You are on page 1of 12

Remarriage, Cohabitation, and Changes in Mothering Behavior

Author(s): Elizabeth Thomson, Jane Mosley, Thomas L. Hanson, Sara S. McLanahan


Source: Journal of Marriage and Family, Vol. 63, No. 2 (May, 2001), pp. 370-380
Published by: National Council on Family Relations
Stable URL: http://www.jstor.org/stable/3654598
Accessed: 27/01/2010 00:38

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=ncfr.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

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.

National Council on Family Relations is collaborating with JSTOR to digitize, preserve and extend access to
Journal of Marriage and Family.

http://www.jstor.org
ELIZABETH
THOMSONUniversityof Wisconsin-Madison

JANEMOSLEY MidwestResearchInstitute*

THOMASL. HANSON WestEd**

SARAS. MCLANAHANPrincetonUniversity***

andChangesin
Cohabitation,
Remarriage,
MotheringBehavior

Weuseddatafrom two wavesof the NationalSur- mothersin partnershipsat the second interview,
vey of Families and Household to investigate comparedwith childrenwhose mothersremained
changes in motheringbehavior associated with single or whose newpartnershiphad ended.Only
remarriageor cohabitationby single mothers.We a smallpart of the differencesin harshdiscipline,
considered three dimensions of mothering:(a) and none of the otherobserveddifferences,could
timeand supervision,(b) harshdiscipline,and (c) be explainedby maternalorfamilycharacteristics
relationshipquality.Mothersand childrenagreed or by motheringbehaviorand relationshipsin the
that motherswho remainedin new partnerships first interview.Althoughcohabitingpartnerships
used harshdisciplinelessfrequentlythanmothers were more likely to end than were marriages,we
who remainedsingle or whose new partnership found no differencesin effects of cohabitingor
had ended by the second interview.Mothersre- maritalpartnerships,net of theirstatusat the sec-
ported less supervisionif they had experienceda ond interview.
disruptedpartnership,whereaschildrenreported
less supervisionif their mothers remainedin a
newpartnershipat the second interview.Children Demographictrendsin cohabitation,marriage,di-
but not mothersreportedbetterrelationshipswith vorce, andremarriagehave changedchildren'sex-
perienceof family life (Bumpass,1994; Cherlin,
Departmentof Sociology, 1180 ObservatoryDrive,University 1992; Manning& Lichter,1995). Gone for most
of Wisconsin,Madison,WI 53706 (thomson@ssc.wise.edu). childrenarethe days of living with two biological
parentsuntil age 18 (Bumpass,Raley, & Sweet,
*Midwest Research Institute, 425 Volker Avenue, Kansas 1995; Graefe & Lichter, 1997). Many children
City, MO 64110. whose parentsdivorceor end theircohabitingre-
**WestEd, 4665 Lampson Avenue, Los Alamitos, CA lationshipor who were born to unmarried,non-
90720. cohabitingmotherswill experiencefurtherfamily
change when a parentremarriesor cohabitswith
***Office of PopulationResearch,Notestein Hall, 21 Pros- a new partner (Cherlin; Norton & Moorman,
pect Avenue, PrincetonUniversity, Princeton,NJ 08540.
1987).
Key Words: cohabitation, divorce, parent-child relation- In this article,we use panel datafromthe Na-
ships, remarriage. tional Survey of Families and Households

370 Journalof Marriageand Family63 (May 2001): 370-380


Cohabitation and Changes in Mothering Behavior 371

(NSFH)to estimateeffects of mothers'new part- to children(Hetherington& Jodl, 1994). Remar-


nerships,cohabitingor marital,on motheringbe- riage or cohabitationmay mean a change of res-
havior.We examinethreedimensionsof children's idence for the motherand her child(ren),and at
experienceswith theirmothers:(a) mother'stime the very least a change in daily routines.Such
and supervision;(b) harshdiscipline,and (c) re- changes may be stressful for both mother and
lationshipquality.Ourstudymakestwo important child(ren),producingpoorermotheringandmoth-
contributions.First,the panel design of NSFH al- er-childrelationships.Becausesingle mothersand
lowed us to measure mothers' behaviors and their childrenoften develop a particularlyclose
mother-childrelationshipsbefore mothersform a and interdependentrelationship(Ganong & Co-
new partnership.We could thereforecontrol for leman, 1994), new partnersmay be viewed as
potential selection of particulartypes of single competitionfor the mother'saffectionandperhaps
mothers into partnerships,going beyond cross- also to theirrelationshipwith a nonresidentfather
sectional analyses of differencesbetween single (Crosbie-Bumett& Ahrons, 1985). As a result,
andremarriedor repartnered mothers.Second,be- they may behave badly and make the mother's
cause both the motherand a focal child were in- parentingtask more difficult.Finally, cohabiting
terviewed, we had a more complete picture of partnershipsand higher ordermarriagesmay be
motheringat the secondinterviewthanwe would less stable than first marriages.If a relationship
have with only mothers'reports. ends, the stressof two majorfamily changesin a
Why might one expect motheringto change relativelyshortperiod of time may adverselyaf-
when a woman enters a new partnership?First, fect motheringand the mother-childrelationship.
new partnersbringaddedincome into the house- In addition,if the new partnershipprovidedim-
hold. Remarriagein particularis associatedwith provementsin mothering,its disruptioncould re-
a large increase in family economic well-being verse those effects.
(McLanahan& Sandefur,1993). Low levels of in- Cross-sectionalanalysesusuallyfinddifferenc-
come increaseparentalstress, which in turnmay es in parentingor parent-childrelationshipsacross
lead to poorerparenting(Elder,1974;McLeod& family types (e.g., Kurdek& Fine, 1988; Peek,
Shanahan,1994; McLoyd & Wilson, 1991). An Bell, Waldren,& Sorrell, 1993; but see Amato,
increasein income and economic securityshould 1987). Furstenbergand Spanier(1984) reported,
lower stressand therebyimprovemothering. for example, that remarriedparents were more
New partnersalso bringa secondpairof eyes, likely thanparentsin originalmarriagesto report
ears, and hands to the childrearingtask. Cross- parentingburdens,while at the same time report-
sectionalanalyses show that the presenceof an- ing that remarriagehelped in the raisingof their
otheradultin the householdis associatedwith in- children. Thomson, McLanahan, and Curtin
creased supervision (Thomson, McLanahan& (1992) foundthat single mothersspentmoretime
Curtin,1992). A new partnermay engagedirectly with children than did remarriedor cohabiting
in childrearingtasks or take over some of the mothersbut that remarriedmothershad stricter
household and daily life tasks so that a mother rules and supervisedchildrenmore closely than
can devote more time to children. single or cohabitingmothers.
Third,a spouseor partneroffers social support Finding differences in parenting behavior
for the mother (Bronfenbrenner, 1979). He may acrossfamily types does not, however,tell us that
be helpful in making difficult childrearingdeci- family structureor the associatedexperienceof
sions and in strengtheningher authority.He may family formationor separationcauses changesin
be a sourceof comfortwhenparentingis stressful. parentingbehavior.Instead,we could be observ-
All of these supportsmay benefitthe mother'spar- ing an associationproducedby selection of par-
enting and her relationshipwith her child(ren). ticulartypes of parentsinto separation,new part-
New partnersmay also inhibitnegativeresponses nerships, or both. Whatever unobserved
towardchildren,simplybecausethey are thereto characteristicsunderlieparentingbehaviorcould
observethem.Motheringis in a sense morepublic also make parentsmore attractivepartnersor less
in a new relationshipthan when mothers live interestedin living alone.
alone with theirchildren. Data collected by Hetheringtonand her col-
Not all of the hypothesizedeffects of new part- leagues provide some of the limited information
nershipsare positive,however.New partnersnec- on remarriageand motheringbehaviorusing lon-
essarilytake time away from motheringand may gitudinaldata (Hetherington,1989; Hetherington
not transferthe same amountsof their own time & Clingempeel,1992;Hetherington,Cox, & Cox,
372 Journal of Marriage and Family

1982). They arguedthatthe additionalstressorsof living with the focal child but withouta partner
remarriageoverridein the shorttermany benefits at the firstinterview;who participatedin the fol-
due to additionalincome or the presenceof an- low-up survey during 1992-1994 (NSFH2) and
otheradult.They reportedthatthe relationshipbe- whose focal child was living with them and was
tween mothersand childrendeclinedafterremar- age 10-17 at thattime.
riage but recoveredwithin two years. They also About four fifths (81%) of the single mothers
found that remarriedmotherswere less involved at NSFH1 were interviewedin NSFH2. As ex-
than divorced single mothersin their children's pected, almost90% of focal childrenaged 10-17
lives (Hetherington& Clingempeel).The limita- at NSFH2 were living with their mothers(N =
tions of theirwork are thatthe samplesare rather 417). About three fourthsof these childrenalso
small and composed predominantlyof White, participatedin the NSFH2 survey (N = 316). We
middle-classfamilies. investigatedpotentialnonresponsebias with lo-
In an earlieranalysis,we used paneldatafrom gistic regressionmodelsof maternalresponseand
the NSFH to simulatechanges in motheringbe- of focal child response, contingenton maternal
haviorbeforeandafterdivorce,includingchanges response. Education and home ownership in-
associatedwith the cohabitationor remarriageof creasedthe likelihoodof maternalresponse,and
divorcedmothers(Hanson,McLanahan,& Thom- AfricanAmericanmotherswere morelikely to re-
son, 1998). Beginningwith a sample of married spondthanwereWhitenon-Hispanicmothers.No
mothers,we followed them throughdivorce,co- differencesin response were found by mother's
habitation,remarriageor a combinationthereof. marital experience, including recentness of di-
Divorce between interviewswas associatedwith vorce, focal child'sage or sex, sibshipsize or sex
increased frequency of child activities and re- composition,maternalemployment,religion, or
duced supervision.Motherswho subsequentlyre- presence of anotheradult (not a partner)in the
marriedreportedthe same elevatedfrequencyof householdat the firstinterview.Two indicatorsof
child activitiesbut less supervisionthandid moth- motheringbehaviormeasuredin the first survey
ers who remainedsingle. Thus, remarriageor co- were also associatedwith responseto the second
habitationdid not lead to a "recovery"of predi- survey:Enjoyabletimes with the focal child were
vorce mothering. associatedwith a lowerlikelihoodof response,fo-
In this article,we investigatethese inferredef- cal child supervisionwith a higher likelihoodof
fects of cohabitationor remarriageby directlyob- response.Focal child responsewas also positively
servingchangesin motheringamongwomenwho associatedwith mother'seducationand activities
were single at the firstNSFH interview.This de- with children.As for mothers,child responsewas
sign allows us to observe potential changes in negativelyassociatedwith maternalreports(in the
motheringbehaviorandmother-childrelationships firstinterview)of enjoyabletimes with the child.
for women who had never been marriedas well We investigatedthe potentialbias in models
as those who had been divorcedfor a longer pe- estimatedfrom panel data by includingthe pre-
riod of time than was the case in our previous dicted likelihoodof responsein our models. For
analysis.The design also enablesus to controlfor children'sreports,we includedtwo separatepre-
motheringbehavior observed during singlehood dictedvalues,one for maternal,the otherfor child
but before remarriageor cohabitation,thus con- response. Estimates of other structuraleffects
trollingfor the potentialselectionof differentsorts were not altered when predictedresponse vari-
of mothersinto new partnerships. ables were included. We thereforepresent esti-
matesfrom models withoutthose variables.
We includedin our analysismeasuresof three
SAMPLEANDMEASURES
dimensionsof mothering:investmentsof time and
We used data from the two waves of the NSFH, supervision, harsh discipline, and relationship
a nationallyrepresentativesurvey of U.S. adults quality.In preliminaryanalyses,we also investi-
with a double sample of single-parentfamilies. gated indicatorsof milder forms of discipline,
The responserate to the initial survey conducted findingno significantdifferencesby mother'spart-
during 1987-1988 (NSFH1) was 74% (Sweet, nership experience. (Analyses available on re-
Bumpass,& Call, 1988). If respondentsreported quest.) Because we did not have good indicators
any childrenunderage 19 living in the household, of mild disciplinein the firstsurvey,we focus in
one child was randomlyselected as a focal child. this article on the dimensionsof motheringfor
Ouranalyticsampleconsistsof motherswho were which we have repeatedor similarreportsfrom
Cohabitation and Changes in Mothering Behavior 373

the motherin both surveysandparallelresponses we constructeda scale identicalto thatconstructed


from the motherand child in the second survey. fromthe secondinterview.An additionalindicator
We attemptedto develop a smaller numberof of prior supervision(not asked in the second in-
scales thanpresentedhere, but exploratoryfactor terview)is also includedin our analysis.Mothers
analyses showed that very few indicatorswere reportedhow often the focal child was requiredto
sufficientlystronglyassociatedto representa sin- inform parentsof her or his whereaboutswhen
gle latent construct.To create multi-itemscales, away from home, rangingfrom hardlyever = 1
we averagedvalid item scores; respondentswho to all of the time = 4.
providedvalid responsesfor at leasthalf the items
in a scale receiveda valid score. In the following
paragraphs,we firstdescribemeasuresof each di- HarshDiscipline
mensionof motheringin the secondinterviewand
then parallelor similarindicatorsof motheringin In the follow-up interview,motherswere asked
the firstinterview. "When (child) does something especially bad,
how often do you yell at him or her? Spank or
Mother'sTimeand Supervision slap him or her?"and "How muchof the time do
In the follow-upinterview,mothersandfocal chil- you yell or shout to get (child) to do what you
drenwere asked to reporton the amountof time want or not to do what you don't want?" Re-
together,thatis, how manyhourseach week they sponse options for both questions ranged from
never = 1 to always = 5, with abouthalf the time
spenttime together,one-on-one,up to a maximum
of 50 hours.Because the tail end of the distribu- at the midpoint.Mothers were also asked how
tion was long andthin,we truncatedthe maximum many times within the past week they had
numberof hoursat 10 per week. Amongmothers, spankedor hit the focal child when he or she be-
13%reportedzero hours,20% 10 or more hours, haved badly, scored from none = 1 to three or
with a relativelyeven distributionbetween 1 and more times = 4. We averagedpairs of items re-
9 hours per week. Mothers were also asked to ferringto yelling and to spankingor hitting,pro-
indicatewhetherthe focal child was allowedto be ducing scales with estimatedreliabilitiesof .80
at home alone in the afternoon,after school, late and .92, respectively.
in the evening,or overnight.Supervisionat home Focal childrenwere askedhow often theirpar-
is scored 1 if the child was allowed home alone ent(s) used yelling to influencethe child'sbehav-
overnight,2 if allowed home alone late in the ior, with responsesrangingfrom never = 1 to all
evening but not overnight, 3 if allowed home the time = 5 and how many times they had been
alone in the afternoonbut not in the evening or spankedor hit in the past week. Only 5% of the
overnight,and4 if not allowedhome alone at any children reportedrecent spankingor hitting, so
of these times. Focal childrenwere askedhow of- this indicatorwas dichotomized.
ten they were left at home alone during these Indicatorsof parentaldisciplinewere extreme-
times;we categorizedeachresponseas anytimeor ly limitedin the initialinterview.Mothersreport-
never and constructeda scale parallelto that for ed how frequentlythey yelled at or spankedor hit
mothers. one of the children,responsesrangingfromnever
Measuresof time and supervisionin the first = 1 to very often = 4. In addition,they were
interview were not completelyparallel.Mothers
were not asked the same questionaboutone-on- askedhow manytimes they had spankedthe focal
one time with the focal child. They were, how- child in the past week (for focal childrenunder
ever, asked to reporthow often they engaged in age 12), with no spankingor hitting= 1 andthree
four differenttypes of activitieswith one or more or more times = 4. Yelling was thereforemea-
of their children (leisure activities away from sured with a single indicatorfrom the initial in-
home, at home workingon a projector playing terview, spankingby an average of the two re-
together,having privatetalks, helping with read- sponses,havingan estimatedreliabilityof .57. As
ing or homework),with responseoptionsranging notedabove,we do not presentanalysesfor mild-
from never = 1 to almost every day = 6. The er forms of discipline in our analyses (e.g., de-
averagedscale, child activities,has an estimated nying privileges, explaining why behavior is
reliabilityof .70. The same questionsabout su- wrong) because we do not have good indicators
pervision were asked in the first interview,and of those forms in the firstinterview.
374 Journal of Marriage and Family

TABLE 1. MOTHERS'NEW PARTNERSHIPS


BETWEEN vorced women were more likely to form a new
SURVEYS
partnership,more likely to marry,and less likely
PartnershipExperience Number Percent to separatefrom a new partnerthanwere women
who had not marriedby the time of the first in-
Remained single 229 54.9
Cohabited,partnershipended 51 12.2 terview (dataavailableon request).
Married,marriageended 23 5.5
Cohabited,partnershipintact 29 7.0
ANALYSES AND RESULTS
Married,marriageintact 85 20.4
Total 417 100.0
Exceptfor the modelof child'sreportof spanking,
estimatedwith logistic regression,models were
estimatedwith ordinaryleast-squaresregression.
As shown in Table 1, we could classify mothers'
Relationship Quality
partnershipexperience along two dimensions-
In the follow-up interview,motherswere asked: whethera partnershipwas cohabitingor marital
"Duringthe last 30 days, how often did you and and whetherit had ended by or was intactat the
(child) talk about somethingthat was worrying second interview.Net of partnershipstatus (dis-
him or her? .... thathe or she was excited about rupted,intact),we foundno significantdifferences
or interested in?" (1 = never ... 6 = almost betweeneffects of cohabitingand maritalpartner-
every day). Childrenwere asked: "If you had a ships. Ouranalysisthereforefocuses on effects of
majordecision to make.... felt depressedor un- disruptedandintactpartnerships, ignoringthe dis-
happy ...., how likely would you be to talk to tinctionbetweenmarriageandcohabitation.Note,
your mother?" (1 = definitely wouldn't .... 5 = however,that cohabitingpartnershipswere more
definitelywould).We averagedeach pairof items likely to end thanwere marriages.
to measurethe extentto which the motherserved We conductedpreliminaryanalysesto identify
as a confidantfor the child, producingscales with demographicand socioeconomic characteristics
estimatedreliabilitiesof .51 and .71, respectively. that were significantlyassociatedwith mothering
As a measureof theirglobal relationshipmothers at NSFH2. We considerednot only focal child's
and childrenwere asked: "Takingall things to- sex andage at NSFH2but also measuresof family
gether,on a scale from 0 to 10 where 0 is really and maternalcharacteristicsat NSFH1: sibship
bad and 10 is absolutelyperfect,how would you size and sex compositionandage of youngestsib-
describeyour relationshipwith (child/yourmoth- ling; mother'smaritalstatusand recentnessof di-
er)?" vorce; mother'seducation,home ownership,eth-
In the firstinterview,mothersreportedthe fre- nicity, and religion; and whether anotheradult
quencyof "enjoyabletimes" with the focal child (not a partner) lived with the mother and
duringthe past month,rangingfromnever = 1 to child(ren).Parallelmeasuresat NSFH2 were not
almost every day = 6. They also provided an includedbecausethey could be the consequences
overall ratingof their relationshipwith the focal ratherthan causes of partnershipexperiencebe-
child, ranging from very poor = 1 to excellent = tween interview.Final models includeonly char-
7. And finally,they indicatedthe frequencywith acteristicsthat were statisticallysignificantlyas-
which they praisedor hugged a child (not neces- sociated with NSFH2 motheringor that altered
sarily the focal child), from 1 (never) to 4 (very coefficients for partnershipexperiencewhen in-
often), producinga scale of warmthwith an esti- cluded in the model. We note in particularthat
matedreliabilityof .48. motherswho hadbeen divorcedfor shorteror lon-
At NSFH2, mothersprovidedinformationon ger periodsat NSFH1 did not differin mothering
all cohabitingand maritalpartnershipsbetween at NSFH2, net of othereffects; significantdiffer-
interviews. Table 1 shows the distributionof ences werefoundbetweenmotherswho hadnever
union experience.More than half of the mothers been marriedand those who were divorced at
remainedsingle throughoutthe periodbetweenin- NSFH1. Parallelanalyses were conductedto se-
terviews, about one fifth cohabitedand did not lect appropriate controlsfor motheringat NSFH1.
marry,andmorethanone quartermarried(usually We considerednot only indicatorsfrom the same
precededby a period of cohabitation).One fifth domain of mothering,but also indicatorsfrom
of the marriagesand morethanhalf of the cohab- otherdomains.In most cases, the strongestor only
iting partnershipsthat did not result in marriage statisticallysignificantassociationswere between
had dissolved before the second interview. Di- parallelmeasuresat NSFH1 and NSFH2.
Cohabitation and Changes in Mothering Behavior 375

TABLE 2. MOTHERING BEHAVIOR AND MOTHER-CHILD RELATIONSHIPS BY PARTNERSHIP EXPERIENCE

Remained Disrupted Intact


Single Partnership Partnership All Mothers
M SD M SD M SD M SD N
NSFH1 Mothering
Child activities 4.68 0.90 4.65 0.95 4.70 0.89 4.68 0.90 402
Supervision at home 3.63 0.58 3.61 0.59 3.60 0.60 3.62 0.58 348
Parentsinformed 3.89 0.43 3.98 0.13 3.92 0.31 3.91 0.37 357
Yelling 2.91 0.75 2.86 0.78 2.75 0.72 2.86 0.75 400
Spanking/hitting 1.37* 0.71 1.54* 0.74 1.25* 0.68 1.37 0.72 417
Global relationship 6.48 0.81 6.47 0.83 6.42 0.82 6.46 0.81 377
Enjoyable times 5.13 1.14 4.95 1.43 5.10 1.00 5.09 1.16 416
Warmth 3.77 0.36 3.80 0.33 3.81 0.33 3.79 0.35 401
NSFH2 Mothering
Time together
mother 4.56 3.46 4.27 3.59 4.09 3.43 4.37 3.46 361
child 3.21 4.40 3.27 4.11 2.57 4.52 3.02 4.39 296
Supervision at home
mother 2.71 0.97 2.59 0.94 2.70 0.97 2.68 0.97 361
child 2.79 1.11 2.79 1.18 2.58 1.00 2.73 1.09 311
Yelling
mother 3.06* 1.19 3.25* 1.13 2.60* 1.12 2.97 1.18 363
child 2.45* 1.18 2.65* 1.16 2.02* 0.94 2.35 1.13 307
Spanking/hitting
mother 1.29* 0.52 1.31* 0.61 1.09* 0.22 1.23 0.48 363
child 0.01 0.24 0.10 0.31 0.00 0.15 0.01 0.23 308
Confidant
mother 3.94 1.21 4.14 1.15 4.02 1.00 4.00 1.14 413
child 3.81 1.02 3.77 0.95 3.98 0.87 3.86 0.97 309
Global relationship
mother 8.43 1.52 8.36 1.72 8.44 1.57 8.42 1.57 413
child 8.16* 2.10 8.67* 1.52 8.72* 1.41 8.41 1.85 306
*p < .05.

In both interviews,a relativelylarge number motheringis that we can control for mothering
of respondents(10-15%) did not provideanswers behaviorbefore the formationor dissolutionof a
to one or anotherquestionpresentedin self-enu- new partnership. Motheringbehaviormay varyby
merated questionnaires.Unfortunately,several a woman's capacity for interpersonalrelation-
questions about motheringwere presentedthis ships,herrelativeinterestin findinga new partner,
way, ratherthan in the personalinterview. For and a varietyof otherdispositionsthat could de-
some mothers,moreover,the focal child was a terminewhethershe entersa new partnershipand
year or so too young (supervision)or too old whetherthat partnershipsurvives to the second
(spanking)at the first interviewfor the question interview.If we controlfor motheringat the initial
to have been asked. To maximizethe size of our interview,we are in effect controllingfor a major
analytic sample, we used scores predictedfrom form of selectioninto andout of new partnerships
demographicand socioeconomic characteristics andcan makestrongerinferencesaboutthe effects
(includingthe child's sex and age at NSFH1) to of partnershipexperienceon motheringat the sec-
substitutefor missing data on motheringat the ond interview than when only socioeconomic
first interview and includeddummyvariablesin characteristicsare controlled.
ourmodelsto identifycases withpredictedscores. To identifypotentialselectionof differentsorts
Remainingdifferencesbetween the total and an- of mothersinto and out of new partnerships,we
alytic samples are due to nonresponseon mea- estimatedassociationsbetweenmotheringbehav-
sures of motheringat NSFH2 or, in a very small ior and mother-childrelationshipsat the first in-
numberof cases, to missing informationon edu- terview and the mother'spartnershipexperience
cation,employment,or home ownership. between interviews.We estimatedthe zero-order
The primaryadvantageof paneldatafor study- association and associations adjusted for the
ing effects of partnership
experienceand statuson child'sage at the firstinterview,andwe compared
376 Journal of Marriage and Family

TABLE3. EFFECTS
OFPARTNERSHIP TIMEANDSUPERVISION
ONMOTHERS'
EXPERIENCE
Time Together Supervision at Home
Mother's Report Child's Report Mother'sReport Child's Report
B SE B SE B SE B SE

Disruptedpartnership -.484 .503 .352 .739 -.325** .101 189 .160


Intact partnership -.645 .421 -.658 .581 -.008 .085 -.338** .125
Child age in years -.007 .083 .161 .123 -.224** .024 -.216** .036
Female child -.486 .366 .133 .871 -.006 .074 .003 .182
Never marriedNSFH1 -.416 .413 .396 .600 .001 .097 -.006 .150
NSFH1 Mothering
Child activities .551** .220
Supervision at home .328** .080 .007 .116
Parentsinformed .181* .106 .294** .149
Global relationship .409 .332 .008* .048
Enjoyable times .157 .167
Praise, hug .874 .569 -.300 .163
Constant -1.174 2.734 -3.656 3.357 3.649** .782 7.144** 1.303
Adjusted R2 .060** .028* .527** .304**
Valid cases 345 287 341 297
Note: Ordinaryleast squares regression estimates. Other control variables measured at NSFH1 include time together,
mother'sreport(any religion); time together,child's report(mother'semployment,sibship gender composition);supervision,
mother's report (mother's education, race/ethnicity, any religion, sibship size); supervision, child's report (same as for
mother's reportplus sibship gender composition). All models include indicatorsof predictedNSFH1 motheringvariables.
*p < .10 (two-tailed). **p < .05 (two-tailed).

results for the observed indicatorsand measures nershipbetween interviews, whether or not the
that includedpredictedvalues for initial mother- couple separated.
ing behaviorand relationships.Very few associ- Table 3 reportskey model parametersfor in-
ations were found. Mothers'partnershipexperi- dicatorsof mothers'time and supervision.As in
ence was not associatedwith measuresof time Table 2, coefficientssuggest less time with chil-
and supervisionor the mother-childrelationship drenfor motherswhose new partnerships werein-
at the firstinterview.As shownin Table2, we did tact at the second interview.Again, however,we
find evidence of selection on harsh discipline. cannotrejectthe hypothesisof no difference.The
Motherswhose new partnerships wereintactat the measureof mothers' activities with all children
secondinterviewhadreportedthe lowest levels of has a relativelylargeeffect on hourswiththe focal
spankingor hitting and yelling (not statistically child, showingcontinuityin mothers'investments
significant)at the first interview, whereasthose of time in childrearing.Otherindicatorsof moth-
who formedand dissolved a partnershipbetween ering do not have significanteffects net of mater-
interviewshad reportedthe highestlevels of harsh nal and family characteristics.
discipline. Accordingto mothers,partnerships thatformed
The bottomhalf of Table2 reportszero-order and dissolved between interviewsreducedchild
associationsbetween mothers'partnershipexpe- supervision; according to children, however,
rience and motheringat the second interview. motherswho remainedin partnershipsat the sec-
Here we find that harshparentingis also greater ond interviewprovidedthe least supervision.The
at the second interview for motherswho ended differencebetweenthese models and the zero-or-
new partnerships,as well as for motherswho re- der associationspresentedin Table2 is primarily
mainedsinglebetweeninterviews.Althoughzero- due to controlsfor child'sage. Becauseolderchil-
order associationsare not significantlydifferent dren are less strictly supervisedand also inhibit
from zero, parameterestimates suggest reduced the formationof new partnerships, the directneg-
time with children-and perhaps less supervi- ative effect of partnershipexperience is sup-
sion-when mothers'new partnershipsare intact pressedin the zero-orderassociation.Indicatorsof
at the second interview.On the otherhand,focal supervisionat the initialinterviewhavethe largest
childrenreport significantlyhigher quality rela- direct effects on supervisionat the second inter-
tionshipswith motherswho formed a new part- view.
Cohabitation and Changes in Mothering Behavior 377

TABLE4. EFFECTS
OFPARTNERSHIP ONHARSHDISCIPLINE
EXPERIENCE

Yelling Spanking/Hitting
Mother's Report Child's Report Mother's Report Child's Report
B SE B SE B SE B SE

Disruptedpartnership .209 .165 .186 .189 .000 .062 .143 .664


Intact partnership -.315** .139 -.314** .150 -.162** .053 -.963 .805
Child age in years -.001 .026 .002 'A0 -.004** .010 -.193 .137
Female child .389** .193 .266 .219 .006 .045 -.583 .567
Never marriedNSFH1 .188 .134 -.001 .173 .118** .057 .769 .595
NSFHI mothering
Yelling .467** .085 .177* .095
Spanking/hitting .159** .032 .314 .358
Constant 1.185** .597 1.391** .629 1.577** .174 -2.199 2.358
AdjustedR2 .125** .064** .207** .055*a
Valid cases 352 300 354 298
Note: Ordinaryleast squares regression estimates, except for child's report of spanking (logistic regression estimates).
Other control variables measured at NSFH1 include yelling, mother's report (mother's education, sibship size and gender
composition); yelling, child's report (home ownership, mother's race/ethnicity, sibship gender composition); spanking,
mother's report (mother'srace/ethnicity);spanking, child's report (mother's education, employment). Models also include
indicatorsof predictedNSFH1 motheringvariables.
aCox and Snell estimate.
*p < .10 (two-tailed). **p < .05 (two-tailed).

Table4 shows thatmothersand childrenagree only for the past week. Althoughcoefficientsfor
abouteffects of new partnershipson harshdisci- disrupted partnershipsare positive, yelling or
pline. Motherswhose partnerships remainedintact spankingor hitting were not significantlymore
yelled and spanked or hit less thanthose who re- commonfor those mothersthanfor motherswho
mained single between interviews.The fact that remainedsingle betweeninterviews.Harshdisci-
coefficients from the logistic regressionmodels pline at the first interviewis stronglyassociated
for children'sreports do not attain significance with harshdisciplineat the second interviewand
could be due to poor measurementin the child's accounts for about one quarterof the effect of
interview,wherespankingor hittingwas reported partnershipexperienceon harshdiscipline.(Mod-

TABLE5. EFFECTS
OFPARTNERSHIP ONTHEMOTHER-CHILD
EXPERIENCE RELATIONSHIP

Confidant Global Relationship


Mother's Report Child's Report Mother's Report Child's Report
B SE B SE B SE B SE

Disruptedpartnership .239 .153 -.006 .160 -.122 .187 .448 .294


Intact partnership .002 .130 .218* .124 .194 .161 .631** .227
Child age in yrs .000 .025 -.008** .026 -.000 .045 -.211** .048
Female child .423** .183 .115 .109 -.007 .136 -.247 .201
Never marriedNSFH1 -.007 .128 -.003 .143 .208 .177 .009 .263
NSFHI mothering
Enjoyable times .132** .050 .008 .048 .229** .061 .157* .090
Praise, hug .334* .174
Global relationship .549** .092 .374** .135
Supervision at home .471** .148
Constant 1.529* .853 4.508** .533 2.417* 1.281 8.397** 1.331
AdjustedR2
.046** .075** .274** .168**
Valid cases 404 302 401 299
Note: Ordinaryleast squaresregressionestimates. Othercontrol variablesmeasuredat NSFH1 include: confidant,mother's
report (sibship sex composition); confidant,child's report (mother'srace/ethnicity);relationship,mother's report (mother's
race/ethnicity,other adult in household);relationship,child's report(mother'srace/ethnicity).Models also include indicators
of predictedNSFH1 motheringvariables.
*p < .10 (two-tailed). **p < .05 (two-tailed).
378 Journal of Marriage and Family

els that controlonly for child's age and sex and suits is that cross-sectionalanalyses include sin-
maternalsocioeconomiccharacteristicsproduced gle-motherand stepparentfamiliesof varyingdu-
estimatedmeandifferencesvery close to thosere- rations,whereaspanelanalyseslimit the degreeof
portedin Table2.) variationat the secondpointof observation.In our
As shownin Table5, childrenbut not mothers currentand previous(Hansonet al., 1998) anal-
reportcloser and bettermother-childrelationships yses, partnershipsobserved at the second inter-
when the motherremainedin a new partnership view are of relativelyshortduration.After a lon-
at the secondinterview,comparedwith remaining ger period of time, we might observe greater
single or separatingfroma new partner.Indicators supervisionin the stable remarriedor cohabiting
of relationshipqualityat the first interviewhave families,comparedwith those of long-termsingle
directeffects on both mothers'and children'sre- mothers.(Sample size constraintspreventedour
ports of relationshipquality at the second inter- distinguishingeffects of partnershipsformed or
view. In addition,motherswho reportedgreater disruptedmoreor less recentlyin referenceto the
supervisionwhen the child was younger report point at which motheringwas measured.)On the
better relationshipsin adolescence.The positive otherhand,Thomsonet al. (1992) did not findany
coefficientfor intact partnershipsis largerin the differencesin family effects linked to the timing
full model thanin the model controllingonly for of partnershipformationor disruption.
child'sage and sex (analysesnot shown).This re- We expectedthat our use of panel datawould
sult is consistentwith a selectionprocessin which show the extent to which selection of different
close mother-childrelationshipsinhibit the for- sorts of mothersinto cohabitingpartnershipsor
mationof new partnerships. marriageand,for some, into a subsequentdisrup-
tion, could account for parentingvariationsby
DISCUSSION AND CONCLUSIONS family structureobservedin cross-sectionalanal-
yses. Althoughwe found that motherswho used
Is remarriageor repartneringgood for mothers less harsh discipline were more likely to be in
and theirchildren?The answeris yes, if our pri- stable partnershipsat the second interview,other
maryconcernis mothers'harshdisciplineandthe dimensionsof motheringwere not associatedwith
mother-childrelationship.Ourclearestresultsare partnership experience.Andharshdisciplineat the
thatmothersless often yell, spank,or hit children firstinterviewaccountedfor only a quarterof the
if they are living with a new husbandor partner. association between partnershipexperience and
In addition,from the child'spoint of view, a new harshdisciplineat the secondinterview.Thesere-
intact partnershipimprovesthe mother-childre- sults imply that analyses of cross-sectionaldata
lationship. may providereasonableestimatesof partnership
If we are more interestedin supervision,how- effects on motheringbehaviorand mother-child
ever,remarriageor repartnering may not be good relationships.They could also, however, reflect
for children.We foundthatsupervisionis greatest poor measurementof mothering(and therefore
in stable single-motherfamilies,but mothersand poor estimatesof a selectionprocess) at the first
childrendifferin reportingthatcontinuingor dis- interview.
ruptedpartnerships producethe least supervision. We were surprisedto findno differencesin ef-
The comparisonwith intactpartnershipsdoes not fects of cohabitationor remarriageon mothering.
correspondto results from cross-sectionalanaly- We know that cohabitingpartnersmay not have
ses in which stepfamiliesprovide greatersuper- the samecommitmentto a mother'schildrenas do
vision thando single-parentfamilies(Thomsonet stepfathers,andwe observedthatcohabitingpart-
al., 1992). The results are, however, consistent nershipswere less stablethanmarriages.(See also
with our simulatedchangeanalysis(Hansonet al., Bumpass & Sweet, 1989.) It could be that our
1998) in which cohabitationor remarriageaccel- sampleis too small to detectdifferencesbetween
erateddeclines in supervisionfollowing divorce. maritaland cohabitingpartnerships.Only a small
It may be thatfamily stability,ratherthannumber numberof cohabitingpartnerships remainedintact
of parents,facilitatesarrangementsfor children's withouthaving been transformedinto marriages,
supervisionat home. Residentialandotherhouse- and only a small numberof marriagesendedbe-
hold changesassociatedwith the formationof new fore the second interview.With a largersample,
partnerships may disruptwell-establishedpatterns it maybe possibleto identifydifferencesin effects
of supervision(Hansonet al.). of partnershipformationor disruptionfor cohab-
Anotherpossible explanationfor differentre- iting and marriedcouples.
Cohabitation and Changes in Mothering Behavior 379

The fact thatNSFHincludesbothmothers'and What we don't know is how partnerpresence


children'sreportsof motheringand the mother- might exert such an effect. Differentmechanisms
childrelationshipstrengthensourfindingsin some have differentimplicationsfor single mothers.If
respects.Mothers'andchildren'sreportsproduced partnersinhibit harsh discipline simply as wit-
the samepatternsfor harshdiscipline;butmothers nesses to the mother'sbehavior,otheradultsin the
and childrendid not agree on the particularex- householdmight serve the same purpose.If part-
periencesassociatedwith reducedsupervisionor ners' economic contributionto the householdre-
relationship quality. Limiting the analysis of duces maternalstress,we can envisionalternative
mothers' reportsto those whose child had also supportsto single mothers,includingdirectpublic
participateddid not alterthese results.The differ- transfers,good jobs and high-quality,inexpensive
ence between the two reports of supervision- childcare.If it is directhelp with or indirectsup-
mothers'reportingon rules, childrenon events- port for the mother'schildrearingresponsibilities
may account for differences in those results. that makes motheringeasier,we may need better
Womenwhose partnershipsended may admit to community resources for single mothers and
less strict rules but in fact leave their children greaterpublicsupportfor the difficultjob they do.
aloneno less often thando motherswho remained
single. Womenin partnershipsmay in fact leave NOTE
theirchildrenalone more often thantheir "rules"
This researchwas supportedby GrantNos. HD29601
dictate.On the otherhand,no such differencesin andHD19375fromthe Centerfor PopulationResearch,
measurementcould accountfor the fact thatwom- NationalInstituteof Child Health and HumanDevel-
en in partnerships don'tperceivethe enhancedre- opment.The NationalSurvey of Families and House-
lationshipqualityperceivedby theirchildren. holds was supportedby NIH/NIAGrantNo. HD21009.
Anothersource of differencesin resultsbased
on mothers'and children'sreportsis thatwe have REFERENCES
only mothers'reportsof motheringat the firstsur- Amato, P R. (1987). Family processes in one-parent,
vey. In almost every case, the estimatedcoeffi- stepparent,and intact families: The child's point of
cients for indicatorsof NSFH1 motheringwere view. Journal of Marriage and the Family, 49, 327-
largerfor mothers'than for children'sreportsof 337.
the same underlyingvariable.To the extent that Bronfenbrenner, U. (1979). The ecology of human de-
we have inadequatelycontrolledfor selection of velopment: Experiments by nature and design. Cam-
bridge,MA: HarvardUniversityPress.
differentsortsof mothersinto andout of new part- Bumpass,L. (1994). The decliningsignificanceof mar-
nerships,our estimatesof partnershipeffects may riage: Changing family life in the United States.
be biased. NSFH WorkingPaper66. Madison:Centerfor De-
It is importantto keep in mindthroughoutthis mographyand Ecology, Universityof Wisconsin.
Bumpass,L. L., Raley, R. K., & Sweet, J. A. (1995).
analysisthatwe arenot comparingsingle mothers The changingcharacterof stepfamilies:Implications
with mothersin originaltwo-parentfamilies.Once of cohabitationand nonmaritalchildbearing.Demog-
a motheris single-through divorceor an unpart- raphy, 32, 425-436.
neredbirth-she does not usuallyhave the choice Bumpass,L. L., & Sweet, J. A. (1989). Nationalesti-
mates of cohabitation. Demography, 26, 615-625.
of reforminga partnershipwith her child(ren)'s Cherlin, A. J. (1992). Marriage, divorce, remarriage.
biological father.The choice is to remainsingle Cambridge,MA: HarvardUniversityPress.
or to cohabitwith or marrya new partner.Stable Crosbie-Burnett,M., & Ahrons, C. (1985). From di-
vorce to remarriage:Implicationsfor therapywith
single motherhoodmay provide advantagesin families in transition. Psychotherapy and the Family,
termsof supervision,avoidingthe risk of a sub-
1, 121-137.
sequent partnershipdisruption with its added Elder, G. (1974). Children of the Great Depression: So-
stresson mothersandchildren.On the otherhand, cial change in life experience. Chicago: University of
marriagesor stable partnershipsappearto offer ChicagoPress.
advantagesfor childrenin reducingharsh disci- Furstenberg,F, & Spanier,G. B. (1984). Recyclingthe
family: Remarriage after divorce. Beverly Hills:
pline and increasingchildren'spositive feelings Sage.
abouttheirmothers. Ganong,L. H., & Coleman,M. (1994). Remarriedfam-
Our results for harsh discipline appearto be ily relationships. Thousand Oaks, CA: Sage.
the most consistent,robust,and theoreticallysen- Graefe,D. R., & Lichter,D. T. (1997). Life coursetran-
sitions of Americanchildren:Parentalcohabitation,
sible. They imply thatit is the presenceof a part-
marriage and single motherhood.Population Re-
ner ratherthanthe processof formingor dissolv- searchInstituteWorkingPaper,97-08. StateCollege:
ing a partnershipthat inhibits harsh discipline. PennsylvaniaStateUniversity.
380 Journal of Marriage and Family

Hanson, T. L., McLanahan,S. S., & Thomson, E. McLanahan,S. S., & Sandefur,G. D. (1994). Living
(1998). Windowson divorce:Beforeandafter.Social with a single parent: What hurts, what helps. Cam-
Science Research, 27, 329-349. bridge,MA: HarvardUniversityPress.
Hetherington,E. M. (1989). Coping with family tran- McLeod,J. D., & Shanahan,M. (1993). Poverty,par-
sitions:Winners,losers and survivors.ChildDevel- entingand children'smentalhealth.AmericanSocio-
opment, 60, 1-14. logical Review, 58, 351-366.
Hetherington,E. M., & Clingempeel,W G. (1992). McLoyd,V., & Wilson,L. (1991). The strainof living
Coping with maritaltransitions.Monographsof the poor: Parenting,social support and child mental
Society for Research in Child Development, 57, Serial health.In A. Huston(Ed.), Childrenin poverty(pp.
no. 227. 105-135). Cambridge,England:CambridgeUniver-
E. M., Cox, M., & Cox, R. (1982). Effects
Hetherington, sity Press.
of divorceon parentsandchildren.In M. Lamb(Ed.), Norton,A. J., & Moorman,J. E. (1987). Currenttrends
Nontraditional families: Parenting and child devel- in marriageand divorce among Americanwomen.
Journal of Marriage and the Family, 49, 3-14.
opment(pp. 233-285). Hillsdale,NJ: Erlbaum. Peek, C. W., Bell, N. J., Waldren,T, & Sorrell,G. T
E. M., & Jodl,K. M. (1994). Stepfamilies
Hetherington,
as settingsfor child development.In A. Booth & J. (1988). Patternsof functioningin families of remar-
ried and first-marriedcouples. Journal of Marriage
Dunn (Eds.), Stepfamilies: Who benefits? Who does and the Family, 50, 699-708.
not? (pp. 55-79). Hillsdale,NJ: Erlbaum. Sweet, J., Bumpass,L., & Call, V. (1988). The design
Kurdek,L. A., & Fine, M. A. (1993). The relationbe- and contentof the NationalSurvey of Familiesand
tween family structureand young adolescents'ap- Households.NSFHWorkingPaper1, Madison:Cen-
praisals of family climate and parentingbehavior. ter for DemographyandEcology,Universityof Wis-
Journal of Family Issues, 14, 279-290. consin.
Manning,W D., & Lichter,D. T. (1995). Parentalco- Thomson,E., McLanahan,S. S., & Curtin,R. B. (1992).
habitationand children'seconomicwell-being.Jour- Family structureand parentalsocialization.Journal
nal of Marriage and the Family, 58, 998-1010. of Marriage and the Family, 54, 368-378.

You might also like