the causal effects of father absence -...
Post on 01-Mar-2020
9 Views
Preview:
TRANSCRIPT
SO39CH20-McLanahan ARI 27 June 2013 14:34
The Causal Effectsof Father AbsenceSara McLanahan,1 Laura Tach,2
and Daniel Schneider3
1Office of Population Research, Princeton University, Princeton, New Jersey 08544;email: mclanaha@princeton.edu2Department of Policy Analysis and Management, Cornell University, Ithaca,New York 14853; email: lauratach@cornell.edu3Department of Sociology and Robert Wood Johnson Scholars in Health Policy ResearchProgram, University of California, Berkeley, California 94720;email: djschneider@berkeley.edu
Annu. Rev. Sociol. 2013. 39:399–427
The Annual Review of Sociology is online athttp://soc.annualreviews.org
This article’s doi:10.1146/annurev-soc-071312-145704
Copyright c© 2013 by Annual Reviews.All rights reserved
Keywords
divorce, single motherhood, education, mental health, labor force,child well-being
Abstract
The literature on father absence is frequently criticized for its use ofcross-sectional data and methods that fail to take account of possibleomitted variable bias and reverse causality. We review studies that haveresponded to this critique by employing a variety of innovative researchdesigns to identify the causal effect of father absence, including studiesusing lagged dependent variable models, growth curve models, indi-vidual fixed effects models, sibling fixed effects models, natural experi-ments, and propensity score matching models. Our assessment is thatstudies using more rigorous designs continue to find negative effects offather absence on offspring well-being, although the magnitude of theseeffects is smaller than what is found using traditional cross-sectional de-signs. The evidence is strongest and most consistent for outcomes suchas high school graduation, children’s social-emotional adjustment, andadult mental health.
399
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39CH20-McLanahan ARI 27 June 2013 14:34
INTRODUCTION
A long tradition of sociological research has ex-amined the effects of divorce and father absenceon offspring’s economic and social-emotionalwell-being throughout the life course.1 Over-all, this work has documented a negative asso-ciation between living apart from a biologicalfather and multiple domains of offspring well-being, including education, mental health, fam-ily relationships, and labor market outcomes.These findings are of interest to family sociolo-gists and family demographers because of whatthey tell us about family structures and familyprocesses; they are also of interest to scholarsof inequality and mobility because of what theytell us about the intergenerational transmissionof disadvantage.
The literature on father absence has beencriticized for its use of cross-sectional dataand methods that fail to account for reversecausality, for omitted variable bias, or for het-erogeneity across time and subgroups. Indeed,some researchers have argued that the negativeassociation between father absence and childwell-being is due entirely to these factors. Thiscritique is well founded because family disrup-tion is not a random event and because the char-acteristics that cause father absence are likely toaffect child well-being through other pathways.Similarly, parents’ expectations about how theirchildren will respond to father absence mayaffect their decision to end their relationship.Finally, there is good evidence that father ab-sence effects play out over time and differ acrosssubgroups. Unless these factors are taken intoaccount, the so-called effects of father absenceidentified in these studies are likely to be biased.
Researchers have responded to concernsabout omitted variable bias and reverse cau-sation by employing a variety of innovativeresearch designs to identify the causal effect
1We use the term “father absence” to refer to children wholive apart from their biological father because of divorce, sep-aration from a cohabiting union, or nonmarital birth. We usethe terms “divorce” and “separation” to talk about change inchildren’s coresidence with their biological fathers.
of father absence, including designs that uselongitudinal data to examine child well-beingbefore and after parents separate, designsthat compare siblings who differ in theirexposure to separation, designs that use naturalexperiments or instrumental variables toidentify exogenous sources of variation infather absence, and designs that use matchingtechniques that compare families that are verysimilar except for father absence. In this article,we review the studies that use one or more ofthese designs. We limit ourselves to articlesthat have been published in peer-reviewedacademic journals, but we impose no restric-tions with regard to publication date (notethat few articles were published before 2000)or with regard to the disciplinary affiliationof the journal. Although most articles makeuse of data from the United States, we alsoinclude work based on data from Great Britain,Canada, South Africa, Germany, Sweden,Australia, Indonesia, and Norway. Using theseinclusion rules, we identified 47 articles thatmake use of one or more of these methods ofcausal inference to examine the effects of fatherabsence on outcomes in one of four domains:educational attainment, mental health, rela-tionship formation and stability, and labor forcesuccess.
In the next section, entitled “Strategiesfor Estimating Causal Effects with Observa-tional Data,” we describe these strategies, theirstrengths and weaknesses, and how they havebeen applied to the study of father absence. Inthe section entitled “Evidence for the CausalEffect of Family Structure on Child Out-comes,” we examine the findings from thesestudies in each of the four domains of well-being. Our goal is to see if, on balance, thesestudies tell a consistent story about the causaleffects of father absence and whether this storyvaries across different domains and across theparticular methods of causal inference that areemployed within each domain. We also notewhere the evidence base is large and where itis thin. We conclude by suggesting promisingavenues for future research.
400 McLanahan · Tach · Schneider
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39CH20-McLanahan ARI 27 June 2013 14:34
STRATEGIES FOR ESTIMATINGCAUSAL EFFECTS WITHOBSERVATIONAL DATA
Identifying causal effects with observationaldata is a challenging endeavor for several rea-sons, including the threat of omitted vari-able bias, the fact that multiple—and oftenreciprocal—causal effects are at work, the factthat the causal treatment condition (such as di-vorce) may unfold over a period of time or thatthere may be multiple treatment conditions,and the fact that the effects of the treatmentmay change over time and across subgroups.Traditional approaches to estimating the effectof father absence on offspring well-being haverelied primarily on ordinary least squares (OLS)or logistic regression models that treat offspringwell-being as a function of father absence plusa set of control variables. These models are at-tractive because the data requirements are min-imal (they can be estimated with cross-sectionaldata) and because they can accommodate com-plex specifications of the father absence ef-fect, such as differences in the timing of fatherabsence (early childhood versus adolescence),differences in postdivorce living arrangements(whether the mother lives alone or remarries),and differences by gender, race, and social class.Studies based on these models typically find thatdivorces that occur during early childhood andadolescence are associated with worse outcomesthan divorces that occur during middle child-hood, that remarriage has mixed effects on childoutcomes, and that boys respond more nega-tively than girls for outcomes such as behaviorproblems (see, for example, Amato 2001, Sigle-Rushton & McLanahan 2004).
Interpreting these OLS coefficients ascausal effects requires the researcher to assumethat the father absence coefficient is uncor-related with the error term in the regressionequation. This assumption will be violatedif a third (omitted) variable influences bothfather absence and child well-being or if childwell-being has a causal effect on father absencethat is not accounted for in the model. Thereare good reasons for believing that both of
these factors might be at work and so theassumption might not hold.
Until the late 1990s, researchers who wereinterested in estimating the effect of father ab-sence on child well-being typically tried to im-prove the estimation of causal effects by addingmore and more control variables to their OLSmodels, including measures of family resources(e.g., income, parents’ education, and age), aswell as measures of parental relationships (e.g.,conflict) and mental health (e.g., depression).Unfortunately, controlling for multiple back-ground characteristics does not eliminate thepossibility that an unmeasured variable is caus-ing both family structure and child well-being.Nor does it address the fact that multiple causalpathways may be at work, with children’scharacteristics and parents’ relationships recip-rocally influencing each other. Adding controlvariables to the model can also create new prob-lems if the control variables are endogenousto father absence. (See Ribar 2004 for a moredetailed discussion of cross-sectional models.)
Lagged Dependent Variable Model
A second approach to estimating the causal ef-fect of father absence is the lagged dependentvariable (LDV) model, which uses the standardOLS model described above but adds a controlfor child well-being prior to parents’ divorce orseparation. This approach requires longitudinaldata that measure child well-being at two pointsin time—one observation before and one af-ter the separation. The assumption behind thisstrategy is that the pre-separation measure ofchild well-being controls for unmeasured vari-ables that affect parents’ separation as well asfuture child well-being.
Although this approach attempts to reduceomitted variable bias, it also has several limi-tations. First, the model is limited with respectto the window of time when father absenceeffects can be examined. Specifically, the modelcannot examine the effect of absences thatoccur prior to the earliest measure of childwell-being, which means LDV models cannot
www.annualreviews.org • The Causal Effects of Father Absence 401
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39CH20-McLanahan ARI 27 June 2013 14:34
be used to estimate the effect of a nonmaritalbirth or any family structure in which a childhas lived since birth. Second, if pre-separationwell-being is measured with error, the variablewill not fully control for omitted variables.Third, lagged measures of well-being do notcontrol for circumstances that change betweenthe two points in time and might influence bothseparation and well-being, such as a parent’sjob loss. Another challenge to LDV studies isthat divorce/separation is a process that beginsseveral years before the divorce/separation isfinal. In this case, the pre-divorce measureof child well-being may be picking up partof the effect of the divorce, leading to anunderestimate of the negative effect of divorce.Alternatively, children’s immediate responseto divorce may be more negative than theirlong-term response, leading to an overestimateof the negative effect of divorce. Both of theselimitations highlight the fact that the LDV ap-proach is highly sensitive to the timing of whenchild well-being is measured before and afterthe divorce. In addition, many of the outcomesthat we care most about occur only once (e.g.,high school graduation, early childbearing),and the LDV strategy is not appropriate forthese outcomes. (See Johnson 2005 for amore detailed technical discussion of the LDVapproach in studying family transitions.)
These advantages and limitations areevident in Cherlin et al.’s (1991) classicstudy employing this method. Drawing onlongitudinal data from Great Britain and theUnited States, the authors estimated howthe dissolution of families that were intact atthe initial survey (age 7 in Great Britain and7–11 in the United States) affected children’sbehavior problems as well as their readingand math test scores at follow-up (age 11 inGreat Britain and 11–16 in the United States).In OLS regression models with controls, theauthors found that divorce increased behaviorproblems and lowered cognitive test scores forchildren in Great Britain and for boys in theUnited States. However, these relationshipswere substantially attenuated for boys andsomewhat attenuated for girls once the authors
adjusted for child outcomes and parentalconflict measured at the initial interviewprior to divorce. By using data that containedrepeated measurements of the same outcome,these researchers argue that they were able toreduce omitted variable bias and derive moreaccurate estimates of the causal effect of familydissolution. This approach also limited theexternal validity of the study, however, becausethe researchers could examine only separationsthat occurred after age 7, when the firstmeasures of child well-being were collected.
Growth Curve Model
A third strategy for estimating causal effectswhen researchers have measures of child well-being at more than two points in time is thegrowth curve model (GCM). This approachallows researchers to estimate two parametersfor the effect of father absence on child well-being: one that measures the difference in ini-tial well-being among children who experiencedifferent family patterns going forward, and an-other that measures the difference in the rate ofgrowth (or decline) in well-being among thesegroups of children. Researchers have typicallyattributed the difference in initial well-being tofactors that affect selection into father absenceand the difference in growth in well-being tothe causal effect of father absence. The GCMis extremely flexible with respect to its abilityto specify father absence effects and is there-fore well suited to uncovering how effects un-fold over time or across subgroups. For exam-ple, the model can estimate age-specific effects,whether effects persist or dissipate over time,and whether they interact with other charac-teristics such as gender or race/ethnicity. Themodel also allows the researcher to conduct aplacebo test—to test whether father absence attime 2 affects child well-being prior to divorce(time 1). If future divorce affects pre-divorcewell-being, this finding would suggest that un-measured variables are causing both the divorceand poor child outcomes.
The GCM also has limitations. First, itrequires a minimum of three observations of
402 McLanahan · Tach · Schneider
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39CH20-McLanahan ARI 27 June 2013 14:34
well-being for each individual in the sample.Second, as was true of the LDV model, itcan examine the effect of divorces that occuronly within a particular window of time—after the first and before the last measure ofchild well-being. Also, like the OLS model,the GCM does not eliminate the possibilitythat unmeasured variables are causing bothdifferences in family patterns and differencesin trajectories of child well-being, includinggrowth or decline in well-being. For example,an unmeasured variable that causes the initialgap in well-being could also be causing thedifference in growth rates. We are more confi-dent in the results of the GCMs if they show nosignificant differences in pre-divorce interceptsbut significant differences in growth rates. Weare also more confident in studies that includeplacebo or falsification tests, such as usingdifferences in future divorce to predict initialdifferences in well-being. If later family disrup-tion is significantly associated with differencesin pre-divorce well-being (the intercept), thisfinding would indicate the presence of selectionbias. [See Singer & Willett (2003) for a moredetailed technical discussion of GCMs andHalaby (2004) for a more detailed discussionof the assumptions and trade-offs among thevarious approaches to modeling panel data.]
Magnuson & Berger’s (2009) analysis of datafrom the Maternal and Child Supplement of theNational Longitudinal Survey of Youth 1979(NLSY79) is illustrative of this approach. Theseauthors used GCMs to examine the relationshipbetween the proportion of time children spentin different family structures between ages 6and 12 and scores on the Peabody IndividualAchievement Test (PIAT) cognitive ability testand the Behavioral Problems Index. They fo-cused on several family types: intact biological-parent families (married or cohabiting), social-father families (married or cohabiting), andsingle-parent families. They found no differ-ences in the initial well-being of the childrenin these different family structures, suggestingthat controls for observable factors had success-fully dealt with problems of selection. In con-trast, they found major differences in children’s
well-being trajectories, with time spent in in-tact biological-parent families leading to morefavorable trajectories than time spent in otherfamily types. The combination of insignificantdifferences in intercepts and significant differ-ences in slopes increases our confidence in theseresults. However, it remains possible that time-varying unobserved characteristics were drivingboth time spent in different family structuresand changes in child behavior and achievement.
Individual Fixed Effects Model
A fourth strategy for estimating causal effectsis the individual fixed effects (IFE) model, inwhich child-specific fixed effects remove alltime-constant differences among children. Thismodel is similar to the LDV and GCM in thatit uses longitudinal data with repeated mea-sures of family structure and child well-being.It is different in that instead of including pre-separation well-being as a control variable, it es-timates the effects of father absence using onlythe associations between within-child changesin family structure and within-child changesin well-being, plus other exogenous covariates(and an error term). The IFE model is estimatedby either including a distinct dummy variableindicator for each child, that absorbs all unob-served, time-constant differences among chil-dren, or by differencing out within-child av-erages from each dependent and independentvariable. In both of these specifications, onlywithin-child variation is used to estimate theeffects of father absence. The advantage of thismodel is that unmeasured variables in the errorterm that do not change over time are sweptout of the analysis and therefore do not bias thecoefficient for father absence. (See Ribar 2004for a discussion of fixed effects models.)
The IFE model also has limitations. As withLDVs and GCMs, IFE models cannot be esti-mated for outcomes that occur only once, suchas high school graduation or a teen birth, or foroutcomes that can be measured only in adult-hood, such as earnings. Also, as with LDVs andGCMs, the IFE model does not control forunobserved confounders that change over time
www.annualreviews.org • The Causal Effects of Father Absence 403
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39CH20-McLanahan ARI 27 June 2013 14:34
and jointly influence change in father presenceand change in child well-being. Third, becausethe model provides an estimate of the effect of achange in a child’s experience of father absence(moving from a two-parent to a single-parentfamily or vice versa), it does not provide anestimate of the effect of living in a stable one-parent family or a stable two-parent family.Unlike the other approaches, the IFE modelestimates the effect of father absence by com-paring before-after experiences for only thosechildren within the treatment group, ratherthan comparing children in the treatment andcontrol groups. Finally, and perhaps mostimportantly, the IFE model is very sensitiveto measurement error because estimates of theeffect of a change in father absence rely heavilyon within-individual changes.
A good illustration of the IFE approach is astudy by Cooper et al. (2011). Using data fromthe first four waves of the Fragile FamiliesStudy, the authors examined the link betweentwo measures of school readiness—verbalability and behavioral problems at age 5—andchildren’s exposure to family instability,including entrances and exits from the house-hold. Using an OLS model, they found thatthe number of partnership transitions wasassociated with lower verbal ability, moreexternalizing behavior, and more attentionproblems, but not more internalizing behavior.These relationships held for both coresiden-tial and dating transitions and were morepronounced for boys than girls. To addresspotential problems of omitted variable bias,the authors estimated a fixed effects model andfound that residential transitions, but not dat-ing transitions, reduced verbal ability amongall children and increased behavior problemsamong boys. The fact that the IFE estimateswere broadly consistent with the OLS estimatesincreases our confidence in the OLS results.
Sibling Fixed Effects Model
A fifth strategy for dealing with omitted vari-able bias is the sibling fixed effects (SFE) model.This model is similar to the previous model in
that unmeasured family-level variables that arefixed (i.e., do not vary among family members)are differenced out of the equation and do notbias the estimates of father absence. In this case,the group is the family rather than the individ-ual, and the difference that is being comparedis the difference between siblings with differ-ent family experiences rather than the changein individual exposure to different family expe-riences. The literature on father absence con-tains two types of SFE models. One approachcompares biological siblings who experience fa-ther absence at different ages. In this case, theestimate of the causal effect of father absenceis based on the difference in siblings’ length ofexposure. For example, a sibling who is age 5at the time of a divorce or separation will ex-perience 12 years of father absence by age 17,whereas a sibling who is age 10 when the sep-aration occurs will experience 7 years of fatherabsence by age 17. In some instances, childrenmay leave home before their parents’ divorce, inwhich case they are treated as having no expo-sure. A second approach compares half-siblingsin the same family, where one sibling is livingwith two biological parents and the other is liv-ing with a biological parent and a stepparentor social father. Both of these strategies sweepout all unmeasured family-level variables thatdiffer among families and could potentially biasthe estimate of the effect of divorce.
Both approaches also have limitations. Thefirst approach assumes that the effect of divorcedoes not vary by the age or temperament of thechild and that there is a dose-response effectof father absence with more years of absenceleading to proportionately worse outcomes,whereas the second approach assumes that thebenefits of the presence of both a biologicalmother and father are similar for children livingwith and without stepsiblings. With respectto the first assumption, as previously noted,both theory and empirical evidence suggestthat, at least for some outcomes, divorcesoccurring in early childhood and adolescencehave more negative effects on child outcomesthan divorces occurring in middle child-hood (Sigle-Rushton & McLanahan 2004).
404 McLanahan · Tach · Schneider
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39CH20-McLanahan ARI 27 June 2013 14:34
Moreover, if siblings differ in their abilityto cope with divorce, and if parents takethis difference into account in making theirdecision about when to divorce, this approachwill lead to an underestimate of the effect of achange in family structure.
The major limitation of the second approachis that it assumes that the benefits of living withtwo biological parents are similar for childrenliving in blended families and children livingin traditional two-parent families. With respectto this assumption, there is good evidence thatstepparent families are less cooperative than sta-ble two-parent families, which means that livingin a blended family is likely to reduce the well-being of all children in the household (Sigle-Rushton & McLanahan 2004). A final limita-tion of the SFE model is that estimates cannotbe generalized to families with only one child.2
Within-family fixed effects models areemployed in Gennetian’s (2005) analysis ofdata on 5- to 10-year-old children interviewedfrom 1986 to 1994 for the children of theNLSY79 study. Gennetian examined howchildren in two-biological-parent families,stepfather families, and single-mother familiesfared on the PIAT cognitive test as well ashow children living with step- or half-siblingscompared to those with only full siblings.In simple comparisons, the data revealed asignificant disadvantage in PIAT scores forchildren in single-mother families, stepfatherfamilies, and blended families relative to thosein two-biological-parent families. Gennetian(2005) then leveraged the data, which includedrepeated measurements over time of family
2Children of twin studies are a variation of the SFE model.These studies, pioneered by D’Onofrio and colleagues (2006,2007), compare the offspring of identical (MZ) twins, frater-nal (DZ) twins, and regular siblings in cases in which onesibling or twin divorces and the other does not. These anal-yses control for family differences that are common to bothsiblings; however, they do not control for within-sibling dif-ferences that lead one sibling to divorce and another to bestably married. Twin studies go one step further, by compar-ing MZ twins (who share identical genetic information) andDZ twins (who have half of their genes identical), allowingresearchers to determine the role of genetics in accountingfor the effect of divorce.
composition and outcomes for all of themother’s children, to estimate models withmother and child fixed effects. These analysesfound very little evidence that children living insingle-mother, stepfather, or blended familieswere disadvantaged on PIAT scores relative tochildren in nonblended two-biological-parentfamilies, although they did indicate thatnumber of years in a single-mother family hada small negative effect on PIAT scores.
Finally, Gennetian further tested the logicof the sibling approach by comparing thewell-being of half-siblings, one of whom wasliving with both biological parents and theother of whom was living with a biologicalparent and a stepparent. The analyses showedthe expected negative effect on PIAT scoresfor children living with stepfathers, with thisrelationship remaining negative (but decliningin size and losing significance) in models withmother and child fixed effects. Importantly,these analyses also revealed a negative effect ofthe presence of a half-sibling on the child whowas living with two biological parents.
Natural Experiments andInstrumental Variables
A sixth strategy is to use a natural experimentto estimate the effect of divorce on child well-being. The logic behind this strategy is to findan event or condition that strongly predicts fa-ther absence but is otherwise unrelated to theoffspring outcome of interest. The natural ex-periment may be an individual-level variable oran aggregate-level measure.
Several studies use parental death as a nat-ural experiment, comparing outcomes for chil-dren whose parents divorced with those whoseparent died. The assumption behind this strat-egy is that experiencing parental death is a ran-dom event and can therefore be used to obtainan unbiased estimate of the effect of father ab-sence. In such analyses, a significant negativerelationship between child outcomes and bothparental death and divorce is taken as evidenceof the causal relationship of divorce on childwell-being, particularly if the divorce and death
www.annualreviews.org • The Causal Effects of Father Absence 405
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39CH20-McLanahan ARI 27 June 2013 14:34
coefficients are not statistically different.3 A ma-jor challenge for these studies is that parentaldeath is rarely random; whatever is causingthe death may also be causing the child out-come. Violent and accident-related deaths, forexample, are selective of people who engage inrisky behaviors; similarly, many illness-relateddeaths are correlated with lifestyles that affectchild outcomes, such as smoking. Children ofdeceased parents are also treated very differ-ently than are children of divorced parents, notonly by their informal support systems but alsoby the government.
Other studies use natural experiments toestimate instrumental variable (IV) models.This strategy involves a two-stage least squaresprocedure. In the first stage, the researcherregresses the endogenous predictor (in thiscase father absence) on the instrument, whichis a source of exogenous variation that is highlycorrelated with father absence, and uses themodel to obtain a predicted father absence(PFA) measure for each individual. Then, in thesecond step, PFA is substituted for the actualfather absence variable in a model predictingoffspring well-being. Because PFA is basedentirely on observed variables, the coefficientfor this variable cannot be correlated withunmeasured variables, thereby removing thethreat of omitted variable bias. For this strategyto work, however, the researcher must find avariable—or instrument—that is a strong pre-dictor of divorce or separation but that is notcorrelated with the outcome of interest exceptthrough its effects on father absence or divorce.This assumption is often violated [for example,see Besley & Case (2000) for a discussion of whystate policies are not random with respect tochild well-being]. A limitation of the IV modelis that it requires a large sample. Because PFAis based on predicted absence rather than actualabsence, it can be measured with a good dealof error, which results in large standard errors
3We only include studies of the effect of parental death onchild outcomes if the author uses one of the causal methodsdescribed below or explicitly uses death as a natural experi-ment for divorce or other types of father absence.
in the child well-being equation and makes itdifficult to interpret results that are not statisti-cally significant. Finally, the IV model requiresa different instrument for each independentvariable, which limits the researcher’s abilityto study different types of father absence.
A good example of the natural exper-iment/IV approach and its limitations isGruber’s (2004) analysis of the effect ofchanges in divorce laws on divorce and childoutcomes. Combining data on state differencesin divorce laws with information from the1960–1990 US Censuses, Gruber found asignificant positive effect of the presence ofunilateral divorce laws—which make divorceeasier—on the likelihood of being divorced.This part of the analysis satisfied the firstrequirement for the IV model, namely, that theinstrument be strongly associated with divorce.He then estimated the effect of living in a state(for at least part of childhood) where unilateraldivorce was available on a host of adult out-comes. These analyses showed that unilateraldivorce laws were associated with early mar-riage and more divorce, less education, loweredfamily income, and higher rates of suicide.Additionally, women so exposed appeared tohave lower labor force attachment and lowerearnings. To distinguish the effect of divorcelaws from other state-level policies, Gruberinvestigated the associations between thepresence of unilateral divorce laws and changesin welfare generosity and education spendingduring this same time period, finding no associ-ations suggestive of bias. He did find, however,that his results were driven in large part byfactors at work in California over this period.
Most importantly, Gruber concluded thatdivorce laws did not pass the second require-ment of the IV model, namely, that they af-fect child well-being only through their effecton parents’ divorce. Instead, he argued that di-vorce laws are likely to affect child well-beingby altering decisions about who marries and byaltering the balance of power among marriedcouples. Gruber’s analysis highlights the diffi-culty of finding a natural experiment that trulysatisfies both assumptions of the IV model.
406 McLanahan · Tach · Schneider
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39CH20-McLanahan ARI 27 June 2013 14:34
Propensity Score Matching
A final strategy used in the literature for ob-taining estimates of the causal effect of divorceis propensity score matching (PSM). Based onthe logic of experimental design, this approachattempts to construct treatment and controlgroups that are similar in all respects except forthe treatment condition, which in this litera-ture is father absence. The strategy begins byestimating the probability of father absence foreach child based on as many covariates as pos-sible observed in the data, and then uses thispredicted probability to match families so thatthey are similar to one another in all respectsexcept for father absence.
This approach has several advantages overthe OLS model. First, researchers may excludefamilies that do not have a good match (i.e., asimilar propensity to divorce), so that we aremore confident that our estimates are based oncomparing “apples to apples.” Second, PSManalyses are more flexible than OLS becausethey do not impose a particular functional formon how the control variables are associated withdivorce. PSM estimation is also more efficientthan OLS because it uses a single variable—predicted probability of divorce—that com-bines the relevant predictive information fromall the potential observed confounders. Finally,it can accommodate the fact that the effects ofdivorce may differ across children by estimatingseparate effects for children in families with lowand high propensities to divorce. Propensityscores may also be used to reweight the data sothat the treatment and control groups are moresimilar in terms of their observed covariates(Morgan & Todd 2008, Morgan & Winship2007).
The PSM approach has limitations as well.First, the model is less flexible than the OLSmodel in terms of the number and complexityof family structures that can be comparedin a single equation. Second, the approachdoes not control for unmeasured variables,although it is possible to conduct sensitivityanalyses to address the potential influence ofsuch variables. For this reason, the approachis less satisfactory than IV models for making
causal inferences. Finally, the strategy reliesheavily on the ability of the researcher tofind suitable matches. If there is not sufficientoverlap in the kinds of people who divorceand the kinds of people who remain stablymarried, the approach will not work. Similarly,by limiting the sample to cases with a match,the researcher also reduces sample size and,more importantly, the generalizability of theresults [see Morgan & Winship (2007), Ribar(2004), and Winship & Morgan (1999) for amore extended technical discussion of the logicand assumptions of matching techniques].
The work of Frisco et al. (2007) serves as anexample of the use of PSM models in the studyof the effects of divorce. Drawing on the AddHealth data, the authors first estimated simpleOLS regressions of the relationship betweenthe dissolution of a marital or cohabiting re-lationship between wave I (when the studentswere ages 7–12) and wave II (the following year)and adolescents’ level of mathematics course-work, change in GPA, and change in proportionof courses failed between the two waves. Thesemodels revealed a significant negative relation-ship between dissolution and the measures ofGPA and course failure but no link to mathe-matics coursework, after controlling for a largenumber of potentially confounding variables.
Next, the authors calculated a propensityto experience dissolution as a function of par-ents’ race, education, income, work, age, re-lationship experience and quality, religiosity,and health and adolescents’ age, gender, andnumber of siblings, and then used this pre-dicted propensity to conduct nearest neighbormatching with replacement and kernel match-ing. Regardless of matching method, the esti-mates from the PSM models accorded very wellwith those from the simple OLS regressions.As in those models, there were significant nega-tive relationships between dissolution and GPAand positive relationships with course failure,and the point estimates were of a very similarmagnitude across models. This study also exam-ined how large the influence of an unobservedconfounder would have had to be in order tothreaten the causal interpretation of the results.
www.annualreviews.org • The Causal Effects of Father Absence 407
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39CH20-McLanahan ARI 27 June 2013 14:34
The study had some unique and some gen-eral limitations. Because of data limitations, theauthors could not separate dissolutions stem-ming from divorce from those attributable toother causes, such as parental death. More gen-erally, because matching is limited to observ-able characteristics, the authors could calculateonly propensities of dissolution based on ob-servable characteristics. To assess the sensitiv-ity of their results to omitted variable bias, theauthors conducted a simulation and discoveredthat an unobserved confounder that is moder-ately associated with dissolution and the out-comes (r < 0.1) could bias their findings.
EVIDENCE FOR THE CAUSALEFFECT OF FAMILY STRUCTUREON CHILD OUTCOMES
In this section, we assess the evidence for acausal effect of father absence on different do-mains of offspring well-being. Empirical stud-ies have used multiple strategies for identifyingcausal effects that each have unique strengthsand weaknesses—as we identified in the previ-ous section—but we are more confident in thepresence of causal effects if we identify con-sistent results across multiple methods. Manyof the articles we examine used more than oneanalytic strategy and/or examined outcomes inmore than one domain. Consequently, our unitof analysis is each separate analysis reportedin an article, rather than the article itself. Forinstance, rather than discussing an article thatincludes both SFE and LDV analyses of testscores and self-esteem as a single entity, wediscuss it as four separate cases. The virtue ofthis approach is that it allows us to discern pat-terns more clearly across studies using similaranalytic strategies and across studies examiningsimilar outcomes. The drawback is that somearticles contribute many cases and some onlyone. If there are strong author-effects, for ar-ticles that contribute many cases, then our un-derstanding of the results produced by a givenanalytic strategy or for a given domain couldbe skewed. We note when this occurs in ourdiscussions below.
Studies in this field measured father absencein several ways, which the reader should keep inmind when interpreting and comparing resultsacross studies. Some studies compared childrenof divorced parents with children of stably mar-ried parents; others compared children whoseparents married after their child’s birth withthose parents who never married; still otherssimply compared two-parent to single-parentfamilies (regardless of whether the former werebiological or stepparents and the latter weresingle through divorce or a nonmarital birth).More recently, researchers have started to useeven more nuanced categories to measure fam-ily structure—including married biological-parent families, cohabiting biological-parentfamilies, married stepparent families, cohabit-ing stepparent families, and single parents bydivorce and nonmarital birth—reflecting thegrowing diversity of family forms in society.Still other studies look at the number of fam-ily structure transitions the child experiencesas a measure of family instability. We did notidentify any studies that used causal methods tostudy the effects of same-sex unions.
Finally, we include studies of father absencethat use data from a range of US and inter-national samples. We should note, however,that what it means to reside in a father-absenthousehold varies a great deal cross-nationally.Children whose parents are not married facestarkly different levels of governmental andinstitutional support and unequal prospects forliving in a stable two-parent family in differentcountries. In fact, both marital and nonmaritalunions in the United States are considerablyless stable than in any other industrializednation (Andersson 2002).
Education
We begin our review of the empirical findingsby looking at studies that attempted to estimatethe causal effect of divorce on school success.We distinguish between studies that lookedat children’s standardized test scores; studiesthat looked at educational attainment; andstudies that looked at children’s attitudes,engagement, and school performance.
408 McLanahan · Tach · Schneider
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39CH20-McLanahan ARI 27 June 2013 14:34
Test scores. We identified 31 analyses thatexamined the relationship between fatherabsence and test scores, including tests ofverbal, math, and general ability. The articlescontaining these analyses are listed and brieflydescribed in the first section of Table 1.Virtually all of the test score analyses usedUS-based samples (only Cherlin et al. 1991used international data). Although the overallpicture for test scores was mixed, with 14finding significant effects and 17 finding noeffect, there were patterns by methodology.4
First, significant effects were most likely in theanalyses using GCMs. Of the GCM studiesfinding significant differences in slopes be-tween children of divorced and intact families,about half found no significant differences inthe pre-divorce intercepts, which made theirsignificant results more convincing. One GCMstudy (Magnuson & Berger 2009) performeda falsification test and found no evidence thatsubsequent divorce predicted intercepts, rulingout the threat of selection bias.
In contrast with analyses based on the GCMdesign, the IFE and SFE analyses rarely foundsignificant effects of family structure on chil-dren’s test scores. In general, standard errorstended to be larger in IFE and SFE analysesthan in OLS analyses, but in virtually all ofthese analyses, the fixed effects coefficientswere markedly reduced in size relative to theOLS coefficients, suggesting that the lack ofsignificant results was not solely due to largerstandard errors.
Several factors may have limited the gen-eralizability of the fixed effects models, how-ever. First, all of these analyses came from com-parisons of siblings in blended families. Theparents in blended families differed from thosein traditional married families because at leastone of the parents had children from a pre-vious relationship, limiting the external valid-ity of these results. Second, the father-absent
4The picture remains mixed even within particular types oftests (math, reading/verbal, or general ability). Most studiesused the PPVT or PIAT Math and Reading tests.
category included children of divorced parentsas well as children of never-married mothers,whereas the father-present category containedboth children whose mothers were married atbirth and children whose mothers married af-ter the child’s birth. We might expect that thebenefit of moving from a single-parent house-hold to a married-parent household would besmaller than the benefit of being born into astably married family. Given these comparisonsand the small samples involved in estimation, itis understandable that we found little evidenceof an impact of family structure on test scoresusing fixed effects models.
Although there were clear patterns in theGCM and fixed effects analyses, LDV studieswere a mixed bag: Half found effects and halfdid not. Sometimes the results were not robusteven within the same paper. For example, bothCherlin et al. (1991) and Sanz-de-Galdeano& Vuri (2007) found significant effects formath scores but not reading scores. Using thesame data as Sanz-de-Galdeano & Vuri (theNational Education Longitudinal Study), Sun(2001) found positive effects for both math andreading tests.
Educational attainment. There is moreconsistent evidence of a causal effect of fatherabsence on educational attainment, particularlyfor high school graduation. Of nine studies ex-amining high school graduation using multiplemethodologies, only one found null effects,and this study used German data to comparesiblings in blended families. There was alsorobust evidence of effects when attainment wasmeasured by years of schooling. Again, the onlystudies that found no effect of father absencewere those that used international samples orcompared siblings in blended families. Finally,there was weak evidence for effects on collegeattendance and graduation, with only one offour studies finding significant results. Takentogether, the evidence for an effect of fatherabsence on educational attainment, particularlyhigh school graduation, is strong in studiesusing US samples, perhaps because of therelatively open structure of the US educational
www.annualreviews.org • The Causal Effects of Father Absence 409
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39CH20-McLanahan ARI 27 June 2013 14:34
Tab
le1
Stud
ies
ofth
eef
fect
sof
fath
erab
senc
eon
educ
atio
n
Ref
eren
ceO
utco
mes
aT
ype
ofan
alys
isb
Fam
ilyty
pes
com
pare
dD
ata
sour
cesc
Age
whe
nou
tcom
esob
serv
edC
ount
ryof
data
Tes
tsc
ores
Aug
hinb
augh
etal
.(2
005)
Mat
hab
ility
(N),
read
ing
abili
ty(N
)G
row
thcu
rves
,IFE
Mar
ried
vers
usdi
vorc
ed,
sing
leve
rsus
gotm
arri
edN
LSY
79C
NL
SY79
4–15
in19
86–2
000
USA
Bac
hman
etal
.(20
09)
Mat
hab
ility
(N),
read
ing
abili
ty(N
)IF
ESi
ngle
vers
usgo
tmar
ried
vers
usst
arte
dco
habi
ting
Thr
ee-c
ityst
udy
10–1
6in
1999
–200
1U
SA
Che
rlin
etal
.(19
91)
Rea
ding
abili
ty(N
),m
ath
abili
ty(Y
)L
DV
Div
orce
dve
rsus
mar
ried
NC
DS
NSC
Age
11in
1969
Gre
atB
rita
in,
USA
Coo
per
etal
.(20
11)
Rea
ding
abili
ty(Y
for
LD
V,N
for
IFE
)L
DV
,IFE
Num
ber
oftr
ansi
tions
into
/out
ofre
latio
nshi
psFF
CW
S3–
5in
2001
–200
5U
SA
Fost
er&
Kal
il(2
007)
Rea
ding
abili
ty(N
)IF
ET
wo
bio,
sing
le-m
othe
r,st
ep,o
rm
ultig
ener
atio
nal
CC
DP
2–5
in19
90–1
996
USA
Hof
fert
h(2
006)
Mat
hab
ility
(Y),
read
ing
abili
ty(Y
)SF
EB
iove
rsus
step
fath
erP
SID
6–12
in19
97U
SA
Mag
nuso
n&
Ber
ger
(200
9)M
ath
abili
ty(Y
),re
adin
gab
ility
(Y)
Gro
wth
curv
esT
wo
bio,
sing
le-m
othe
r,or
step
CN
LSY
795–
12in
1986
–200
2U
SA
Mor
riso
n&
Che
rlin
(199
5)R
eadi
ngab
ility
(N)
LD
VT
wo
bio
pare
nts
vers
usdi
srup
ted
NL
SY79
Ave
rage
age
5–8
in19
86–1
988
USA
Sanz
-de-
Gal
dean
o&
Vur
i(20
07)
Mat
hab
ility
(Y),
read
ing
abili
ty(N
),sc
ienc
eab
ility
(N),
and
soci
alst
udie
sab
ility
(N)
LD
VT
wo
bio
pare
nts
vers
usdi
srup
ted
NE
LS
15–1
8in
1990
–199
2U
SA
Shaf
feta
l.(2
008)
Mat
hab
ility
(Yfo
rne
ver
mar
ried
,Nfo
rdi
vorc
ed),
read
ing
abili
ty(Y
)
Gro
wth
curv
esR
epar
tner
ing
afte
rdi
vorc
eor
nonm
arita
lbir
thC
NL
SY79
5–18
in19
86–1
996
USA
Sun
(200
1)M
ath
abili
ty(Y
),re
adin
gab
ility
(Y),
educ
atio
nal
aspi
ratio
ns(N
),sc
hool
enga
gem
ent(
N)
LD
VT
wo
bio
vers
usdi
vorc
edN
EL
S15
–18
in19
90–1
992
USA
410 McLanahan · Tach · Schneider
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39CH20-McLanahan ARI 27 June 2013 14:34
Sun
&L
i(20
02)
Mat
hab
ility
(Y),
read
ing
abili
ty(Y
),sc
ienc
eab
ility
(Y),
soci
alst
udie
sab
ility
(Y),
educ
atio
nal
aspi
ratio
ns(Y
)
Gro
wth
curv
esT
wo
bio
vers
usdi
vorc
edN
EL
S13
–18
in19
88–1
992
USA
Edu
cati
onal
atta
inm
ent
Bib
larz
&G
otta
iner
(200
0)
Edu
catio
nala
ttai
nmen
t(Y
)N
atur
alex
peri
men
t:pa
rent
alde
ath
Div
orce
dve
rsus
wid
owed
GSS
18+
in19
72–1
996
USA
Bjo
rklu
nd&
Sund
stro
m(2
006)
Edu
catio
nala
ttai
nmen
t(N
)SF
ET
wo
bio
pare
nts
(incl
udin
gco
habi
ting)
vers
usno
t
Nat
iona
lR
egis
try
33–4
8in
1996
Swed
en
Bjo
rklu
ndet
al.
(200
7)E
duca
tiona
latt
ainm
ent
(N)
SFE
Tw
obi
opa
rent
s(in
clud
ing
coha
bitin
g)ve
rsus
not
NL
SY79
PSI
DR
egis
try
NL
SYag
es24
–34
in19
94,P
SID
ages
23–3
3in
1993
,Sw
eden
26–3
6in
1996
USA
,Sw
eden
Cas
e&
Ard
ingt
on(2
006)
Edu
catio
nala
ttai
nmen
t,sc
hool
perf
orm
ance
(Nfo
rfa
ther
deat
h,Y
for
mot
her
deat
h)
Nat
ural
expe
rim
ent:
pare
ntal
deat
hP
aren
tald
eath
vers
usno
deat
hD
SASu
rvey
and
Sout
hA
fric
anC
ensu
s
6–16
in20
03–2
004
Sout
hA
fric
a
D’O
nofr
ioet
al.
(200
6)E
duca
tiona
latt
ainm
ent
(Y)
Tw
ins
Div
orce
dve
rsus
mar
ried
(MZ
vers
usD
Ztw
ins)
Aus
tral
ian
Nat
iona
lT
win
Reg
istr
y
14–3
9in
1992
–199
3A
ustr
alia
Erm
isch
&Fr
ance
scon
i(2
001)
Edu
catio
nala
ttai
nmen
t(Y
for
disr
uptio
nag
e0–
5,N
for
disr
uptio
nag
e6+
)
SFE
Tw
obi
opa
rent
s(in
clud
ing
coha
bitin
g)ve
rsus
not
BH
PS
18–2
9in
1999
Gre
atB
rita
in
Erm
isch
etal
.(2
004)
Edu
catio
nala
ttai
nmen
t(Y
for
disr
uptio
n<
age
11,N
for
disr
uptio
nag
e11
+)
SFE
Tw
obi
opa
rent
sve
rsus
not
BH
PS
18–2
9in
1999
Gre
atB
rita
in
Finl
ay&
Neu
mar
k(2
010)
educ
atio
nala
ttai
nmen
t(N
)O
ther
:IV
(inca
rcer
atio
nra
tes)
Mar
ried
vers
usne
ver-
mar
ried
mot
hers
Dec
enni
alC
ensu
s15
–17
in19
70–2
000
USA
(Con
tinue
d)
www.annualreviews.org • The Causal Effects of Father Absence 411
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39CH20-McLanahan ARI 27 June 2013 14:34
Tab
le1
(Con
tinu
ed)
Ref
eren
ceO
utco
mes
aT
ype
ofan
alys
isb
Fam
ilyty
pes
com
pare
dD
ata
sour
cesc
Age
whe
nou
tcom
esob
serv
edC
ount
ryof
data
Fran
cesc
onie
tal.
(201
0)E
duca
tiona
latt
ainm
ent
(N),
scho
olpe
rfor
man
ce(N
)
SFE
,pro
pens
itysc
ore,
natu
ral
expe
rim
ent:
divo
rce
law
,na
tura
lex
peri
men
t:pa
rent
alde
ath
Tw
obi
opa
rent
sve
rsus
not,
expo
sed
tore
form
eddi
vorc
ela
wve
rsus
not,
pare
ntal
deat
hve
rsus
divo
rce
vers
usm
arri
ed
Ger
man
Soci
o-E
cono
mic
Pan
el
Age
14in
1984
–200
5G
erm
any
Gen
netia
n(2
005)
Edu
catio
nala
ttai
nmen
t(Y
)SF
ET
wo
bio
nonb
lend
edfa
mily
vers
ussi
ngle
-mot
her,
step
,or
blen
ded
fam
ily
NL
SY79
CN
LSY
795–
10in
1986
–199
4U
SA
Ger
tler
etal
.(20
04)
Edu
catio
nala
ttai
nmen
t(Y
)P
rope
nsity
scor
eP
aren
tald
eath
vers
usno
deat
hSu
sena
s6–
20in
1994
–199
6In
done
sia
Gin
ther
&P
olla
k(2
004)
Edu
catio
nala
ttai
nmen
t(N
),m
ath
abili
ty(N
),re
adin
gab
ility
(N)
SFE
Tw
obi
opa
rent
sve
rsus
step
pare
ntP
SID
5–15
in19
94U
SA
Gru
ber
(200
4)E
duca
tiona
latt
ainm
ent
(Y)
Nat
ural
expe
rim
ent:
divo
rce
law
Exp
osed
tore
form
eddi
vorc
ela
wve
rsus
not
Dec
enni
alC
ensu
s20
–50
in19
60,
1970
,198
0,or
1990
Cen
sus
USA
Lan
g&
Zag
orsk
y(2
001)
Edu
catio
nala
ttai
nmen
t(N
),A
FQT
test
(N)
Nat
ural
expe
rim
ent:
pare
ntal
deat
h
Par
enta
ldea
thve
rsus
divo
rce
vers
usm
arri
edN
LSY
7922
–36
in19
93U
SA
Man
skie
tal.
(199
2)E
duca
tiona
latt
ainm
ent
(Y)
Oth
er:l
aten
tva
riab
le,
nonp
aram
etri
cm
odel
s
Tw
obi
opa
rent
sve
rsus
not
NL
SY79
Age
20in
1979
–198
5U
SA
Sand
efur
&W
ells
(199
9)E
duca
tiona
latt
ainm
ent
(Y)
SFE
Tw
obi
opa
rent
sver
suss
ingl
epa
rent
NL
SY79
27–3
5in
1992
USA
412 McLanahan · Tach · Schneider
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39CH20-McLanahan ARI 27 June 2013 14:34
Stee
leet
al.(
2009
)E
duca
tiona
latt
ainm
ent
(N)
Nat
ural
expe
rim
ent:
pare
ntal
deat
h
Par
enta
ldea
thve
rsus
divo
rce
vers
usm
arri
edN
orw
egia
npo
pula
tion
regi
ster
s
16–2
9in
2003
Nor
way
Att
itud
es,p
erfo
rman
ce,a
nden
gage
men
tA
ston
e&
McL
anah
an(1
991)
Edu
catio
nala
spir
atio
ns(N
),sc
hool
enga
gem
ent
(Y),
scho
olpe
rfor
man
ce(N
)
LD
VT
wo
bio
pare
nts
(incl
udin
gco
habi
ting)
vers
usno
tH
S&B
16–1
9in
1980
–198
2U
SA
Bro
wn
(200
6)Sc
hool
enga
gem
ent(
N)
LD
VT
wo
bio
pare
nts
(incl
udin
gco
habi
ting)
,sin
gle-
mot
her,
mar
ried
step
fam
ily,
coha
bitin
gst
epfa
mily
Add
Hea
lth12
–20
in19
94–1
996
USA
Eve
nhou
se&
Rei
lly(2
004)
Scho
olen
gage
men
t(N
),sc
hool
perf
orm
ance
(Yfo
rG
PA
,Nfo
rre
peat
eda
grad
e)
SFE
Tw
obi
opa
rent
sve
rsus
step
pare
ntA
ddH
ealth
12–1
8in
1994
USA
Fris
coet
al.(
2007
)Sc
hool
perf
orm
ance
(Yfo
rG
PA
,Yfo
rco
urse
sfa
iled,
Nfo
rm
ath
cour
ses
com
plet
ed)
Pro
pens
itysc
ore
Tw
obi
opa
rent
sver
suss
ingl
epa
rent
Add
Hea
lth12
–20
in19
94–1
996
USA
a Out
com
eab
brev
iatio
ns:N
,no
effe
ctfo
und;
Y,e
ffect
foun
d.bT
ype
ofan
alys
isab
brev
iatio
ns:I
FE,i
ndiv
idua
lfixe
def
fect
s;IV
,ins
trum
enta
lvar
iabl
e;L
DV
,lag
ged
depe
nden
tvar
iabl
e;SF
E,s
iblin
gfix
edef
fect
s.c D
ata
sour
cesa
bbre
viat
ions
:Add
Hea
lth,N
atio
nalL
ongi
tudi
nalS
tudy
ofA
dole
scen
tHea
lth;B
HP
S,B
ritis
hH
ouse
hold
Pan
elSu
rvey
(199
1–19
99);
CC
DP,
Com
preh
ensi
veC
hild
Dev
elop
men
tP
roje
ct;C
NL
SY79
,Chi
ldre
nof
the
Nat
iona
lLon
gitu
dina
lSur
vey
ofY
outh
1979
;DSA
,dem
ogra
phic
surv
eilla
nce
area
;FFC
WS,
Frag
ileFa
mili
esan
dC
hild
Wel
lbei
ngSt
udy;
GSS
,Gen
eral
Soci
alSu
rvey
;HS&
B,H
igh
Scho
ol&
Bey
ond;
NC
DS
NSC
,Nat
iona
lChi
ldD
evel
opm
entS
tudy
/Nat
iona
lSur
vey
ofC
hild
ren;
NE
LS,
Nat
iona
lEdu
catio
nalL
ongi
tudi
nalS
tudy
;NL
SY79
,N
atio
nalL
ongi
tudi
nalS
urve
yof
You
th19
79;P
SID
,Pan
elSt
udy
ofIn
com
eD
ynam
ics.
www.annualreviews.org • The Causal Effects of Father Absence 413
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39CH20-McLanahan ARI 27 June 2013 14:34
system compared with the more rigid trackingsystems within many European countries.
How might one explain the stronger, moreconsistent evidence base for father absence ef-fects on educational attainment relative to cog-nitive ability? One explanation is that measure-ment error in test scores is to blame for the weakand sometimes inconsistent findings in that do-main. Another explanation is that the meth-ods involved in measuring attainment—siblingmodels and natural experiments—do not con-trol as rigorously for unobserved confoundersas the repeated-measure studies (GCM, LDV,IFE) of cognitive ability.
The lack of strong test score effects is alsoconsistent with findings in the early educationliterature that suggest that cognitive test scoresare more difficult to change than noncognitiveskills and behaviors (see, e.g., HighScope PerryPreschool Project; Schweinhart et al. 2005).Given that educational attainment is based on acombination of cognitive ability and behavioralskills (that are influenced by family structure, aswe describe below), it makes sense that we findstrong evidence of effects on the likelihood ofhigh school graduation but not on test scores.
Attitudes, performance, and engagement.A smaller number of analyses (11) examinedthe effect of father absence on children’s schoolperformance, including GPA, coursework, andtrack placement. Of these analyses, four foundno significant effect on track placement us-ing German data and multiple methodologies(Francesconi et al. 2010). Three analyses camefrom a study in the United States by Friscoet al. (2007) that found effects for GPA andcourses failed, but not for a third, somewhat un-usual measure: years of math coursework com-pleted. It is difficult to draw any conclusionsabout the effects of family structure on schoolperformance across these disparate samples andmeasures.
Finally, seven analyses examined the effectof father absence on educational engagementand aspirations among teenagers in the UnitedStates. Five of the seven analyses found no ef-fect on these noncognitive measures. For ex-
ample, one study (Sun & Li 2002) found pos-itive effects on aspirations, but the other twofound no effect. Similarly, one study (Astone& McLanahan 1991) found positive effects onschool engagement, but the other three foundno effect. The latter findings suggest that ed-ucational aspirations and orientations towardschooling may form at younger ages, and noneof these analyses examined aspirations amongchildren younger than age 12.
Mental Health
After education, the second most commonoutcome examined in the literature is mentalhealth, which is measured as social-emotionaldevelopment when respondents are childrenand adolescents. Mental health and social-emotional development are closely related towhat social scientists call noncognitive skills orsoft skills to distinguish them from cognitiveskills measured by math and reading tests. Re-cent research shows that social-emotional skillsplay an important role in adult outcomes, notonly in influencing mental health but also ininfluencing educational attainment, family for-mation and relationships, and labor market suc-cess (Cunha & Heckman 2008).
Adult mental health. We identified six stud-ies that examined the association betweenparental divorce and adult mental health (seeTable 2). Three of these studies were based onUK data, and three were based on US data. Allof the empirical strategies that we discussed inthe previous section were used to estimate theeffects of divorce and father absence on adultmental health. The findings were quite robust,with four of the six analyses showing a nega-tive effect of parental divorce on adult mentalhealth. Moreover, one of the two null findings(Ermisch & Francesconi 2001) was overturnedin a subsequent paper by the same authors thatdistinguished between early and later exposureto divorce (Ermisch et al. 2004).
Social-emotional problems. Social-emo-tional problems in childhood are typically
414 McLanahan · Tach · Schneider
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39CH20-McLanahan ARI 27 June 2013 14:34
Tab
le2
Stud
ies
ofth
eef
fect
sof
fath
erab
senc
eon
men
talh
ealt
h
Ref
eren
ceO
utco
mes
aT
ype
ofan
alys
isb
Fam
ilyty
pes
com
pare
dD
ata
sour
cesc
Age
whe
nou
tcom
esob
serv
edC
ount
ryof
data
Adu
ltm
enta
lhea
lth
Am
ato
(200
3)A
dult
men
talh
ealth
(Y).
Pro
pens
itysc
ore
Div
orce
dve
rsus
mar
ried
MIO
LC
,19
80–1
997
19+
in19
97(m
edia
nag
e27
)U
SA
Bib
larz
&G
otta
iner
(200
0)A
dult
men
talh
ealth
(Y)
Nat
ural
expe
rim
ent:
pare
ntal
deat
h
Div
orce
dve
rsus
wid
owed
GSS
18+
in19
72–1
996
USA
Che
rlin
etal
.(19
98)
Em
otio
nalp
robl
ems
(Yfo
rgr
owth
curv
es,N
for
fixed
effe
cts)
Gro
wth
curv
es,I
FED
ivor
ced
vers
usm
arri
edN
CD
SA
ge7
in19
72to
age
33in
1998
Gre
atB
rita
in
D’O
nofr
ioet
al.(
2007
)E
mot
iona
ldiffi
culti
es(N
),al
coho
lpro
blem
s(Y
)T
win
sD
ivor
ced
vers
usm
arri
ed(M
Zve
rsus
DZ
twin
s)V
irgi
nia
Reg
istr
y16
–79
(IQ
R30
–40
year
s)U
SA
Erm
isch
&Fr
ance
scon
i(2
001)
Psy
chol
ogic
aldi
stre
ss(N
),sm
okin
g(Y
)SF
ET
wo
bio
pare
nts
(incl
udin
gco
habi
ting)
vers
usno
tB
HP
S18
–29
in19
99G
reat
Bri
tain
Erm
isch
etal
.(20
04)
Psy
chol
ogic
aldi
stre
ss(Y
for
earl
ydi
srup
tions
)and
smok
ing
(Y)
SFE
Tw
obi
opa
rent
sve
rsus
not
BH
PS
18–2
9in
1999
Gre
atB
rita
in
Soci
al-e
mot
iona
lpro
blem
sA
pel&
Kau
kine
n(2
008)
Ant
isoc
ialb
ehav
ior
(Y)
LD
VT
wo
bio
mar
ried
,tw
obi
obl
ende
d,tw
obi
oco
habi
ting,
step
mar
ried
,st
epco
habi
ting,
sing
lepa
rent
,oth
er
NL
SY97
13–1
8in
1997
–199
8U
SA
Aug
hinb
augh
etal
.(2
005)
Beh
avio
rpr
oble
ms
inde
x(N
)G
row
thcu
rves
,IFE
Mar
ried
vers
usdi
vorc
ed,
sing
leve
rsus
gotm
arri
edN
LSY
79C
NL
SY79
4–15
in19
86–2
000
USA
Bac
hman
etal
.(20
09)
Ext
erna
lizin
gbe
havi
ors
(N),
inte
rnal
izin
gbe
havi
or(N
)
IFE
Sing
leve
rsus
gotm
arri
edve
rsus
star
ted
coha
bitin
gT
hree
-city
stud
y10
–16
in19
99–2
001
USA
Bou
twel
l&B
eave
r(2
010)
Self-
cont
rol(
N)
Pro
pens
itysc
ore
Tw
obi
opa
rent
sve
rsus
not
FFC
WS
Age
3in
2001
–200
3U
SA
Bro
wn
(200
6)D
elin
quen
cy(Y
for
stab
lesi
ngle
vers
usst
able
two
pare
nt)
LD
VT
wo
bio
pare
nts
(incl
udin
gco
habi
ting)
,sin
gle-
mot
her,
mar
ried
step
fam
ily,
coha
bitin
gst
epfa
mily
Add
Hea
lth12
–20
in19
94–1
996
USA (C
ontin
ued
)
www.annualreviews.org • The Causal Effects of Father Absence 415
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39CH20-McLanahan ARI 27 June 2013 14:34
Tab
le2
(Con
tinu
ed)
Ref
eren
ceO
utco
mes
aT
ype
ofan
alys
isb
Fam
ilyty
pes
com
pare
dD
ata
sour
cesc
Age
whe
nou
tcom
esob
serv
edC
ount
ryof
data
Cav
anag
h&
Hus
ton
(200
8)E
xter
naliz
ing
beha
vior
s(Y
),pe
erco
mpe
tenc
y(Y
),pe
erlo
nelin
ess
(Y)
LD
VN
ine
fam
ilyty
pes
(bio
/ste
p,co
habi
ting/
mar
ried
/sin
gle)
,nu
mbe
rof
fam
ilyst
ruct
ure
chan
ges
NIC
HD
SEC
CY
D6–
11in
1998
–200
3U
SA
Che
rlin
etal
.(19
91)
Ext
erna
lizin
gpr
oble
mbe
havi
ors
(Yfo
rgi
rls)
LD
V,I
FED
ivor
ced
vers
usm
arri
edN
CD
SN
SCA
ge11
in19
69G
reat
Bri
tain
,U
SAC
oope
ret
al.(
2011
)E
xter
naliz
ing
beha
vior
san
dat
tent
ion
prob
lem
s(Y
for
boys
)
LD
V,I
FEN
umbe
rof
tran
sitio
nsin
to/o
utof
rela
tions
hips
FFC
WS
3–5
in20
01–2
005
USA
D’O
nofr
ioet
al.(
2005
)B
ehav
iora
lpro
blem
s(Y
),in
tern
aliz
ing
beha
vior
s(Y
),dr
ugan
dal
coho
luse
(Y)
Tw
ins
Div
orce
dve
rsus
mar
ried
(MZ
vers
usD
Ztw
ins)
Aus
tral
ian
Nat
iona
lTw
inR
egis
try
14–3
9in
1992
–199
3A
ustr
alia
D’O
nofr
ioet
al.(
2006
)In
tern
aliz
ing
beha
vior
s(Y
),ag
eat
first
drug
use
(Y)
Tw
ins
Div
orce
dve
rsus
mar
ried
(MZ
vers
usD
Ztw
ins)
Aus
tral
ian
Nat
iona
lTw
inR
egis
try
14–3
9in
1992
–199
3A
ustr
alia
Eve
nhou
se&
Rei
lly(2
004)
Del
inqu
entb
ehav
ior
(N),
popu
lari
ty(N
),ps
ycho
logi
cald
istr
ess
(N),
drug
use
(N)
SFE
Tw
obi
opa
rent
sve
rsus
step
pare
ntA
ddH
ealth
12–1
8in
1994
USA
Fost
er&
Kal
il(2
007)
Ext
erna
lizin
gbe
havi
ors
(N),
inte
rnal
izin
gbe
havi
ors
(N),
adap
tive
soci
albe
havi
orin
vent
ory
(N)
IFE
Tw
obi
o,si
ngle
-mot
her,
step
,or
mul
tigen
erat
iona
lC
CD
P2–
5in
1990
–199
6U
SA
Hao
&M
atsu
eda
(200
6)E
xter
naliz
ing
beha
vior
(N)a
ndin
tern
aliz
ing
beha
vior
s(N
)
SFE
Tw
obi
opa
rent
sve
rsus
not
NL
SY79
CN
LSY
796–
14du
ring
1986
–199
6U
SA
Hao
&X
ie(2
002)
Mis
beha
vior
(Y)
IFE
Tw
obi
opa
rent
s,si
ngle
pare
nt,s
tep,
unst
able
coha
bitin
g
NSF
H6–
17in
1987
–199
4U
SA
Hof
fert
h(2
006)
Ext
erna
lizin
gbe
havi
or(N
)and
inte
rnal
izin
gbe
havi
ors
(Y)
SFE
Bio
vers
usst
epfa
ther
PSI
D6–
12in
1997
USA
416 McLanahan · Tach · Schneider
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39CH20-McLanahan ARI 27 June 2013 14:34
Ker
r&
Mic
hals
ki(2
007)
Hyp
erac
tivity
(Y)a
ndat
tent
ion
prob
lem
s(Y
)G
row
thcu
rves
Tw
obi
o,si
ngle
-mot
her,
step
,or
divo
rced
Can
adia
nN
LSY
0–11
in19
94th
roug
h6–
17in
2000
Can
ada
Mag
nuso
n&
Ber
ger
(200
9)B
ehav
ior
prob
lem
sin
dex
(Y)
Gro
wth
curv
esT
wo
bio,
sing
le-m
othe
r,or
step
CN
LSY
795–
12in
1986
–200
2U
SA
Mor
riso
n&
Che
rlin
(199
5)B
ehav
ior
prob
lem
sin
dex
(Yfo
rbo
ys)
LD
VT
wo
bio
pare
nts
vers
usdi
srup
ted
CN
LSY
79A
vera
geag
e5–
8in
1986
–198
8U
SA
Mor
riso
n&
Coi
ro(1
999)
Beh
avio
rpr
oble
ms
inde
x(Y
)L
DV
Tw
obi
opa
rent
sve
rsus
disr
upte
dN
LSY
794–
17in
1986
–199
4U
SA
Pag
anie
tal.
(199
8)D
elin
quen
cy(Y
for
rem
arri
ed)
Gro
wth
curv
esT
wo
bio
pare
nts
vers
usdi
vorc
edve
rsus
rem
arri
edM
ontr
eal
Lon
gitu
dina
lSt
udy
11–1
5in
1989
–199
3C
anad
a
Stro
hsch
ien
(200
5)A
ntis
ocia
lbeh
avio
r(N
),an
xiet
y/de
pres
sion
(Y)
Gro
wth
curv
esD
ivor
ced
vers
usm
arri
edC
NL
SY79
4–11
in19
94–1
998
Can
ada
Sun
(200
1)B
ehav
ior
prob
lem
s(Y
),se
lf-es
teem
(N),
locu
sof
cont
rol(
Y),
subs
tanc
eab
use
(mix
ed)
LD
VT
wo
bio
vers
usdi
vorc
edN
EL
S15
–18
in19
90–1
992
USA
Sun
&L
i(20
02)
Self-
este
em(Y
)and
locu
sof
cont
rol(
Y)
Gro
wth
curv
esT
wo
bio
vers
usdi
vorc
edN
EL
S13
–18
in19
88–1
992
USA
a Out
com
eab
brev
iatio
ns:N
,no
effe
ctfo
und;
Y,e
ffect
foun
d.bT
ype
ofan
alys
isab
brev
iatio
ns:I
FE,i
ndiv
idua
lfixe
def
fect
s;L
DV
,lag
ged
depe
nden
tvar
iabl
e;SF
E,s
iblin
gfix
edef
fect
s.c D
ata
sour
cesa
bbre
viat
ions
:Add
Hea
lth,N
atio
nalL
ongi
tudi
nalS
tudy
ofA
dole
scen
tHea
lth;B
HP
S,B
ritis
hH
ouse
hold
Pan
elSu
rvey
(199
1–19
99);
CC
DP,
Com
preh
ensi
veC
hild
Dev
elop
men
tP
roje
ct;C
NL
SY79
,Chi
ldre
nof
the
Nat
iona
lLon
gitu
dina
lSur
vey
ofY
outh
1979
;FFC
WS,
Frag
ileFa
mili
esan
dC
hild
Wel
lbei
ngSt
udy;
GSS
,Gen
eral
Soci
alSu
rvey
;MIO
LC
,Mar
ital
Inst
abili
tyO
ver
the
Life
Cou
rse;
NC
DS
NSC
,Nat
iona
lChi
ldD
evel
opm
entS
tudy
/Nat
iona
lSur
vey
ofC
hild
ren;
NE
LS,
Nat
iona
lEdu
catio
nalL
ongi
tudi
nalS
tudy
;NIC
HD
SEC
CY
D,
Nat
iona
lIns
titut
eof
Chi
ldH
ealth
and
Hum
anD
evel
opm
entS
tudy
ofE
arly
Chi
ldC
are
and
You
thD
evel
opm
ent;
NL
SY79
,Nat
iona
lLon
gitu
dina
lSur
vey
ofY
outh
1979
;NL
SY97
,Nat
iona
lL
ongi
tudi
nalS
urve
yof
You
th19
97;N
SFH
,Nat
iona
lSur
vey
ofFa
mili
esan
dH
ouse
hold
s;P
SID
,Pan
elSt
udy
ofIn
com
eD
ynam
ics.
www.annualreviews.org • The Causal Effects of Father Absence 417
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39CH20-McLanahan ARI 27 June 2013 14:34
measured using the Child Behavior Checklist(CBCL) (Achenbach & Edelbrock 1981),which includes behaviors such as aggression,attention, anxiety, and depression. Someresearchers use the full CBCL scale, whereasothers use subscales that distinguish betweenexternalizing behavior (aggression and atten-tion) and internalizing behavior (anxiety anddepression).
For adolescents, researchers often use adelinquency scale or a measure of antisocialbehavior, which overlaps with some of theitems on the externalizing scale. A few of thestudies we examined looked at other psycho-logical outcomes, such as locus of controland self-esteem, and several studies looked atsubstance use/abuse.
We identified 27 separate analyses thatexamined the association between parental di-vorce and some type of externalizing behavioror delinquency. These analyses were based ondata from four countries: the United States, theUnited Kingdom, Canada, and Australia. Ofthese, 19 analyses found a significant positiveeffect of divorce or father absence on problembehavior for at least one comparison group,whereas 8 found no significant association.The findings varied dramatically by method,with the LDV approach yielding the mostsignificant results and the two fixed effectsapproaches yielding the fewest significant find-ings. Two analyses found effects among boysbut not girls (Cooper et al. 2011, Morrison &Cherlin 1995), and one analysis found effectsamong girls but not boys (Cherlin et al. 1991).
Of the analyses reporting null findings, sev-eral had characteristics that might accountfor the lack of significant findings. One com-bined cohabiting parents with married parents(Boutwell & Beaver 2010), which likely weak-ened the effect of father absence on child out-comes, as prior research shows that disrup-tions of cohabiting unions are less harmful forchildren than are disruptions of marital unions(Brown 2006). A second controlled for familyincome, which is partly endogenous to divorce(Hao & Matsueda 2006). And a third used asmall, school-based sample (Pagani et al. 1998).
Six analyses examined internalizing behav-ior in children, including studies that measuredloneliness and difficulty making friends. Threeof these analyses reported significant effectsof father absence, whereas the other three re-ported no effects. As was true of the externaliz-ing analyses, the internalizing analyses relied onmultiple strategies. Also, as before, the analy-ses reporting null effects had characteristics thatmight account for their lack of strong findings.Two of the analyses that used IFE models werebased on low-income samples (Bachman et al.2009, Foster & Kalil 2007), and a third studycontrolled for income (Hao & Matsueda 2006).In addition, the Bachman analysis comparedsingle mothers who married with those who re-mained single. Finally, five analyses looked atlow self-esteem and low self-control, which aresometimes treated as markers of depression orpsychological distress. The findings from thesestudies were mixed.
Substance use. We identified six analyses thatexamined substance use, measured as cigarettesmoking and drug and alcohol use. The evi-dence for this set of outcomes was very ro-bust, with only one analysis reporting a null ef-fect (Evenhouse & Reilly 2004). Furthermore,the findings were consistent across multiplestrategies, including SFE models, which oftenshowed no effects for other outcomes.
Labor Force
We found only a few analyses that examinedthe effect of father absence on children’s laborforce outcomes in adulthood (see Table 3).In part, this is because earnings, employment,and welfare receipt in adulthood do not lendthemselves to analysis using IFEs, GCMs, orLDVs, which require observations before andafter the divorce. Indeed, all the analyses ofthis domain of outcomes used SFE models ornatural experiments.
However, in many other respects, thereis limited comparability between the stud-ies. Although several studies used data fromthe United States (Biblarz & Gottainer 2000,
418 McLanahan · Tach · Schneider
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39CH20-McLanahan ARI 27 June 2013 14:34
Tab
le3
Stud
ies
ofth
eef
fect
sof
fath
erab
senc
eon
the
labo
rfo
rce
Ref
eren
ceO
utco
mes
aT
ype
ofan
alys
isb
Fam
ilyty
pes
com
pare
dD
ata
sour
cesc
Age
whe
nou
tcom
esob
serv
edC
ount
ryof
data
Em
ploy
men
tB
ibla
rz&
Got
tain
er(2
000)
Occ
upat
iona
lsta
tus
(Yfo
rst
eppa
rent
fam
ilies
,Nfo
rsi
ngle
mot
her)
Nat
ural
expe
rim
ent:
pare
ntal
deat
h
Div
orce
dve
rsus
wid
owed
GSS
18+
in19
72–1
996
USA
Cor
ak(2
001)
Em
ploy
men
t(Y
for
men
,Nfo
rw
omen
),fa
mily
inco
me
(Yfo
rm
en,N
for
wom
en),
earn
ings
(N),
unem
ploy
men
tins
uran
ce(N
),in
com
eas
sist
ance
(Y)
Nat
ural
expe
rim
ent:
pare
ntal
deat
h
Par
enta
ldea
thve
rsus
divo
rce
vers
usm
arri
edT
axre
cord
s25
–32
in19
95C
anad
a
Erm
isch
&Fr
ance
scon
i(20
01)
Nei
ther
wor
king
nor
insc
hool
(Yfo
rdi
srup
tion
age
0–5,
Nfo
rdi
srup
tion
age
6+)
SFE
Tw
obi
opa
rent
s(in
clud
ing
coha
bitin
g)ve
rsus
not
BH
PS
18–2
9in
1999
Gre
atB
rita
in
Erm
isch
etal
.(20
04)
Nei
ther
wor
king
nor
insc
hool
(Yfo
rdi
srup
tion
<ag
e11
,Nfo
rdi
srup
tion
age
11+)
SFE
Tw
obi
opa
rent
sver
susn
otB
HP
S18
–29
in19
99G
reat
Bri
tain
Gru
ber
(200
4)E
mpl
oyed
(Yfo
rw
omen
,Nfo
rm
en),
earn
ings
(Yfo
rw
omen
,N
for
men
),pe
rca
pita
inco
me
(Y),
pove
rty
(N)
Nat
ural
expe
rim
ent:
divo
rce
law
Exp
osed
tore
form
eddi
vorc
ela
wve
rsus
not
Dec
enni
alC
ensu
s20
–50
in19
60,
1970
,198
0,or
1990
Cen
sus
USA
Inco
me/
earn
ings
Bjo
rklu
ndet
al.
(200
7)E
arni
ngs
(N)
SFE
Tw
obi
opa
rent
s(in
clud
ing
coha
bitin
g)ve
rsus
not
NL
SY79
PSI
DR
egis
try
NL
SYag
es24
–34
in19
94,P
SID
ages
23–3
3in
1993
,Sw
eden
26–3
6in
1996
USA
,Sw
eden
Lan
g&
Zag
orsk
y(2
001)
Wag
es(N
)N
atur
alex
peri
men
t:pa
rent
alde
ath
Par
enta
ldea
thve
rsus
divo
rce
vers
usm
arri
edN
LSY
7922
–36
in19
93U
SA
a Out
com
eab
brev
iatio
ns:N
,no
effe
ctfo
und;
Y,e
ffect
foun
d.bT
ype
ofan
alys
isab
brev
iatio
ns:S
FE,s
iblin
gfix
edef
fect
s.c D
ata
sour
ces
abbr
evia
tions
:BH
PS,
Bri
tish
Hou
seho
ldP
anel
Surv
ey(1
991–
1999
);G
SS,G
ener
alSo
cial
Surv
ey;N
LSY
79,N
atio
nalL
ongi
tudi
nalS
urve
yof
You
th19
79;P
SID
,Pan
elSt
udy
ofIn
com
eD
ynam
ics.
www.annualreviews.org • The Causal Effects of Father Absence 419
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39CH20-McLanahan ARI 27 June 2013 14:34
Bjorkland et al. 2007, Gruber 2004, Lang &Zagorsky 2001), many of these analyses werederived from estimates based on British orCanadian data. Further, the Gruber (2004) andCorak (2001) studies, which contributed 9 ofthe 14 analyses, differed in the ages and peri-ods examined, with Gruber using data from alonger time period (1960–1990), a wider rangeof ages (20–50), and so a much larger set ofcohorts (births 1910–1970) than Corak (2001):ages 25–32 and births 1963–1972. The remain-ing analyses, with the exception of Biblarz &Gottainer (2000), accorded with Corak (2001)insofar as they used data from the mid to late1990s and focused on respondents in their 20sand early 30s.
The findings for effects of father absencewere, however, consistent. Both Gruber (2004),using changes in US state laws to allow forunilateral divorce, and Corak (2001), usingparental death in Canada, found that divorcewas associated with lower levels of employ-ment. The studies disagreed, however, aboutfor whom these effects were most pronounced,with Gruber’s (2004) analyses suggesting thatfemale children of divorce were less likely towork and Corak (2001) finding that male chil-dren exposed to parental loss had lower la-bor force participation. Similarly, using SFEmodels with British data, Ermisch and coau-thors (Ermisch & Francesconi 2001, Ermischet al. 2004) found evidence of higher levels oflabor force inactivity among those who expe-rienced divorce in early childhood. Lookingat adult occupational status rather than sim-ply employment status, Biblarz & Gottainer(2000) found that although children growingup in divorced-mother households fared worsethan those growing up in stable two-parenthouseholds, there was no significant disadvan-tage to growing up in widow-mother house-holds. However, these researchers did find thatchildren growing up in stepparent householdswere disadvantaged regardless of whether fa-ther absence was due to divorce or widowhood.
The results of analyses of the effect of di-vorce on income and earnings were less con-sistent than the results for employment. Again,
Gruber (2004) and Corak (2001) contributedmost of the analyses for these outcomes, withGruber finding evidence of negative effects ofdivorce on income per capita and on women’searnings (but not poverty), and Corak find-ing negative effects of divorce on men’s fam-ily income (but minimal impacts on earnings).Corak’s result is consistent with analyses byLang & Zagorsky (2001) who, using parentaldeath as a natural experiment, found no effectof father absence on wages and by Bjorklandet al. (2007) who, using SFE models with USand Swedish data, found no effects on earnings.Corak (2001) also investigated how divorce wasrelated to the receipt of unemployment insur-ance and income assistance in Canada, findinga higher probability of receiving income assis-tance but not unemployment assistance.
Family Formation and Stability
Like the evidence base for labor force out-comes, there is relatively little research on howfamily structure affects patterns of offspring’sown family formation and relationship stability.The lack of research in this domain is some-what surprising, given that these outcomes areclosely related to the causal effect under consid-eration. The dearth of studies may be becausethese outcomes do not lend themselves to LDV,GCM, or IFE analyses.
Marriage and divorce. Virtually everythingwe know about the effects of father absenceon marriage and divorce comes from just threestudies (see Table 4), all of which used a naturalexperiment design, with the experimental vari-able being parental death (Corak 2001, Lang& Zagorsky 2001) or changes in divorce laws(Gruber 2004). All three studies examined mar-riage as an outcome but came to different con-clusions. Lang & Zagorsky found that parentaldeath and divorce reduced the likelihood thatsons will marry but found no effect on daugh-ters. Using parental death as a natural experi-ment, Corak found no evidence of a causal effectof father absence on marriage for either sonsor daughters. Finally, using divorce laws as a
420 McLanahan · Tach · Schneider
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39CH20-McLanahan ARI 27 June 2013 14:34
Tab
le4
Stud
ies
ofth
eef
fect
sof
fath
erab
senc
eon
fam
ilyfo
rmat
ion
and
stab
ility
Ref
eren
ceO
utco
mes
aT
ype
ofan
alys
isb
Fam
ilyty
pes
com
pare
dD
ata
sour
cesc
Age
whe
nou
tcom
esob
serv
edC
ount
ryof
data
Mar
riag
ean
ddi
vorc
eC
orak
(200
1)M
arri
age
(N),
divo
rce
(N),
and
sepa
ratio
n(Y
for
wom
en,N
for
men
)
Nat
ural
expe
rim
ent:
pare
ntal
deat
h
Par
enta
ldea
thve
rsus
divo
rce
vers
usm
arri
edT
axre
cord
s25
–32
in19
95C
anad
a
Gru
ber
(200
4)M
arri
age
(Y),
divo
rce
(N),
and
sepa
ratio
n(Y
)N
atur
alex
peri
men
t:di
vorc
ela
w
Exp
osed
tore
form
eddi
vorc
ela
wve
rsus
not
Dec
enni
alC
ensu
s20
–50
in19
60,
1970
,198
0,or
1990
Cen
sus
USA
Lan
g&
Zag
orsk
y(2
001)
Mar
riag
e(Y
for
men
,Nfo
rw
omen
)N
atur
alex
peri
men
t:pa
rent
alde
ath
Par
enta
ldea
thve
rsus
divo
rce
vers
usm
arri
edN
LSY
7922
–36
in19
93U
SA
Ear
lych
ildbe
arin
gE
rmis
ch&
Fran
cesc
oni(
2001
)E
arly
child
bear
ing
(Yfo
rdi
srup
tion
age
0–5,
Nfo
rdi
srup
tion
age
6+)
SFE
Tw
obi
opa
rent
s(in
clud
ing
coha
bitin
g)ve
rsus
not
BH
PS
18–2
9in
1999
Gre
atB
rita
in
Erm
isch
etal
.(20
04)
Ear
lych
ildbe
arin
g(Y
for
disr
uptio
n<
age
11,N
for
disr
uptio
nag
e11
+)
SFE
Tw
obi
opa
rent
sve
rsus
not
BH
PS
18–2
9in
1999
Gre
atB
rita
in
a Out
com
eab
brev
iatio
ns:N
,no
effe
ctfo
und;
Y,e
ffect
foun
d.bT
ype
ofan
alys
isab
brev
iatio
ns:S
FE,s
iblin
gfix
edef
fect
s.c D
ata
sour
ces
abbr
evia
tions
:BH
PS,
Bri
tish
Hou
seho
ldP
anel
Surv
ey(1
991–
1999
);N
LSY
79,N
atio
nalL
ongi
tudi
nalS
urve
yof
You
th19
79.
www.annualreviews.org • The Causal Effects of Father Absence 421
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39CH20-McLanahan ARI 27 June 2013 14:34
natural experiment, Gruber found that grow-ing up under the newer, relaxed divorce lawsactually increased the likelihood of marriagefor youth. The evidence for an effect of fatherabsence on marital stability was more con-sistent, with both Corak and Gruber findingevidence of a positive effect on separation butnot on divorce.
Early childbearing. We identified only twoanalyses that examined the effect of fatherabsence on early childbearing (Ermisch &Francesconi 2001, Ermisch et al. 2004). Theseanalyses were conducted by the same researchteam, they used the same SFE model, andthey used the same data—the British House-hold Panel Survey data in Great Britain. Bothanalyses found a positive association betweenparental absence and early childbearing, withdivorce in early childhood having a stronger ef-fect than divorce in middle childhood.
CONCLUSIONS
The body of knowledge about the causal ef-fects of father absence on child well-being hasgrown during the early twenty-first century asresearchers have increasingly adopted innova-tive methodological approaches to isolate causaleffects. We reviewed 47 such articles and findthat, on the whole, articles that take one ofthe more rigorous approaches to handling theproblems of omitted variable bias and reversecausality continue to document negative effectsof father absence on child well-being, thoughthese effects are stronger during certain stagesof the life course and for certain outcomes.
We find strong evidence that father absencenegatively affects children’s social-emotionaldevelopment, particularly by increasing exter-nalizing behavior. These effects may be morepronounced if father absence occurs duringearly childhood than during middle childhood,and they may be more pronounced for boysthan for girls. There is weaker evidence of aneffect of father absence on children’s cognitiveability.
Effects on social-emotional developmentpersist into adolescence, for which we findstrong evidence that father absence increasesadolescents’ risky behavior, such as smoking orearly childbearing. The evidence of an effecton adolescent cognitive ability continues to beweaker, but we do find strong and consistentnegative effects of father absence on high schoolgraduation. The latter finding suggests that theeffects on educational attainment operate by in-creasing problem behaviors rather than by im-pairing cognitive ability.
The research base examining the longer-term effects of father absence on adult outcomesis considerably smaller, but here too we see thestrongest evidence for a causal effect on adultmental health, suggesting that the psychologi-cal harms of father absence experienced duringchildhood persist throughout the life course.The evidence that father absence affects adulteconomic or family outcomes is much weaker.A handful of studies find negative effects on em-ployment in adulthood, but there is little con-sistent evidence of negative effects on marriageor divorce, on income or earnings, or on collegeeducation.
Despite the robust evidence that fa-ther absence affects social-emotional outcomesthroughout the life course, these studies alsoclearly show a role for selection in the relation-ship between family structure and child out-comes. In general, estimates from IFE, SFE,and PSM models are smaller than those fromconventional models that do not control forselection bias. Similarly, studies that compareparental death and divorce often find that evenif both have significant effects on well-being,the estimates of the effect of divorce are largerthan those of parental death, which can also beread as evidence of partial selection.
The Virtues and Limitations of theKey Analytic Strategies
Although we are more confident that causal ef-fects exist if results are robust across multiplemethodological approaches, such robustness iselusive, given the wide range of strategies for
422 McLanahan · Tach · Schneider
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39CH20-McLanahan ARI 27 June 2013 14:34
addressing bias. Each of these strategies has im-portant limitations and advantages. AlthoughGCMs, LDV designs, and PSM models allowfor broad external validity, these approaches doless to adjust for omitted variables than do IFEand SFE models. Yet such fixed effects modelsrequire one to assume that biological parentsin blended families are just like parents in non-blended families and that the age at which chil-dren experience father absence does not affecttheir response. In general, studies that employmore stringent methods to control for unob-served confounders also limit the generalizabil-ity of their results to specific subpopulations,complicating efforts to draw conclusions acrossmethods.
In many ways, the natural experiment strat-egy is appealing because it addresses concernsabout omitted variable bias and reverse causal-ity. In practice, however, these models are diffi-cult to implement. Approaches that use parentaldeath must make assumptions about the exo-geneity of parental death and the comparabil-ity of the experiences of father absence due todeath and divorce. Similarly, approaches thatuse instruments such as divorce law changesand incarceration rates must make a convincingcase that such policies and practices affect childoutcomes only through their effects on familystructure.
Some of these methodological approachesare better suited to examining one set of out-comes rather than others. For instance, GCM,LDV, and IFE designs do not lend themselvesto the investigation of the effects of father ab-sence on adult outcomes. In contrast, althoughnatural experiments and PSM models can beused to examine a wider range of outcomes,they are much less flexible in how father ab-sence can be measured, generally using dichoto-mous measures of absence rather than the moredetailed categorical measures of family type ormeasures that seek to capture the degree of in-stability experienced by children.
Because of these differences by method inthe domains that are examined and the defi-nitions of family structure that are used, it isdifficult to discern if some methods seem more
apt than others to find evidence for or againstthe effect of father absence on children. But ourimpression is that LDV and GCM designs tendto find stronger evidence of effects of fatherabsence on education and, particularly, social-emotional health than do the other designs.The evidence on the effects of father absence ismore mixed in studies using IFE and SFE. Therelatively smaller number of papers that usePSM designs also return a split verdict. Amongthose studies using natural experiments, thereis some evidence of negative effects of fatherabsence from changes in divorce laws, weak ev-idence when incarceration is used as an instru-ment, and mixed evidence from studies usingparental death.
Areas for Future Research
Looking across studies, we see that father ab-sence can affect child well-being across the lifecourse. But, within any one study, there is rarelyan attempt to understand how these differenttypes of outcomes are related to one another.For instance, studies separately estimate the ef-fect of father absence on externalizing behav-ior, high school completion, and employment,and from these analyses we can tell that familydisruption seems to have effects on each out-come. But it is also plausible that the effect offather absence on high school completion oper-ates through an effect on externalizing behavioror that the effect on employment is attributableto the effect on high school completion. Stateddifferently, the articles reviewed here do a goodjob of attempting to estimate the causal effectsof father absence on particular outcomes, butthey do not tell us very much about why or howthese effects come about. This omission reflectsa fundamental tension, extending beyond ourparticular substantive topic, between the goal ofestimating causal effects versus the goal of un-derstanding the mechanisms and processes thatunderlie long-term outcomes (Moffitt 2003).
Few of the studies reviewed here investigatewhether the effects of father absence vary bychild age, but those that do find importantdifferences, with effects concentrated among
www.annualreviews.org • The Causal Effects of Father Absence 423
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39CH20-McLanahan ARI 27 June 2013 14:34
children who experienced family disruption inearly childhood (Ermisch & Francesconi 2001,Ermisch et al. 2004). New developments inthe fields of neuroscience and epigenetics arerapidly expanding our understanding of howearly childhood experiences, including in uteroexperiences, have biological consequences,and sociologists would benefit from a betterunderstanding of these dynamics as theyrelate to a wide range of potential outcomes,especially health in adulthood (Barker 1992,Miller et al. 2011). Similarly, although therehas been some attention to how boys and girlsmay respond differently to father absence,researchers should continue to be attentive tothese interactions by gender.
We found surprisingly little work on inter-actions between father absence and race or class.Given that African American and low-incomechildren experience higher levels of fatherabsence than their white and middle-class coun-terparts, a differential response to absence couldserve to mitigate or further exacerbate inequal-ities in childhood and adult outcomes. Morework, particularly using the methods of causalinference discussed here, remains to be done onthis topic. We also suggest that more researchis needed to understand if the effects of father
absence on child well-being may have changedover time. We might expect that if stigma haslessened, as father absence has become morecommon, then the negative effects may havediminished. Alternatively, diminishing socialsafety net support and rising workplace inse-curity could have served to make the economicconsequences of father absence more severeand the negative effects more pronounced.
Finally, emerging research on familycomplexity shows that children raised apartfrom their biological fathers are raised in amultitude of family forms—single-motherfamilies, cohabiting-parent families, stepparentfamilies, blended families, multigenerationalfamilies—many of which are often very unsta-ble (McLanahan 2011, Tach et al. 2011, Tach2012). Indeed, stable single-mother householdsare quite rare, at least among children born tounmarried parents, which means that unstableand complex families may be the most commoncounterfactual to the married two-biological-parent family. Thus, studies of the causalimpact of father absence should not treat fatherabsence as a static condition but must distin-guish between the effect of a change in familystructure and the effect of a family structureitself.
DISCLOSURE STATEMENT
The authors are not aware of any affiliations, memberships, funding, or financial holdings thatmight be perceived as affecting the objectivity of this review.
ACKNOWLEDGMENTS
Support for S. McLanahan was provided by NICHD through grants R01HD36916,R01HD39135, and R01HD40421. L. Tach acknowledges support from the Robert Wood JohnsonFoundation Health & Society Scholars Program. D. Schneider received support from PrincetonUniversity and the Robert Wood Johnson Foundation Scholars in Health Policy Research Pro-gram. We are grateful to Anne Case, Andrew Cherlin, Stephen Morgan, and David Ribar for theircomments.
LITERATURE CITED
Achenbach TM, Edelbrock CS. 1981. Behavioral problems and competencies reported by parents of normaland disturbed children aged four through sixteen. Monogr. Soc. Res. Child Dev. 46(1):1–82
Amato PR. 2001. Children of divorce in the 1990s: an update of the Amato and Keith 1991 meta-analysis.J. Fam. Psychol. 15:355–70
424 McLanahan · Tach · Schneider
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39CH20-McLanahan ARI 27 June 2013 14:34
Amato PR. 2003. Reconciling divergent perspectives: Judith Wallerstein, quantitative family research, andchildren of divorce. Fam. Relat. 52:332–39
Andersson G. 2002. Children’s experience of family disruption and family formation: evidence from 16 FFScountries. Demogr. Res. 7:343–64
Apel R, Kaukinen C. 2008. On the relationship between family structure and antisocial behavior: parentalcohabitation and blended households. Criminology 46:35–70
Astone NM, McLanahan SS. 1991. Family structure, parental practices and high school completion. Am.Sociol. Rev. 56:309–20
Aughinbaugh A, Pierret CR, Rothstein DS. 2005. The impact of family structure transitions on youth achieve-ment: evidence from the children of the NLSY79. Demography 42(3):447–68
Bachman HJ, Coley RL, Chase-Lansdale PL. 2009. Is maternal marriage beneficial for low income adolescents?Appl. Dev. Sci. 13(4):153–71
Barker DJ. 1992. The fetal and infant origins of adult disease. BMJ 310:1111Besley T, Case A. 2000. Unnatural experiments? Estimating the incidence of endogenous policies. Econ. J.
110(467):F672–94Biblarz TJ, Gottainer G. 2000. Family structure and children’s success: a comparison of widowed and divorced
single-mother families. J. Marriage Fam. 62(2):533–48Bjorklund A, Ginther D, Sundstrom M. 2007. Family structure and child outcomes in the USA and Sweden.
J. Popul. Econ. 20:183–201Bjorkland A, Sundstrom M. 2006. Parental separation and children’s educational attainment: a siblings ap-
proach. Economica 73(292):605–24Boutwell B, Beaver K. 2010. The role of broken homes in the development of self-control: a propensity score
matching approach. J. Crim. Justice 38:489–95Brown SL. 2006. Family structure transitions and adolescent well-being. Demography 43:447–61Case A, Ardington C. 2006. The impact of parental death on school outcomes: longitudinal evidence from
South Africa. Demography 43:401–20Cavanagh SE, Huston AC. 2008. The timing of family instability and children’s social development. J. Marriage
Fam. 70:1258–70Cherlin AJ, Chase-Lansdale PL, McRae C. 1998. Effects of parental divorce on mental health throughout the
life course. Am. Sociol. Rev. 63:239–49Cherlin AJ, Furstenberg FF, Chase-Lansdale PL, Robins PK, Morrison DR, Tietler JO. 1991. Longitudinal
studies of effects of divorce on children in Great Britain and the United States. Science 252:1386–89Cooper C, Osborne C, Beck A, McLanahan S. 2011. Partnership instability, school readiness, and gender
disparities. Sociol. Educ. 84:246Corak M. 2001. Death and divorce: the long-term consequences of parental loss on adolescents. J. Labor Econ.
19(3):682–715Cunha F, Heckman JJ. 2008. Formulating, identifying and estimating the technology of cognitive and noncog-
nitive skill formation. J. Hum. Resour. 43(4):738–82D’Onofrio BM, Turkheimer E, Emery RE, Maes HH, Silberg J, Eaves LJ. 2007. A children of twins study of
parental divorce and offspring psychopathology. J. Child Psychol. Pyschiatr. 48:667–75D’Onofrio BM, Turkheimer E, Emery RE, Slutske WS, Heath AC, et al. 2005. A genetically informed study
of marital instability and its association with offspring psychopathology. J. Abnorm. Psychol. 114(4):570–86D’Onofrio BM, Turkheimer E, Emery RE, Slutske WS, Heath AC, et al. 2006. A genetically informed study
of the processes underlying the association between parental marital instability and offspring adjustment.Dev. Psychol. 42(3):486–99
Ermisch JF, Francesconi M. 2001. Family structure and children’s achievements. J. Popul. Econ. 14:249–70Ermisch J, Francesconi M, Pevalin DJ. 2004. Parental partnership and joblessness in childhood and their
influence on young people’s outcomes. J. R. Stat. Soc. Ser. A 167:69–101Evenhouse E, Reilly S. 2004. A sibling study of stepchild well-being. J. Hum. Resour. 39:248–76Finlay K, Neumark D. 2010. Is marriage always good for children? Evidence from families affected by incar-
ceration. J. Hum. Resour. 45(4):1046–88Foster EM, Kalil A. 2007. Living arrangements and children’s development in low-income White, Black, and
Latino families. Child Dev. 78:1657–74
www.annualreviews.org • The Causal Effects of Father Absence 425
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39CH20-McLanahan ARI 27 June 2013 14:34
Francesconi M, Jenkins SP, Siedler T. 2010. Childhood family structure and schooling outcomes: evidencefor Germany. J. Popul. Econ. 23(3):1432–75
Frisco ML, Muller C, Frank K. 2007. Parents’ union dissolution and adolescents’ school performance: com-paring methodological approaches. J. Marriage Fam. 69:721–41
Gennetian L. 2005. One or two parents? Half or step siblings? The effects of family composition on youngchildren’s achievement. J. Popul. Econ. 1818(3):415–36
Gertler P, Levine DI, Ames M. 2004. Schooling and parental death. Rev. Econ. Stat. 86(1):211–25Ginther DK, Pollak RA. 2004. Family structure and children’s educational outcomes: blended families, stylized
facts, and descriptive regressions. Demography 41:671–96Gruber J. 2004. Is making divorce easier bad for children? The long-run implications of unilateral divorce.
J. Labor Econ. 22(4):799–833Halaby C. 2004. Panel models in sociological research: theory into practice. Annu. Rev. Sociol. 30:507–44Hao L, Matsueda RL. 2006. Family dynamics through childhood: a sibling model of behavior problems. Soc.
Sci. Res. 35:500–24Hao L, Xie G. 2002. The complexity and endogeneity of family structure in explaining children’s misbehavior.
Soc. Sci. Res. 31:1–28Hofferth SL. 2006. Residential father family type and child well-being: investment versus selection. Demography
43:53–77Johnson D. 2005. Two wave panel analysis: comparing statistical methods for studying the effects of transitions.
J. Marriage Fam. 67:1061–75Kerr DW, Michalski JH. 2007. Family structure and children’s hyperactivity problems: a longitudinal analysis.
Can. J. Sociol. 32:85–112Lang K, Zagorsky JL. 2001. Does growing up with a parent absent really hurt? J. Hum. Resour. 36:253–73Magnuson K, Berger LM. 2009. Family structure states and transitions: associations with children’s well-being
during middle childhood. J. Marriage Fam. 71:575–91Manski CF, Sandefur GD, McLanahan S, Powers D. 1992. Alternative estimates of the effect of family structure
during adolescence on high school graduation. J. Am. Stat. Assoc. 87:25–37McLanahan S. 2011. Family instability and complexity after a nonmarital birth: outcomes for children in
Fragile Families. In Social Class and Changing Families in an Unequal America, ed. M Carlson, P England,pp. 108–133. Stanford, CA: Stanford Univ. Press
Miller G, Chen E, Parker K. 2011. Psychological stress in childhood and susceptibility to the chronic diseasesof aging: moving toward a model of behavioral and biological mechanisms. Psychol. Bull. 137:959–97
Moffitt R. 2003. Causal analysis in population research: an economist’s perspective. Popul. Dev. Rev. 29(3):448–58
Morgan SL, Todd JJ. 2008. A diagnostic routine for the detection of consequential heterogeneity of causaleffects. Sociol. Methodol. 38:231–81
Morgan SL, Winship C. 2007. Counterfactuals and Causal Inference: Methods and Principles for Social Research.Cambridge, UK: Cambridge Univ. Press
Morrison DR, Cherlin AJS. 1995. The divorce process and young children’s well-being: a prospective analysis.J. Marriage Fam. 57:800–12
Morrison DR, Coiro MJ. 1999. Parental conflict and marital disruption: Do children benefit when high-conflict marriages are dissolved? J. Marriage Fam. 61:626–37
Pagani L, Tremblay RE, Vitaro F, Kerr M, McDuff P. 1998. The impact of family transition on the develop-ment of delinquency in adolescent boys: a 9-year longitudinal study. J. Child Psychol. Pyschiatr. 39:489–99
Ribar DC. 2004. What do social scientists know about the benefits of marriage? A review of quantitative methodologies.IZA Disc. Pap. #998, Inst. Study Labor (IZA), Bonn, Ger. http://ftp.iza.org/dp998.pdf
Sandefur G, Wells T. 1999. Does family structure really influence educational attainment? Soc. Sci. Res.28(4):331–57
Sanz-de-Galdeano A, Vuri D. 2007. Parental divorce and students’ performance: evidence from longitudinaldata. Oxf. Bull. Econ. Stat. 69:321–38
Schweinhart LJ, Montie J, Xiang Z, Barnett WS, Belfield CR, Nores M. 2005. Lifetime Effects: The HighScopePerry Preschool Study Through Age 40. Ypsilanti, MI: HighScope
426 McLanahan · Tach · Schneider
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39CH20-McLanahan ARI 27 June 2013 14:34
Shaff KA, Wolfinger N, Kowaleski-Jones L, Smith K. 2008. Family structure transitions and child achieve-ment. Sociol. Spectrum 28:681–704
Sigle-Rushton W, McLanahan S. 2004. Father absence and child wellbeing: a critical review. In The Future ofthe Family, ed. DP Moynihan, TM Smeeding, L Rainwater, pp. 116–55. New York: Russell Sage Found.
Singer J, Willett J. 2003. Applied Longitudinal Data Analysis: Modeling Change and Event Occurrence. Oxford,UK: Oxford Univ. Press
Steele F, Sigle-Rushton W, Kravdal Ø. 2009. Consequences of family disruption on children’s educationaloutcomes in Norway. Demography 46:553–74
Strohschien L. 2005. Parental divorce and child mental health trajectories. J. Marriage Fam. 67(5):1286–1300Sun Y. 2001. Family environment and adolescents’ well-being before and after parents’ marital disruption: a
longitudinal analysis. J. Marriage Fam. 63:697–713Sun Y, Li Y. 2002. Children’s well-being during parents’ marital disruption process: a pooled time-series
analysis. J. Marriage Fam. 64(2):472–88Tach L. 2012. Family complexity, childbearing, and parenting stress: a comparison of mothers’ and fathers’ experiences.
Fragile Fam. Work. Pap. WP12-15-FF, Fragile Families and Child Wellbeing Study, Princeton, NJ/NewYork. http://crcw.princeton.edu/workingpapers/WP12-15-FF.pdf
Tach L, Edin K, McLanahan S. 2011. Multiple partners and multiple partner fertility in Fragile Families. FragileFam. Work. Pap. WP11-10-FF, Fragile Families and Child Wellbeing Study, Princeton, NJ/New York.http://crcw.princeton.edu/workingpapers/WP11-10-FF.pdf
Winship C, Morgan S. 1999. The estimation of causal effects from observational data. Annu. Rev. Sociol.25:659–707
www.annualreviews.org • The Causal Effects of Father Absence 427
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39-FrontMatter ARI 5 June 2013 17:48
Annual Reviewof Sociology
Volume 39, 2013Contents
FrontispieceCharles Tilly � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � xiv
Prefatory Chapter
Formations and Formalisms: Charles Tilly and the Paradoxof the ActorJohn Krinsky and Ann Mische � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 1
Theory and Methods
The Principles of Experimental Design and Their Applicationin SociologyMichelle Jackson and D.R. Cox � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �27
The New Sociology of MoralitySteven Hitlin and Stephen Vaisey � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �51
Social Processes
Social Scientific Inquiry Into Genocide and Mass Killing:From Unitary Outcome to Complex ProcessesPeter B. Owens, Yang Su, and David A. Snow � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �69
Interest-Oriented ActionLyn Spillman and Michael Strand � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � �85
Drugs, Violence, and the StateBryan R. Roberts and Yu Chen � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 105
Healthcare Systems in Comparative Perspective: Classification,Convergence, Institutions, Inequalities, and Five Missed TurnsJason Beckfield, Sigrun Olafsdottir, and Benjamin Sosnaud � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 127
Institutions and Culture
Multiculturalism and Immigration: A Contested Fieldin Cross-National ComparisonRuud Koopmans � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 147
Sociology of Fashion: Order and ChangePatrik Aspers and Frederic Godart � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 171
v
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39-FrontMatter ARI 5 June 2013 17:48
Religion, Nationalism, and Violence: An Integrated ApproachPhilip S. Gorski and Gulay Turkmen-Dervisoglu � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 193
Formal Organizations
Race, Religious Organizations, and IntegrationKorie L. Edwards, Brad Christerson, and Michael O. Emerson � � � � � � � � � � � � � � � � � � � � � � � � � 211
Political and Economic Sociology
An Environmental Sociology for the Twenty-First CenturyDavid N. Pellow and Hollie Nyseth Brehm � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 229
Economic Institutions and the State: Insights from Economic HistoryHenning Hillmann � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 251
Differentiation and Stratification
Demographic Change and Parent-Child Relationships in AdulthoodJudith A. Seltzer and Suzanne M. Bianchi � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 275
Individual and Society
Gender and CrimeCandace Kruttschnitt � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 291
White-Collar Crime: A Review of Recent Developments andPromising Directions for Future ResearchSally S. Simpson � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 309
From Social Structure to Gene Regulation, and Back: A CriticalIntroduction to Environmental Epigenetics for SociologyHannah Landecker and Aaron Panofsky � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 333
Racial Formation in Perspective: Connecting Individuals, Institutions,and Power RelationsAliya Saperstein, Andrew M. Penner, and Ryan Light � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 359
The Critical Sociology of Race and Sport: The First Fifty YearsBen Carrington � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 379
Demography
The Causal Effects of Father AbsenceSara McLanahan, Laura Tach, and Daniel Schneider � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 399
International Migration and Familial Change in Communitiesof Origin: Transformation and ResistancePatricia Arias � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 429
Trends and Variation in Assortative Mating: Causes and ConsequencesChristine R. Schwartz � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 451
vi Contents
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
SO39-FrontMatter ARI 5 June 2013 17:48
Gender and International Migration: Contributions andCross-FertilizationsGioconda Herrera � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 471
LGBT Sexuality and Families at the Start of the Twenty-First CenturyMignon R. Moore and Michael Stambolis-Ruhstorfer � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 491
Urban and Rural Community Sociology
Housing: Commodity versus RightMary Pattillo � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 509
Indexes
Cumulative Index of Contributing Authors, Volumes 30–39 � � � � � � � � � � � � � � � � � � � � � � � � � � � 533
Cumulative Index of Article Titles, Volumes 30–39 � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � � 537
Errata
An online log of corrections to Annual Review of Sociology articles may be found athttp://soc.annualreviews.org/errata.shtml
Contents vii
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
AnnuAl Reviewsit’s about time. Your time. it’s time well spent.
AnnuAl Reviews | Connect with Our expertsTel: 800.523.8635 (us/can) | Tel: 650.493.4400 | Fax: 650.424.0910 | Email: service@annualreviews.org
New From Annual Reviews:Annual Review of Organizational Psychology and Organizational BehaviorVolume 1 • March 2014 • Online & In Print • http://orgpsych.annualreviews.org
Editor: Frederick P. Morgeson, The Eli Broad College of Business, Michigan State UniversityThe Annual Review of Organizational Psychology and Organizational Behavior is devoted to publishing reviews of the industrial and organizational psychology, human resource management, and organizational behavior literature. Topics for review include motivation, selection, teams, training and development, leadership, job performance, strategic HR, cross-cultural issues, work attitudes, entrepreneurship, affect and emotion, organizational change and development, gender and diversity, statistics and research methodologies, and other emerging topics.
Complimentary online access to the first volume will be available until March 2015.TAble oF CoNTeNTs:•An Ounce of Prevention Is Worth a Pound of Cure: Improving
Research Quality Before Data Collection, Herman Aguinis, Robert J. Vandenberg
•Burnout and Work Engagement: The JD-R Approach, Arnold B. Bakker, Evangelia Demerouti, Ana Isabel Sanz-Vergel
•Compassion at Work, Jane E. Dutton, Kristina M. Workman, Ashley E. Hardin
•ConstructivelyManagingConflictinOrganizations, Dean Tjosvold, Alfred S.H. Wong, Nancy Yi Feng Chen
•Coworkers Behaving Badly: The Impact of Coworker Deviant Behavior upon Individual Employees, Sandra L. Robinson, Wei Wang, Christian Kiewitz
•Delineating and Reviewing the Role of Newcomer Capital in Organizational Socialization, Talya N. Bauer, Berrin Erdogan
•Emotional Intelligence in Organizations, Stéphane Côté•Employee Voice and Silence, Elizabeth W. Morrison• Intercultural Competence, Kwok Leung, Soon Ang,
Mei Ling Tan•Learning in the Twenty-First-Century Workplace,
Raymond A. Noe, Alena D.M. Clarke, Howard J. Klein•Pay Dispersion, Jason D. Shaw•Personality and Cognitive Ability as Predictors of Effective
Performance at Work, Neal Schmitt
•Perspectives on Power in Organizations, Cameron Anderson, Sebastien Brion
•Psychological Safety: The History, Renaissance, and Future of an Interpersonal Construct, Amy C. Edmondson, Zhike Lei
•Research on Workplace Creativity: A Review and Redirection, Jing Zhou, Inga J. Hoever
•Talent Management: Conceptual Approaches and Practical Challenges, Peter Cappelli, JR Keller
•The Contemporary Career: A Work–Home Perspective, Jeffrey H. Greenhaus, Ellen Ernst Kossek
•The Fascinating Psychological Microfoundations of Strategy and Competitive Advantage, Robert E. Ployhart, Donald Hale, Jr.
•The Psychology of Entrepreneurship, Michael Frese, Michael M. Gielnik
•The Story of Why We Stay: A Review of Job Embeddedness, Thomas William Lee, Tyler C. Burch, Terence R. Mitchell
•What Was, What Is, and What May Be in OP/OB, Lyman W. Porter, Benjamin Schneider
•Where Global and Virtual Meet: The Value of Examining the Intersection of These Elements in Twenty-First-Century Teams, Cristina B. Gibson, Laura Huang, Bradley L. Kirkman, Debra L. Shapiro
•Work–Family Boundary Dynamics, Tammy D. Allen, Eunae Cho, Laurenz L. Meier
Access this and all other Annual Reviews journals via your institution at www.annualreviews.org.
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
AnnuAl Reviewsit’s about time. Your time. it’s time well spent.
AnnuAl Reviews | Connect with Our expertsTel: 800.523.8635 (us/can) | Tel: 650.493.4400 | Fax: 650.424.0910 | Email: service@annualreviews.org
New From Annual Reviews:
Annual Review of Statistics and Its ApplicationVolume 1 • Online January 2014 • http://statistics.annualreviews.org
Editor: Stephen E. Fienberg, Carnegie Mellon UniversityAssociate Editors: Nancy Reid, University of Toronto
Stephen M. Stigler, University of ChicagoThe Annual Review of Statistics and Its Application aims to inform statisticians and quantitative methodologists, as well as all scientists and users of statistics about major methodological advances and the computational tools that allow for their implementation. It will include developments in the field of statistics, including theoretical statistical underpinnings of new methodology, as well as developments in specific application domains such as biostatistics and bioinformatics, economics, machine learning, psychology, sociology, and aspects of the physical sciences.
Complimentary online access to the first volume will be available until January 2015. table of contents:•What Is Statistics? Stephen E. Fienberg•A Systematic Statistical Approach to Evaluating Evidence
from Observational Studies, David Madigan, Paul E. Stang, Jesse A. Berlin, Martijn Schuemie, J. Marc Overhage, Marc A. Suchard, Bill Dumouchel, Abraham G. Hartzema, Patrick B. Ryan
•The Role of Statistics in the Discovery of a Higgs Boson, David A. van Dyk
•Brain Imaging Analysis, F. DuBois Bowman•Statistics and Climate, Peter Guttorp•Climate Simulators and Climate Projections,
Jonathan Rougier, Michael Goldstein•Probabilistic Forecasting, Tilmann Gneiting,
Matthias Katzfuss•Bayesian Computational Tools, Christian P. Robert•Bayesian Computation Via Markov Chain Monte Carlo,
Radu V. Craiu, Jeffrey S. Rosenthal•Build, Compute, Critique, Repeat: Data Analysis with Latent
Variable Models, David M. Blei•Structured Regularizers for High-Dimensional Problems:
Statistical and Computational Issues, Martin J. Wainwright
•High-Dimensional Statistics with a View Toward Applications in Biology, Peter Bühlmann, Markus Kalisch, Lukas Meier
•Next-Generation Statistical Genetics: Modeling, Penalization, and Optimization in High-Dimensional Data, Kenneth Lange, Jeanette C. Papp, Janet S. Sinsheimer, Eric M. Sobel
•Breaking Bad: Two Decades of Life-Course Data Analysis in Criminology, Developmental Psychology, and Beyond, Elena A. Erosheva, Ross L. Matsueda, Donatello Telesca
•Event History Analysis, Niels Keiding•StatisticalEvaluationofForensicDNAProfileEvidence,
Christopher D. Steele, David J. Balding•Using League Table Rankings in Public Policy Formation:
Statistical Issues, Harvey Goldstein•Statistical Ecology, Ruth King•Estimating the Number of Species in Microbial Diversity
Studies, John Bunge, Amy Willis, Fiona Walsh•Dynamic Treatment Regimes, Bibhas Chakraborty,
Susan A. Murphy•Statistics and Related Topics in Single-Molecule Biophysics,
Hong Qian, S.C. Kou•Statistics and Quantitative Risk Management for Banking
and Insurance, Paul Embrechts, Marius Hofert
Access this and all other Annual Reviews journals via your institution at www.annualreviews.org.
Ann
u. R
ev. S
ocio
l. 20
13.3
9:39
9-42
7. D
ownl
oade
d fr
om w
ww
.ann
ualr
evie
ws.
org
by U
nive
rsity
of
Bri
tish
Col
umbi
a on
03/
29/1
4. F
or p
erso
nal u
se o
nly.
top related