Annu. Rev. Psychol. 2000.

51:1–27
Copyright q 2000 by Annual Reviews. All rights reserved

PARENTING AND ITS EFFECTS ON CHILDREN:

On Reading and Misreading Behavior Genetics

Eleanor E. Maccoby
Department of Psychology, Stanford University, Building 420, Jordan Hall, Stanford,
California 94305–2130; e-mail: maccoby@psych.stanford.edu
Abstract There is clear evidence that parents can and do influence children. There
is equally clear evidence that children’s genetic makeup affects their own behavioral
Downloaded from www.annualreviews.org. Guest (guest) IP: 180.75.250.144 On: Fri, 08 Nov 2024 04:09:10

characteristics, and also influences the way they are treated by their parents. Twin and
adoption studies provide a sound basis for estimating the strength of genetic effects,
although heritability estimates for a given trait vary widely across samples, and no one
estimate can be considered definitive. This chapter argues that knowing only the strength
of genetic factors, however, is not a sufficient basis for estimating environmental ones
and indeed, that attempts to do so can systematically underestimate parenting effects.
Children’s genetic predispositions and their parents’ childrearing regimes are seen to be
closely interwoven, and the ways in which they function jointly to affect children’s devel-
opment are explored.

CONTENTS
Introduction ......................................................................................... 1
How Strong is the Connection Between Parent and Child Behaviors................. 5
The Challenge from Behavior Genetics ...................................................... 9
The Focus on Variation .......................................................................... 9
The Claim for Substantial Genetic Effects .....................................................11
Estimating the Size of Environmental Effects ..................................................12
Shared and Unshared Environmental Effects ..................................................13
Interpreting Parent-Child Covariance .........................................................17
The Interaction of Genetic and Environmental Factors...................................19
G 2 E Interactions in Animal Studies .........................................................19
G 2 E Interactions in Adoption Studies.......................................................20
Studies of Interactions with Temperament......................................................21
Overview ............................................................................................22

INTRODUCTION
What are the forces that affect when and how children will change as they grow
older? Can development be seen as a progressive process whereby children move
toward a specifiable outcome or end state that we can call maturity? What conditions

0084–6570/00/0201–0001$12.00 1
 2 MACCOBY

determine differences among children in their rates of development or their ultimate

outcomes? These questions have been at the heart of much of the work in develop-
mental psychology since the inception of the field. In pursuing the answers, the broad
forces of nature and nurture, and the interplay between them, have been of central
concern. It has long been clear that there are powerful maturational time-tables gov-
erning developmental change: e.g. the progression in infancy from sitting to crawling
to standing to walking, or in the acquisition of language, the transition from rudi-
mentary one-word utterances through intermediate phrases to the production of full,
well-formed sentences. However, it has been equally obvious that children are learn-
ing many things through their daily experiences in interacting with the physical and
social world, and that what is learned is not encoded in the genes. Some of the
experiences children have are random—not planned or organized by any outside
Downloaded from www.annualreviews.org. Guest (guest) IP: 180.75.250.144 On: Fri, 08 Nov 2024 04:09:10

agency—but some occur according to what might be called a socialization time table.
It is here that parenting has its place.
All societies prescribe certain characteristics that their members are expected to
possess and certain things people must not do, if they are to function adequately as
members of their society. Some of these prescriptions and proscriptions are nearly
universal across cultures, such as the requirement for parents, or specified parent
surrogates, to provide nurturance and protection for children. Other standards and
values vary greatly from one cultural setting to another. In all societies, training of
children occurs, and social controls are in place to ensure that children are
socialized—that is, brought up in such a way that each new generation acquires the
prescribed patterns of beliefs and behaviors. Of course, cultures do change, either
slowly or rapidly, so that the cross-generational transmission is by no means absolute.
A new generation may need to adapt to conditions that the parent generation did not
face. And transmission of values, even when they continue to be appropriate for
succeeding generations, is not always successful. Some children in every cohort may
be seen to be inadequately socialized by the criteria that the society applies.
Not all socialization occurs in childhood. People are socialized into the customs
and standards of an occupational culture when they take up an entry-level job. Social-
ization and resocialization occur when adults enter into new life roles (e.g. marriage,
parenthood). In considering the role of parents, however, we are mainly concerned
with childhood socialization. Some of the socialization that occurs throughout child-
hood is in a sense anticipatory, in that it functions to prepare children for adaptation
to a fairly wide range of life roles and the various contexts children will encounter
as they grow older. But childhood socialization also concerns the training of children
in modes of behavior that are acceptable for the stage of childhood they currently
occupy. Societies set different standards for people at different stages of their life
cycle, and there are requirements that loom especially large in childhood. These
include requirements for children to comply with adult demands, to avoid irritating
adults or disrupting their activities, to accept age-appropriate responsibility, and to
function as a pleasant, cooperative family member.
In modern societies, there are least three major contexts in which childhood social-
ization takes place: families, peer groups, and out-of-home contexts such as school
 PARENTING AND ITS EFFECTS ON CHILDREN 3

classrooms or day-care centers in which the daily experiences of children are struc-
tured and overseen by adults. The enormous body of literature on childhood social-
ization has strongly emphasized the role of parents. This emphasis has a long and
deep tradition. The idea that ‘‘as the twig is bent, so grows the tree’’ can be traced
at least as far back as Greek and Biblical times—(probably earlier), and in most
societies parents are the ones assigned primary responsibility for ‘‘bending’’ the chil-
dren in desirable directions, by supervising, teaching, and disciplining them as they
grow up. Early childhood in particular has long been thought to be a period in the
life cycle when humans are especially plastic—a time when children are especially
open to social influences on characteristics they will carry with them long after they
have left their family of origin. Things thought to be especially vulnerable to influence
in the first 5–7 years of children’s lives include the language they speak, their food
Downloaded from www.annualreviews.org. Guest (guest) IP: 180.75.250.144 On: Fri, 08 Nov 2024 04:09:10

preferences, their religious beliefs, and certain enduring personality traits.

In the twentieth century, assumptions about the importance of within-family child-
hood socialization have been part of the fabric of mainstream psychological theories.
From roughly the 1920s through the 1960s, behaviorist learning theories held sway,
emphasizing the ‘‘blank slate’’ status of infants and the power of adults to teach
young children, for good or ill, what they must learn. Parents, of course, were seen
as the most available teachers, and the ones responsible for carrying out the training
of their children. The physiological drive states (hunger, fatigue) with which children
are innately endowed were not ignored in the learning theories of the time, so there
was some blending of nature and nurture, but the major emphasis was on the control
of learning processes exercised by environmental inputs. Psychoanalytic theories of
this period emphasized the importance of early in-family experience in determining
subsequent inner conflicts, defense mechanisms, and internalization of values. In
more recent decades, as the cognitive revolution took hold and learning theory (as it
related to socialization) was reformulated as cognitive social learning theory, the
active role of children as participants in their own socialization was increasingly
stressed. Currently, there is increasing emphasis on the role of parents’ and children’s
mutual perceptions and understandings about each other’s dispositions and intentions
as determiners of their influence upon one another. But none of these theoretical
shifts has greatly affected the underlying assumption that parents have a powerful
impact on the characteristics children develop and the directions their lives take. The
child development research literature has continued to include a wide range of studies
on such things as (a) familial risk factors (i.e. aspects of family functioning that are
related to the subsequent development of externalizing or internalizing disorders in
children); (b) social conditions that affect such parenting practices as how well parents
are able to monitor their children, or how warm and responsive they are; and (c)
parenting behaviors as mediators of the connection between societal risk factors (e.g.
poverty or dangerous neighborhoods) and children’s adjustment.
In recent decades, there has been a countervailing ground swell of research and
theorizing about nature—the genetic endowment of parents and children—as exerting
a powerful influence on the characteristics that children develop. Of course, for many
decades, elementary psychology textbooks have carried tables comparing identical
 4 MACCOBY

and fraternal twins with respect to their degree of similarity on IQ or other traits.
Studies of adopted children were also widely reported many years ago, and inferences
were routinely drawn from both twin and adoption studies concerning the importance
of genetic factors in development. Still, for many years, thinking remained largely
compartmentalized, and readers continued to believe in both the importance of
genetic factors and the importance of socialization factors as though they were in no
way incompatible. In recent years, however, there has been more sophisticated work
in behavior genetics, and there are insistent voices claiming that the findings from
this work are indeed incompatible with many widely-held views about the power of
within-family socialization.
These messages from behavior genetics have been picked up and synthesized with
other misgivings about the weaknesses of socialization research into a more broad-
Downloaded from www.annualreviews.org. Guest (guest) IP: 180.75.250.144 On: Fri, 08 Nov 2024 04:09:10

based attack on traditional assumptions concerning parenting and its effects. Rowe’s
book, The Limits of Family Influence (1994), stated the case strongly, and Harris’s
more popular book The Nurture Assumption (1998) attracted a flurry of media atten-
tion to the issues. These authors have drawn together the findings from some well-
known studies of parenting effects and findings from behavior genetics to make the
following claims:
1. The connections that studies have found between the way parents deal with their
children and how the children turn out are actually quite weak and have proved
difficult to replicate. When parent ‘‘effects’’ are found, they tend to be effects
on the way children behave at home and the relationships they develop with
their parents. There is little carry-over from at-home experiences to the way
children function in out-of-home contexts
2. When studies do establish connections between parenting and children’s attri-
butes, these are correlational findings. An example is Baumrind’s early finding—
now widely replicated—that the children of parents who are both responsive
and firm tend to be more competent and cooperative than children of parents
who are either authoritarian or permissive (Baumrind & Black 1967). Such
findings have traditionally been interpreted as showing that authoritative par-
enting has beneficial effects on children, ignoring the possibility that the causal
connection may run the other way—i.e. that competent, cooperative children
may make it easier for their parents to be firm and responsive. In fact, the critics
argue, parent behavior is substantially driven by the behavior of children, and
much if not most of the parent/child correlation can be accounted for by the
child’s genetic predispositions.
3. Parental influence has been emphasized at the expense of sources of influence
that in fact have great—or perhaps greater—importance in shaping children’s
development. Two kinds of influence which critics argue have been underem-
phasized are genetic predispositions and the influence of peers.
In the popular media, these critiques have been condensed into the oversimplified
message ‘‘Parents don’t matter’’ or ‘‘matter very little’’—news bites that, on their
face, have little relation to reality as it is experienced daily in family life. Often,
 PARENTING AND ITS EFFECTS ON CHILDREN 5

reports in the popular media do not reflect what the cited authors actually said. For
example, late in her book, Harris (1998) says she believes parents can foster the
development of specific talents (e.g. by providing music lessons) and can influence
such things as children’s leisure time activities, their food preferences, their religious
beliefs and practices, and the acquisition of knowledge and skills and preferences
that will contribute to their ultimate choice of a profession. Yet, the burden of her
book is to down-play such influences and stress the respects in which parents are not
influential. Rowe says: ‘‘ . . . parents in most working to professional-class families
may have little influence on what traits their children may eventually develop as
adults.’’ (1994:7). His use of the word ‘‘may’’ does not greatly soften the import of
his message. He goes on to say that he doubts whether any undesirable trait displayed
Downloaded from www.annualreviews.org. Guest (guest) IP: 180.75.250.144 On: Fri, 08 Nov 2024 04:09:10

by a child can be significantly modified by anything a parent does. Scarr (1992)

expresses a similarly skeptical view about the possible effects of interventions. Such
views, of course, when picked up and simplified in the popular press, can have serious
implications for public policies concerning whether to invest in remedial or supportive
programs for children and families.
These critiques constitute serious efforts to present a point of view that is clearly
different from the traditional emphasis on the importance of parenting. They cite
large bodies of data and have attracted the support of highly reputable psychologists.
They deserve to be taken seriously. Nonetheless, I believe that they are out of date
with respect to both the genetic studies and the parenting-effects studies they cite,
and that they seriously misinterpret the pertinent body of research.
I turn first to the question of how strong the connections are between what parents
do with their children and how the children turn out. I then turn to issues of genetic
factors, in particular the ways in which these factors may determine or limit how we
can interpret parenting effects.

HOW STRONG IS THE CONNECTION BETWEEN

PARENT AND CHILD BEHAVIORS?

As noted above, critics charge that interpreters of traditional socialization studies

have exaggerated the importance of parenting in children’s lives—that in fact, the
effect sizes reported in many widely-cited studies are really quite small. Indeed,
reviews of research done before the mid-1980s did show weak correlations between
parenting processes and children’s characteristics (e.g. Maccoby & Martin 1983).
Since then, many studies have come up with more robust findings, no doubt reflecting
improvements in the ways in which parent and child characteristics are assessed.
Leading researchers no longer rely on a single measure, such as a parent or child
interview or a parent or child self-report scale, as a measure of parent or child attrib-
utes. Instead information is obtained from multiple sources—from parents, children,
teachers, school records, sometimes from children’s peers and police records as
well—and importantly, from direct observation of parent-child interactions
 6 MACCOBY

and of children in out-of-home settings. When several measures such as these are
aggregated, associations between parent attributes and children’s behavior can be
quite substantial. Parenting variables have typically accounted for 20% to 50% of
the variance in child outcomes (Conger & Elder 1994, Reiss et al 1995). Exception-
ally robust connections are reported in the recent large-scale study of adolescents in
never-divorced and step-families, Hetherington and colleagues (Hetherington et al
1999). Using composite scores for both parenting styles and children’s attributes,
report a concurrent coefficient of 0.76 between mothers’ ‘‘authoritative parenting’’
and adolescents’ ‘‘social responsibility’’ (the coefficient for fathers is 0.49). Parental
negativity has very strong connections for both parents with adolescents’ depression
and internalizing behavior.1 Patterson and colleagues have also found substantial
Downloaded from www.annualreviews.org. Guest (guest) IP: 180.75.250.144 On: Fri, 08 Nov 2024 04:09:10

correlations between parental characteristics (e.g. disciplinary practices and monitor-

ing) and children’s antisocial behavior (Patterson & Forgatch 1995). They are able
to show connections between parental behaviors and the children’s negative, coercive
behavior both at home and in out-of-home contexts.
Concurrent correlations are usually considerably larger than predictive ones. Lon-
gitudinal studies present the opportunity to examine the connections, if any, between
child-rearing styles at one point in time and subsequent attributes of the child. The
strength of the connections that have been found depends on many things, such as
what ‘‘packages’’ or clusters of parent and child variables are considered, the way
they are measured, the length of time between predictive and outcome measures, and
whether background variables are statistically controlled. A few examples will illus-
trate the range of findings. Kochanska 1997b:94 has been able to show that aspects
of early parenting account for a significant but moderate (Beta coefficient 0.29, F
9.96) portion of the variance in young children’s self-regulation and internalization
assessed a year later. Pettit and colleagues (1997:908) found some—but fewer and
weaker—predictive relationships between parenting as assessed at the beginning of
the kindergarten year and children’s adjustment and academic performance seven
years later, in the sixth grade. Strong predictive power of family interaction processes
over much longer spans of time have been found in longitudinal studies of antisocial
behavior (see Loeber & Dishion 1983). In current socialization studies, simple first-
order correlations between parenting characteristics and child outcomes are seldom
relied on. Indeed, sometimes they are not even reported. Instead, multivariate analyses
are used to investigate such questions as whether a given aspect of parenting has
different effects on different kinds of children or in families living in different cir-
cumstances; or whether different aspects of parenting have independent, addi-
1
The coefficient connecting mothers’ negative/conflictual behavior with children’s exter-
nalizing behavior was 0.82; for fathers this coefficient was 0.79. These coefficients are
path coefficients in structural models in which two aspects of parenting (authoritative
parenting parental negativity/conflict) are considered simultaneously along with measures
of sibling relationships. Scores derived from videotapes of parent-child interactions were
a contributing element (though a minor one) in the composite scores of both parenting
characteristics and child outcomes.
 PARENTING AND ITS EFFECTS ON CHILDREN 7

tive effects, whether they are interchangeable, or whether they interact so that the
effects of one depend on the level of another.
In longitudinal work, the initial level of a child’s characteristic at time 1 is some-
times statistically controlled to determine whether a time-l parent attribute is asso-
ciated with subsequent change change in the child’s behavior. As an example,
Patterson & Bank (1989) studied families when their sons were in grade school, and
again when the boys were adolescents. They found that changes in parenting during
these years were strongly related to the chances of a boy’s being arrested for delin-
quent activities in adolescence, even after the boy’s anti-social tendencies at grade-
school age were controlled. We see, then, that a variety of questions are being asked
in current and recent research—questions to which simple parent/child correlations,
Downloaded from www.annualreviews.org. Guest (guest) IP: 180.75.250.144 On: Fri, 08 Nov 2024 04:09:10

either concurrent or time-lagged, will not provide answers.

A word should be said, too, about how large a correlation between some aspect
of parenting and a child outcome is required for the relationship to be considered
important or meaningful. Along with the rest of the psychological discipline, devel-
opmental psychologists are currently turning away from reporting the outcomes of
studies primarily (or only) in terms of significance levels (p values) that indicate
degree of departure from the null hypothesis. Instead, results are beginning to be
reported in terms of effect sizes. For purposes of policy decisions in the medical
arena, correlations as small as 0.03 between the use of a medication and reduction
of disease have been considered strong enough to justify FDA approval of the drug
(Rosenthal 1999). The importance of a medical intervention can be estimated in terms
of such outcomes as the number of heart attacks averted or the number of people for
whom a debilitating, chronic disease can be arrested or reversed. In the past, corre-
lations in the 0.20s or 0.30s between aspects of family functioning and children’s
outcomes have often been dismissed as inconsequential. But when translated into the
number of children who are at risk, for example, for failing in school or becoming
delinquent or seriously depressed, predictive coefficients of this magnitude can be
seen as by no means trivial. From the standpoint of social policy, the issue becomes
one of how much importance a society attaches to social/behavioral outcomes, as
compared with medical ones. This is obviously a matter of values, not statistics.
Studies continue to vary considerably with respect to the size of first-order cor-
relations between parent and child characteristics. Clearly, a given parent behavior
may have different effects on different children, depending on such things as age,
sex, temperament, and distinctive prior experiences. If such differential effects exist,
aggregating data across a whole sample of children will wash out parent/child
effects—effects that might be quite robust within sub-groups of children. (See section
on interactions, below.) It is not possible to arrive at any general rule as to when
dividing by subgroups will increase or decrease a parent-child correlation. That will
depend on the researcher’s theoretical and empirical skill in identifying what the
pertinent groupings might be. The use of more sophisticated statistical methods has
contributed significantly to the ability of present-day researchers to identify parenting
effects within the matrix of other factors with which they often co-vary.
 8 MACCOBY

Not only have methods of assessment been improved, but current socialization
research includes a broader array of parenting attributes and focuses on a set of
parenting processes that were not so clearly delineated in times past. One aspect of
parental skill that has emerged in several recent studies as related to children’s well-
being is household organization; another concerns the ability of some parents to
develop a reciprocal form of interaction with their children (e.g. shared positive affect,
mutual responsivity). Studies of the predictive power of parent-child reciprocation in
early childhood have yielded quite robust parenting effects (See Kochanska &
Thompson 1997 for a review of this work). These examples illustrate the ways in
which the field of family-impact studies has been growing in conceptual as well as
methodological strength. Nevertheless we must be reconciled to the fact that there
Downloaded from www.annualreviews.org. Guest (guest) IP: 180.75.250.144 On: Fri, 08 Nov 2024 04:09:10

are important aspects of parenting that will never be revealed in studies that, by
necessity, try to encapsulate parental characteristics into measurable clusters or traits.
There are the memorable little socialization moments when the members of a parent/
child dyad are, for some reason, especially attuned to one another—when the child,
perhaps by virtue of having encountered a new and salient issue, is ready to both
explain and listen. At such a moment, the parent may do or say something that makes
a deep impression and can have a lasting influence. Conversely, a broken promise or
a revealed deception may break the prevailing relationship of trust between the two,
changing the nature of the influence that is possible between them. Such moments
are unique to a dyad and may not be captured in socialization studies, even though
our awareness of them is highlighted in biographies, autobiographies, and fiction.
I do not want to claim too much for the strength of parental influence in children’s
lives. Critics are right in pointing out that we have overemphasized these influences
at the expense of other kinds of environmental influences. To what extent early
childhood is a time of especially great plasticity, during which environmental inputs
will be more likely to have a lasting influence than inputs later in life is an open
question. Probably the answer will vary, depending on what domain of children’s
development we are talking about. (See for example, Neville’s finding [Neville 1995]
that the openness to influence by early experience differs between the semantic and
syntactic language systems). Because parents are usually the ones who spend the
most time with young children over extended periods of time, these questions of
changing plasticity do matter in our efforts to understand the parental realm of influ-
ence. Still, parents are never the only source of influence on children, and as children
grow older, they are more and more subject to the influence of peers, of schools and
teachers, and of television. Also, there are the random events—a serious illness or
accident, an unexpected success, a residential move, an environmental catastrophe—
that can alter the trajectory of a child’s life in ways that have little to do with parenting.
Of course, when we do see robust correlations between parent and child attri-
butes, the question of the direction of effects arises at once. In making their argu-
ment that we may be seeing child-to-parent effects rather than the reverse, critics
 PARENTING AND ITS EFFECTS ON CHILDREN 9

have relied heavily on the findings of behavior genetics2, especially on studies of

twins and adopted children. They have also relied on these findings to urge that
nonparental aspects of a child’s environment have greater weight than parental inputs
in determining how a child will develop.

THE CHALLENGE FROM BEHAVIOR GENETICS

Some of the major findings of behavior genetics are powerful and require students
of socialization to rethink some of their assumptions. Many of these findings are well
known, and I do not summarize them in any detail here, but focus on the main lines
Downloaded from www.annualreviews.org. Guest (guest) IP: 180.75.250.144 On: Fri, 08 Nov 2024 04:09:10

of argument that bear on the issue of parenting effects.

The Focus on Variation

Behavior geneticists seek to understand the sources of variation in some human trait
or characteristic. Their approach is to be distinguished from that of evolutionary
psychologists, who seek to understand the genetic underpinnings of characteristics
that are relatively uniform across a species.
There are important effects of both genes and environment that are overlooked in
studies that focus on the variation of a characteristic within a given population. A
human characteristic such as being born with two eyes is entirely genetic, yet its
heritability would be computed as zero in a twin or adoption study since it is a
characteristic that does not vary within the population studied. Similarly, there may
be an environmental factor that affects the mean level of a characteristic—raising or
lowering all scores to a similar degree—without greatly disturbing the rank-order of
individuals on the characteristic. Thus, adoption studies have found that the corre-
lation of adopted children’s IQs with those of their biological parents can remain
substantial, while at the same time the average IQ of the adopted children is higher
than that of their natural parents, as though children receive an IQ bonus from being
adopted into relatively stable, middle-class homes, while nevertheless continuing to
differ from each other according to their genetic endowment. In a study of French
children adopted at about the age of 5, it was found that the amount of increase in
their IQs (assessed again in adolescence) was considerably greater for children
adopted into affluent, well-educated families than for those adopted into underpriv-
ileged homes (Duyme et al 1999).
Secular trends illustrate the same point. The ‘‘Flynn effect’’ (Flynn 1987, 1999)—
the substantial, monotonic rise in mean IQ scores over many decades in

2
The term ‘‘behavior genetics’’ is a commonly used term for twin studies, adoption studies,
and epidemiological studies of family resemblance. Currently, since molecular geneticists
also study certain ‘‘behavioral’’ phenotypes in their relation to genes, the term quantitative
genetics is sometimes used to distinguish studies that rely on statistical genetic analyses
of family resemblance rather than on molecular gene identifiers. However the term ‘‘behav-
ior genetics’’ is used here because it is more familiar to readers.
 10 MACCOBY

Western industrialized countries—is well known. There has been a substantial rise
in the rates of smoking among American women in the last several decades, and the
rates of drinking alcohol dropped during prohibition. These changes, of course, have
occurred during periods of time that are much too short to reflect any genetic changes
and they have occurred despite the fact that heritability estimates for IQ, drinking,
and smoking have remained quite stable over the same time periods during which
the average levels were changing. A similar phenomenon is seen in some migration
studies, in which second-generation immigrants are on average quite different from
their foreign-born grandparents, even on highly heritable traits such as height (Angoff
1988) or obesity (Price at all 1993, Ravussin et al 1994). The implication of these
phenomena for parenting effects is this: There may have been secular changes in
Downloaded from www.annualreviews.org. Guest (guest) IP: 180.75.250.144 On: Fri, 08 Nov 2024 04:09:10

parenting—triggered perhaps by such things as changes in family structure or overall

economic level—that have had widespread effects on children without affecting her-
itability estimates for the outcome characteristics being affected.
These powerful environmental effects are missed in the estimates of E (environ-
ment) derived from behavior genetics studies of twins and adopted children. Another
way of putting this point is to note that high heritability of a trait does not imply that
it is not also subject to the influence of environmental factors, or that it cannot be
changed by alterations in environmental conditions. It is for this reason that, when
comparing group means (by race, sex, or socioeconomic status) it is not legitimate
to interpret any group differences in terms of estimates of genetic or environmental
effects derived from quantitative behavior genetic studies.

Experimental Interventions with Parents If large-scale environmental events can

change mean levels of a characteristic without greatly changing the rank-order of
individuals, it follows that experimental interventions might do the same. It is difficult
to change actual parenting practices through parent-training programs, and then to
document that program-induced changes in parenting change the mean levels of
children’s characteristics. Such programs must be longitudinal, of course, and must
have an untreated control group for comparison. Studies that intervene with the
parents but do not simultaneously treat the children, and that have random assignment
of families to treatment or control groups, are understandably rare, but several have
clearly shown that when treatment is able to change parental behavior toward children
in specified ways, the behavior of children changes correspondingly (e.g. Patterson
& Forgatch 1995, Van den Boom 1994, Forehand et al 1980). Dishion et al (1992)
were able to show that it was indeed the reduction of parent-to-child coercive behav-
ior, brought about by a parent-training intervention with a randomly assigned exper-
imental group, that produced declining levels of antisocial behavior in a group of
aggressive children. An intervention program that changes the mean of a group of
parents (and consequently, of their children’s behavior as well) may or may not
change the initial rank order of the children. Researchers commonly find that some
parents are influenced more than others by an intervention, and some children are
affected more than others by improvements in parental disciplinary or monitoring prac-
 PARENTING AND ITS EFFECTS ON CHILDREN 11

tices. These differential effects might either increase or decrease the range of outcome
scores in the treatment group, depending on whether it was the initially better-func-
tioning or poorly-functioning families who were most affected by the intervention.
However, expanding or shrinking the range of outcome scores does not necessarily
change the initial rank-order. The point here is that changes in a mean can be inde-
pendent of any changes in rank order. Thus, changes in a mean can clearly demon-
strate an environmental effect, quite apart from any correlational information (based
on rank orders of individuals) that might be used to compute genetic or environmental
effects in a genetic analysis. The environmental effects revealed by the mean change
would go undetected in a correlational analysis.

The Claim for Substantial Genetic Effects

Downloaded from www.annualreviews.org. Guest (guest) IP: 180.75.250.144 On: Fri, 08 Nov 2024 04:09:10

In traditional behavior genetic research, data from studies of twins and adopted chil-
dren are used to compute heritability estimates (h2), which are interpreted as estimates
of the proportion of variance accounted for by genetic factors. Many such studies
have yielded substantial heritability estimates. Identical twins have been found to be
more similar to each other than are same-sex fraternal twins with respect to a wide
range of characteristics, including susceptibility to certain diseases, intelligence, tem-
perament, and a number of personality characteristics. The inference is that this must
be due to their greater genetic similarity, because the important aspects of their
environments—parenting received, neighborhood, presence of a same-age, same-sex
sibling—are presumably equally similar for the two kinds of twin pairs. Adopted
children have been found to be more similar to their biological parents than to their
adoptive parents with respect to a selected set of characteristics for which researchers
have been able to obtain measures from both biological and adoptive parents.
In a general sense, the behavior geneticists have made their case. Children’s
genetic endowments do clearly affect how individuals will develop—in comparison
to other children—to a much greater extent than was thought to be the case during
the years of the ascendancy of reinforcement learning theories and psychodynamic
theories (the middle decades of the twentieth century.)
How substantial is this genetic contribution? Critics have argued that estimates
derived from twin studies systematically overestimate the genetic contribution to a
trait because identical twins in fact have more similar environments than do same-
sex fraternal twins. Identical twins (compared with fraternals) are treated more sim-
ilarly by their parents, spend more time together (and hence constitute a greater
proportion of each other’s social environment), and more often share the same friends
(Dunn & Plomin 1986, Plomin et al 1988, Reiss et al 1999, Rowe 1983). Probably,
the greater similarity in the environments of identical twins is not sufficiently strong
to negate the findings on genetic effects, but it does weaken them. Very likely, it
helps to account for the fact that heritability estimates are usually larger in twin studies
than in adoption studies.
Of course, the genetic contribution might be expected to be greater for some
human attributes than others. It appears to be more substantial for measures of intel-
fjlectual abilities than for social or personality attributes. However, it is difficult to
establish a reliable, generalizable estimate for any given trait. For one thing, estimates
 12 MACCOBY

vary depending on the source of information for measuring a trait. When children’s
characteristics are assessed through parents’ ratings, heritability estimates are often
considerably higher than when assessments are derived from behavioral observations
of the children, from children’s self-reports, or from teacher ratings. It appears that
parents see their children as more different from one another than other sources of
information find them to be (a contrast effect). In a recent review of studies of the
heritability of aggressive behavior, Cadoret and colleagues (Cadoret et al 1997) report
a very wide range of heritability coefficients, (from near zero to over 0.70), with the
higher figures coming from studies using parent report measures, and the lower ones
from observational studies. Miles & Carey (1997), in a meta-analysis of 24 twin and
adoption studies, report substantially greater values for h2 based on parent reports
Downloaded from www.annualreviews.org. Guest (guest) IP: 180.75.250.144 On: Fri, 08 Nov 2024 04:09:10

than for those based on adolescent self-reports.

Especially important is the fact that the size of a heritability coefficient depends
greatly on the range of both genetic and environmental factors in the population
being studied (G. Patterson, under review). Estimates of the heritability of a given
trait can change considerably when a new estimate is based on a culturally different
population, or especially when a new estimate includes families from a wider range
of subcultures and socioeconomic levels.
All this means that while the fact of a genetic contribution to human variability
is not in doubt, the size of this contribution is indeterminate for any given trait. More
specifically, the size of a heritability estimate cannot be generalized from the specific
population—in its specific environment—assessed with the specific set of measures
used in a given study.

Estimating the Size of Environmental Effects

In twin and adoption studies, estimates of the power of environmental factors are
derived by adopting the additive assumption, i.e. by assuming that that the sources
of variation in a trait can be separated into independent genetic (G) and environmental
(E) components that together (along with error variance) add to l00% of the variance
to be accounted for. On the basis of this assumption, the heritability coefficient can
be subtracted from 100% to yield an estimate of the environmental contribution to
variance. Estimating E in this way can be done without utilizing any direct measures
of environmental factors. Obviously, if the estimates of h2 are indeterminate, so are
the estimates of E derived by subtracting h2 from 100%.
The validity of the additive assumption has been widely challenged (Feldman &
Lewontin 1975, Gottlieb l995, Block 1995, Rose 1995, Turkheimer 1998) A number
of these critiques have appeared in connection with the controversy over Hernstein
& Murray’s book The Bell Curve (1994), but they are equally pertinent to the current
debate over parenting effects. If one adopts the additive assumption, it follows that
when h2 is large, the effects of all environmental factors—including parenting—must
be correspondingly small. A major counter-argument has been that in fact, everything
that human beings are or do must be a joint function of both their genes and their
life experiences. The pathway between genes and phenotypes is a long one,
 PARENTING AND ITS EFFECTS ON CHILDREN 13

with G and E being interwoven all along the way (see Elman et al 1996). The effects
of genes depend on environmental triggers or enabling conditions, and the effects of
different environments depend on the genetic characteristics of the individuals
encountering an environment. When genes and environment act jointly, this can
emerge empirically in behavior genetics studies in the form of either G 2 E corre-
lations or G 2 E interactions. In estimating environmental effects, much depends
on how these joint processes are handled (or not handled). Both kinds of coaction
are considered below, but the main point here is that neither G 2 E covariances nor
interactions fit into an additive model

Shared and Unshared Environmental Effects

Downloaded from www.annualreviews.org. Guest (guest) IP: 180.75.250.144 On: Fri, 08 Nov 2024 04:09:10

In twin and adoption studies, once an overall estimate of E has been derived by
subtracting h2 from 100%, E can be further subdivided into two environmental com-
ponents: Es (shared environment) and Eus (unshared environment). Once again, this
can be done without utilizing direct measures of either. If fraternal twins are quite
similar—more similar than would be expected from their shared genetics alone—or
if adopted children are more similar to the parents or siblings in their adopted families
than they are to adults or children in other households, this would imply an effect of
their rearing environment, including of course the parents’ child-rearing methods. Es
is estimated from sibling similarities, and any variance still unexplained after the
effects of G and Es have been accounted for are attributed to unshared environment
or error of measurement.
An especially surprising finding emerging from the body of behavior genetics
work has been that the effects of nonshared environment appear to be much greater
than those of shared environment (see Plomin & Daniels 1987 and Plomin et al l994).
Recent estimates of nonshared environmental effects are much reduced when mea-
surement error is taken into account (Rutter et al 1999). And a number of studies of
social behavior or pathology have found substantial shared-environment effects. Nev-
ertheless, shared environmental effects consistently emerge as small, and indeed are
often reported as being close to zero (Plomin & Bergeman 1991). Adopted children
do not appear to resemble their adoptive siblings or parents any more closely than
they resemble children growing up in different households. Also,in many respects
fraternal twins—or ordinary siblings, for that matter—do not greatly resemble each
other or their parents.
Critics have urged that it is not valid to estimate environmental effects, either
shared or unshared, without measuring them (Goodman 1991, Hoffman 1991, Rose
1995, Stoolmiller 1999, Patterson 1975). Recent work has involved designs in which
both genetic and environmental factors have been directly assessed. For example, a
group of leading behavior geneticists and leading students of parent-child interaction
collaborated in a study comparing children of different degrees of genetic relatedness
(twins, full siblings, half siblings, step siblings), in which parental child-rearing inputs
were assessed through observations of parent-child interactions, as well as through
parent and child reports (Reiss 1997, Reiss et al 1999, Hetherington et al 1999).
 14 MACCOBY

Relying on the additive assumption, these investigators have partitioned the variance
in child adjustment outcomes into the three components: G, Es, and Eus, reporting
substantial contributions from genetics. Effects of shared environment are variable,
making clear contributions to some outcomes but not others; in general, though, they
are considerably smaller than the substantial contributions from unshared environ-
ment. Unfortunately the design of this study confounds genetic similarities and family
structures: the group in which siblings are most genetically unlike (stepsiblings) is
also the group in which the two siblings receive the most discordant parenting (Heth-
erington et al 1999). It should be noted, too, that the range of environmental variation
is restricted in this study. Thus, many of the reported findings of this important study
are difficult to interpret.
Downloaded from www.annualreviews.org. Guest (guest) IP: 180.75.250.144 On: Fri, 08 Nov 2024 04:09:10

The inference of behavior geneticists’ claims concerning shared and unshared

environments might be that children are not greatly affected by the characteristics of
the household in which they are growing up. The weak shared-environment effects
have been interpreted to mean that such factors as the parents’ income or education,
parental pathology, the level of harmony or conflict between the parents, or the
neighborhood where the family lives must have little impact on how well the child
will do in school, how socially competent the children will be, and so forth (Plomin
et al 1994, Scarr & Grajek 1982).
These findings on weak shared-environment effects are startling, considering how
consistently studies of parenting effects have found substantial relationships between
these family characteristics and child outcomes. As an example, McLoyd (1998)
made a strong case for the mediating role of parenting in the deleterious effects of
poverty. In McLoyd’s analysis, it emerges that the great stresses on impoverished
parents—stresses stemming from the day-to-day struggle to find the resources to pay
for food and rent, and the stresses of trying to cope with living in crowded housing
and deteriorated, dangerous neighborhoods—bring about a weakening of parenting
skills and a disorganization of family life. It is the deterioration of parenting, McLoyd
found, that in its turn is responsible for many of the adjustment difficulties of children
growing up in impoverished families. (See also Conger et al 1994 and Pettit et al
1997 for findings supporting this conclusion.)
It is difficult to reconcile findings such as these with the claim that the aspects of
family environments that are shared by siblings do not affect their development. An
obvious possibility is that while the family environment does have an effect on each
child, its effects are different for different children. There probably was an unspoken
assumption, in traditional socialization work, that the effects of shared environments
would be to make siblings similar to one another. What the behavior geneticists are
telling us is that any influences of familial circumstances—such as parental illness
or health, economic prosperity or adversity, good or poor parenting—often function
to make siblings different rather than similar. It is possible that a dysfunctional family
environment may have effects on both members of a sibling pair, but that the effects
are not such as to make siblings more alike, but indeed might function to make them
more different. We know from Elder’s work on effects of the Great Depression (Elder
1974) that when a father loses his job, the effects on the child will depend on the
 PARENTING AND ITS EFFECTS ON CHILDREN 15

age and sex of the child at the time that this stressful event occurs. Even for same-
sex twins, we can imagine that if they were adolescents at the time, one might react
to a father’s job loss by going out to get an after-school job to help support the family
while the other might distance himself from the family and spend more time ‘‘hanging
out’’ with friends. Both children would be affected by the change in the family
environment, but differently.
Any familial or parental factors that serve to make siblings different rather than
similar to one another are assigned, in behavior genetics, to the unshared rather than
the shared environmental component when computing environmental effects. Behav-
ior geneticists have never said that estimates of unshared environments did not include
parent effects, but they argue that if parenting does have effects it must take one of
Downloaded from www.annualreviews.org. Guest (guest) IP: 180.75.250.144 On: Fri, 08 Nov 2024 04:09:10

two forms: parents must be treating different children in their families differently (or
providing different environments for them), or different children in the same family
who are exposed to similar parenting must react to the same parental inputs
differently.
A considerable body of recent work has focussed on the question, What is it that
makes siblings different from one another? (see Hetherington et al 1994). In these
studies evidence is presented that siblings tend to join different peer groups and that
siblings have considerably different experiences within the context of the sibling
relationship itself. The question of how differently they are treated by their parents
remains open. Studies done during a single time period often show that two siblings
are treated differently by their parents (see summary by Brody & Stoneman 1994)
However, in a longitudinal study Dunn found that parents were fairly consistent in
how they treated children at a specific age. That is, a second child, when reaching
the age of four, is treated in a similar way to the way his/her older sibling was treated
at that age, even though the older sibling may now be receiving different treatment.
Thus, over the span of the ‘‘growing up’’ years, different children in the same family
received comparable treatment. This fact, of course, would be missed in any study
that did not look for it longitudinally; the extent of differential treatment is likely to
be overestimated in cross-sectional studies (except in the case of twins). Whether or
not children actually are treated differently over the whole span of childhood, there
is reason to believe that children’s perceptions of how differently they are treated
may be of considerable importance in children’s development, so concurrent differ-
ences are important in their own right (Dunn & McGuire 1994).
In general, the exploration of siblings’ unshared environments has been a pro-
ductive and instructive enterprise. We now know that the environments of children
growing up in the same family can indeed be different. But this does not solve the
problem of how to interpret aspects of the environment that are truly shared, such as
a parental illness, family income, parents’ education, or the neighborhood where the
family lives—factors that have an impact even when they function to make siblings
different rather than alike. As noted above, behavior geneticists tend to conclude that,
since it is clear that these aspects of environment are truly shared, they must not be
having an effect because Es effects are negligible. As Plomin and colleagues say,
‘‘So often, we have assumed that the key influences on children’s development are
 16 MACCOBY

shared: their parents’ personality and childhood experiences, the quality of their par-
ents’ marriage relationship, children’s educational background, the neighborhood in
which they grow up, and their parents’ attitude to school or to discipline. Yet to the
extent that these influences are shared, they cannot account for the differences we
observe in children’s outcomes’’ (Plomin et al 1994:23).
On the contrary, it seems plausible that these shared factors may indeed have
powerful effects that do not show up in computations of shared environmental effects
because of the requirement that only an environmental factor that makes siblings
more similar can be called ‘‘shared.’’ A behavior geneticists might say about the
effect of a shared environmental factor that makes siblings different, ‘‘Oh, but we
are calling those unshared effects.’’ But to call an environmental input unshared even
Downloaded from www.annualreviews.org. Guest (guest) IP: 180.75.250.144 On: Fri, 08 Nov 2024 04:09:10

though it is experienced by all children in a family (e.g. a father’s job-loss, a mother’s

depression, a move to a better neighborhood) is an unfortunate distortion of the simple
meaning of the word ‘‘shared.’’ We could see this as only a trivial matter of termi-
nology choice, but it can lead to serious misunderstandings of behavior geneticists’
findings. By definition, they have ruled out the possibility that a truly shared aspect
of the environment could have a significant effect on at least one child, when the
effects on different children are not the same.
When we deal with a shared environmental factor that impacts different children
in the family differently, it could be argued—and behavior geneticists do so argue—
that the effect stems from the fact that some children are more genetically vulnerable
to an environmental event than others. In the usual computations of heritability, such
an effect would then be assigned to the G component of the equation, rather than to
the environmental one. Surely, it is equally plausible that both G and E are important
here. Risk factors, such as poverty, a father’s unemployment, or a mother’s depres-
sion, are indeed environmental conditions that are shared by all the children in a
family. In large population studies they will rightly emerge as having a negative
impact on children, even though some children are more vulnerable to them than
others. In the extreme case, we could imagine that in every two-child family, one of
the children would show the deleterious effects of poverty and the other would not
(perhaps because of genetic differences between them). Across many families, there
would be a very powerful effect of poverty and it would be rightly identified as a
strong risk factor, even though the shared environment effect would be computed at
zero. The obvious danger here is that low estimates for Es can be interpreted as
meaning that family environmental conditions that children share do not have an
impact on their development, whereas in fact the opposite can be true, and often is.
The findings from behavior genetics on shared and unshared environments have
profound implications for the way we think about child-rearing practices and their
effects. For one thing, they focus attention on sibling differences. This is something
that traditional research on child-rearing—almost always involving only one child
per family—did not deal with. It should be noted that there is nothing about the
findings of these traditional studies that is invalidated by their having studied only
one child. The connections identified between the parental inputs to this child and
the child’s characteristics can be reliable, replicable ones, even though if we had
 PARENTING AND ITS EFFECTS ON CHILDREN 17

studied a different parent-child pair in the same family we might have gotten a
different constellation of parenting and outcomes. The picture emerging from aggre-
gating data across a set of one-child cases is valid as well, though the findings are
surely attenuated by the within-family sibling variation. Still, we get a less differ-
entiated picture than the one that emerges from the study of siblings. Family systems
theorists have alerted us to ‘‘niche-picking’’ by different children in a family—the
effort of children to find distinctive roles. Evolutionary theorists have argued that
there is natural competition among siblings for parental attention and other resources
provided by parents. In short, there is reason to believe that there are forces motivating
children to differentiate themselves from their siblings, and these may counterbalance,
or transform, the effects of parental inputs that might otherwise function to make
Downloaded from www.annualreviews.org. Guest (guest) IP: 180.75.250.144 On: Fri, 08 Nov 2024 04:09:10

them the same. Of course, some of the differentiation between siblings can come
directly from differential treatment by the parents, or it can stem from differential
reactions by different children to the same parental inputs.

INTERPRETING PARENT-CHILD COVARIANCE

As noted above, quantitative geneticists have raised serious questions concerning the
direction of effects when parental behaviors and child characteristics are found to be
correlated. They point out that parent-child correlations could stem from genetic
predispositions shared by parents and children that are directly transmitted from one
generation to the next. In addition, evocative covariance occurs when children with
different genetic predispositions elicit correspondingly different reactions from their
parents. Thus, when a child is predisposed to be resistive or distractible and does not
pay attention to the parent, the parent reacts by becoming more authoritarian, whereas
a cooperative child will evoke a different reaction. (See Ge et al 1996, which shows
clearly how the parenting by adoptive parents is affected by the predispositions of
their adopted children.) Active covariance occurs when children select from a range
of potential environmental influences only certain features with which to engage—
certain TV programs, certain friends, certain sports—presumably on the basis of their
own predispositions. Although children do not have the freedom to choose their
parents, they do have some power to select which aspects of parental inputs they will
attend to. Children with different genetic predispositions no doubt react differently
to the same parental input, depending either on what they attend to, how they interpret
their parents’ actions, or what behavioral predisposition of their own has been trig-
gered. In twin and adoption studies, all these forms of covariance between parent
and child are thought to imply that genetics—either the child’s own or the genes
shared with parents—are driving the parental behaviors. For these reasons, it has
seemed reasonable, in behavior genetic analyses, to assign parent-child covariances
to the genetic component in the G ` E 4 l00% equation.
I would argue that to assign parent-child covariance to G systematically under-
estimates the strength of parenting effects. It does so by ignoring the feed-back
 18 MACCOBY

loop whereby parents, in reacting to a given child’s distinctive input, reciprocate with
counter influences of their own.
The fact that parents respond differently to children with different predispositions
is not in doubt, and it has been one of the contributions of behavior genetics to bring
this fact into the foreground of our thinking. Socialization researchers, too, have for
some time been centrally aware of this issue and in the past several decades have by
no means ignored the problem of direction of effects. A great deal of effort has been
devoted to examining the processes whereby parents and children influence one
another. The predominant modern viewpoint among students of socialization is an
interactionist one, in which it is assumed that in any ongoing relationship, each
member of an interacting pair is a significant feature of the other’s environment to
Downloaded from www.annualreviews.org. Guest (guest) IP: 180.75.250.144 On: Fri, 08 Nov 2024 04:09:10

which each must adapt. In addition, it has become clear that the developmental level
of a child is a powerful determiner of what kind of socialization inputs a parent will
provide and what kind of receptiveness, resistance, or negotiation the child will bring
to a parent-child encounter. We cannot expect to find generalizations about the nature
and effects of specific parent/child interactions that will span all the ages and stages
of a child’s development.
From an interactionist perspective, the idea that in a long-standing relationship
such as the one between a parent and child, the child would be influencing the parent
but the parent would not be influencing the child is absurd. While it is entirely
reasonable to assign the child’s part in parent-child covariance (i.e. evocative effects)
to the genetic component, it is not reasonable to assign the reciprocal parent contri-
bution to the child’s genetics. The parent’s response is surely a function not only of
the child’s initiative but also of the parent’s genetics, learned modes of behavior,
perceptions of the child’s needs and characteristics, and socialization objectives. And,
just as surely, the parent’s response to the child’s initiatives is a central element in
the child’s environment. Thus, to assign the whole of parent-child covariance to G
is surely to overestimate G and underestimate E.
A recent study from the Rutter-Plomin research group in London (O’Connor et
al 1998) beautifully identifies the contributions of correlated G and E factors to
developmental outcomes. Using longitudinal data from the Colorado Adoption Study,
these researchers identified two groups of adoptees: one at genetic risk for anti-social
behavior ( i.e. a history of anti-social behavior in the biological mother) and the other
not at risk. At several points during the adoptees’ childhood, both the children’s
characteristics and the adoptive parents’ child-rearing methods were assessed. Find-
ings were that children carrying a genetic risk for antisocial behavior were more
likely to receive negative socialization inputs from their adoptive parents—an evoc-
ative effect. But parental negative behavior made an independent contribution to
children’s externalizing, over and above the children’s genetic predispositions.
This study illustrates what an interactionist perspective would lead us to expect:
Parent-child covariance reflects the reciprocal effects of both parent and child inputs
to a relationship. The issue here is not to compare G and E effects to see which is
stronger. Instead, it is to explore how they intersect or how one mediates the effect
of the other. Such issues remain largely unexplored. The relative strength of each con-
 PARENTING AND ITS EFFECTS ON CHILDREN 19

tribution is difficult to assess and is almost entirely unknown in the large body of
research literature on within-family socialization. The study by O’Connor et al illus-
trates the futility of efforts to compartmentalize the variance in children’s character-
istics into separate G and E components without getting independent measures of
each. What this study shows is that G and E operate jointly to produce an outcome.

THE INTERACTION OF GENETIC AND

ENVIRONMENTAL FACTORS

Interactions are found when a given environment has different effects on an organism,
Downloaded from www.annualreviews.org. Guest (guest) IP: 180.75.250.144 On: Fri, 08 Nov 2024 04:09:10

depending on the organism’s genetic traits. Interactions are also seen when organisms
with a given set of genetic traits react in one way under one set of environmental
conditions, but another way under different environmental conditions. Plant biologists
are able to point to dramatic examples, such as when there are two genetic strains of
a grain, and strain 1 grows taller than strain 2 at high altitudes and shorter than strain
2 at low altitudes.

G @ E Interactions in Animal Studies

A careful review of animal studies that looked for G 2 E interactions (Plomin 1986)
reported that though interactions were sometimes found, they were not consistent
within or across studies and accounted for only a small portion of variance. Since
that review, some progress has been made in the difficult enterprise of mapping the
complex processes that intervene between genotype and phenotype, and recently
there has been some success in uncovering interactions with respect to these better-
defined processes. In several mammalian species, it is now known that there are
genetic factors underlying variation in ‘‘reactivity,’’ that is, in the tendency to become
emotionally aroused and fearful. Different levels of reactivity in rats are associated
with both neuroendocrine and behavioral functioning (Caldji et al 1998, Liu et al
1997). Reactive animals appear jittery and hesitate to explore novel environments.
In Rhesus monkeys, a gene has been isolated one of whose alleles is associated with
the emergence of a reactive temperament (Suomi 1999). It has been found that young
animals carrying the ‘‘reactive’’ allele are particularly vulnerable to variations in early
rearing experience. If they are subjected to maternal deprivation during their first six
months (reared with peers but no adult females) their neuroendocrine functioning is
affected and they display a variety of pathological symptoms into adulthood, includ-
ing incompetence in social interactions, low status in peer groups, and incompetence
in mothering their own offspring (Suomi 1997). By contrast, young animals who do
not carry the genetic risk factor are much less affected by maternal deprivation. In
current work, genetically reactive newborn monkeys are being cross-fostered to non-
reactive mothers, and preliminary observations indicate that calm mothering does
indeed buffer them from the development of strongly reactive behavior. Cross-fos-
tering work with rodents is also showing the positive effects of rearing
 20 MACCOBY

genetically at-risk infants by a nurturant mother (Anisman et al 1998).We see here

that the effects of a genetic predisposition are strongly seen under one set of envi-
ronmental (rearing) conditions but not another.

G @ E Interactions in Adoption Studies

Of course it is not possible to carry out systematic experiments of this sort with
humans, but quantitative genetic studies can be used to test for G 2 E interactions.
However, in such studies it is no longer possible to bypass measures of the environ-
ment and estimate E effects only as a residual after G effects have been estimated
Downloaded from www.annualreviews.org. Guest (guest) IP: 180.75.250.144 On: Fri, 08 Nov 2024 04:09:10

and subtracted out. Instead, there must be direct measures of both G and E. In most
twin studies, the environments of twin pairs are too homogeneous to permit good
estimates of G 2 E interactions, and there are difficulties in interpreting the differ-
ences between identical and fraternal twins in interaction terms. As Plomin said, ‘‘
. . . it is difficult if not impossible to use the twin design to estimate the overall
contribution of genotype-environment interaction to phenotypic variance’’ (Plomin
1986:96).3 Since that time, there have been some innovations in utilizing twin studies
to study interactions. One method is to use one twin’s characteristic as an index to
the co-twin’s genetic risk; when the two are not highly concordant for the trait, their
respective environments can then be examined for clues as to the origins of their
non-genetic differences. Another method is simply to compare the heritability esti-
mates found in two different environments.
In studies of adopted children, adoptive families vary with respect to the kind of
environments they provide (though the range of environmental variation is usually
consistently narrower than in unselected populations), and interactions can be effec-
tively studied. In a large-scale study of adopted children in Finland (Tienari et al
1994), children with a schizophrenic biological parent were contrasted with adopted
children who did not carry this genetic risk factor. It was found that the at-risk children
were more likely to develop a range of psychiatric problems, but only if they were
adopted into dysfunctional adoptive families. A study of adopted children whose
biological parents did or did not have a history of criminality (Bohman 1996) yielded
similar results: Among adoptees who carried a risk factor from their biological par-
ents, those who had been adopted into dysfunctional homes were over three times
more likely to become petty criminals than those whose adoptive parents had pro-
vided a stable, supportive environment.
These findings from adoption studies are consistent with studies of cross-
generational transmission of psychiatric disorders (Ge et al 1996, Downey & Walker
1992). These studies point to a mediating role of parenting: Children whose parents
suffer from a psychiatric disorder are usually no more likely than children with normal
parents to develop psychiatric disorders, unless the children are exposed directly

3
In studies of twins reared apart more variation in environments is of course usually present
but the Ns for such studies are small and environmental information fragmentary.
 PARENTING AND ITS EFFECTS ON CHILDREN 21

to harsh parenting and/or an otherwise dysfunctional family environment by the

parents who are rearing them.
Taken together, these studies indicate that genetic risks may or may not become
manifest, depending on the quality of the parenting children receive. In other words,
whatever genetic risks a child carries can require an environmental trigger to emerge
into phenotypic expression. Well-functioning parents can buffer children against the
emergence of negative genetic potentials.

Studies of Interactions with Temperament

Downloaded from www.annualreviews.org. Guest (guest) IP: 180.75.250.144 On: Fri, 08 Nov 2024 04:09:10

It is possible to approximate the study of G 2 E interactions even when no direct

information on children’s genetics is available. Since several dimensions of temper-
ament are known to have a significant genetic component,4 researchers have identified
children with different temperaments, and studied how they differ in the way they
interact with their parents and in the impact parental inputs have on them. Children’s
temperamental characteristics appear to set the stage for the kind of bi-directional
processes that will emerge between them and their parents (Collins et al 1999).
Evidence has been emerging that a given parental practice can have different effects
on children with different temperaments. Kochanska (1995, 1997a) studied the devel-
opment of conscience in young children. She reported that for shy, temperamentally
fearful children, parental power-assertion does not appear to promote conscience
development—gentler techniques are called for. But with bold assertive children,
effective parenting involves firmness, along with maternal responsiveness and the
formation of a close emotional bond with the child. In a similar way, it has been
found that for children who are initially difficult, impulsive, and/or resistive, parental
firmness and restrictiveness are more important ingredients in preventing the subse-
quent development of externalizing behavior than is the case for children with easier
temperaments (Bates et al 1998). Other studies finding interactions between children’s
temperament and parenting effects are those by Belsky et al (1997) and Deater-
Deckard & Dodge (1997).
In Plomin’s (1986) review of the studies on G 2 E interactions in children that
were available at that time, significant interactions were found to be quite rare. It is
possible at this time to be more positive—though guardedly so—concerning the
prevalence and power of these interactions. They may be more prevalent with respect
to personality dimensions and psychopathology than they are with respect to cog-
nitive dimensions, but they obviously cannot be detected by using the traditional
additive approach to partitioning variance between G and E. Indeed, the presence of
interactions constitutes strong evidence against the validity of this approach. In molec-
ular genetics, it is axiomatic that interactions are the rule, not the exception and that
efforts to partition variance into the two traditional components are counterproductive.

4
Temperament is currently defined as ‘‘ . . . constitutionally based individual differences
in reactivity and self-regulation . . . ’’ (Rothbart & Ahadi 1994:54).
 22 MACCOBY

OVERVIEW

Behavior genetics studies have made substantial contributions to our understanding

of the factors that underlie the variation among children in their intellectual and
personality characteristics. Studies of twins and adopted children have shown beyond
reasonable doubt that a wide range of children’s attributes are influenced substantially
by the genes they inherit from their biological parents. These studies first began to
appear in the 1930s, and work done since that time has continued to confirm the
power of genetic factors. The precise magnitude of the genetic contribution to a given
trait, however, has proved to be difficult to establish: Heritability estimates vary
widely, and indeed there is no reason to expect that there exists any one valid number
Downloaded from www.annualreviews.org. Guest (guest) IP: 180.75.250.144 On: Fri, 08 Nov 2024 04:09:10

for any given trait. Instead, heritability inevitably depends on the range of variation
within a given sample being studied, and on the socio-cultural milieu in which the
studied population lives. No single estimate can ever be taken as definitive.
I have argued that when genetic factors are strong, this does not mean that envi-
ronmental ones, including parenting, must be weak. The relation between the two is
not a zero-sum game, and the additive assumption is untenable. There are environ-
mental factors that can affect a group or population without greatly rearranging the
rank order of individuals within that group. In such a case, estimates of heritability
can remain high while at the same time powerful environmental forces are at work.
For this reason, it is not legitimate to extrapolate G or E estimates derived from a
behavior genetic analysis to differences between groups (e.g. between races, social
classes, or genders) that differ in their environmental milieu.
Experimental interventions have been designed to change children’s behavior by
means of changing the child-rearing practices of their parents. These intervention
programs have amply demonstrated that parenting does have direct effects on how
children behave, both inside and outside the home. When families are randomly
assigned to an intervention group, the children show a reduction in problem behaviors
by comparison with an untreated control group, and these effects are clearly inde-
pendent of any genetic contribution to the outcome behavior being studied. Equally
important is the presence of interactions between genes and environment, such that
an environmental trigger is needed to evoke a genetic predisposition. Included here
would be instances in which competent, supportive parenting protects a child from
developing a dysfunction for which he or she is genetically predisposed. Such inter-
actions have been largely ignored in traditional behavior genetic studies. What I argue
here is that while the contribution of genetic factors to children’s characteristics has
been solidly documented in behavior genetics work, the contribution of environmen-
tal factors as derived from these studies has not.
A crucially important contribution of behavior genetics has been to draw our
attention to the unlikeness of siblings. While we may have been marginally aware
of sibling disparities, the traditional studies of childhood socialization included only
one child per family, and there was an implicit assumption that parents treated their
various children much alike and that the effects of what they did would be similar
 PARENTING AND ITS EFFECTS ON CHILDREN 23

for all their children. We must now seriously reexamine these assumptions. We now
know that the correlations between siblings with respect to many of their character-
istics are very low—indeed, sometimes lower than their genetic relatedness would
predict. Is the unlikeness of siblings due to their being treated differently by their
parents? To some extent, yes, though findings are not consistent across studies. What
the behavior geneticists have shown is that the genetic predispositions of different
children often drive the responses of parents, determining to some degree the kind
of parenting a child will receive. Understandably then, behavior geneticists have
assigned correlations between parent and child behaviors to the child’s genetics, but
I argue that this is a mistake, in that it ignores the return feed-back loop whereby a
parent, whose behavior has been triggered by the child, responds with actions which
Downloaded from www.annualreviews.org. Guest (guest) IP: 180.75.250.144 On: Fri, 08 Nov 2024 04:09:10

in their turn influence the child. To ignore this reciprocal influence is to seriously
underestimate parenting effects.
The unlikeness of siblings continues to be something we do not fully understand.
It has been interpreted to mean that aspects of environment which siblings share—
amount of inter-parental conflict, good or poor neighborhoods, poverty or affluence,
level of parental education or the ‘‘cultural’’ level of the home environment, house-
hold organization or disorganization, the amount of good humor characterizing the
family atmosphere—all these things must have very little influence on children’s
development. This interpretation flies in the face of the large body of research on
risk factors, which repeatedly finds strong relationships between these aspects of
family functioning and children’s outcomes. I argue that the risk-factor findings are
indeed valid, but that they need not have the same effects on all children in a family
nor function to make siblings more alike. It begins to seem likely that there are strong
factors pushing siblings toward differentiation from one another, including perhaps
competition for parental attention or other resources, ‘‘niche picking,’’ counteriden-
tification, and differential perceptions of the sibling relationship on the part of the
participants in it. Such factors could function as counter forces, working against
parental inputs that might otherwise make siblings more alike. But this is speculation.
Much remains to be learned about this complex matter.
Many factors other than parents’ actions influence how children grow and develop.
As children grow beyond the preschool years, they are exposed more and more to
other adult socialization agents (teachers, coaches) and, of course, to individual
friends and larger peer groups. Within the matrix of factors that affect children’s
development, it is clear that parenting effects are real, though they often combine
with genetic effects in influencing an outcome. Along with many other students of
these phenomena, I urge that we give up the effort to partition the causal factors
influencing children’s development into two separate ‘‘nature’’ and ‘‘nurture’’ com-
ponents, and that we abstain from asking ourselves which is more important. The
two are inextricably interwoven all along the pathway from birth to maturity. So be
it. Let us not underestimate either, but concentrate on the ways in which they function
jointly.
 24 MACCOBY

ACKNOWLEDGMENTS
During the time I have been working on this chapter, I have also been participating
with four colleagues in writing a closely related paper. These colleagues are
W Andrew Collins, E Mavis Hetherington, Lawrence Steinberg, and Marc Bornstein.
I gratefully acknowledge their help in searching out references and the clarifying
value of our discussions. Others who have been helpful in providing materials for
this chapter are Michael Rutter, Andrew Heath, Gerald Patterson, Stephen Suomi,
Rich Weinberg, Megan Gunnar, John Flavell, Grazyna Kochanska, Robert Cairns,
and Tom Dishion.

Visit the Annual Reviews home page at www.AnnualReviews.org.

Downloaded from www.annualreviews.org. Guest (guest) IP: 180.75.250.144 On: Fri, 08 Nov 2024 04:09:10

LITERATURE CITED
Angoff WH. 1988. The nature-nurture debate, Cadoret RJ, Leve LD, Devor E. 1997. Genetics
aptitudes and group differences. Am. Psychol. of aggressive and violent behavior. Psychol.
43:713–20 Clin. N. Am. 20:301–22
Anisman H, Zaharia MD, Meaney MJ, Merali Z. Caldji C, Tannenbaum B, Sharma S, Francis D,
1998. Do early-life events permanently alter Plotsky PM, Meaney MJ. 1998. Maternal care
behavioral and hormonal responses during infancy regulates the development of
to stressors. Int. J.Dev. Neurosci. 16:149– neural systems mediating the expression of
64 fearfulness in the rat. Proc. Nat. Acad. Sci.
Bates J, Pettit G, Dodge K, Ridge B. 1998. Inter- 95:5335–40
action of temperamental resistance to control Collins WA, Maccoby EE, Steinberg L, Hether-
and restrictive parenting in the development ington EM, Bornstein M. 1999. Contempo-
of externalizing behavior. Dev. Psychol. rary research on parenting: the case for nature
34:982–95 and nurture. Am. Psychol. In press
Baumrind D, Black AE. 1967. Socialization prac- Conger RD, Elder GH. 1994. Families in Trou-
tices associated with dimensions of compe- bled Times: Adapting to Change in Rural
tence in preschool boys and girls. Child Dev. America. Hawthorne, NY: Aldine
38:291–327 Conger RD, Ge X, Elder GH, Lorenz FO, Simons
Belsky J, Hsieh K, Crnic K. 1997. Mothering, R. 1994. Economic stress, coercive family
fathering and infant negativity as predictors process, and developmental problems of ado-
of boys’ externalizing problems and inhibi- lescents. Child Dev. 65:541–61
tion. Dev. Psychopathol. 10:301–19
Deater-Deckard K, Dodge K. 1997. Spare the
Block N. 1995. How heretability misleads about
rod, spoil the authors: emerging themes in
race. Cognition 56:99–128
research on parenting. Psychol. Inq. 8:230–35
Bohman M. 1996. Predispositions to criminality:
Dishion TJ, Patterson GR, Kavanagh K. 1992.
Swedish adoption studies in retrospect. In
An experimental test of the coercion model:
Genetics of Criminal and Anti-Social
linking theory, measurement, and interven-
Behavior, Ciba Found. Symp. 194, ed. GR
tion. In The Interaction of Theory and Prac-
Bock, JA Goode, pp. 99–114. Chickester/
New York: Wiley. 283 pp. tice: Experimental Studies of Interventions,
Brody G, Stoneman Z. 1994. Sibling relation- ed. J McCord, R Trembly, pp. 253–82. New
ships and their association with parental dif- York: Guilford. 29 pp.
ferential treatment. See Hetherington et al Downey G, Walker E. 1992. Distinguishing
1994, pp. 129–42 family-level and child-level influences on
 PARENTING AND ITS EFFECTS ON CHILDREN 25

the development of depression and aggres- Hetherington EM, Reiss D, Plomin R, eds. 1994.
sion. Dev. Psychopathol. 4:81–96 Separate Social Worlds of Siblings. Hillsdale,
Dunn J, McGuire S. 1994. Young children’s non- NJ: Erlbaum. 232 pp.
shared experiences: a summary of studies in Hetherington EM, Henderson SH, Reiss D. 1999.
Cambridge and Colorado. See Hetherington Adolescent Siblings in Stepfamilies: Family
et al 1994, pp. 111–28 Functioning and Adolescent Adjustment.
Dunn J, Plomin R. 1986. Determinants of mater- Monog. Soc. Res. Child Dev. In press
nal behavior toward three-year-old siblings. Hoffman LW. 1991. The influence of the family
Br. J. Dev. Psychol. 57:348–56 environment on personality: accounting for
Duyme M, Dumaret AC, Stanislaw T. 1999. How sibling differences. Psychol. Bull. 110:187–
can we boost IQs of ‘‘dull’’ children?: a late 203
adoption study. Proc. Nat. Acad. Sci. In press Kochanska G. 1995. Children’s temperment,
Elder GH. 1974. Children of the Great Depres- mothers’ discipline, and security of attach-
Downloaded from www.annualreviews.org. Guest (guest) IP: 180.75.250.144 On: Fri, 08 Nov 2024 04:09:10

sion. Chicago: Univ. Chicago Press. 400 pp. ment: multiple pathways to emerging inter-
Elman JL, Bates EA, Johnson MH, Karmiloff- nalization. Child Dev. 66:597–615
Smith A, Parisi D, Plunkett K. 1996. Rethink- Kochanska G. 1997a. Multiple pathways to con-
ing Innateness: A Connectionist Perspective science for children with different tempera-
on Development. Cambridge, MA: MIT Press ments: from toddlerhood to age five. Dev.
Feldman MW, Lewontin RC. 1975. The herita- Psychol. 33:228–40
bility hangup. Science 190:1163–68 Kochanska G. 1997b. Mutually responsive ori-
Flynn JR. 1999. Searching for justice: the dis- entation between mothers and their young
covery of IQ gains over time. Am. Psychol. children: implications for early socialization.
54:5–20 Child Dev. 68:908–23
Flynn JR. 1987. Massive IQ gains in 14 nations: Kochanska G, Thompson RA. 1997. The emer-
what IQ tests really measure. Psychol. Bull. gence and development of conscience in tod-
101:171–91 dlerhood and early childhood. In Parenting
Forehand R, Wells KC, Griest DL. 1980. An and Children’s Internalization of Values, ed.
examination of the social validity of a parent JE Grusec, L Kuczunski pp. 53–77. New
training program. Behav. Ther. 11:488–502 York: Wiley
Ge X, Conger R, Cadoret R, Neiderhiser J, Yates Liu D, Diorio J, Tannenbaum B, Cladji C, Francis
W, et al. 1996. The developmental interface D, et al. 1997. Maternal care, hippocampal
between nature and nurture: a mutual influ- glucocorticoid receptors and hypothalamic-
ence model of child antisocial behavior and pituitary-adrenal responses to stress. Science
parent behavior. Dev. Psychol. 32:574–89 277:1659–62
Goodman R. 1991. Growing together and grow- Loeber R, Dishion TJ. 1983. Early predictors of
ing apart: the non-genetic forces on children male delinquency: a review. Psychol. Bull.
in the same family. In The New Genetics of 94:68–99
Mental Illness, ed. R McGuffin, R Murray, Maccoby EE, Martin JA. 1983. Socialization in
pp. 212–24. Oxford: Oxford Univ. Press the context of the family: parent-child inter-
Gottlieb G. 1995. Some conceptual deficiencies action. In Handbook of Child Psychology,
in ‘‘developmental’’ behavior genetics. Hum. Vol. 4, ed. PH Messen, EM Herrington, 4:1–
Dev. 38:131–41 102. New York: Wiley. 4th ed. pp.1–102
Harris JR. 1998. The Nurture Assumption: Why McLoyd VC. 1998. Socioeconomic disadvantage
Children Turn Out the Way They Do. New and child development. Am. Psychol. 53:185–
York: Free Press. 462 pp. 204
Hernnstein RJ, Murray C. 1994. The Bell Curve: Miles D, Carey G. 1997. Genetic and environ-
Intelligence and Class Structure in American mental architecture of human aggression. J.
Life. New York: Free Press Pers. Soc. Psychol. 72:207–17
 26 MACCOBY

Neville HJ. 1996. Developmental specificity in Ravussin E, Bennett PH, Valencia ME, Schulz
neurocognitive development in humans. In LO, Esparaz J. 1994. Effects of a traditional
The Cognitive Neurosciences, ed. M Gazza- lifestyle on obesity in Pima Indians. Diabetes
niga, pp. 219–31 Cambridge, MA: MIT Press Care 17:1067–74
O’Connor TG, Deater-Deckard K, Fulker D, Rut- Reiss D. 1997. Mechanisms linking genetic and
ter M, Plomin R. 1998. Genotype-environ- social influences in adolescent development:
ment correlations in late childhood and early beginning a collaborative search. Curr. Dir.
adolescence: antisocial behavioral problems Psychol. Sci. 6:100–6
and coercive parenting. Dev. Psychol. Reiss D, Hetherington EM, Plomin R, Howe
34:970–81 GW, Simmens SJ, et al. 1995. Genetic ques-
Patterson GR. 1975. Multiple evaluations of a tions for environmental studies: differential
parent-training program. In Applications of parenting and psychopathology in adoles-
Downloaded from www.annualreviews.org. Guest (guest) IP: 180.75.250.144 On: Fri, 08 Nov 2024 04:09:10

Behavior Modification, ed. T Thompson, W cence. Arch. Gen. Psychol. 52:925–36

Dockens, pp. 299–322. New York: Academic Reiss D, Neiderhiser J, Hetherington EM, Plomin
Patterson GR, Bank LI. 1989. Some amplifying R. 1999 The Relationship Code: Deciphering
mechanisms for pathologic processes in fam- Genetic and Social Patterns in Adolescent
ilies. In Systems and Development: The Min- Development. Cambridge, MA: Harvard
nesota Symposium on Child Psychology, ed. Univ. Press. In press
MR Gunnar, E. Thelen pp. 167–209. Hills- Rose R. 1995. Genes and human behavior. Annu.
dale, NJ Rev. Psychol. 46:625–54
Patterson GR, Forgatch M. 1995. Predicting Rosenthal R. 1999. Discussion on effect sizes at
future clinical adjustment from treatment out- symposium on, ‘‘Does Child Care Quality
come and process variables. Psychol. Assess. matter?’’ Presented at Biennial Meet. Soc.
7:275–85 Res. Child Dev., April, Albuquerque, NM
Pettit GS, Bates JE, Dodge KA. 1997. Supportive Rothbart M, Ahadi S. 1994. Temperament and
parenting, ecological context, and children’s the development of personality. J. Abnorm.
adjustment: a seven-year longitudinal study. Psychol. 103:55–66
Child Dev. 68:908–23 Rowe D. 1994. The Limits of Family Influence:
Plomin R. 1986. Development, Genetics, and Genes, Experience, and Behavior. New York:
Psychology. Hillsdale, NJ: Erlbaum. 372 pp. Guilford
Plomin R, Bergman CS. 1991. The nature of nur- Rowe D. 1983. A biometrical analysis of percep-
ture: genetic influences on ‘‘environmental’’ tions of family environment: a study of twin
measures. Behav. Brain Sci. 14:1–15 and singleton sibling kinship. Child Dev.
Plomin R, Chipuer HM, Neiderhiser JM. 1994. 54:416–23
Behavioral genetic evidence for the impor- Rutter M, Silberg J, O’Connor T, Simonoff E.
tance of nonshared environment. See Heth- 1999. Genetics and child psychiatry: I.
erington et al 1994, pp. 1–31 Advances in quantitative and molecular
Plomin R, Daniels D. 1987. Why are children in genetics. Child Psychol. Psychol. 40:3–
the same family so different from each other? 18.
Behav. Brain Sci. 10:1–16 Scarr S. 1992. Developmental theories for the
Plomin R, DeFries J, Fulker D. 1988. Nature and 1990’s: development and individual differ-
Nurture During Infancy and Early Childhood. ences. Child Dev. 63:1–19
Pacific Grove, CA: Brooks-Cole Scarr S, Grajek S. 1982. Similarities and differ-
Price RA, Charles MA, Pettit DJ, Knowler WC. ences among siblings. In Sibling Relation-
1993. Obesity in Pima Indians: large increases ships: Their Nature and Significance Across
in post- World War II birth cohorts. Am. J. the Lifespan, ed. ME Lamb, B Sutton-Smith,
Phys. Anthropol. 92:473–79 pp. 357–82. Hillsdale, NJ: Erlbaum
 PARENTING AND ITS EFFECTS ON CHILDREN 27

Stoolmiller M. 1999. Implications of the Adult Psychopathology, ed. MS Kesheven,

restricted range of family environments for RM Murra, pp. 104–16. Cambridge: Cam-
estimates of heritability and nonshared envi- bridge Univ. Press
ronment in behavior genetic adoption studies. Tienari P, Wynne LC, Moring J, Lahti I, Naarala
Psychol. Bull. In press M, et al. 1994. The Finnish adoptive family
Suomi SJ. 1999. A biobehavioral perspective on study of schizophrenia: implications for fam-
developmental psychopathology: excessive ily research. Br. J. Psychiatry 23(Suppl.
aggression and serotonergic dys- 164):20–26
function in monkeys. In Handbook of Devel- Turkheimer E. 1998. Heritability and biological
opmental Psychopathology, ed. AJ Samaroff, explanation. Psychol. Rev. 105:1–10
M Lewis, S Miller. New York: Plenum. In Van den Boom DC. 1994. The influence of tem-
press perament and mothering on attachment and
Suomi SJ. 1997. Long-term effects of different exploration: an experimental manipulation of
Downloaded from www.annualreviews.org. Guest (guest) IP: 180.75.250.144 On: Fri, 08 Nov 2024 04:09:10

early rearing experiences on social, emotional sensitive responsiveness among lower-class

and physiological development in non-human mothers with irritable infants. Child Dev.
primates. In Neurodevelopmental Models of 65:1457–77
Downloaded from www.annualreviews.org. Guest (guest) IP: 180.75.250.144 On: Fri, 08 Nov 2024 04:09:10