RSDSi + (co +fi ) their distress does not appear to spill over into their
MAngt + c RSDSi x MAngt + rt + ei (5) parenting. We test this possibility explicitly.
TABLE 1. MEAN DAILY DISTRESS FOR MOTHERS AND CHILDREN BY FAMILY TYPE
Mother Child
Daily Distress M SD M SD F M SD M SD F
Anger 1.70 (.52) 1.31 (.42) 43.19*** 1.32 (.43) 1.31 (.41) .02
Anxiety 1.74 (.53) 1.31 (.36) 46.65*** 1.31 (.43) 1.25 (.39) 1.37
Depression 1.71 (.52) 1.21 (.28) 59.33*** 1.17 (.33) 1.17 (.29) .03
***p < .001.
Table 3), indicating that RSDS mothers were RSDS families. For control families, the coefficient
less
for the total association between maternal anger
likely than control mothers to engage in negative
parenting when they felt angry. and child's anger is .11, whereas it is -.001 for the
RSDS families (Table 4, Model A). For control
We conducted a similar analysis with positive
parenting as the dependent variable. In this families,
case, the part of the total anger-transmission
the interaction term was not significant (beffect= .06,that is mediated by negative parenting is .05,
F(1,79) = 1.2, ns). Next, we examined whether maternal
the anger -> negative parenting coefficient
anger levels of children of RSDS mothers(.22, Model B) x negative parenting -> child anger
were
less responsive than those of control children coefficient
to the (.24, Model D), and the part that is un-
mother's negative parenting. We tested thismediated
possi- is .06 (Model D). For RSDS families,
bility by estimating a model in which child's the part of the total anger-transmission effect that
anger
is mediated by negative parenting is .015, maternal
was the dependent variable, and negative parent-
anger -> negative parenting coefficient (.11,
ing, RSDS status, and a term for the interaction
between negative parenting and RSDS status Model
were B) x negative parenting -> child anger co-
efficient (.14, Model D), and the part that is un-
the independent variables. A marginally significant
mediated is -.014 (Model D). Although the links
negative interaction term indicated that child's
between maternal anger and negative parenting
anger was less strongly associated with negative
(Model B) and between negative parenting and
parenting in RSDS families than in control families
child's anger (Model D) are significant for RSDS
(b = -.17, F(1,66) = 3.3, p < .08). When a similar
analysis was conducted for positive parenting, families,
the they are about half the magnitude that
interaction term was nonsignificant (bthey
= .01,
are for control families.
F(1,80) = .01, ns). When we controlled for mothers' daily anxiety
or depressed mood, the findings reported in Table 4
In sum, these findings indicate that the anger-
transmission model depicted in Figure 1 is weredamp-
not altered. Moreover, controlling for between-
ened in RSDS families. To more clearly show family
thevariables on which RSDS families and
differences between RSDS families and control control families differed significantly (parental
age, father's education, and the number of children
families in anger transmission via negative parent-
ing, we recomputed the relevant models in Tablein the
2 family) did not account for why the process
separately for both groups (i.e., Models A, B, ofD).
anger transmission differed in RSDS families
and control
The results of this analysis are reported in Table 4 families.
and depicted graphically in Figure 2, where the
coefficients for control families are above the line, DISCUSSION
and the coefficients for RSDS families are below the
line. The results show that the anger-transmission The model of distress transmission depicted in Fig-
model operates for control families but not for ure 1 has received considerable support in between-
subjects designs in which parental distress is may operate differently, depending on the
model
treated as a stable, individual difference. Our results
source of parental distress.
show support for the model with data from a
within-subjects design in which the focus is on
Why Is the Transmission Process Dampened in
establishing whether the distress levels of mothersFamilies in Which the Mother Has RSDS?
and their adolescent children covary across days.
However, our findings also suggest a numberOne of potential explanation is that the mothers and
their children may simply have adapted to chroni-
refinements that need to be made to the general
model of distress transmission. cally high levels of maternal anger. We found no
First, we found evidence of a unique correspon- support for this explanation because the process of
dence between mother's anger and child's anger at anger transmission was not dependent on the
the daily level. This correspondence was mediatedmother's mean level of anger across the diary
partially by harsh mother-child interactions but notperiod. An alternative explanation is based on the
by a reduction in positive mother-child interactions. assumption that the effect of people's anger on
We did not find a unique correspondence between their behavior toward others depends on the attri-
mother and child in daily depressed mood or anxiety. butions they make for their anger (Cervone, 1994;
Second, the overall support for the model in theSchachter & Singer, 1962; Schwartz & Clore,
case of anger masked important differences in the 1988). Because pain provides a ready explanation
transmission process between typical families and for RSDS mothers' anger, these mothers may be
families in which the mother experienced chroni- less likely than control mothers to attribute their
cally heightened distress because of RSDS. The anger to the actions of their children. In addition to
model of anger transmission operated in typicalhaving a heightened awareness of the role of pain
families, but it did not operate in RSDS families.in their anger, they also may be more practiced at
This supports the hypothesized model of distressinhibiting the translation of angry thoughts and
containment for these families. The within-dayfeelings into harsh, coercive parenting. Thus, their
correspondence between mother's anger andthreshold for reacting harshly to their child when
child's anger was significantly reduced in RSDS they feel angry may be higher than that of control
families, relative to typical families, because RSDS women. Testing these explanations requires assess-
mother's anger was less likely to spill over into ing mothers' attributions for their feelings each day
negative parenting and because the children ofand their efforts to regulate the impact of their mood
RSDS mothers were marginally less reactive to on their behavior. Such information would help to
their mother's anger. The reduced transmission ofestablish whether control mothers transmit anger to
anger from mother to child helps explain why chil- their children to a greater extent than RSDS moth-
dren with RSDS mothers did not differ from their ers because the control mothers are more likely
counterparts in control families in daily levels of than RSDS mothers to attribute their anger to their
anger, despite having mothers who reported signif- child's behavior and to let the child know that they
icantly higher levels of daily anger than the control are angry at him or her.
mothers reported. Similar explanations may account for why chil-
In sum, the results of our research on the dren of RSDS mothers are marginally significantly
process of distress transmission underscore the less responsive to their mother's negative parent-
value of distinguishing among different types ofing. These children may be more likely to explain
emotion and of addressing the possibility that the their mother's negativity as resulting from her
pain than from her negative personality. This ex- Second, we rely on mothers' reports of their
planation is illustrated by the response given by
own parenting behavior. However, a similar pattern
the child of a RSDS mother. Asked how she of results emerged when we substituted the child's
would interpret a hypothetical parental behavior
report of the occurrence of mother-child conflict for
that embarrassed her in front of a friend the
(i.e., if
mother's report of negative parenting. Specifi-
the parent angrily ordered the child to tidy up her
cally, the association between maternal anger and
room in front of a friend), the child stated, children's
"If my reports of conflict with their mothers was
father did it, it would be because he's beingdampened
a jerk; in families with RSDS mothers. In future
if my mother did it, it would be because she is in it will be important to complement self-
studies,
pain because of her illness." The more benevolent
reports of parenting with reports by other family
explanation given to the behavior of themembersparent or with direct observations. (See for ex-
with RSDS is likely to modulate the child'sample,
affec-Repetti & Wood, 1997b.)
tive and behavioral response. Third, it should not be concluded from our find-
It is also possible that family-level characteris-
ings that the mother's pain condition has no detri-
tics-beliefs, routines-may help to structure daily
mental effect on family life. A different picture
life in ways that facilitate anger containment
emerged in from RSDS family members' qualitative
RSDS mothers. Qualitative reports from study par-of how the illness had changed their fam-
accounts
ticipants suggest that the families of women ily. with
They spoke of dashed dreams for the future, of
RSDS often believe that they should make every worries, of the feeling that family life had
financial
effort to minimize family stress and conflict, which
become centered around the illness. Nonetheless,
they believe exacerbate pain symptomatology.
the toll of the illness does not emerge in the daily
Thus, members of these families may work to min- levels of children. Yet, qualitative data re-
distress
imize stress and conflict by establishing vealed
familythat, when asked to focus more directly on
routines and using proactive coping strategies.
how theIn illness has changed their lives, children
addition, RSDS mothers' awareness of the limits
readily identified its disruptive impact.
on their energy and activity may prompt them to
Finally, our criteria for the sample may have
direct their scarce resources toward their most im-
inadvertently selected RSDS families with a rela-
portant goals, such as parenting effectively. tively more adaptive approach to coping with the
mother's health crisis, and thus our sample may be
Caveats and Conclusions unrepresentative of RSDS families, in general.
Caveats notwithstanding, this study illustrates
Several caveats need to be considered when evalu-
that it is sometimes possible to function ade-
ating our results. First, the hypothesized unidirec-quately in important social roles, despite being
tional model linking maternal anger with childdistressed. Understanding the circumstances under
anger via parenting clearly does not capture the which anger, depressed mood, or anxiety do and
bidirectional nature of anger transmission in parent-
do not translate into impaired interpersonal func-
child dyads (Dix, 1991). Moreover, because we ex- tioning is clearly a topic for further research.
amined same-day associations between parent and
child distress, it was not possible to determine un-
NOTE
ambiguously the direction of causality in the asso-
ciations that we documented. One way around this This research was supported by grants from the National
would be to examine whether mother's current Institute of Mental Health (R29-MH51113) and the
anger predicted change from one day to the next Harry
in Frank Guggenheim Foundation and by a W. T.
Grant Faculty Scholar Award to the first author. We
parenting or child anger. Analyses revealed no
thank the families who participated in our research, the
cross-day effects either from mother to child RSDS
or Association, and regional support groups for help
from child to mother, however. Thus, the effect recruiting
of families for the study. For their assistance in
mothers' anger on children or vice versa was conducting
re- the study, we thank Miranda Koss, John
Mathews, and Claudia Rinc6n. Niall Bolger, Scott Feld-
stricted to the same day. This finding suggests that
man, Antonio Freitas, Reed Larson, and David Almeida
the 24-hour period covered by daily diaries may
provided insightful comments on earlier drafts. Special
not be optimal for detecting changes in child's anger
thanks are due Barbara Schaffer who inspired this re-
that are caused by mother's anger and associated
search and who assisted every step of the way.
negative parenting, a prerequisite to demonstrating
causality. Detecting such effects may require re-
ports made many times during the day.