Youth Appraisals of Marital Conflict and Genetic Risk for Attention-Deficit Hyperactivity
Disorder: Examination of Gene x Environment Interactions Using Behavioral and Molecular
Genetic Methodologies
By
Molly A. Nikolas

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Psychology
2011

ABSTRACT
YOUTH APPRAISALS OF MARITAL CONFLCIT AND GENETIC RISK FOR
ATTENTION-DEFICIT HYPERACTIVITY DISORDER: EXAMINATION OF GENE X
ENVIRONMENT INTERACTIONS USING BEHAVIORAL AND MOLECULAR GENTEIC
METHODOLOGIES
By
Molly A. Nikolas

Identifying the specific etiological factors that contribute to the development of attentiondeficit hyperactivity disorder (ADHD) holds great promise for future innovations regarding the
conceptualization of the disorder as well as prevention and treatment measures. A wealth of
evidence has demonstrated that genetic factors make large contributions to ADHD, yet numerous
environmental risk factors have also been identified. Uncovering the nature of the exchange
processes that are involved in the development of ADHD via investigation of gene x
environment interactions (GxE) represents an important step forward in research involving the
causal mechanisms of the disorder.
Risk factors related to the family environment may be particularly important for the
development of behavioral and emotional regulation capabilities. In particular, conflict in the
home has emerged as an important correlate of both symptom severity and impairment for child
problems and represents a potential putative environmental risk factor for ADHD.
The current research examined the potential etiological role of children‟s cognitive
appraisals self-blame in relation to their parents‟ marital conflict in ADHD via tests of GxE
effects using two complementary methodologies: behavioral and molecular genetics. However,
prior to these tests, the phenotypic relationships among children‟s appraisals of marital conflict
and externalizing behaviors were examined. In Study 1, the unique relationships between

appraisals of self-blame and ADHD symptoms was replicated, indicating that self-blame was
related to ADHD symptoms even when oppositional and conduct are controlled.
In Study 2, behavioral genetic methods for testing GxE effects were conducted in a twin
sample of 248 twin pairs. Self-blame emerged as a significant moderator of latent genetic and
environmental influences on parent rated ADHD symptoms on the Child Behavior Checklist
DSM-IV ADHD Scale, such that genetic influences decreased but non-shared environmental
influences increased with higher reports of self-blame. In Study 3, tests of GxE effects involving
a specific genetic marker, the promoter polymorphism of the serotonin transporter gene
(5HTTLPR) were conducted in a completely independent sample of n=304 youth, of whom
n=151 had ADHD. That analysis revealed significant interactions, such that increases in teacherrated DSM-IV ADHD symptoms corresponded with increases in self-blame, but only for
individuals with the low and high serotonin-activity genotypes.
Findings from both studies are complementary and suggest that self-blame may indeed
have a specific role in the etiology of ADHD via moderating of genetic effects. Results suggest
that different exchange processes (i.e., genetic main effects, environmental main effects, GxE
interaction effects) may be differentially important for the etiology of ADHD and provide
support for an etiological role of self-blame in ADHD. More broadly, these findings offer an
innovative approach for understanding the interactional processes between genetic and
environmental risk factors and their contributions to ADHD.

Dedicated to my loving parents, Ann and Ron Nikolas, for their invaluable support and
encouragement for this and for all of my adventures…

iv

ACKNOWLEDGMENTS

I would like to thank my guidance committee, Dr. Alex Burt, Dr. Joel Nigg, Dr. Kelly
Klump, and Dr. Karen Friderici for their support and enthusiasm for this project. I also wish to
thank my fellow clinical psychology graduate students for their unwavering help and
encouragement throughout our time in graduate school.

v

TABLE OF CONTENTS
LIST OF TABLES…………………………………………………………….……………viii
LIST OF FIGURES….……………………………………………………….……………..ix
CHAPTER 1…………………………………………………………………………..…….1
BACKGROUND……………………………………………………………………1
ETIOLOGICAL STRUCTURE OF ADHD...............................................................1
THEORETICAL BASIS FOR GXE EFFECTS FOR ADHD…………………........4
SPECIFICATION OF GENETIC AND ENVIRONMENTAL RISK FACTORS....7
CANDIDATE GENE SELECTION...............................................................7
ENVIRONMENTAL CANDIDATE SELECTION.......................................9
METHODOLOGIES FOR EXAMINATION OF GXE EFFECTS FOR ADHD…11
OUTLINE OF CURRENT STUDIES.......................................................................15
CHAPTER 2
PHENOTYPIC RELATIONSIPS.........................................................................................17
INTRODUCTION.....................................................................................................17
METHOD..................................................................................................................18
RESULTS..................................................................................................................22
DISCUSSION............................................................................................................27
CHAPTER 3
TESTS OF GXE: BEHAVIORAL GENETICS....................................................................31
INTRODUCTION......................................................................................................31
METHOD...................................................................................................................37
RESULTS...................................................................................................................44
DISCUSSION ............................................................................................................49
CHAPTER 4
TESTS OF GXE: MOLECULAR GENETICS.....................................................................56
INTRODUCTION......................................................................................................56
METHOD...................................................................................................................62
RESULTS...................................................................................................................67
DISCUSSION.............................................................................................................72
CHAPTER 5
SUMMARY AND DISCUSSION.........................................................................................78
GENERAL DISCUSSION.........................................................................................78
INTEGRATION OF GXE RESULTS........................................................................83
IMPLICATIONS........................................................................................................84
LIMITATIONS...........................................................................................................89
OVERALL CONCLUSION.......................................................................................90
vi

APPENDICES: TABLES AND FIGURES...........................................................................92
REFERENCES........................................................................................................................115

vii

LIST OF TABLES

Table 1. Internal consistency estimates of four CPIC factors across the age range....................92
Table 2. Bivariate correlations among CPIC factors and externalizing behaviors:
Full sample..................................................................................................................................93
Table 3. Bivariate correlations among CPIC factors and externalizing behaviors:
Ages 5-7......................................................................................................................................94
Table 4. Bivariate correlations among CPIC factors and externalizing behaviors:
Ages 8-10....................................................................................................................................95
Table 5. Bivariate correlations among CPIC factors and externalizing behaviors:
Ages 11-13..................................................................................................................................96
Table 6. Bivariate correlations among CPIC factors and externalizing behaviors:
Ages 14-16..................................................................................................................................97
Table 7. Hierarchical linear models examining CPIC factors as predictors of ADHD, ODD, and
CD behaviors among 5-17 year old twins...................................................................................98
Table 8. Hierarchical linear models examining CPIC factors as predictors of ADHD, ODD, and
CD behaviors among 5-17 year old twins: Specificity of effects................................................99
Table 9: Intraclass correlations for mother-rated CBCL ADHD Score: Overall and by level of
CPIC self-blame...........................................................................................................................100
Table 10. Fit statistics for nested “straight” GxE models............................................................101
Table 11. Unstandardized path and moderator estimates in the best-fitting “straight” GxE
model............................................................................................................................................102
Table 12. Unstandardized path and moderator estimates for the GxE in the presence of rGE
model............................................................................................................................................103
Table 13. Demographic and descriptive statistics for ADHD cases and non-ADHD controls...106
Table 14. 5HTTLPR allele frequencies and genotypes by self-reported ethnic group………..107
Table 15. ADHD and externalizing symptoms: Main effect tests of high, intermediate, and low
activity 5HTTLPR genotypes......................................................................................................108

viii

LIST OF FIGURES
Figure 1. Unstandardized genetic (A), shared environmental (C), and non-shared environmental
variance contributions to ADHD by level of self-blame: “Straight” GxE model........................104
Figure 2. Unstandardized genetic (A), shared environmental (C), and non-shared environmental
variance contributions to ADHD by level of self-blame: GxE in the presence of rGE model…105
Figure 3. Relationship between CPIC self-blame and ADHD Rating Scale total symptom score
(teacher report) by 5HTTLPR genotype group............................................................................109
Figure 4. Relationship between CPIC self-blame and ADHD Rating Scale Inattention Symptoms
(teacher report) by 5HTTLPR genotype group............................................................................110
Figure 5. Relationship between CPIC self-blame and ADHD Rating Scale Hyperactivity
Symptoms (teacher report) by 5HTTLPR genotype group..........................................................111
Figure 6. Relationship between CPIC self-blame and Conners‟ ADHD Index score (teacher
report) by 5HTTLPR genotype group..........................................................................................112
Figure 7. Relationship between CPIC self-blame and Conners‟ Cognitive Problems (teacher
report) by 5HTTLPR genotype group..........................................................................................113
Figure 8. Relationship between CPIC self-blame and Conners‟ Hyperactivity (teacher report) by
5HTTLPR genotype group .........................................................................................................114

ix

CHAPTER 1: BACKGROUND AND RATIONALE FOR GXE INVESTIGATION
OF ADHD
Attention-deficit hyperactivity disorder (ADHD) is one of the most commonly diagnosed
disorders of childhood, affecting approximately 3-5% of school-aged children. Under DSM-IV,
ADHD is characterized as a behavioral syndrome that includes two correlated behavioral
dimensions: inattention-disorganization and hyperactivity-impulsivity (DSM-IV-TR, APA,
2000). Symptoms of the inattention dimension include difficulties with sustaining attention and
listening, forgetfulness, and disorganization while symptoms of hyperactivity include
restlessness, fidgeting, and excessive motor activity/ talking. Children with ADHD display these
characteristics to an excessive degree that is inappropriate for their age and development and
often result in a wide range of impairments across multiple settings (Barkley, 2006). While
ADHD is currently classified into three subgroups under DSM-IV (APA, 2000), taxometric work
has indicated that the disorder is likely best captured as extremes along these behavioral
dimensions as opposed to a discrete category (Haslam et al., 2006). Uncovering the etiological
pathways of ADHD has remained a top research priority as discovery of key casual mechanisms
for the disorder will likely allow for early intervention and prevention measures as well as the
development of more sophisticated treatment approaches.
Etiological Structure of ADHD
Twin and family studies have provided a wealth of converging evidence as to the degree
and type of etiological processes involved in the development of ADHD. Family studies of
ADHD have demonstrated that risk for the disorder is significantly elevated among family
members of ADHD probands, regardless of proband sex and ethnicity (Biederman et al., 1992;
Faraone et al., 2000; Samuel et al., 1999). Twin and adoption studies have estimated the genetic
1

liability for ADHD to be between 70 and 90 percent, indicating that a large portion of the
variance between individuals is due to genetic factors (Levy et al., 1997; Nadder et al., 2002;
Nikolas & Burt, in press; Sherman, McGue, & Iacono, 1997; Sprich et al., 2000; van den Oord,
Boomsma & Verhulst, 1994; Willcutt, Pennington, & DeFries, 2000). Importantly, these high
estimates of heritability have been found at both the moderate and extreme ends of the
behavioral dimensions of inattention and hyperactivity-impulsivity (Willcutt, Pennington, &
DeFries, 2000), indicating genetic effects are strongly influencing the entire range of ADHD
behaviors. Further, recent meta-analyses have indicated that genetic factors make strong
contributions to the disorder regardless of age, sex, informant, and measurement method. By
contrast, contributions from the non-shared environment are small to moderate, and the influence
of shared environmental factors is virtually nil (Bergen, Gardner, & Kendler, 2007; Burt, 2009;
Nikolas & Burt, in press).
Despite the consistent evidence of moderate to large genetic influences on ADHD, the
etiology of the disorder can also be conceptualized within a developmental psychopathology
framework as the result of exchanges between multiple risk processes occurring within and
across development (Campbell, 2000; Nigg & Nikolas, 2008). While these risk processes will
likely include genetic and biological vulnerabilities, many theorists have posited that factors
related to the family environment are also particularly influential on the development of ADHD
symptoms since the family provides the basic structure and context for the development of
behavioral regulation capacities (Campbell, 1994; Campbell et al., 1994; Johnson & Mash, 2001;
Sonuga-Barke et al., 2005). That is, a developmental psychopathology model of ADHD would
posit that some children may have a chaotic or unresponsive family environment that serves to
further enhance genetically-influenced ADHD symptoms to clinically significant levels (Carlson,
2

Jacobvitz, & Sroufe, 1995). Thus, in addition to genetic influences, components of the family
environment may also influencing the development of ADHD behaviors.
In line with this, empirical investigations have also noted many differences among
families of children with ADHD compared to families of typically-developing children. First,
numerous investigations have found that ADHD children are more likely come from low
socioecononmic status homes (Biederman et al., 1995) and are less likely to live in intact
families (Johnston & Mash, 2001). Parents of ADHD children have reported higher stress, lower
social support, lower parenting and marital satisfaction, and increased marital and parent-child
conflict (Biederman et al., 2002; Johnston & Mash, 2001; Lange et al., 2005; Pressman et al.,
2006; Whalen & Henker, 1999). Further, parents of ADHD children have been observed to be
more negative and demanding and to use less authoritative parenting compared to the parents of
control children (Buhrmester et al., 1992; Campbell et al., 1986; DuPaul et al., 2001).
In their seminal review, Johnston and Mash (2001) noted that while there are descriptive
differences among families of ADHD children when compared to their non-disordered
counterparts, there are multiple mechanisms by which characteristics of the family environment
may be related to child behavior. First, characteristics of the family environment may be directly
involved in the casual processes that give rise to ADHD (or they might exacerbate or maintain
symptoms once they develop). Alternatively, differences in the family environment may be the
consequence of having a child with ADHD (i.e., the differences in the family environment are
child-driven). In addition, the relationship among family characteristics and child behavior may
be the result of shared genes among parents and children.
Overall, the influence of family environment factors on ADHD remains difficult to
explain within the context of large and robust genetic effects and small to moderate main
3

environmental effects. Further, several potential mechanisms have been hypothesized to account
for the relationship between the family environment and ADHD. Thus, it appears that
examination of the interplay between genetic and family environmental risk processes via gene x
environment interactions may be a crucial component to understanding the transactional and
ongoing nature of risk and protective factors that may be influencing ADHD.
Theoretical Basis for GxE Effects for ADHD
GxE effects are specifically defined as genetically-modulated individual differences in
sensitivity to environmental risk factors, such that specific genetic variants exert effects on a
disorder only within particular environmental contexts (Rutter, Moffitt, & Caspi, 2006; Rutter &
Silberg, 2002). Recent reviews of theoretical issues regarding gene x environment interplay
have noted that the previous view that GxE effects were rare for psychopathology is very likely
false because of past failures to specify and measure G and/or E (Rutter, Moffitt, & Caspi, 2006).
Importantly, the authors note that these “black box” analyses were, in essence, examining a
biologically improbable omnibus interaction between all genes and all environments.
Furthermore, as GxE effects involve genetically-influenced sensitivities to specific
environmental risks, it has been argued that such sensitivities may only apply to subgroups
within the overall population (Rutter & Silberg, 2002). Thus, the importance of specifying and
measuring the G and the E is emphasized for current studies of GxE effects for psychiatric
disorders.
Rutter and colleagues (2006) also specified conditions under which GxE effects are more
likely to influence the etiology of multifactorial disorders. These include (1) the presence of
substantial environmental risk factors but also substantial heterogeneity as to an individual‟s
vulnerability for developing a particular disorder if exposed to environmental risk, (2) substantial
4

genetic risk for a disorder operating via indirect risk pathways, and (3) discordance within
monozygotic twin pairs. These three conditions are very likely to be applicable to a wide range
of illnesses and mental disorders, which nearly all involve multifactorial etiological processes.
Even so, determining that these conditions are met for ADHD remains important for
understanding the likely mechanisms involved in GxE for ADHD. Below, I consider each
condition for ADHD specifically.
With regard to the first condition, namely the existence of environmental risk factors that
result in heterogeneous outcomes with respect to ADHD, previous reports have identified
numerous candidates. These include exposure to interparental conflict (Wymbs et al., 2008),
poor parenting and family dysfunction (Biederman et al., 2002), low parental socioeconomic
status (Ornoy, 2003), maternal smoking and alcohol consumption (Banerjee et al., 2007; Knopik
et al., 2005; Thapar et al., 2003), and traumatic brain injury (Bloom et al., 2001). For each of
these risk factors, there is demonstrated risk for ADHD, yet substantial heterogeneity in terms of
overall outcome.
In reference to the second condition (i.e., substantial genetic effects operating via indirect
causal pathways), twin and family studies have estimated a large magnitude of genetic effects of
all common mental disorders of childhood and adolescence, including ADHD (Bergen, Gardner,
& Kendler, 2007; Burt, 2009). Further, endophenotype models for ADHD have posited that
genetic influences on ADHD involve changes in neural transmission in key brain regions
important for the development of cognitive control, response inhibition, and working memory,
which in turn, give rise to the behavioral symptoms of ADHD (Doyle et al., 2005). Initial work
examining genotype differences in cognitive performance have shown some positive results
(Bellgrove et al., 2005; Cornish et al., 2005; Waldman et al., 2006), suggesting that genetic
5

effects for ADHD may operate via an indirect causal pathway involving impairments in neural
circuitry underlying cognitive control and executive functioning processes.
Lastly, Rutter and colleagues (2006) also specified that GxE effects are more likely to
exert risk for a disorder if there is discordance within monozygotic twin pairs. Studies of
concordance rates among monozygotic (MZ) twins have demonstrated moderate to high
concordance rates (55-75%) among MZ twins using both DSM criteria and empirically-assigned
subclasses based on symptom profiles (Neuman et al., 2001; Sherman et al., 1997; Volk et al.,
2005). These concordance rates are rather high, and in some cases, appear to be approaching the
reliability ceiling. Despite this, important differences have been noted among identical twins
discordant for ADHD. First, a recent longitudinal investigation of MZ twin pairs discordant for
ADHD found that the affected twin had significantly lower birth weight and delayed physical
and motor maturation compared to their unaffected co-twin (Lehn et al., 2007). Secondly,
studies of neural anatomy have indicated that the volume of the caudate is significantly smaller
in affected twins when compared to their co-twin (Castellanos et al., 2003). Thus, while
concordance rates for ADHD among MZ twins appear to be moderate to high, there is some
initial evidence suggesting early environmental moderation of genetic risk for ADHD. Further,
the discovery of neural differences between discordant MZ twins suggests that environmental
factors influencing brain development play a significant role in the etiology of ADHD.
In summary, there is good evidence that both genetic and environmental factors
contribute to the development of ADHD (Thapar et al., 2005). Examination of geneenvironment interplay thus emerges as one potentially fruitful strategy for understanding
associations between ADHD and a variety of environmental risk factors within the context of
large genetic effects on the disorder. Furthermore, ADHD appears to meet the conditions set
6

forth by Rutter and colleagues (2006) for probable GxE effects, namely (1) the presence of
environmental risk factors that result in heterogeneous outcomes (e.g., interparental conflict,
familial adversity, maternal smoking and alcohol use), (2) substantial genetic effects that likely
operate through indirect causal pathways (e.g., via altered effects on neural systems involved in
executive functioning and cognitive control), and (3) discordance for ADHD among MZ twin
pairs as well as important differences among discordant MZ pairs.
Specification of Genetic and Environmental Risk Factors
As noted, the likelihood of detecting GxE effects increases when G and E contributors
are specified and measured (Rutter & Silberg, 2002). Furthermore, Moffitt and colleagues
(2005) argued for strategic, hypothesis-driven selection of both genetic and environmental
candidates in GxE studies, such that each component is rooted firmly in empirical research.
Given this, it is important to specify the genetic and environmental risk factor(s) that may be
involved as well as to review why each component may be important in the etiology of ADHD.
I will begin with consideration of candidate genes followed by a discussion of candidate
environmental factors.
Candidate Gene Selection
In terms of specific genes of influence, it is most likely that several genes, each of small
effect, contribute to ADHD (Smalley et al., 2002). In recent years, work examining molecular
genetic association with ADHD has grown exponentially, with numerous positive and negative
findings. Genes of the dopamine and noradrenergic transmission systems have received much
study due to the cognitive and behavioral abnormalities associated with catecholamine
dysfunction as well as the influence of psychostimulants such as methylphenidate (the
pharmacological treatment of choice for ADHD) on these systems (Spencer et al., 2000). Several
7

catecholamine candidate gene markers have shown replicated association with ADHD, including
the dopamine transporter gene (DAT1), and the dopamine D4 and D5 receptor genes (DRD4,
DRD5), (Brookes et al., 2006; Friedel et al., 2007; see Faraone et al., 2005).
These genes remain theoretically important candidates because of their association with
cognitive components related to ADHD (e.g., reward processing, cognitive control). However,
an integrated theoretical model of the disorder has posited that ADHD reflects deficiencies in
both behavioral and emotional regulation capabilities that are underpinned by neural circuitry
involved in both reward response as well as emotional processing (Nigg & Casey, 2006). Thus,
from this perspective, genes from additional systems, such as the serotonin neurotransmission
system, may also be relevant to ADHD.
Genes from the serotonin transmission system have also been hypothesized to play a role
in ADHD because of findings demonstrating that altered serotonergic activity is involved in
impulse control and aggressive behavior (Coccaro et al., 1989; Evans et al., 2000; Manuck et al.,
2002; O‟Keane et al., 1992, Retz et al., 2004). In line with this, several markers from multiple
genes in the serotonin system have shown association with ADHD in both case-control and
family-based analyses (Faraone et al., 2005; Hawi et al., 2002; Kent et al., 2002, Manor et al.,
2001). However, the promoter polymorphism of the serotonin transporter gene (5HTTLPR) has
emerged as a particularly important candidate in a recent meta-analysis (Gizer, Ficks, &
Waldman, 2009). Importantly, the authors also found significant heterogeneity of effects across
studies, which suggests the potential for GxE processes (Gizer, Ficks, & Waldman, 2009). In
fact, one potential explanation for the inconsistent findings and failures to replicate previous
association results is that genetic effects do vary with different levels of environmental risk. In
other words, the presence of GxE effects may be influencing the ability to detect the main effects
8

of generic factors in molecular genetic association studies.
In addition, the serotonin transporter gene itself is expressed in brain regions often
implicated in emotion regulation, memory, attention, and motor control (Frankle et al., 2004),
areas which are also relevant to ADHD (Nigg & Casey, 2006). Furthermore, the known
functionality of 5HTTLPR (i.e., the resultant decrease in transcription efficiency associated with
the “short” allele) as well as previous work demonstrating that genetic regulation of serotonin
neurotransmission is sensitive to environmental context (Bennet et al., 2002; Manuck et al.,
2004) also make it a prime candidate for GxE investigation.
Environmental Candidate Selection
As mentioned earlier, factors related to the family environment are thought to be
particularly important for the development and maintenance of ADHD symptoms within a
developmental psychopathology framework. In line with this, animal studies have shown that
exposure to chronic stress during early development may alter cortical functioning via effects on
neurotransmission (Arnsten & Goldman-Rakic, 1998). More specifically, conflict within the
family was recently been shown to be a robust predictor of ADHD symptom severity and
accounted for 40% of the variance in impairment (Pressman et al., 2006). Thus, stress within the
family environment, particularly stress related to conflict, may prove to be a critical
environmental moderator of genetic risk for ADHD.
In regard to conflict within the family environment, inter-parental conflict has been found
to be a robust predictor of child adjustment and has been associated with ADHD and
externalizing behavior problems in numerous studies (Buehler, Lange, & Franck, 2007; Counts
et al. 2005; Cummings & Davies, 1994; El-Sheikh & Harger, 2001; Gerard et al., 2005; Grych &
Fincham, 1990; Grych et al., 2000; Grych, Harold, & Miles, 2003; Jouriles et al., 2000; Skopp et
9

al., 2005; Wymbs et al., 2008). Furthermore, studies of inter-parental conflict have
demonstrated that conflict is more predictive of child externalizing problems than internalizing
problems (Cummings & Davies, 1994), suggesting some specificity of effects on youth behavior
and adjustment problems.
Two prominent hypotheses (and not necessarily mutually exclusive) have emerged to
explain how inter-parental conflict influences child externalizing problems. First, the exposure
hypothesis posits that mere exposure to negative exchanges among parents, regardless of
intensity of negative emotions or topic of conflict results in an increase in externalizing
behaviors among children (Cummings, Goeke-Morey, & Papp, 2004). This work posits that
social learning processes influence how children imitate and carry forward tactics they observe
during their parents‟ marital disputes (Davies et al., 2002). Second, more recent work has also
supported emotional-security theorists, who suggest that the topics of marital disputes are
differentially related to children‟s reactions and behaviors, such that conflicts about the child are
linked to greater behavioral dysregulation in children (Cummings et al., 2004; Cummings &
Davies, 2002; Harold et al.,1997).
In line with the view of emotional-security theorists, Grych and Fincham (1990)
convincingly argued that the child‟s perceptions and appraisals of marital conflict often play a
critical determining role as to the effect of marital conflict on youth behavior problems. They
advocated strongly for the assessment of inter-parental conflict from the child‟s perspective as
youth reports of inter-parental conflict have been shown to be more predictive of externalizing
behavior problems than the parents‟ own report of their own marital conflict and satisfaction
(Cummings, Davies, & Simpson, 1994; Grych, Seid, & Fincham, 1992). The development of the
Children‟s Perception of Interparental Conflict scale (CPIC) provided an important measurement
10

tool for quantifying youth perception and appraisals of the marital disputes of their parents‟
(Grych, Seid, & Fincham, 1992). The authors proposed nine subscales reflecting perceptions of
conflict frequency, intensity, content, resolution, and triangulation. However, in a recent factor
analysis of the CPIC in over 1000 children, four scales reliably emerged, reflecting youth
perceptions of conflict frequency and intensity (i.e., properties of conflict),
triangulation/stability, self-blame, and perceived threat (Nigg et al., 2009). Moreover, when
these CPIC scales were examined together with other family environment variables (SES,
familial cohesion, maternal depression), children‟s appraisals of self-blame in relation to their
parents‟ marital conflict remained a robust predictor of ADHD symptoms (Miller et al., in
preparation), indicating that self-blame may be particularly important for ADHD. Importantly,
however, the psychometric properties of these newly-identified CPIC scales as well as the
specificity of the relationships among the scales and with ADHD and other externalizing
behaviors remains to be tested in additional samples.
Overall, the promoter polymorphism of the serotonin transporter gene (5HTTLPR)
appears to be a good candidate for GxE effects because of its involvement in emotional
regulation processes and expression in key brain regions. Furthermore, inter-parental conflict
appears to be relevant for the development of externalizing problems and ADHD. Moreover,
there is evidence that this association may be partially dependent on the topic of marital conflict
(Cummings et al., 2004), as well as on youth perceptions and cognitive appraisals of their
parents‟ conflict. Given all this, testing for interactions involving 5HTTLPR and cognitive
appraisals of self-blame appears to be well-grounded empirically and offers a biologicallyplausible mechanisms that may be operating in the etiology of ADHD .
Methodologies for Examination of GxE Effects for ADHD
11

The next critical consideration for examination of GxE effects in the relationship between
youth appraisals of self-blame and ADHD involves the selection of methodologies by which to
examine potential differences in genetic effects. While previous work involving twin designs
and molecular genetic studies of ADHD have typically involved separate lines of investigation
(e.g., quantifying the magnitude of omnibus genetic effects versus examination of association of
specific genetic markers, as explained below in detail), the use of the two methodological
approaches represents a particularly powerful way to examine possible GxE. Because of scaling
effects of both the environmental mediator and the outcome in question, several investigators
have warned against potential false-positives in GxE designs (Eaves, 2006; Moffitt et al., 2005;
Rutter, Moffitt, & Caspi, 2006). Thus, examination of potential G x self-blame interactions for
ADHD using both behavioral genetic and molecular genetic methodology allows for some
protection against false-positive results. Moreover, initial examination of any impact of the
environmental risk factor on the latent genetic and environmental variance on a trait using
behavioral genetic methodology provides a type of “omnibus” test to determine if GxE effects
are likely with a particular environmental moderator. Molecular genetic techniques can then
specify particular genetic markers that may be relevant to the disorder and examine whether or
not association between the genetic marker and the trait changes with different levels of the
environmental moderator. The current project will take advantage of the complementary nature
of these two GxE methodologies to evaluate whether youth appraisals of self-blame in relation to
their parents‟ marital conflict moderates genetic risk effects for ADHD.
Behavioral Genetic Approaches. Historically, twin studies have focused upon
quantifying the degree of genetic, shared, and non-shared environmental influence on a particular
disorder. Importantly, however, more recent research has suggested that, contrary to earlier
12

thinking, these influences are necessarily composed solely of main effects. Instead, they likely
contain GxE as well. In particular, Purcell (2002) reasoned that genetic (A) x shared
environment (C) interaction loads on A based on the following: shared environmental effects
result in similar sibling correlations across different levels of genetic relatedness. Interactions
between A and C would thus contribute differentially to sibling similarities with degree of
genetic relatedness. Because MZ twins share a greater proportion of their genes than do DZ
twins (100% versus 50%), AxC interactions would result in greater MZ than DZ correlations,
and would thus load on A term in traditional behavioral genetic modeling. By contrast, because
non-shared environmental factors (E) suppress both MZ and DZ twin correlations, AxE
interactions would not contribute to similarities in twins, but to variance that is unique across
sibling pairs regardless of degree of genetic relatedness. AxE interactions would thus emerge as
part of the E term in behavior genetic modeling.
In addition to this important theoretical work, Purcell (2002) also described a
methodology for entering a measured environmental variable (M) in a means model, so as to
estimate latent genetic and environmental path coefficients as a function of a particular
environmental variable. Moreover, Purcell‟s (2002) model allowed researchers to examine these
latent GxE in the presence of a genetic correlation between the moderator and the trait (referred
to as a gene-environment correlation, or rGE). This was a crucial analytical advance. If rGE
effects are not considered, it becomes impossible to determine if the multiplicative effect in
question is due to an actual interaction of the two independent sources of variance (i.e. G and E),
or if it stems from an underlying genetic relationship between G and E. For example, several
studies have reported GxE interactions between genetic factors and stressful life events for
symptoms of depression, such that particular genetic markers show greater association with
13

depression for individuals with higher levels of stressful life events (Caspi., et al., 2003; Laucht
et al., 2009; Taylor et al., 2006). Critically, however, recent work has also noted that genetic
factors also contribute to the occurrence of stressful life events (see Kendler & Baker, 2007).
Given this, it is certainly possible that the genetic factors that influence stressful life events also
influence the development of depressive symptoms, and thus what appears to be GxE is in fact
rGE “in disguise”. In short, rGE can resemble GxE, and thus our ability to account for rGE
when examining GxE addresses a major confound in GxE research.
Molecular Genetic Approaches. Over the past 10-15 years, there have been remarkable
advances in the area of genomics, which has allowed for direct tests of effects between variations
in genetic markers and complex traits, including ADHD. Recent GxE work for ADHD has
examined differential relationships between putative environmental risk factors and ADHD
across different genotype groups. Several environmental variables have been examined,
including parental marital status (Waldman, 2007), prenatal exposure to tobacco and alcohol
(Langley et al., 2008; Neuman & Todd, 2007), and low birth-weight (Thapar et al., 2005). In
these designs, the presence of a particular allele in combination with increased environmental
risk significantly predicted ADHD diagnostic status and symptom severity, an effect that was
detected via straightforward moderated regression analyses. Thus, one potentially fruitful model
for examining whether youth appraisals of marital conflict moderates genetic risk on ADHD is
to directly examine their relationship across different genotype groups.
As mentioned earlier, while catecholamine genes have been frequently examined in
molecular genetic association studies of ADHD, the role of genes in the serotonin transmission
system have also been posited to influence the development of ADHD as dysregulated
serotonergic function has been implicated in impulsive and aggressive behavior in studies of
14

animals, adults, and children (Coccaro et al., 1989; Evans et al., 2000; Halperin et al., 1994;
Halperin et al., 1997; Manuck et al., 2002; Schulz et al., 2002). Moreover, the functional 44-bp
promoter polymorphism of the serotonin transporter gene (5HTTLPR) has been associated with
ADHD as well as comorbid phenotypes, including persistent aggression and conduct disorder
(Kent et al., 2002; Manor et al., 2001; Sakai et al., 2006). 5HTTLPR presents as an ideal
candidate for examination of GxE effects for ADHD for several reasons. These include its
functionality, its sensitivity to environmental regulation, and its expression in key brain regions
often implicated in ADHD.
Outline of Current Studies
The aim of the current set of studies is to investigate the potential etiological role of
appraisals of self-blame related to their parents‟ marital conflict in ADHD. Importantly, the
investigation will implement two concurrent lines of investigation. First, I will make use of a
behavioral genetic approach in which latent G (and environmental) effects for ADHD are
allowed to be moderated by youth reports of self-blame (measured E). In this way, I can
determine whether self-blame generally moderates genetic (and environmental) influences on
ADHD. Furthermore, while self-blame is likely influenced at least in part by constitutional
factors, these analyses will aid in determining whether self-blame influences ADHD at the
family-wide level (i.e., via shared environmental factors), at the child-specific level (I.e., via
unique environmental factors), or perhaps at both levels. Second, I will make use of a molecular
genetic approach to examine potential interactions between a measured G (i.e. 5HTTLPR
genotype) and the same measured E (youth appraisals of self-blame). In this way, I hope to
identify one of the genes interacting with self-blame in the prediction of ADHD (as indicated in
the behavioral genetic GxE study). These complimentary lines of investigation thus allow for a
15

strong test of GxE effects in ADHD and thereby strengthen the ability to accurately detect any
potentially real interactions.
Importantly, however, the phenotypic relationships among the newly-identified scales of
the CPIC (see Nigg et al., 2009) as well as their relationships with ADHD and other
externalizing behaviors have not been well-examined. Thus, prior to the main tests of GxE
effects outlined above, I will first examine relationships among all of the CPIC scales and
externalizing disorders in order to accomplish three key goals: (1) to examine the psychometric
properties of these scales both in general and across a wide age range, (2) to quantify the strength
of relationships among the CPIC scales to determine if higher reports of conflict correspond with
higher reports of self-blame, and (3) to determine if relationships among the CPIC scales are
specific to ADHD or if they related to externalizing behaviors more generally.
The current project will thus include three studies. Study 1 will involve examination of
the phenotypic relationships among scales of the CPIC as well as their relationships with
externalizing behaviors within a community-based twin sample. Next, Study 2 will provide the
first tests of GxE effects by examining self-blame as a moderator of latent genetic and
environmental influences on ADHD. Study 3 will then examine interactions between a specific
genetic marker (i.e., 5HTTLPR) and youth appraisals of self-blame within a clinically-diagnosed
sample of ADHD cases and non-ADHD control youth. Following presentation of these three
empirical studies will be an integrative summary and discussion of findings as well as potential
implications for future work involving investigation of GxE effects on ADHD.

16

CHAPTER 2: PHENOTYPIC RELATIONSHIPS AMONG CPIC FACTORS AND
EXTERNALIZING BEHAVIORS
Developmental and family process research has consistently pointed to a robust
relationship between marital conflict and child adjustment and behavior problems (Buehler et al.,
1997; Grych & Fincham, 2001; Rhoades, 2008). Importantly, prior work has also demonstrated
that it is not mere exposure to conflict that contributes to later behavioral problems, as some
children exposed to inter-parental conflict do not experience behavioral and adjustment
difficulties (Jouriles, Murphy, O‟Leary, 1989). Rather, several theorists have agreed that
additional proximal processes are responsible for the link between interparental conflict and
youth behavior problems (Cummings & Davies, 1994; Grych & Fincham, 1990).
Children‟s responses to interparental conflict are a potential set of proximal processes
that likely influence the relationship between conflict exposure and later behavior problems.
Children‟s reactions to marital conflict, including cognitive appraisals, affective regulation,
behavioral responses, and physiological reactivity have all been demonstrated to mediate the
association between exposure to conflict and resultant internalizing and externalizing problems
in children and adolescents (Rhoades, 2008). Of these, cognitive appraisals have emerged as
particularly strong predictors of child adjustment and behavior problems (Harold et al., 1997;
Grych et al., 2000; Rhoades, 2008). That said, several questions remain regarding relationships
among the various components of child response to martial conflict as well as their associations
with behavioral phenotypes. These include the relationships between various aspects of conflict
severity and youth cognitive appraisals as well as the reliability and validity of the measures of
youth appraisals related to martial conflict.

17

The purpose of the following set of analyses was to examine the phenotypic associations
of these various components of child response to marital conflict with child behavior problems.
Youth perceptions were assessed via the Children‟s Perception of Interparental Conflict scale
(CPIC; Grych, Seid, & Fincham, 1992), which was designed to specifically tap into youth
perception of conflict based on the cognitive-contextual framework hypothesis (Grych &
Fincham, 1990). The original CPIC scale was constructed to assess nine different dimensions
related to interparental conflict, which included both objective reports on conflict (frequency,
intensity) and cognitive appraisals related to conflict (e.g., self-blame, coping efficacy). Despite
the potential theoretical relevance of the original nine dimensions of the CPIC, recent factor
analytic work (which included data from the sample examined herein) has demonstrated that the
item-level data on the CPIC was best represented by four factors (Nigg, Nikolas, Miller, Burt,
Klump, & von Eye, 2009), at least in samples like the one being studied here. Analyses
primarily focused upon the psychometric properties of the four CPIC scales across the age range,
their inter-relationships, and the specificity of their associations with youth behavior problems.
METHOD
Participants
Participants were child and adolescent twin pairs and their primary caregiver who were
assessed as part of the Michigan State University Twin Registry (MSUTR), an ongoing project
examining genetic and environmental contributions to both internalizing and externalizing
psychopathology (Klump & Burt, 2006). Families were recruited through the use of state birth
records in collaboration with the Michigan Department of Community Health (MDCH) and the
Michigan Bureau of Integration, Information, and Planning Services (MBIIP; for a full
description of recruitment procedures for the MSUTR, see Klump & Burt, 2006). Parents gave
18

informed consent for both themselves and their children and children provided informed assent.
All research protocol was approved by the Michigan State University Institutional Review
Board.
The current sample consisted of 214 child and adolescent monozygotic (MZ) and
dizygotic (DZ) twin pairs (total n= 428 twins). This sample was composed of 107 MZ twin pairs
(52 male-male, 54 female-female) and 108 DZ twin pairs (59 male-male, 49 female-female) that
ranged in age from 5-16 years (M=10.5, SD=2.6 years). Participating families in the MSUTR
were notably representative of individuals living in the mid-Michigan region in terms of racial
identification (see Culbert et al., 2008); approximately 88% self-identified as Caucasian, 8%
identified as African-American, 3% identified as Latino, and 1% identified as Asian-American.
Perceptions of Marital Conflict
Perceptions and appraisals of marital conflict were assessed with the CPIC (Grych, Seid,
& Fincham, 1992). Each twin completed a separate CPIC during the computerized assessment at
the MSU laboratory. The 48 CPIC items were rated by participating twins on a three-point scale
(1-3: true, sort of true, and false). Because the CPIC was designed to be completed by schoolaged children, the questionnaire was read to participating twins with reading levels under 5th
grade in order to ensure correct comprehension and completion of the items. Based on
exploratory and confirmatory analysis of the 48 items (Nigg et al., 2009), four empirically
derived CPIC scale scores were computed. These scales included Conflict Properties,
Triangulation/Stability, Self-Blame, and Threat (described below).
Items loading on the Conflict Properties scale (n=11 items, α = .82) assessed the
perceived frequency and intensity of the observed conflict. Sample items from this scale include
“My parents get really mad when they argue” and “My parents hardly ever argue.” The
19

Triangulation/Stability scale (n=13 items, α = .88) required children to report on the extent to
which they feel caught in the middle of conflict as well as if their parents‟ conflict is an
entrenched and enduring part of family life. Representative items from this scale include “I feel
like I have to take sides when my parents argue” and “Even after my parents argue, they stay
mad at each other.” The Self-Blame and Threat scales assessed cognitive appraisals regarding
inter-parental conflict. Items on the Self-Blame scale (n=9 items, α = .85) assessed the extent to
which children blame themselves for conflict they have observed between their parents. Sample
items from the CPIC self-blame scale include “My parents usually argue about something that I
do”; “It is usually my fault when my parents argue”; and “I am to blame when my parents
argue.” Lastly, items loading on the Threat scale (n=6 items, α = .84) assessed youth perceptions
regarding the negative implications their parents‟ marital conflict may have for them. Sample
items on this scale include “When my parents argue, I worry about what will happen to me” and
“I get scared when my parents argue.”
Child Externalizing Behaviors and ADHD
Mothers of the twin participants completed the Child Behavior Checklist (CBCL;
Achenbach & Rescorla, 2001) to assess child externalizing behaviors. The CBCL is a
standardized questionnaire for parents to report the frequency of 118 problem behaviors
exhibited by their children during the past six months. Mothers rated how often particular
behaviors occurred for each twin using a 3-point Likert scale (0=never true;
1=sometimes/somewhat true; 2=often true). These 118 behaviors were then summed into the 8
empirically-validated syndrome scales and 6 DSM oriented scales. The DSM oriented scales for
ADHD, ODD, and CD were selected for analyses as the items most closely map onto DSM-IV
criteria for externalizing psychopathology. Note that these scales do not represent clinical
20

diagnoses of externalizing disorders but rather encompass ratings of these behaviors. The DSMoriented scales have demonstrated good reliability and validity in terms of predicting clinically
significant levels of externalizing disorders (Hudziak et al., 2004). In this sample, internal
consistency estimates were adequate for all three scales (ADHD scale α=.88; ODD scale α=.86;
CD scale α=.89). While age and sex based norms are available for CBCL scores, raw scale
scores were used in the current study, as recommended by the authors for population/community
studies to avoid potential age and gender confounds (Achenbach & Rescorla, 2001).
Data Analysis
The current analyses aimed to address three primary questions regarding the properties of
the CPIC scale within a community-based twin sample. First, as noted in Nigg et al. (2009), the
age-invariance factor analytic model provided some evidence indicating that the CPIC factors
may vary in their psychometric properties and relationship across the age-span. This is
particularly important for application of the CPIC four-factor solution to this (and any other)
large sample that includes a wide age range of children. Thus, internal consistency estimates
were examined across the age range of the sample. The sample was divided into four age ranges:
5-7 years (n=66 twins), 8-10 years (n=162 twins), 11-13 years (n=136 twins), and 14-16 years
(n=64 twins) and alpha was calculated for each factor at each age range.
Secondly, the analyses set out to examine the inter-relationships among the four CPIC
factors within the sample. The particular question guiding these analyses centered on whether or
not children reporting higher objective levels of conflict (e.g., frequency) also reported more
negative cognitive appraisals of conflict (e.g., higher self-blame and higher perceived threat).
Thus, bivariate correlations among the four CPIC scales were examined in order to assess the

21

nature and strength of the relationships among the scales. These associations were also
examined at different age ranges to further test the possibility of age-invariance of the CPIC.
Lastly, the relationships among the four CPIC scales and measures of youth behavior
problems (e.g., externalizing behaviors) were examined. Hierarchical linear models (HLM) were
used to examine differences in relationships among the four CPIC scales and measures of
externalizing behaviors in twins. Hierarchical linear models are an extension of general linear
models in which factors are assumed to have a linear relationship with the dependent variable.
HLM analyses were selected for use with this sample because the non-independence of the twin
data can be accounted for by nesting a level 1 variable (individual twins) within a level 2
variable unit (families). As recommended (Kenny, Kashy, & Cook, 2006), categorical predictors
(e.g., gender) were effect coded (1 for males, -1 for females) and continuous variables (e.g.,
CPIC factors, age) were centered at zero. Models were first run with each of the four CPIC
factor scales as predictor variables (ADHD, ODD, and CD scale scores each served as outcome
variables). Next, in order to test for specificity of effects, HLM analyses were re-run while
controlling for the overlap between the externalizing behaviors (e.g., ODD and CD behaviors
were entered as predictors when the ADHD score served as the outcome variable).
RESULTS
Psychometric Properties of CPIC Factors
As noted in the methods section, the internal consistency estimates for each of the four
CPIC factors within the overall sample were adequate (Conflict Properties alpha=.82;
Triangulation/Stability, alpha=.88; Self-Blame, alpha=.85; Threat, alpha=.84). Examination of
internal consistency estimates across the age range of the sample revealed that, in general, all
scales demonstrated adequate reliability at all ages (see Table 1). However, the internal
22

consistency of .66 for Conflict Properties among children ages 5-7 years suggests that even with
reading assistance, young children may not be as reliable in answering questions regarding
conflict characteristics such as frequency and intensity. Despite this, estimates of internal
consistency were adequate in this age range for the other three scales (alphas range from .77.84). Furthermore (and as expected), reliability estimates generally improved slightly with age,
indicating that older youth may be somewhat more reliable reporters of both objective aspects of
interparental conflict as well as cognitive appraisals of self-blame and threat.
Inter-relationships among CPIC Factors
Full Sample. Table 2 shows the bivariate correlations among the four CPIC factors. As
seen there, the four CPIC scales generally demonstrated moderate to strong and significant
relationships with each other. The relationship between Conflict Properties and
Triangulation/Stability appears to be especially strong (r=.59, p<.01), a finding that is not
surprising given that both scales are hypothesized to tap more “objective” reports of conflict
(e.g., frequency, intensity, stability). The relationship between Conflict Properties and cognitive
appraisals of conflict, including Self-Blame (r=.21) and Threat (r=.40) were significant, but
smaller in magnitude. Importantly, inter-scale correlations with Self-Blame in particular appear
to be somewhat attenuated (although still significant) when compared to inter-correlations
among the rest of the scales (e.g., Self-Blame correlations with other scales range from .21-.36 as
compared to .40-.59 across the other scales). Indeed, correlations with Self-Blame were
significantly smaller than the inter-correlations among the other scales (Fisher z, all p < .04).
Overall, these relationships suggest some commonalities among the four CPIC scales.
However, statistical comparisons of the correlations between the scales revealed that the
relationships between each of the scales and Self-Blame was significantly weaker than the inter23

relationships among the remaining scales. This pattern of results suggests that while there is
overlap among the scales, the frequency, intensity, and stability of marital conflict do not
necessarily correspond with self-blame regarding the conflict.
Age. Bivariate correlations were also examined for each age range (e.g., ages 5-7, 8-10,
11-13, 14-16; see Tables 3-6). The overall pattern of relationships remained largely the same for
all groups. The relationship between Conflict Properties and Triangulation/Stability was robust
across the age range (r‟s ranged from .41-.77). Furthermore, correlations between Self-Blame
and other scales were again somewhat attenuated in both younger and older children (rs range
between .03-.35), with the exception of the association with Triangulation/Stability in younger
children (r=.57 for children ages 5-7). Examination of differences among the pattern of intercorrelations revealed that the relationship between Conflict-Properties and Self-Blame was
significantly weaker than relationships among Conflict Properties and the other two factors for
children in the three youngest age ranges (all ps <.05, two tailed). For youth ages 14-16,
however, the inter-correlations among all 4 CPIC factors were, for the most part, moderate in
nature. In general, however, I would conclude that the overall pattern of inter-relations among
the CPIC factor scales appeared to persist across the age range of the sample.
CPIC Factors and Externalizing Behaviors
Full Sample. Next, bivariate correlations were computed to examine relationships
between each of the four CPIC factors and parent-reports of child externalizing behavior (i.e.,
ADHD, ODD, and CD, see Table 2). The four scales demonstrated different patterns of
relationships with externalizing behaviors: Conflict Properties was significantly related to all
measures of externalizing behaviors (ADHD, ODD, and CD scores), however correlation
coefficients were rather small (rs range .10-.17). Triangulation/Stability was also significantly
24

related to all three measures and correlations were a bit larger in magnitude (rs range from .15.23). The Self-Blame scale showed the largest correlations of any of the CPIC scales with
ADHD, ODD, and CD behaviors (rs range from .16-.28). By contrast, the Threat scale showed
no relationship with any of the measures of externalizing behaviors (rs range from -.06 to .04).
However, tests of significant differences yielded few significant differences in the strength of the
relationships between the CPIC factors and ADHD, ODD, and CD scores. Overall, the
correlations generally pointed to significant and positive relationships between the CPIC factors
and ADHD, ODD, and CD behaviors (outside of the Threat scale which showed no significant
relationship with any of the behavioral measures).
Correlations were further examined across the age range to evaluate whether these
relationships were specific to any particular developmental stage (Tables 3-6). As in the overall
results, the Threat scale was not significantly related to measures of ADHD and externalizing
behaviors at any of the age ranges. Next, the relationship between Conflict Properties and the
ADHD score remained modest throughout the age range, whereas the relationships between this
scale and ODD and CD each became somewhat stronger in older children (although this effect
was primarily driven by the negative associations with CD and ODD behaviors in the youngest
cohort). Relationships between the Triangulation/Stability and the ODD and CD scores also
significantly increased among older youth (ages 14-16) relative to the younger age groups
(ps<.03, two tailed). This may be a function of the improved reliability within this age range. In
contrast, the relationship between Triangulation/Stability and ADHD did not show significant
changes across the age-range. Similarly, the Self-Blame scale showed positive and significant
relationships with both ADHD and CD behaviors across all ranges with no significant
differences in the strength of the correlations by age. The relationship between Self-Blame and
25

ODD behaviors showed less of a discernable pattern and was only significant for children ages
5-7 and for those ages 11-13.
CPIC Factors as Predictors of Externalizing Behaviors
Next, HLM analyses were conducted in order to assess the potential specificity of the
relationships between the CPIC factors and ADHD, ODD, and CD behaviors (see Table 7). As
mentioned earlier, HLM analyses are preferred here in order to account for the non-independence
of twin data. All four CPIC scales were entered as predictors and each externalizing measure
served as a separate outcome variable. Age and gender were included as covariates. Results
revealed that when all scales were entered as predictors, both the Triangulation/Stability (b=.08,
p=.02) and Self-Blame (b=.19, p<.001) scales remained significantly related to ADHD behaviors
(associations with the other CPIC scales were no longer significant). These same two scales also
emerged as significant predictors of ODD behaviors (Triangulation/Stability, b=.06, p=.03; SelfBlame, b=.08, p=.02), and CD behaviors (Triangulation/Stability, b=.10, p=.004; Self-Blame,
b=.23, p<.001). Interestingly, the Threat scale demonstrated significant negative relationships
with ODD and CD. This is may be due to a suppression effect. Overall, Triangulation/Stability
and Self-Blame appeared to show relationships with all measures of externalizing behaviors.
Specificity of Relationships
In order to assess the specificity of the relationships between Triangulation/Stability and
Self-Blame with externalizing behavior, the HLM analyses were conducted again (see Table 8).
As before, the CPIC factor scales were entered as predictors into the model and the ADHD,
ODD, and CD scores again each served as outcome variables. However in these models, the
other externalizing behaviors were also entered as predictors in order to parse out the shared

26

variance among ADHD, ODD, and CD scores. For example, when ADHD behaviors served as
the outcome measure, both the ODD and CD scores were entered as predictors.
For ADHD, results revealed that only the Self-Blame scale remained a significant
predictor (b=.08, p=.05) after controlling for the overlap with the other externalizing behaviors.
None of the CPIC factors remained significant predictors of ODD behaviors after accounting for
its overlap with ADHD and CD (all ps >.13). Lastly for CD, Self-Blame and Threat remained
significant predictors after the shared variance with ADHD and ODD was accounted for (b=.12,
p=.009 and b=-.09, p=.03, respectively).
Such results suggest that while Conflict Properties was moderately correlated with each
measure of externalizing psychopathology, it was not a unique predictor of ADHD, ODD, or CD
behaviors. The Threat scale was not correlated with any of the externalizing behaviors and was
not a significant predictor of any of the externalizing measures, with the exception of CD
behaviors. Somewhat surprisingly however, the relationship between Threat and CD behaviors
was negative, indicating that higher reports of Threat were corresponded with lower CD scores.
There appeared to be a relationship between Triangulation/Stability and externalizing behaviors
in general, however, this scale was not specifically related to any one measure of
psychopathology. Importantly, the Self-Blame scale was revealed to be a robust predictor of
externalizing behaviors in general and ADHD and CD scores in particular.
DISCUSSION
The CPIC was originally developed to examine the child‟s perspective of interparental
conflict as a unique method of measuring the potential impact of marital discord on child
behavior problems and adjustment. Recent methodological work by our group has suggested
that the measure may be best captured with four factors: Conflict Properties,
27

Triangulation/Stability, Self-Blame, and Threat (Nigg et al., 2009). Conflict Properties and
Triangulation/Stability have been hypothesized to tap into more objective aspects of interparental
conflict (e.g., frequency, intensity, and stability), whereas Self-Blame and Threat are posited to
measure children‟s cognitive (and partially emotional) appraisals regarding marital conflict.
These cognitive appraisals have emerged as a particularly robust predictor of both internalizing
and externalizing problems (Rhoades, 2008) and have been proposed a potential moderator of
etiological influences on ADHD.
The purpose of these analyses was to address three primary questions regarding the
phenotypic relationships of the CPIC‟s four factors. First, confirmatory factor analytic models
reported by Nigg et al (2009) suggested potential age-invariance of the CPIC scales. Despite
this, all four CPIC scales demonstrated adequate internal consistency in the full sample of 5-16
year old twins as well as across four different age ranges (5-7 years, 8-10 years, 11-13 years, 1416 years). The only notable exception that emerged was the low reliability estimate (α=.66) for
the Conflict Properties scale among 5-7 year olds. Overall then, it appeared that youth of all
ages in the sample were, in general, reliable reporters on the four CPIC scales.
Secondly, the analyses aimed to examine the relationships among the four CPIC scales in
order to determine if higher reports of objective aspects of the conflict (e.g., Conflict Properties)
corresponded with more negative cognitive and emotional appraisals of conflict (e.g., higher
scores of Self-Blame and/or Threat). Results revealed a general pattern of moderate and positive
associations among the 4 CPIC scales that persisted across the age-range. Specifically, the
correlation between Conflict Properties and Self-Blame was significantly lower than correlations
involving Conflict Properties, Triangulation-Stability, and Threat, a pattern which persisted for
each age group. Overall, while the relationship among Conflict Properties and Self-Blame was
28

positive, the low correlation indicated that the frequency and intensity of interparental conflict do
not map directly onto child reports of self-blame regarding the conflicts. In other words, the
Self-Blame evidences only a modest association with the level of conflict frequency and
intensity.
Lastly, the phenotypic analyses aimed to examine the four CPIC scales as predictors of
maternal-report of ADHD, ODD, and CD behaviors. Furthermore, I aimed to examine any
potential specificity of the relationships among the four CPIC scales and measures of
externalizing behaviors. Results indicated that the Triangulation/Stability and Self-Blame scales
significantly predicted ADHD, ODD and CD behaviors. Yet, when the covariance among the
externalizing behaviors was taken into account, Self-Blame remained the only significant and
unique predictor of both ADHD and CD. None of the CPIC scales appeared to be uniquely
related to ODD.
Taken together, these results suggest that cognitive appraisals of self-blame in regard to
marital discord are acting as a significant predictor of both ADHD and CD. Furthermore, the
level of self-blame appears to be partially independent of the level of overall martial conflict
(i.e., the frequency, intensity, or stability of the conflict). These results indicate that self-blame
in regard to parents‟ martial conflict may have a unique relationship with measures of
externalizing behavior. However, while these analyses have demonstrated a significant and
unique relationship among self-blame and ADHD (as well as CD), additional work is necessary
in order to determine the etiological role of self-blame in these behaviors (if any). For example,
longitudinal work may be useful in understanding the predictive nature of youth appraisals
related to martial conflict and the onset (or maintenance) of externalizing behavior problems.
Furthermore, having a child with behavior problems may in fact result in increased marital
29

conflict. Children may then (perhaps correctly) observe conflict that is related to their behavior,
resulting in increased reports of self-blame.
The unique role of self-blame as a predictor of both ADHD and CD behaviors may be
somewhat unexpected under social learning theory, which would posit that mere exposure to
higher levels of conflict would correspond to higher levels of behavior problems. That said,
under the cognitive-contextual framework (Grych & Fincham, 1990), the child‟s interpretation of
the conflict, particularly their role in the conflict, often plays a determining role as to the impact
of marital discord on child behavior problems. These results support the latter view and indicate
that negative appraisals of self-blame correspond with an increase in inattention and
hyperactivity as well as aggressive behaviors.
While the exact mechanisms remain unknown, recent work investigating the LHPA axis
(limbic-hypothalamic-pituitary-adrenal axis via cortisol levels), a system believed to be
intimately involved in the stress response, have indicated that children‟s observed distress level,
rather than just exposure to conflict, is related to an increase in cortisol (Davies et al., 2008).
Further, additional research has indicated that children‟s internal representations of family
relationships mediated the relationship between exposure to conflict and attention problems
specifically (Sturge-Apple et al., 2008). Thus, children‟s cognitions regarding marital conflict
may be playing an important etiological role in the development of behavioral and emotional
regulatory capacities, both of which are often impaired in youth with externalizing problems
such as ADHD and CD. A formal test of a potential etiological role for self-blame in ADHD
could thus provide additional insight regarding the role of environmental variables in the causal
pathways that underpin the disorder.

30

CHAPTER 3: THE IMPACT OF YOUTH APPRAISALS OF MARTIAL CONFLCIT ON
GENETIC AND ENVIRONMENTAL CONTRIBUTIONS TO ATTENTION-DEFICIT
HYPERACTIVITY DISORDER: EXAMINATION OF GXE EFFECTS IN A TWIN
SAMPLE
Attention-deficit hyperactivity disorder (ADHD) is one of the most commonly diagnosed
psychiatric disorders of childhood, affecting approximately 3-5% of school-aged children. The
disorder is characterized by excessive and developmentally inappropriate symptoms of
inattention, hyperactivity, and impulsivity that impair functioning across multiple settings and
within a variety of domains (Barkley, 2006, Nigg, 2006). While the diagnostic criteria for
ADHD continue to evolve, taxometric analyses have revealed that ADHD may be best
understood as extremes along a behavioral dimension as opposed to a discrete category (Haslam
et al., 2006).
Research involving the etiology of ADHD has provided strong evidence of genetic
influence, with heritability estimates ranging between 70-90% (Bergen, Gardner, & Kendler,
2007; Burt, 2009). Importantly, these very high levels of heritability have been found at both the
moderate and extreme ends of the behavioral dimensions of inattention and hyperactivityimpulsivity (Willcutt, Pennington, & DeFries, 2000). In contrast to these high estimates of
genetic influence, contributions from the environment appear to be much smaller, only
accounting for 10-30% of the variance in ADHD (Burt, 2009). Multiple behavioral genetic
investigations and meta-analyses have consistently observed no contribution from shared
environmental factors (Bergen, Gardner, & Kendler, 2007; Burt, 2009, Nikolas & Burt, in press).

31

Thus, these environmental influences appear to be exclusively non-shared environmental in
origin.
Because the non-shared environment estimate also includes measurement error and its
overall contribution to the variance in ADHD is small, it is perhaps not surprising that far less
attention has been given to identifying the unique environmental effects that may be operating on
ADHD. Even so, there is compelling evidence that non-shared environmental factors are
important in the development of the disorder and likely reflect more than measurement error.
For example, work examining identical twin pairs discordant for ADHD has revealed important
differences between the affected and unaffected siblings. First, a recent longitudinal
investigation found that the affected twin had significantly lower birth weight and delayed
physical and motor maturation compared to their unaffected co-twin (Lehn et al., 2007).
Secondly, studies of neural anatomy have indicated that the volume of the caudate is
significantly smaller in affected twins when compared to their co-twin (Castellanos et al., 2003).
Thus, while high genetic effects are consistently reported for ADHD, environmental factors also
play an important role in the etiology of ADHD.
Developmental research has also demonstrated the importance of environmental factors
to the etiology of ADHD (Biederman, Faraone, & Monuteaux, 2003; Johnston & Mash, 2001).
Specifically, interparental conflict has been shown to be a robust predictor of child adjustment
and behavior problems, including externalizing behaviors (Buehler, Lange, & Franck, 2007;
Counts et al. 2005; Cummings & Davies, 1994; El-Sheikh & Harger, 2001; Gerard et al., 2005;
Grych & Fincham, 1990; Grych et al., 2000; Grych, Harold, & Miles, 2003; Jouriles et al., 2000;
Skopp et al., 2005) and ADHD (Wymbs et al., 2008). One prominent theory for the mechanism
underlying this relationship posits that the topics of marital disputes are differentially related to
32

children‟s reactions and behaviors, such that conflicts about the child and his/her behavior are
linked to greater behavioral dysregulation in the child than are other sources of martial conflict
(e.g., financial concerns; Cummings et al., 2004; Cummings et al., 2002; Harold et al., 2004).
Building on this latter line of research, it has also been suggested that youth perceptions
and appraisals of marital conflict play a critical determining role as to the effect of marital
conflict on youth behavior problems (Grych & Fincham, 1990). From this view, assessing interparental conflict from the perspective of the child is critical for determining the actual impact of
the conflict on the child‟s behavior (Grych, Seid, & Fincham, 1992). In line with this, children‟s
distress in relation to their parents‟ marital conflict, rather than conflict exposure per se, has been
associated with an increased stress response in the child (Davies et al., 2008). This view has
been further supported by empirical work demonstrating that youth reports of inter-parental
conflict are more predictive of their externalizing behavior problems than are the parents‟ reports
of their own marital conflict and satisfaction (Cummings, Davies, & Simpson, 1994; Grych,
Seid, & Fincham, 1992).
Recent meta-analytic work has further indicated that specific cognitive appraisals of selfblame and perceived threat made by youth in relation to their parents‟ marital conflict are robust
predictors of child behavior problems (Rhoades, 2008). While these appraisals are also likely
influenced by constitutional factors (e.g., temperament), they have been argued to represent a set
of proximal process by which family environmental factors exert their influence on child
behavior (Grych & Fincham, 1990). Indeed, children‟s appraisals of self-blame in relation to
their parents‟ marital conflict appear to be a particularly robust predictor of ADHD symptoms
(Counts et al., 2005), even after controlling for other family factors such as socioeconomic
status, parental psychopathology, and parenting stress. Thus, self-blame appears to be an
33

important “environmental” factor related to ADHD, yet its potential etiological role in the
disorder remains untested.
In sum, both genetic and environmental factors appear to play important roles in the
development of ADHD. That said, crucial questions remain regarding how to understand these
relationships among family environment variables within the context of large and robust
estimates of genetic effects. In particular, how are the relationships between family environment
variables and ADHD accounted for in behavioral genetic models that consistently demonstrate
large genetic main effects, small non-shared environmental influences, and zero shared
environmental effects?
One likely answer involves gene x environment interactions (GxE), in which
environmental influences moderate genetic effects on ADHD. Indeed, recent theoretical
advances within behavioral genetics (see Purcell, 2002) have shown that the genetic (a2) and
non-shared environmental (e2) components of variance likely contain not only their respective
main effects but also variance due to GxE effects. In particular, Purcell (2002) noted that there
are two types of GxE interactions that are captured in the traditional genetic and non-shared
environmental variance components, respectively: genetic x shared environment interactions (or
AxC) and genetic x non-shared environment interactions (or AxE).
The first type of interaction (i.e., AxC) represents interactions between genetic and
shared environmental effects, and is captured in the additive genetic variance term (a2).
Monozygotic (MZ) twins share 100% of their segregating genes, whereas dizygotic (DZ) twins
share, on average, 50%. However, all twin pairs (regardless of zygosity) share 100% of the
shared environmental factors (indeed, the correlation of shared environmental factors between
twins within pairs is set to unity in all biometric twin models). Should these common shared
34

environmental factors activate genes of risk, they would then do so more for MZ twins as
compared to DZ twins (i.e., because MZ twins share 100% of their genes, interactions between A
and C would always result in increased MZ correlations, a pattern that would not necessarily
hold for DZ correlations). Because genetic influences are inferred by the difference between MZ
and DZ correlations, anything that increases the MZ correlation relative to the DZ correlation
would “load” on the genetic proportion of variance. Accordingly, AxC interactions would
emerge as part of the genetic variance component (a2) in behavioral genetic models.
By contrast, AxE represents interactions between genetic and non-shared environmental
factors, which would be captured in the non-shared environmental variance term. Non-shared
environmental factors are necessarily different across twins (i.e., they are shared at 0% in
biometric models). Accordingly, if only one MZ twin within a pair is exposed to a given
environmental moderator, then only the genes of that twin will be activated. This will increase
sibling differences, regardless of zygosity, and would thus emerge as part of the unique
environmental variance component (e2).
Overall, the differential GxE loadings on genetic and non-shared environmental
proportions of variance are thus dependent on whether the environmental moderator of interest
represents a shared environmental factor (C) that is common across siblings or a non-shared
environmental factor (E) that is unique to each sibling. For example, maternal warmth may
emerge as a shared environmental factor if both twins in the family experience and are
influenced by equal levels of warmth and nurturing from their mother. Should warmth activate
genes of risk, it would therefore do so more for MZ twins than DZ twins, and thus load on the
genetic proportion of variance. However, should maternal warmth emerge as a non-shared
environmental factor (E) that is unique to each sibling (i.e., a mother may be particularly warm
35

and nurturing to only one of the twins, as described in Caspi et al., 2004), maternal nurturing
would then interact only with the genes of one child within the twin pair and load on the nonshared environmental component of variance. Finally, should warmth function as both a C and
an E (i.e., maternal warmth may both color the global home environment that surrounds the
children and differentially impact each child individually), then moderation of genetic and nonshared environmental effects would be expected.
This biometric framework for understanding GxE has a number of key advantages for
studies of GxE (both in general and for ADHD in particular). First, the application of this
framework may be particularly relevant for ADHD, for which numerous environmental risk
factors have been identified (Banerjee et al., 2007; Nigg, 2006) even though the environmental
component of variance in behavioral genetic studies is consistently small to moderate in
magnitude. Second, it provides an omnibus test of GxE effects for a particular environmental
variable (i.e., children‟s appraisals of inter-parental conflict), as genetic influences are examined
at the latent or composite level of analysis. This advantage stands in contrast to the more
specific molecular genetic GxE analyses, which typically examine only one specific
polymorphism (Langley et al., 2008; Waldman, 2007) and are thus likely to be explaining only a
very small part of the causal chain of polygenic disorders like ADHD. Finally, the biometric
framework detailed above would provide critical new information regarding the specific
manifestation of any GxE between children‟s appraisals of inter-parental conflict and ADHD. In
other words, if appraisals of self-blame are involved in the etiology of ADHD via GxE effects,
the biometric framework would be able to clarify whether interactions occur at the family-level
(e.g., AxC interactions) or at the child-specific level (e.g., AxE interactions) or both. While there
is an established phenotypic link between self-blame and ADHD, it is unknown whether self36

blame functions as a shared environmental factor (and therefore creates sibling similarity),
whether it is operating as a unique environmental factor (resulting in sibling differences), or
perhaps both. In other words, application of this methodology to ADHD can help us not only
confirm the existence of GxE in ADHD, but also illuminate how environmental variables, such
as appraisals of self-blame related to interparental conflict, interact with genetic risk for ADHD.
Somewhat surprisingly, however, this type of methodology has yet to be applied to ADHD.
The purpose of the current study was to do just this by examining how youth appraisals
of self-blame in relation to their parents‟ martial conflict influence the magnitude of genetic and
environmental contributions to ADHD. Specifically, the current study aimed to examine how
the genetic, shared, and non-shared environmental variance components of ADHD shifted as a
function of children‟s report of self-blame. Evidence of significant changes in genetic and
environmental contributions to ADHD could provide an omnibus test of GxE effects involving
children‟s appraisals of self-blame and clarify the specific manifestation of these GxE effects.

METHOD
Participants
Participants were child and adolescent twin pairs from the Michigan State University
Twin Registry (MSUTR), an ongoing project examining genetic and environmental contributions
to both internalizing and externalizing psychopathology (Klump & Burt, 2006). Participants
were recruited through the use of state birth records in collaboration with the Michigan
Department of Community Health (MDCH) and the Michigan Bureau of Integration,
Information, and Planning Services (MBIIP; for a full description of recruitment procedures for
the MSUTR, see Klump & Burt, 2006). Parents gave informed consent for both themselves and
37

their children and children provided informed assent. All research protocol was approved by the
Michigan State University Institutional Review Board.
The current sample consisted of 246 child and adolescent monozygotic (MZ) and
dizygotic (DZ) twin pairs (total n = 492 twins). This sample was composed of 120 MZ twin
pairs (55 male-male, 65 female-female) and 126 DZ twin pairs (69 male-male, 57 femalefemale) that ranged in age from 5-16 years (M=10.2, SD=2.6 years). Participating families in the
MSUTR were representative of individuals living in the mid-Michigan region in terms of racial
identification (see Culbert et al., 2008); approximately 88% self-identified as Caucasian, 8%
identified as African-American, 3% identified as Latino, and 1% identified as Asian-American.
Zygosity Determination
Zygosity was established using physical similarity questionnaires administered to the
twins‟ primary caregiver (Peeters, Van Gestel, Vlietinck, Derom, & Derom, 1998), as well as a
research assistant who independently evaluated the twins on physical similarity indices.
Zygosities were then compared between the participant and research assistant reports.
Discrepancies were resolved through review of questionnaire data and twin photographs (when
available) by one of the MSUTR principal investigators (KLK or SAB) or by DNA markers. On
average, the physical similarity questionnaires used by the MSUTR have accuracy rates of 95%
or better (Peeters et al., 1998).
ADHD Behaviors
Mothers of the twin participants completed the Child Behavior Checklist (CBCL;
Achenbach & Rescorla, 2001) to assess behaviors relating to ADHD. The CBCL is a
standardized questionnaire for parents to report the frequency of 118 problem behaviors
exhibited by their children during the past six months. Mothers rated how often particular
38

behaviors occurred for each twin using a 3-point Likert scale (0=never true;
1=sometimes/somewhat true; 2=often true). These 118 behaviors were then summed into the 8
empirically-validated syndrome scales and 6 DSM oriented scales. The 7-item DSM oriented
scale for ADHD was selected for analyses since its items have demonstrated adequate predictive
power for DSM-IV ADHD (Hudziak et al., 2004). The items included “fails to finish things
he/she starts,” “can‟t concentrate or pay attention for long”, “can‟t sit still, restless or
hyperactive”, “impulsive or acts without thinking”, “inattentive or easily distracted”, “talks too
much,”, and “unusually loud”. The CBCL has demonstrated good reliability and validity in
terms of predicting clinically significant problems related to ADHD and other disorders
(Hudziak et al., 2004). In this sample, internal consistency estimates were adequate (α=.88).
While age and sex based norms are available for CBCL scores, the raw scores for the
DSM-ADHD scale were used in the current study to avoid potential gender and age bias that can
occur with the use of standardized scores in population studies (Achenbach & Rescorla, 2001).
Thus, the ADHD measure for each twin was the total raw score added across the seven ADHD
items. In order to better approximate normality, the final ADHD scores from the CBCL were
log-transformed prior to analysis.
Perceptions of Marital Conflict
Perceptions and appraisals of marital conflict were assessed with the Children‟s
Perception of Marital Conflict scale (CPIC; Grych, Seid, & Fincham, 1992). Each twin
completed a separate CPIC during the assessment. The 48 CPIC items were rated by
participating twins on a three-point scale (1-3: true, sort of true, and false). Because the CPIC
was designed to be completed by school-aged children, the questionnaire was read to
participating twins with reading levels under 5th grade in order to ensure correct comprehension
39

and completion of the items. Based on exploratory and confirmatory analysis of the 48 items
(Nigg et al., 2009), four empirically derived CPIC scale scores were computed – the main scale
of interest being the self-blame scale. Sample items from the CPIC self-blame scale include “My
parents usually argue about something that I do”; “It is usually my fault when my parents argue”;
and “I am to blame when my parents argue.” Internal consistency measures indicated good
reliability for the nine-item CPIC self-blame scale in the full sample (α =.85), as well as across
various age groupings (ages 5-7, α =.81; ages 8-10, α =.83; ages 11-13, α =.86; ages 14-16, α
=.87).
Because GxE analyses estimate genetic and environmental influences at each level of the
moderator, small cell sizes across the range of continuous values (i.e., 9-26) would result in
imprecise estimates. Thus, the self-blame scale was trichotomized into low, moderate, and high
levels of self-blame, so as to more meaningfully estimate genetic and environmental effects at
each broad level of self-blame. The lowest third of the distribution was assigned a score of zero
(n=108); the middle third was assigned a score of 1 (n=157), and the highest third was assigned a
score of 2 (n=162). This trichotomized self-blame variable was used in all model-fitting
analyses.
Data Analyses
Behavioral genetic analyses make use of the difference in the proportion of genes shared
between reared-together siblings. Utilizing these differences, the variance within observed
behaviors or characteristics (i.e., phenotypes) is partitioned into three components, additive
genetic (a2), shared environment (c2), and non-shared environment plus measurement error (e2).
The additive genetic component (a2) is the effect of individual genes summed over loci, and acts
to increase twin correlations relative to the amount of genes shared. The shared environment (c2)
40

is that part of the environment common to siblings that acts to make them similar to each other.
The non-shared environment (e2) encompasses environmental factors (and measurement error)
differentiating twins within a pair. As noted earlier, however, these genetic and environmental
estimates are not composed solely of main effects – they also contain GxE effects. Thus, a2
includes both the main effects of additive genetic factors as well as gene x shared-environment
interactions (AxC). The unique environmental variance term (e2) consists of main effects due to
non-shared environmental factors, gene x non-shared environment interactions (AxE), and
measurement error.
The question of interest in the current study is whether self-blame in relation to
interparental conflict moderates genetic and environmental contributions to ADHD. Whereas
GxE is characterized as environmental sensitivity to genetic risk, gene-environment correlation
(rGE) represents genetic control of exposure to different environments. For example, a child
with ADHD may be more likely to evoke conflict between his or her parents (Whalen & Henker,
1999), and thus be more likely to attribute blame regarding this conflict to him or herself. Thus,
a child‟s genetic proclivities could elicit an “environmental” response that is consistent with
his/her genetic make-up (referred to as an evocative rGE). rGE is particularly important for GxE
researchers to consider because it can resemble GxE in moderator models (i.e., genetic
influences on ADHD could vary across levels of self-blame because the “moderator” is
correlated with genetic risk for ADHD). In short, genetic covariation between ADHD and selfblame (i.e., rGE) could resemble GxE if rGE is not simultanesouly considered.
Given this, Purcell (2002) proposed two GxE models: the first model examines GxE
regardless of rGE (i.e., the univariate or “straight” GxE model); the second model examines GxE
in the presence of rGE. In order to test for GxE effects while also considering the possible
41

impact of rGE in any GxE that were uncovered, two sets of analyses were conducted. First, the
univariate or “straight GxE” model was examined to estimate genetic and environmental
influences on ADHD at each level of the moderator. This model provides a test of GxE but does
not allow us to consider possible genetic overlap between the moderator and the outcome. Next,
the GxE in the presence of rGE model was conducted, thereby allowing us to both confirm the
presence of GxE, and evaluate the impact of rGE on these effects. The use of both models
thereby allows us to accomplish three critical tasks for examination of GxE effects. These
included (1) a test of moderation of the genetic and environmental influences on ADHD by selfblame (GxE), (2) a gauge as to whether rGE effects are present and may be influencing any
observed moderation, and most importantly, (3) a method for controlling for rGE while
simultaneously re-evaluating GxE effects. It is important to note that the “straight” GxE and the
GxE in the presence of rGE models are not nested, and thus their results cannot be statistically
compared. However, conceptual comparisons of the results could prove useful for furthering
understanding of the issues mentioned above.
The “straight” GxE model encompasses three nested moderator models. The first and
least restrictive model allows for both linear and non-linear moderation of the genetic, shared,
and non-shared environmental contributions (i.e., a, c, e) to ADHD. At each age, linear (i.e., A1,
C1, E1) and non-linear (i.e., A2, C2, E2) moderators are added to these genetic and environmental
paths using the following equation: Unstandardized VarianceTotal = (a + A1(self-blame) +
A2(self-blame2))2 + (c + C1(self-blame) + C2(self-blame2))2 + (e + E1(self-blame) + E2(selfblame2))2. A series of more restrictive moderator models for each measure were then fit to the
data, constraining the moderators for each source of etiological influence to be zero and
evaluating the reduction in model fit. As recommended by previous work, the current models
42

were run a minimum of 5 times using multiple start values to ensure that all the estimates
obtained minimized the -2lnL value.
Next, GxE effects were evaluated while accounting for rGE. This GxE in the presence of
rGE model is essentially a re-formulation of the standard behavioral genetic bivariate model,
such that the moderator is entered twice – as a dependent variable and as a moderator variable.
Influences on the outcome measure (ADHD) are then partitioned into two sources of genetic
influence (as well as from two sources of shared and two sources of non-shared environmental
influence): (1) variance shared with the moderator, and (2) variance that is unique to ADHD
(i.e., that residual variance in ADHD that does not overlap with self-blame). The moderator is
then allowed to moderate both the genetic and environmental covariance paths between selfblame and ADHD as well as with the genetic and environmental paths unique to ADHD (i.e.,
that variance remaining in ADHD once the overlap/rGE with self-blame has been partialled out)
(Purcell, 2002). Only the latter index “true” GxE (i.e., controlling for rGE).
Because these interaction models effectively involve fitting a separate biometric model
for each individual as a function of their self-blame score, they both require the use of FullInformation Maximum-Likelihood raw data techniques (FIML). Mx, a structural-equation
modeling program (Neale, 1997), was used to fit models to the transformed raw data. When
fitting models to raw data, variances, covariances, and means of those data are freely estimated
by minimizing minus twice the log-likelihood (-2lnL). For the “straight” GxE model, the
minimized value of -2lnL in the full moderation model is compared with the -2lnL obtained in
more restrictive moderator models to yield a likelihood-ratio 2 test for the significance of the
moderator effects. Non-significant changes in chi-square indicate that the more restrictive model
(i.e., that model with fewer parameters and thus more degrees of freedom) provides a better fit to
43

the data. The chi-square was then converted to the Bayesian Information Criterion (BIC) so as to
measure model fit relative to parsimony. The lowest BIC among a series of nested models is
considered best. BIC was used to determine the best-fitting model as it is one of the most
commonly used indices within the field of behavioral genetics (Markon & Krueger, 2004) and
because it weighs parsimony most heavily.

RESULTS
Descriptive Statistics
Sex. Overall levels of ADHD and self-blame and their relationship across sex and age
were first examined. Mean levels of ADHD varied significantly across gender, such that boys
had higher scores on the CBCL ADHD scale (M=3.4, SD=3.1) than did girls (M=2.9, SD=2.8).
Boys also had significantly higher scores of self-blame (M=12.1, SD=3.3) than did girls
(M=11.0. SD=2.7), even after controlling for levels of ADHD behaviors. Despite these mean
differences, there was a significant and positive relationship between self-blame scores and
ADHD for both boys and girls (boys r=.29, p<.001; girls r=.18, p=.02). Although the relationship
appeared somewhat stronger for boys, the difference in the correlations was not statistically
significant (Fisher z = 1.29, p=.19 two tailed).
Age. Bivariate correlations revealed a modest association between age and ADHD score
(r = -.10, p=.04), such that ADHD behaviors generally decreased with age. In contrast, selfblame was not significantly related to age (r = -.07, p=.17). Importantly, however, the
association between self-blame and ADHD remained moderate and significant across the age
range of the sample (ages 5-7, r=.29, p<.01; ages 8=10, r=.20, p<.001; ages 11-13, r=.26, p<.001;

44

ages 14-16, r=.32, p<.001). These correlations were not statistically different from one another
(all ps>.25).
Overall then, the associations between self-blame and ADHD behaviors appeared to be
robust across both age and sex. It is thus thought to be unlikely that the moderation of ADHD by
self-blame would be restricted to a particular age range or to one gender only. That said,
however, the current sample size did not offer adequate power for examining latent estimates
separately by sex or age. To adjust for potential differences in the moderation analyses, the
transformed ADHD score was regressed on age and gender, a very common practice in the field
of behavioral genetics (Burt, 2009; McGue & Bouchard, 1984). The residuals were then utilized
in the moderation analyses that provided the central test of gene x environment interaction.
Intraclass Correlations
Intraclass correlations for the CBCL ADHD score were first compared across MZ and
DZ pairs in order to gauge the relative influence of genetic and environmental influences for
ADHD (see Table 9). For the overall sample, the MZ correlation was significantly greater than
the DZ correlation. This difference is highly suggestive of genetic influences on ADHD.
However, in order to assess potential etiological moderation of ADHD by self-blame, intraclass
correlations at various levels of self-blame were also examined. To do so, analyses were
restricted to those twin pairs who were concordant for moderator level (i.e., both twins were
classified as either low, moderate, or high self-blame). Given this, the sample sizes are small
relative to the overall sample. Of note, while this technique was necessary to examine potential
moderating effects using intraclass correlations, the twins do not have to be concordant on value
of the moderator when using structural equation modeling techniques in Mx (and thus the full
sample was used for the final analyses).
45

Results indicate that when both twins reported low levels of self-blame, the MZ-DZ
correlation difference, and thus the estimate of genetic influences, was large. As reports of selfblame increased to moderate and high, the DZ correlation dropped moderately, whereas the MZ
correlation decreased substantially. The decreasing difference between the MZ and DZ
correlations implies that genetic influences may decrease with increasing self-blame. Moreover,
the decreasing MZ correlation also implies that non-shared environmental influences increase
with increasing levels of self-blame. Such results collectively suggest that self-blame may
indeed act to moderate genetic and environmental influences on ADHD.
”Straight” GxE Analyses
Test statistics for the “straight” GxE analyses are reported in Table 10. Results indicate
that the linear moderation model best fit the data (as indicated by the non-significant change in χ2
from the non-linear model, but the significant change in χ2 as compared to the main effects
model, as well as the lowest BIC value). Thus, self-blame appears to be a significant linear
moderator of the genetic and environmental contributions to ADHD.
In order to examine the nature of this etiological moderation, we used the estimated paths
and moderators from the best-fitting linear model (see Table 11) to calculate and plot the
unstandardized genetic, shared, and non-shared environmental variance components at each level
of self-blame. Unstandardized parameter estimates are favored here in order to examine the
absolute (rather than proportional) shifts in each parameter across each level of the moderator (as
recommended by Purcell, 2002). Figure 1 displays the unstandardized estimates of genetic,
shared, and environmental variance components for ADHD at different levels of self-blame. As
seen there, when self-blame scores are low, genetic influences on ADHD are quite large, with
small to moderate contributions from the non-shared environment. Yet, as self-blame scores
46

increase, the absolute genetic variance appears to decline sharply, whereas the unique
environmental variance appears to increase substantially. Moreover, as indicated by moderator
estimates whose confidence intervals do not overlap with zero (see Table 3), both the increase in
non-shared environmental factors and the decrease in genetic factors were statistically
significant. In other words, increasing levels of self-blame appear to significantly decrease
genetic contributions, while also increasing non-shared environmental contributions, to ADHD.
Estimates of shared environmental contributions to ADHD were zero at all levels of self-blame.
GxE in the Presence of rGE
We next examined potential GxE effects while simultanesouly considering potential rGE
confounds. As a reminder, in this model, the moderator (self-blame) is entered twice so that it
can be examined as both a dependent variable and as a moderator. Moderation is estimated for
the genetic and environmental overlap between self-blame and ADHD (i.e., the covariance
paths) as well as for the genetic and environmental contributions to ADHD that do not overlap
with those for self-blame. The key estimates of interest in the current study involve the latter
(i.e., the unique variance estimates for ADHD as well as the moderation of those unique
estimates). Examination of these parameters is crucial for determining if self-blame significantly
moderates genetic and environmental influences on ADHD, even after accounting for the
presence of possible rGE between self-blame and ADHD. Of note, unlike the former model, the
GxE in the presence of rGE model only allows for linear moderation.
The full ACE GxE in the presence of rGE model provided an adequate fit to the data (2lnL=2109.682, df=831, BIC = -1170.821), as indicated by the negative BIC value. The
resultant path estimates and moderators for this model are presented in Table 12. As can be seen
there, there was evidence of significant genetic covariance between self-blame and ADHD (.489,
47

p<.05). Such findings indicate that there is significant genetic overlap between self-blame and
ADHD (i.e., rGE) which may be influencing the previously observed “GxE”. In order to rule out
this possibility, etiological moderation of that variance that is unique to ADHD was next
examined (see Figure 2). Importantly, results are generally similar to those reported above. The
shared environmental estimates again remained at zero for all levels of self-blame. Examination
of the confidence intervals revealed significant contributions from both genetic and non-shared
environmental factors to ADHD at low levels of self-blame. Moreover, genetic influences on
ADHD decreased as reports of self-blame in relation to parents‟ martial disputes increased.
Critically, however, the effect was clearly more muted than in the prior model (and was no
longer significantly greater than zero, as seen by the non-significant A1 value in Table 4). Most
importantly, however, non-shared environmental influences on ADHD still appeared to increase
substantially (and significantly so) with increasing levels of self-blame.
The lack of significant change in the estimates of genetic variance for ADHD deserves
some additional comment. This difference may be due to differences in the absolute value of
genetic variance in ADHD across the two models. As can be seen from Figures 2 and 3, the
absolute level of genetic variance within ADHD appears to be smaller after accounting for
genetic overlap between ADHD and self-blame (using a visual comparison only, since the two
models are not nested and thus cannot be statistically compared). For example, at low levels of
self-blame, the unstandardized or absolute genetic variance is .79 in the “straight” GxE model
(which does not parse out the covariance between the moderator and the outcome), but only .41
in the model accounting for rGE effects. There thus appears to be some genetic overlap between
self-blame and ADHD, a conclusion that is bolstered by the bivariate results reviewed above.

48

However, even when taking into account this genetic overlap, the moderation on the unique
environmental effects for ADHD by self-blame remained significant.
Moreover, as can be seen in Figure 2, the shared environmental path estimates as well as
moderation paths of the shared environmental variance unique to ADHD were all estimated to be
zero. Thus, the rGE in the presence of GxE model was re-examined, fixing all the shared
environmental paths (Cholesky elements as well as common and unique moderation paths) to
zero. Fixing these paths resulted in a slight improvement in fit, as indicated by the more
negative BIC value (-2lnL = 2119.733, df=836, BIC= -1179.328). Unsurprisingly, the
moderation of E remained significant when C was constrained to be zero. However, moderation
of the unique genetic variance in ADHD became statistically significant, results that are likely a
function of the tightened confidence intervals following the dropping of five paths, especially
given the small sample size. In any case, such results indicate that there may be some (albeit
small-to-moderate) moderation of genetic effects on ADHD by self-blame even after controlling
for rGE.
DISCUSSION
Along with strong estimates of genetic effects for ADHD, prior research has consistently
demonstrated that the family environment, particularly children‟s appraisals of interparental
conflict, remain important predictors of ADHD. One potential way of understanding
environmental associations within the context of large genetic effects of ADHD may be via GxE
effects. The aim of the current study was to investigate the potential etiological role of selfblame related to interparental conflict as a moderator of the genetic, shared, and non-shared
environmental influences on ADHD. Two sets of moderator analyses were conducted. First, a
“straight” GxE model was conducted that investigated potential moderation of genetic and
49

environmental influences for ADHD by level of self-blame, but did not consider possible rGE.
Results indicated that non-shared environmental influences on ADHD significantly increase with
higher levels of self-blame, while genetic influences on ADHD significantly decrease with
increasing self-blame. A second moderator model was examined to clarify the role of rGE in the
above findings, and in this way, confirm the results. As before, results indicated that the
contribution of non-shared environmental factors to ADHD significantly and substantially
increased with increasing levels of self-blame. There was also a simultaneous decrease in
genetic contributions to the variance in ADHD with increasing levels of self-blame (although
only significantly so when shared environmental effects were constrained to be zero). Overall
then, these results indicate that non-shared environmental influences on ADHD increase, while
genetic influences may decrease somewhat, with increasing levels of child self-blame regarding
interparental conflict.
The notion of an environmental risk factor reducing genetic effects on ADHD stands in
contrast to previous reports of GxE effects, which have reported increased risk for the disorder
within particular environmental contexts (Todd & Neuman, 2007; Waldman et al., 2007). While
these lines of research have examined different environmental moderators of genetic risk (e.g.
prenatal alcohol and cigarette exposure, parental marital status), the findings have suggested
stronger relationships between particular gene variants and ADHD in the high-risk
environmental context. By contrast, the current results suggested lower genetic and higher
unique environmental effects in the high-risk environmental context (e.g., high levels of selfblame). While it remains unclear what accounts for these discrepancies, one possibility is that
this environmental risk factor simply functions differently than other risk factors. This possibility
is bolstered by previous work on GxE in personality, which found that for some environmental
50

risk factors, higher levels of risk may actually attenuate as opposed to enhance genetic influences
(Burt, 2008). That is, some environmental risk factors may act as “main effects” for
psychopathology even without contributions from genetic factors. This may be particularly true
for psychosocial risk factors, like self-blame, which likely reflect the proximal end result of
multiple risk processes.
While these results require replication, the current project was strengthened by several
factors. First, self-blame regarding interparental conflict was assessed from the child‟s
perspective. This is advantageous both because the child‟s understanding of marital conflict is
more predictive than parental report of marital conflict (as previously discussed), but also
because the use of child self-reports for the moderator allowed us to circumvent shared informant
variance (since we examined maternal-reports of ADHD). Moreover, because each child
reported on his or her own self-blame, the moderator was free to vary across twins, allowing us
to examine and control for possible rGE. As noted above, rGE remains a significant confound in
GxE research as these effects can resemble GxE. The explicit examination of rGE provides
additional assurances that the GxE effects observed here are “real” (or at least not due to rGE).
Similar analyses would not be possible if only parent report were used, as it would not vary
across twins. This represents an important step forward in behavioral genetic modeling of GxE
effects.
Implications
The results of the current study demonstrate that as scores of self-blame increase, the
genetic influences on ADHD appear to decrease somewhat. This finding suggests that the large
genetic influences previously reported for ADHD may vary, at least somewhat, as a function of
the level of environmental risk. One possible explanation for these findings is that genetic
51

effects are more likely to be expressed as main effects in the absence of environmental
challenges (as discussed in Burt, 2008). In other words, genetic factors may make the greatest
contribution to ADHD when environmental risk factors, like appraisals of self-blame, are low.
This could have particular importance for molecular genetic investigations that are attempting to
identify associations between DNA variants and ADHD. Furthermore, future molecular GxE
investigations of ADHD may also benefit from examining the strength of association between
genetic markers and ADHD in a low-risk environmental context.
Next, these results indicated a substantial increase in the non-shared environmental
variance in ADHD with increasingly levels of self-blame, even when controlling for rGE. One
possible explanation for this increase in non-shared environmental effects is that it may reflect an
overall increase in the importance of unique environmental factors on ADHD. This idea does
have some theoretical backing, as prior work examining the development of psychopathology
has demonstrated that individual environmental risk factors are not as predictive of child
psychopathology as are the aggregate of multiple environmental risk factors (Rutter, 1999).
Thus, for ADHD, high levels of one environmental risk factor (cognitive appraisals of selfblame), may signal the presence of other child-specific environmental risk factors exerting main
effects on ADHD. For example, youth reporting higher levels of self-blame may also experience
greater conflict overall within the family as well as may experience more inconsistent discipline
from parents, both of which have been linked to ADHD.
A second possible explanation for the increase in e2 is that at high levels of self-blame,
there are greater contributions from AxE interactions. As illustrated earlier, AxE interactions
load on the unique environmental variance component (e2). Thus, the increase in e2 with higher
levels of self-blame may indicate that there are additional child-specific interactional processes
52

that are influencing ADHD behaviors. This may make sense, given developmental and family
process research demonstrating biological, psychological, and emotional changes in children
who report high self-blame and threat in relation to their parents‟ marital conflict (Rhoades et al.,
2008). For example, it may be that with higher levels of self-blame, children have increased
problems with behavioral regulation, such that they have difficulties disengaging from the
negative cognitive appraisals of blame, leading to greater symptoms of inattention. That is,
youth who experience more negative cognitive appraisals may experience more difficulties with
cognitive control and set-shifting abilities, which may emerge behaviorally as inattention.
Furthermore, these cognitive appraisals may be influencing physiological responses to stress,
including increased arousal and cortisol release, which then may result in greater symptoms of
hyperactivity and impulsivity. Indeed, recent work has shown elevated stress responses (e.g.,
cortisol) in youth who presented with high distress levels while witnessing interparental conflict
(Davies et al., 2008). While still preliminary, this work does suggest that cognitive appraisals of
conflict may indeed influence biological mechanisms underlying ADHD (among other things),
systems which should also be influenced by the child‟s genes. Given this, it may be that high
appraisals of self-blame by one twin (but not the other) differentially activate only that particular
twin‟s genes, thereby increasing the non-shared environmental variance within ADHD. Such
processes would result in an increase in unique environmental influences on ADHD with higher
levels of self-blame, which was in fact observed in the current study.
Limitations
There are some limitations of the current work that are important to note. The sample
was underpowered to examine estimates of latent genetic and environmental influences by sex or
by age. As a result, effects of sex and age were regressed out of ADHD prior to analysis (a
53

common practice in the field of behavioral genetics; Burt, 2009; McGue & Bouchard, 1984).
Future work should seek to examine potential sex-specific and age-specific effects in a larger
sample. Next, the study examined the DSM-oriented ADHD scale on the CBCL rather than
DSM-IV symptoms of ADHD. While high scores on DSM ADHD scale have been shown to
align with diagnosis of ADHD (Hudziak et al., 2004), the use of the CBCL scale did not allow us
to examine ADHD symptom dimensions of inattention-disorganization and hyperactivityimpulsivity separately. Examination of each of the ADHD symptom dimensions as outcomes
may be warranted, as work in behavioral genetics, neuroscience, and temperament have
demonstrated that there may be important etiological differences between inattention and
hyperactivity-impulsivity (Martel & Nigg, 2006; Nikolas & Burt, under review; Sonuga-Barke,
2005).
Lastly, this study relied in part on the GxE in the presence of rGE model advanced by
Purcell (2002). The mathematical accuracy of this model has recently been questioned by
Rathouz and colleagues (2008), who suggested that a correlated factors model was more
appropriate for examining GxE in the presence of high levels of rGE (although this model has
yet to be thoroughly tested). At this time, it remains unclear whether small-to-moderate levels of
rGE (such as those examined here) also lead to mathematically imprecise GxE estimates
(Rathouz, personal communication, July, 2009); future studies are planned to examine this
possibility. Regardless, these results are not likely to be a function of any potential problems
with the Purcell (2002) rGE model, as the conclusions extended to the univariate or “straight
GxE” model results as well (which is not affected by the potential problems noted by Rathouz et
al., 2008).
Conclusions
54

In sum, the data convincingly showed that negative cognitive appraisals (e.g. self-blame)
made by children in relation to their parents‟ marital conflict enhanced non-shared environmental
contributions but reduced genetic contributions to ADHD. Such findings not only support
additional examinations of family factors as etiological moderators of ADHD, but also highlight
the need for an evaluation of self-blame as a moderator of genetic risk in molecular genetic GxE
studies of ADHD.

55

CHAPTER 4: GENE X ENVIRONEMTN INTERACTIONS FOR ADHD:
SYNGERGISTIC EFFECT OF 5HTTLPR GENOTYPE AND YOUTH APPRAISALS OF
MARITAL CONFLICT
Attention-deficit hyperactivity disorder (ADHD) is one of the most commonly diagnosed
disorders or childhood, affecting anywhere between 3-10% of school-aged children. The
behavioral symptoms of ADHD likely reflect the interplay of complex developmental processes
involving both genetic liability and family environment factors that are believed to shape the
gradual, yet staged development of emotional and behavioral regulation during early to middle
childhood (Nigg, Hinshaw, & Huang-Pollock, 2006). Whereas the interplay of these processes is
emphasized in theoretical work, prior work has often examined the main effects of genetic and
environmental influences on ADHD separately. Yet, examination of interactions between
specific genetic or biological contributors and family risk factors is sorely needed in order to test
hypotheses regarding the roles of each in the development of ADHD via behavioral and
emotional dysregulation in children.
Examination of gene x environment interactions (GxE) for ADHD provides one
potentially fruitful method for testing hypotheses regarding current conceptualizations of the
interplay of liability and environmental risk in the development of the disorder. GxE effects may
be characterized as genetically-modulated individual differences in sensitivity to environmental
risk factors, such that specific genetic variants exert risk for a disorder only in particular
environments or after specific environmental exposures have occurred (Rutter & Silberg, 2002).
Quantitative behavioral genetic research has consistently shown that genetic contributions to
ADHD are moderate to large (Bergen, Gardner, & Kendler, 2007). The association of specific
candidate genes with ADHD have been replicated in molecular genetic investigations (Faraone et
56

al., 2005; Gizer, Ficks, & Waldman, 2009), but effects account for only a small fraction of the
heritable component. Importantly, theoretical work has suggested that the genetic and nonshared environmental variance components in traditional behavioral genetic models contain main
effects as well as interactions (Purcell, 2002). Further, given that there are also many confirmed
environmental risk factors for ADHD (Banerjee, Middleton, & Faraone, 2007; Nigg, 2006), the
notion of GxE effects operating for ADHD is a compelling possibility (Pennington et al., 2009).
The next steps then involve selection of specific genetic markers and specific family
environment risk factors that may likely be operating synergistically in the development of
behavioral and emotional regulatory capacities, deficiencies of which are believed to contribute
to ADHD symptoms. Because of the hazards of arbitrary selection of variables, selection of the
gene and environment variables to study depends heavily on theoretical as well as empirical
considerations. Genes are considered first, followed by environments.
Selection of Candidate Gene for GxE in ADHD
Genes of the dopaminergic and noradrenergic neurotransmitter systems have received
much attention due to the presumed centrality of the influence of psychostimulants and
noradrenergic reuptake inhibitors on these systems (Spencer et al., 2000). However, they have
been primarily associated theoretically with ADHD‟s cognitive and reward-processing elements,
rather than emotional dysregulation (Nigg & Casey, 2006). Genes coding for proteins involved
in emotional regulation and impulse control systems are needed to test an integrated regulation
model of ADHD (Nigg & Casey, 2006; Nigg, Hinshaw, & Huang-Pollock, 2006). A particularly
attractive neural system from this perspective is the serotonergic system. It has been
hypothesized to play a role in ADHD via its global contribution to externalizing problems such
as impulse control and aggressive behavior (Evans et al., 2000; Moffitt et al., 1998).
57

Furthermore, at the genetic level, the serotonin transporter gene (5HTT) is expressed in brain
regions often implicated in emotion regulation, memory, attention, and motor control (Frankle et
al., 2004). It is hypothesized that it plays a role in the development of emotional and behavioral
regulation and for this reason will be relevant to GxE effects in ADHD.
The functional 44-bp promoter polymorphism of the serotonin transporter gene
(5HTTLPR) has been frequently studied for psychiatric conditions as the “short” allelic variant
results in reduced transcription efficiency and lower uptake activity (Lesch et al., 1996;
Greenberg et al., 1999). In puzzling contrast, the “long” (or “high activity” allele) that has been
associated with ADHD in case-control and within-family studies as well as in two meta-analytic
investigations (Faraone et al., 2005; Gizer et al., 2009). Yet, Gizer and colleagues (2009)
reported substantial heterogeneity in effect sizes across studies in their meta-analysis of main
effect association between 5HTTLPR and ADHD.
One potential explanation for the heterogeneity in the mains effects of 5HTTLPR for
ADHD is the presence of unmeasured environmental risk factors that may be serving to activate
(or attenuate) the genetic association with the disorder. GxE investigations of 5HTTLPR have
been encouraged by evidence that the genetic regulation of serotonin neurotransmission is
sensitive to environmental context (Bennet et al., 2002; Manuck et al., 2004). In addition,
genome-wide investigations of GxE interactions for ADHD found some suggestive evidence of
an interaction between parental warmth and a SNP in the serotonin transporter gene (p=0.008)
(Sonuga-Barke et al., 2008). Thus, 5HTTLPR may be particularly well-suited for investigation
of GxE effects for ADHD.
Selection of Environmental Risk Factors for GxE

58

The only genome-wide association GxE study of ADHD suggested the serotonin
transporter may interact with family distress measures (Songa-Barke et al., 2008). This is not
surprising given that the development of emotional and behavioral regulation in children is
dependent on a range variables related to the family environment. Familial conflict in particular
has emerged as particularly good candidate domain, due to its high salience for children and
findings that chronic emotional stress during development can alter cortical functioning (Arnsten
& Goldman-Rakic, 1998). Family distress or conflict has been assessed in numerous ways over
many decades of work. However, recent work consistently has emphasized the role of marital
conflict in child adjustment, including ADHD and externalizing behavior problems (Buehler et
al., 2007; Counts et al., 2005; Cummings & Davies, 1994; Gerard et al., 2005; Grych &
Fincham, 1990; Grych et al., 2000; Jouriles et al., 2000; Skopp et al., 2005; Wymbs et al., 2008).
Yet, the mechanism of this relationship is thought to depend on more than mere exposure to
marital conflict but also on child liability and construal.
Grych and Fincham (1990) highlighted that the child‟s appraisals of marital conflict (and
not just exposure) often play a determining role as to the effect of marital conflict on youth
behavior problems. That is, the frequency of the conflict may not be as important for the
development and persistence of attention and behavior problems as the extent to which children
blame themselves or feel threatened by their parents‟ disagreements. In support of this view,
youth reports of marital conflict have been shown to be more predictive of behavior problems
than the parents‟ own report of their own marital conflict and satisfaction (Cummings, Davies, &
Simpson, 1994; Grych, Seid, & Fincham, 1992). Furthermore, children‟s observed distress level
while witnessing conflict has been especially associated with biological changes, including
increased stress response (Davies et al., 2008) and alterations in parasympathetic nervous system
59

activity (El-Shiekh et al., 2009). Additionally, a recent meta-analysis found that along with
physiological and emotional reactivity, cognitive appraisals of self-blame in regard to marital
discord emerged as a particularly salient predictor of children‟s internalizing and externalizing
problems (Rhoades, 2008). Thus, construals of self-blame in relation to marital conflict emerges
from the literature as a particularly potent experiential moderator that may interplay with genetic
liability in child dysregulation, and thus guided the hypothesis about how GxE may influence
ADHD in the context of a dysregulation conception.
Additional Considerations: The Role of Serotonin in ADHD
Despite the appeal of 5HTTLPR and self blame as a model micro-system for examining
inputs to regulatory problems and ADHD, the role of serotonin in the development of behavioral
and emotional regulation in ADHD requires a bit more discussion due to complexities involved
in the gene structure itself. First, while the “long” allele has been implicated in risk for ADHD,
the “short” or “low-activity” allele of 5HTTLPR has been associated with numerous conditions
often comorbid with ADHD including childhood disruptive behavior problems, persistent
aggression, and DSM-IV Conduct Disorder (Beitchman et al., 2006; Hallikainen et al., 1999;
Sakai et al., 2006). Secondly, studies of peripheral and central serotonergic functioning in
children have produced somewhat contradictory results. Both low (Clarke et al., 1999; Flory et
al., 2007; Kruesi et al., 1990; Kruesi et al., 1992) and high (Castellanos et al., 1994; Halperin et
al., 1994; Stoff et al., 1992) serotonergic functioning have shown relationships with impulsivity
and aggression in children with ADHD, whereas studies of animals and adult humans have
implicated low serotonergic activity only for these behaviors (Coccaro et al., 1989; Manuck et
al., 2002; O‟Keane et al., 1992; Retz et al., 2004). Thus, there are surprisingly conflicting
findings regarding the direction of association of serotonin activity and ADHD or related
60

behaviors. This gives pause in terms of the expectation of direction of effects and even the
linearity of effects.
A further complexity is the recent discovery of an A>G substitution contained in a subset
of the repeat sequences of 5HTTLPR. The A>G substitution has been shown to have functional
significance, such that the long allele with the G substitution (Lg) functions similarly to the short
allele (i.e., reduced transcription efficiency and decreased expression) (Hu et al., 2005; Kraft et
al., 2005; Nakamura et al., 2000). This resultant triallelic model of 5HTTLPR failed to show
association with ADHD (Wigg et al., 2006), yet the low functioning alleles (short and Lg) have
been associated with conduct problems using case-control and within-family methods (Sakai et
al., 2006).
A few potential explanations have been posited to account for these discrepant results. It
has been suggested that 5HTTLPR may function differently for ADHD and cognitive impulsivity
versus aggressive impulsivity (Oades et al., 2008). Further, differences in etiology of ADHD
symptom dimensions of inattention and hyperactivity may be causing disparate results, such that
5HTTLPR may be associated with one symptom dimension but not the other (Nikolas & Burt, in
press). Thus, examination of ADHD as a unidimensional construct as well as the two constituent
symptom domains of inattention and hyperactivity-impulsivity may aid in clarification of these
relationships.
However, another more provocative possibility suggested by the complex mix of prior
findings is that both the high and low functioning 5HTTLPR genotypes confer risk for ADHD
and related disruptive behaviors. Supporting this notion, the one prior study examining GxE
interactions involving 5HTTLPR and ADHD found a main association effect with the more
efficient “long” allele of 5HTTLPR (high serotonin transporter activity), yet an interaction with
61

psychosocial adversity with the low-efficiency “short” allele of 5HTTLPR (Retz et al., 2008).
That is, it may be that optimal adjustment occurs at the mid-range of serotonergic transcription,
and that at either extreme poor regulatory functioning can occur. This possibility has not been
tested for main effects of 5HTTLPR or in studies of GxE effects involving 5HTTLPR.
The aim of the current study was to examine GxE effects for ADHD within the
framework of a dysregulatory conception of ADHD, and thus focused specifically on examining
a theoretically relevant interaction between the triallelic 5HTTLPR polymorphisms and youth
perception of self-blame in relation to marital conflict. Given the mixed literature as to whether
high or low serotonin activity would be a risk factor, the hypotheses included both linear models
(that either low or high serotonin activity genotypes confer sensitivity to the psychosocial risk
element) and a non-linear model (both low and high serotonin activity genotypes confer
sensitivity to environmental risk, whereas moderate levels confer protection).
METHODS
Participants
Participants were 304 children and adolescents ages 6-18 years (M=14.04, SD=2.70,
56.6% male). The sample was recruited using mass mailings to parents in the local school
districts, public advertisements, and outreach to local clinics in order to screen as broad of a
range of volunteers as possible for the study. A multi-stage screening process was used to
identify cases and non-cases meeting research criteria among those who volunteered. At stage 1,
rule-outs were evaluated by a telephone screen (physical handicap, non-native English speaking,
mental retardation, autistic disorder, and prescription of long-acting psychoactive medications;
stimulant use was not a rule out). Families passed through the telephone screen were then invited
to complete the stage 2 diagnostic screen. Informed consent was obtained from all participating
62

parents; children provided written assent. These studies were approved by the local Institutional
1

Review Board .
For stage 2, parents and teachers completed normative behavioral rating scales, including
(1) the Behavior Assessment Scale for Children (BASC; Reynolds & Kamphaus, 1992) (2) the
Conners‟ (1997) Rating Scale-Revised short form (Conners, 1997), and (3) the DSM-IV ADHD
Rating Scale (Du Paul et al., 1998). One parent (in most cases, the mother) completed the Kiddie
Schedule for Affective Disorders and Schizophrenia-E (KSADS-E; Puig-Antich & Ryan, 1986)
with a trained master‟s level clinical interviewer.
At stage 3, final eligibility and diagnostic assignment were made using a best-estimate
procedure as follows. Data from the KSADS-E and the parent and teacher rating scales, along
with interviewer notes and observations and history of medication use, was presented to a
diagnostic team consisting of a board-certified child psychiatrist and a licensed child clinical
psychologist. Both professionals arrived independently at a clinical decision regarding ADHD
subtype and comorbid diagnoses. Agreement rates were acceptable for all diagnoses (all kappas
>.89). In all cases of disagreement, consensus was able to be reached upon discussion.
Those youths with subthreshold ADHD (5 symptoms, N=10) or situational ADHD (n=6)
were included for analysis of dimensional symptom scores. The final sample consisted of 137
non-ADHD participants, 151 ADHD participants (72 Primarily Inattentive Subtype, 1
Hyperactive-Impulsive Subtype, 78 Combined Subtype), and 16 subthreshold/situational.
Exclusionary Criteria. Youth were excluded if they had mental retardation (based on
having a full-scale IQ <75), head injury with a loss of consciousness, a history of seizures as
1

This study was funded by R01-MH070004 from the National Institutes of Mental Health to K.
Friderici & J. Nigg.
63

ascertained by parent report, autistic or pervasive developmental disorder as reported by the
parent, or KSADS-E diagnosis of current major depressive disorder (viewed as rendering ADHD
symptom ratings difficult to evaluate), lifetime bipolar disorder, lifetime psychosis, or Tourette‟s
syndrome.
Perception of Marital Conflict
To assess marital conflict, children and adolescents completed the Children‟s Perception
of Marital Conflict scale (CPIC; Grych, Seid, & Fincham, 1992). Youth rated the 48 CPIC items
on a three-point scale (1-3: true, sort of true, and false). Recent work has shown that the CPIC
yields 4 factors, two of which capture children‟s appraisals of conflict (Nigg et al., 2009). The
nine-item self-blame scale was the focus of analysis here because of previous work
demonstrating the importance of the child‟s construals and appraisals of marital conflict in
predicting externalizing behavior problems (Grych & Fincham, 1990). Sample items from the
CPIC self-blame scale include “My parents usually argue about something that I do”; “It is
usually my fault when my parents argue”; and “I am to blame when my parents argue.” Internal
consistency reliability for the nine-item self-blame scale was satisfactory (alpha=.83).
DNA Collection and Serotonin Transporter Genotyping
Overview. Buccal DNA samples were requested from all participating children and
adolescents and purified using previously used methods (Meulenbelt et al., 1995).
Serotonin Transporter Promoter Polymorphisms. The 44-bp promoter polymorphism
of the serotonin transporter gene (5HTTLPR) and the rs25531 A>G polymorphism were
genotyped as follows. The “short and long” alleles of the 5HTTLPR were genotyped according
to previous methodology [21] with the following modifications to the primer sets (5‟GACTGAGCTGGACAACCACG-3‟ and 5‟-GGTTGCCGCTCTGAATGCCA-3‟). Genomic
64

DNA (40 to 60 ng) was amplified using the Taq DNA Polymerase kit (Qiagen Inc., Valencia,
CA), standard kit protocol, including 1.5 mM MgCl2, 0.2 mM dNTPs, and 0.7 µM primer.
Polymerase chain reaction (PCR) conditions consisted of an initial denaturing step at 95°C for 3
minutes, followed by 35 cycles of: 95°C denaturation for 30 seconds, 63°C annealing for 30
seconds, and an extension at 72°C for 45 seconds, followed by a final extension step of 4
minutes at 72°C. A portion of the amplified DNA was analyzed using a 2% agarose gel to
determine the L/S alleles. The remainder of the amplification reaction was digested with MspI
endonuclease (New England Biolabs, Ipswich, MA) and examined by 3% agarose gel
electrophoresis. The final products were (340, 120, and 64 bp) for (La), (174, 166, 120, and 64
bp) for (Lg), and 484 bp (short).
Based on previous work (Beitchman et al., 2006; Hu et al., 2005) individuals were
assigned to the high, intermediate and low activity groups based on their genotypes. Those
homozygous for the La allele were classified as “high‟‟ 5HTTLPR activity (n=78). Those with
the La/Lg or La/short genotypes were classified as “intermediate” 5HTTLPR activity (n=137).
Individuals with Lg/Lg, Lg/short, or short/short genotypes were classified as “low” 5HTTLPR
activity (n=89).
ADHD Symptom Outcome Measures
In order to examine effects at all levels of symptomatology and to avoid artifacts
associated with examining GxE effects for categorical outcomes (Eaves, 2006), the use
dimensional ratings of inattention and hyperactive behaviors were used as the primary outcome
measures. The primary dependent measure was teacher-rated symptoms of ADHD for several
reasons. First, it enabled complete disaggregation of the sources of data (which could be partially
confounded if parent ratings of behavior are used, because parents who are distressed by marital
65

conflict may also inflate their ratings of child ADHD symptoms). Put another way, teacher
ratings of child ADHD behaviors are unlikely to be directly influenced by inter-parental conflict
in the home. Teacher reports of attention problems have also been cited as the most robust
predictor of life outcomes when controlling for other environmental variables, including single
parent status, socially disadvantaged community, parental education, and child IQ as well as
disruptive and emotional behavior problems (Breslau et al., 2008; Breslau et al., 2009; Duncan et
al., 2007).
To evaluate whether effects were robust to particular measurement approaches, both
Teacher ADHD Rating Scale (Du Paul et al., 1998) and teacher report on the Conners‟ Rating
Scale-Revised (Conners et al., 1997) were evaluated. Interactions were first examined for the
total DSM-IV ADHD symptom score on the ADHD-Rating Scale (alpha=.88). In a Fisherian
strategy (Keppel & Wickens, 2004), if that effect was significant, it was followed by
examination of the symptom dimensions of inattention and hyperactivity-impulsivity separately
(with p value of .025 as threshold for significance for each dimension in order to adjust for
multiple comparisons).
The analyses were then reproduced using the Conners‟ Rating Scale ADHD Index as the
outcome measure in order to provide a within-study replication of any significant findings. In a
parallel approach, the Conners‟ ADHD Index was first examined, and if it was significant then
again the Cognitive Problems and Hyperactivity scales were examined separately with the same
p value correction as used for the ADHD rating scale (p=.025).
Statistical Analyses
Testing Gene –Environment Interplay for ADHD Symptoms. Prior to examining GxE
effects, we first had to examine potential gene-environment correlation effects (rGE). The
66

presence of an underlying relationship between the genetic and environmental risk factors (e.g.,
5HTTLPR and self-blame) can confound any test of GxE, as rGE effects can potentially emerge
(falsely) as GxE. Thus, in order to examine this possibility, we first examined differences in
reports of self-blame by level of 5HTTLPR genotype. Next, hierarchical linear regression
analyses were used to examine potential gene x environment interactions for ADHD. As
mentioned earlier, models examined both the linear effects of 5HTTLPR (i.e. that either low or
high serotonin activity genotypes exert risk) and non-linear effects of 5HTTLPR (i.e. that both
low and high serotonin activity genotypes exert risk). The interactions (linear x self-blame and
non-linear x self-blame) were evaluated using the following orthogonal coding system. For the
linear effects, high, intermediate, and low activity 5HTTLPR genotypes were coded as 1, 0, -1
respectively. For non-linear effects, high, intermediate, and low activity 5HTTLPR genotypes
were coded -1, 2, -1, respectively.
RESULTS
Sample Characteristics.
Demographic and descriptive statistics of the children with non-intermediate diagnostic
status are presented in Table 13. As expected, youth in the ADHD group were rated as having
significantly more inattentive and hyperactive symptoms across informants and measurement
method. The ADHD group was more predominately male and significantly younger than the
non-ADHD group; thus age and gender were covaried in all models. Children in the ADHD
group were also less likely to be living in two-parent households and their families had
significantly lower annual incomes than the non-ADHD families.
Ethnicity and 5HTTLPR

67

There were no differences between the ADHD and non-ADHD group in terms of
ethnicity. However, ethnic variation is a potentially important variable in genetic studies. Table
14 shows that there were significant differences in the distribution of 5HTTLPR alleles by ethnic
group, such that the La and Lg alleles both occurred more frequently in African-American
children than non-African-American children, indicating that genotyping the A>G substitution in
5HTTLPR is capturing some important differences among these groups (which is consistent with
expectations from the literature, see Hu et al., 2005; Nakamura et al., 2000). Although ethnicity
and ADHD were uncorrelated, as a precaution, ethnicity was covaried as follows. Ethnicity was
first divided into three codes: (1) Caucasian versus non-Caucasian, (2) African-American versus
non-African-American, and (3) Latino versus non-Latino. The three codes were then entered at
step 1 of all regression models.
Simple Main Effects of 5HTTLPR on ADHD
Table 13 also includes the 5HTTLPR allele frequencies in the ADHD and non-ADHD
groups. There were no significant differences in the distribution of 5HTTLPR alleles among
ADHD and control youth (all p>.25). There were also no significant main effects of 5HTTLPR
genotype when dimensional measures of ADHD were examined (see Table 15). As seen there,
5HTTLPR genotype was unrelated to ADHD and externalizing symptoms (KSADS-E) as well as
scores on the ADHD Rating Scale and Conners‟ Rating Scale.
Test for Gene-Environment Correlation
Before examining interactions, the presence of a relationship between 5HTTLPR
genotype and youth appraisals of self-blame was evaluated. There were no significant
differences in reports of self-blame across the three genotype groups (p=.22), suggesting an
absence of gene-environment correlation between 5HTTLPR genotype and children‟s report of
68

self-blame. This is extremely important as the lack of a correlation between this gene and this
environmental measure signals that rGE effects are not likely to emerge as spurious interaction
findings in the present analyses.
Main Analyses of GxE Interaction Effects
DSM-IV ADHD Symptoms
Total ADHD Score. Hierarchical linear regression analyses revealed a significant main
2

effect of self-blame for ADHD [b=.14, 95% confidence interval .05-.21, p=.002, R =.28].
indicating an increase in total ADHD symptoms with higher reports of self-blame. The linear
(low activity as risk) and non-linear (low and high activity as risk) main effects of 5HTTLPR
genotype were nonsignificant as was the linear x self-blame interaction (all ps>.27). However,
the non-linear x self-blame interaction was significant [b=.11, 95% confidence interval .012

.1214, p=.019, R =.31]. Examination of the simple slopes revealed that there was no
relationship between ADHD symptoms and self-blame for those with the intermediate activity
genotypes. In contrast, a significant and positive relationship between self-blame and ADHD
emerged for those with the high (La/La) and low (Lg/Lg, Lg/short, short/short) activity
genotypes (See Figure 3).
ADHD Symptom Dimensions. Next, the ADHD symptom dimensions of inattention and
hyperactivity-impulsivity were examined separately. Teacher report of inattention symptoms on
the ADHD Rating Scale revealed a significant main effect of self-blame [b=.14, 95% confidence
2

interval .05-.23,p=.004, R =.14] and a significant 5HTTLPR non-linear x self-blame interaction
2

[b=.09, 95% confidence interval .01-.14, p=.02, R =.17]. All other effects were non-significant
(all p>.30). Examination of the simple slopes again revealed a significant positive relationship
69

between self-blame and inattention symptoms for those with the high (La/La) and low (Lg/Lg,
Lg/short, short/short) activity 5HTTLPR genotypes, such that those reporting greater self-blame
had increased inattentive symptoms in those genotype groups. There was no relationship
between self-blame and inattentive symptoms for those in the intermediate activity 5HTTLPR
genotype group (La/Lg, La/short - see Figure 3).
Examination of hyperactivity on the Teacher ADHD Rating Scale also revealed
significant main effects of youth appraisals of self-blame [b=.30, 95% confidence interval .032

.57, p=.03, total R =.17]. The linear and non-linear main effects of 5HTTLPR genotype were
nonsignificant (p>.15) as were the linear 5HTTLPR x self-blame interactions (all p>.26). The
non-linear 5HTTLPR x self-blame interaction was just shy of the .025 correction for significance
2

[b=.16, 95% confidence interval .02-.34, p=.03, total R =.21]. For completeness this marginal
effect was also decomposed. The pattern of results was again the same, such that for those with
the high and low functioning 5HTTLPR genotypes, there was a significant positive relationship
between self-blame and hyperactivity (See Figure 3). However, for those in the intermediate
activity genotype, there was no association of self blame with hyperactivity.
Replication of GxE Effects Using an Alternative Measure (Conners’ Rating Scale)
Conners‟ ADHD Index. In order to check for false-positive results, I next attempted to
replicate the non-linear 5HTTLPR x self-blame interactions using a separate but related measure
of ADHD symptomatology, beginning with the omnibus Conners ADHD Index. Regression
analyses again revealed a significant main effect of self-blame [b=.19, 95% confidence interval
2

.08-.29, p=.003, R =.17]. The main effects of 5HTTLPR (both linear and non-linear) were again
not significant nor was the linear 5HTTLPR x self-blame interaction (all ps>.28).
70

As before, the non-linear 5HTTLPR x self-blame interaction was significant [b=.17, 95%
2

confidence interval .06-.24, p<.001, R =.25]. The pattern of results was again identical, such that
scores on the ADHD Index increased with higher levels of self-blame, but only for those with the
low and high serotonin activity genotypes (see Figure 2). Given this replication, results were
then examined separately for the two ADHD symptom dimensions.
Cognitive Problems and Hyperactivity. The model predicting Cognitive Problems
(analog to inattention) on the Teacher Conners‟ Rating Scale revealed a significant main effect of
2

self-blame [b=.21, 95% confidence interval .09-.32, p<.001, R =.07]. There was no main effect
of 5HTTLPR genotype group using either the linear (p=.29) or non-linear (p=.54) coding
schemes. The linear x self-blame interaction was nonsignificant (p=.54). The non-linear x selfblame interaction was also non-significant [b=.07, 95% confidence interval -.01-.14, p=.08,
2

R =.08] (see Figure 4).
Regression results for reports of hyperactivity on the Conners‟ Rating Scale, however,
mirrored other results. The main effect of self-blame was again significant [b=.15, 95%
2

confidence interval .05-.26, p=.005, total R =.15]. The linear and non-linear main effects of
5HTTLPR genotype were again nonsignificant as were the linear 5HTTLPR x self-blame
interactions (all p>.26). As observed with the ADHD Rating Scale, the non-linear 5HTTLPR x
self-blame interaction was again significant [b=.15, 95% confidence interval .08-.23, p<.001,
2

total R =.20]. Examination of the simple slopes again revealed a similar pattern of results. For
youth with the low and high activity serotonin genotypes, there was a significant and positive

71

relationship between self-blame and hyperactivity. Yet, there was no relationship between selfblame and ADHD for those with the intermediate serotonin activity genotypes (see Figure 4).
DISCUSSION
Prior work has demonstrated the potential relevance of both biological and family
environmental factors in the development of emotional and behavioral regulation capabilities. In
particular, serotonin genetic risk has been shown to a particularly important biological marker for
the development of emotional and behavioral regulation whereas children‟s appraisals of blame
in relation to their parents‟ martial conflict have been suggested as particularly salient
environmental risk factor. The interplay of both biological and family environmental risk factors
has been emphasized, and testing for potential GxE interactions provided a test of this hypothesis
of interplay.
The results of the current study provide suggestive evidence of GxE effects for ADHD
involving 5HTTLPR and youth appraisals of self-blame. Additionally, these analytic methods
allowed for a test of a hypothesis previously untested at the genetic level – namely that both high
and low serotonergic activity genotypes exert risk for ADHD symptoms. Findings from both the
functional serotonin literature as well as from molecular genetic association studies have
previously suggested that only one type of serotonergic activity (increased or reduced) is related
to ADHD and other externalizing phenotypes, including impulsivity and aggression. These
results suggest that perhaps at the genetic level, both high and low serotonergic activity
genotypes exert risk and that these risk mechanisms are likely modulated by environmental
stressors.
The results were strikingly consistent across measures and indicated significant nonlinear 5HTTLPR x self-blame interactions for total ADHD symptoms as measured by both the
72

ADHD Rating Scale and the Conners‟. The interactions revealed that youth appraisals of selfblame were significantly related to total ADHD symptoms for children with the low activity
(Lg/Lg, Lg/short, short/short) and high activity (La/La) 5HTTLPR genotypes. For those with the
intermediate activity genotypes, there was no relation between reports of self-blame and ADHD
symptoms. The interactions were significant after including the covariates of age, gender, and
ethnicity, and while taking into account potential main effects of 5HTTLPR genotype and youth
reports of self-blame.
Analysis of the symptom dimensions revealed a more complex picture. After correcting
for multiple comparisons, the non-linear 5HTTLPR x self-blame interaction was significant for
inattentive symptoms (on the ADHD Rating Scale). Again, the same pattern of results emerged,
such that the relationship between self-blame and inattention was significant only for those youth
with the high and low serotonin activity genotypes. However, the attempt to replicate this effect
using the Conners‟ Cognitive Problems scale did not yield a significant interaction. In contrast,
the non-linear 5HTTLPR x self-blame interaction was not significant for DSM symptoms of
hyperactivity, after correcting for multiple comparisons (p=.03). Interestingly, the replication
attempt yielded a significant non-linear 5HTTLPR x self-blame interaction for Conners‟
Hyperactivity. Analysis of the interaction again revealed that for youth with the high and low
serotonin activity genotypes, the relationship between self-blame and hyperactivity was positive
and significant. Overall, the discrepant pattern of results that emerged in the symptom
dimension analysis in combination with the strong results from the overall analysis suggest that
the types of gene x environment interactions observed likely extend to ADHD symptoms
generally and are not specific to either of the two behavioral dimensions of inattention or
hyperactivity.
73

The overall results are also suggestive of a potential heterozygote advantage, which has
been suggested as a potential explanation for some diseases, including mental disorders (Keller
& Miller, 2006). Further, these results suggest that the interplay between serotonin activity and
the family environment are important for the development of behavioral and emotional
regulation capabilities as indicated by the behavioral symptoms of ADHD. Importantly, these
results add to recent evidence that serotonin genetic risk as indexed by 5HTTLPR genotype and
disruptions in the family environment interact together to predict deficits in behavioral and
emotional regulation (Barry et al., 2008; Pauli-Pott et al., 2009).
With regard to the genetic literature for ADHD, these results failed to replicate a main
effect of 5HTTLPR genotype with ADHD symptoms that has been previously reported (Kent et
al., 2003; Manor et al., 2001; Seeger et al., 2001). Unlike these prior studies, however, the A>G
substitution was also genotyped, however, which has been found to have functional significance,
such that the “long” allele with the G substitution functions similarly to the “short” allele. Using
this triallelic configuration of 5HTTLPR genotype, these results are consistent with those of
Wigg and colleagues (2006) who found no main association effect with ADHD.
In regard to the functional serotonin literature, studies have found relationships between
both high and low serotonin functioning and impulsivity and aggression ratings in children
(Halperin et al., 1994). Results are consistent with the possibility that either extreme of serotonin
functioning may exert risk for ADHD. One potential explanation that has been posited is that
ADHD children with and without comorbid aggression differ in serotonin functioning. Thus, the
discrepant results across studies may be due to the degree to which children have comorbid
profiles. I undertook post-hoc analyses in order to examine whether co-occurring oppositional
defiant disorder (ODD) symptoms and conduct disorder (CD) symptoms altered the results.
74

First, there were no significant differences in the distribution of genotype groups among those
with ADHD alone and those with ADHD and comorbid ODD/CD (p=.41). Further, all nonlinear 5HTTLPR x self-blame interactions remained significant after controlling for ODD and
CD symptoms (all ps remained <.035), indicating that potential differences in the degree of
externalizing comorbidity were not influencing the results.
Given the effects of selective serotonin reuptake inhibitor (SSRI) medications, I also
checked whether a history of comorbid depression accounted for the results. When lifetime
depression symptoms (as measured on the KSADS-E) were included as a covariate, the nonlinear interaction terms remained significant (and resulted in an increase in the unstandardized
regression coefficients). This may make sense, given that depression is often associated with
only low serotonin activity and the non-linear model posited that both low and high serotonin
activity genotypes would show the same relationship between self-blame and ADHD symptoms.
These findings also offer broader implications for genetic association studies of ADHD.
These findings, like many of the recently published GxE studies for ADHD, have found that the
relation between specific genetic markers and ADHD varies depending on particular
environmental influences. These findings, coupled with previous work, suggest that the nonreplications of genetic associations may be due in part to differing levels of exposure to various
environmental risk factors across samples. While these findings require replication, they do
indicate that exposure to stressful environments, such as exposure to inter-parental conflict
during childhood and resultant negative cognitive appraisal of that stress (i.e. blaming one‟s self
for the conflict) influences ADHD symptoms.
There are some limitations to this work that are important to note. First, parent DNA was
not available for the majority of the sample, thus the use of family-based analyses was not
75

possible. While unlikely, population stratification effects cannot be ruled out. Second, because
the sample is cross-sectional, the longitudinal relationships between appraisals of self-blame and
ADHD symptoms cannot be examined. It may be the case that more frequent inter-parental
conflict is the result of having a child with more severe ADHD symptoms and that over time,
children view themselves (perhaps correctly) as being responsible for their parents‟ marital
problems. As is always the case with single studies of genetic or environmental effects, the
current findings may be false-positives. Thus, replication of these results in other samples is
necessary.
Yet, the findings are consistent with prior functional work examining the role of
serotonin the development of ADHD and several attempts were made to guard against false
positive results. These included correction for multiple comparisons as well as replication of
findings with an additional teacher-report measure of ADHD symptoms. While only examined
one marker was examined, the findings suggest that children‟s report of self-blame in regard to
their parents‟ marital conflict may represent one potential environmental moderator of the
genetic risk for ADHD. These findings may then help inform future genome-wide association
studies aiming to examine GxE interactions for ADHD. Moreover, the use of a candidate gene
study here provides the opportunity to link and test a particular hypothesis regarding the etiology
of ADHD, namely the role of serotonin activity in the etiology of ADHD. By focusing on a
particular functional serotonin genetic marker (5HTTLPR), the current GxE study was able to
integrate genetic, neurobiological, and psychological theories regarding the processes that
underpin the development of behavioral and emotional regulatory abilities, which are often
disrupted among individuals with ADHD. If this effect is confirmed, it may justify a more
expensive GWAS study of this environmental moderator to identify other genes related to it.
76

Overall, the current study is among the first to examine relationships between 5HTTLPR
and ADHD as well as interaction effects using the triallelic genotype configuration. While
replication of these initial findings is certainly needed, these results suggest that both the lowand high-activity 5HTTLPR genotypes increase risk for ADHD symptoms within the context of
higher levels of youth self-blame in relation to their parents‟ marital conflict.

77

CHAPTER 5: SUMMARY AND DISCUSSION
General Discussion
Within a developmental psychopathology framework, ADHD can be conceptualized as a
result of exchanges between multiple genetic and environmental risk processes occurring
throughout development (Rutter & Sroufe, 2000). Although behavioral genetic investigations
have consistently demonstrated moderate to large genetic influences on ADHD (Bergen,
Gardner, & Kendler, 2007; Burt, 2009), multiple environmental risk factors have also been
identified for the disorder. Therefore, it appears that one potentially fruitful method for
understanding the causal processes that give rise to ADHD involves examination of interplay
between genetic and environmental factors. The current series of studies aimed to examine
interactions between genetic and environmental risk factors (GxE) for ADHD using both
behavioral and molecular genetic approaches.
Youth appraisals of self-blame in regard to their parents‟ martial conflict were selected as
the key environmental variable in tests of GxE effects. As mentioned earlier, these appraisals
have been repeatedly linked to measures of internalizing and externalizing behavior (Rhoades,
2008), and have been argued to represent a proximal set of processes by which marital conflict
exerts influence on child behavior (Grych & Fincham, 1990). While child appraisals are likely
influenced to some degree by genetic/constitutional factors (an intuition that is supported in these
data; see Study 2), examination of GxE effects allowed for a direct investigation of the potential
etiological role of self-blame in ADHD using two complimentary lines of investigation. Recent
review of positive GxE effects in ADHD has been encouraging (Nigg, Nikolas, & Burt, under
review), yet the implementation and concurrent examination of both behavioral and molecular
78

genetic tests of GxE effects for ADHD using the same moderator (e.g., self-blame) could provide
confirming support of any positive findings and also aid in understanding the ways in which GxE
processes may be operating for ADHD.
While prior work has demonstrated a link between appraisals of self-blame and behavior
problems (Rhoades, 2008), recent factor analytic work has reliably identified four factors from
the CPIC (Nigg et al., 2009). Yet, the phenotypic relationships among these newly-identified
scales as well as their relationships with ADHD and other externalizing disorders have not been
well-examined. In order to provide some guiding information regarding relationships among the
scales at the phenotypic level, three main sets of analyses were conducted prior to the main tests
of GxE effects. These analyses aimed to (1) examine the psychometric properties of these scales
across a wide age range, (2) to quantify the strength of relationships among the CPIC scales to
determine if higher reports of conflict correspond with higher reports of self-blame, and (3) to
determine if relationships among the CPIC scales are specific to ADHD or if they related to
externalizing behaviors more generally.
Summary of Results from Study 1. Results of the phenotypic analyses in Study 1 revealed
that, in general, the four factors of the CPIC demonstrated adequate internal consistency across
the wide age range of the twin sample. Additionally, while all the scales were generally
positively correlated with one another, Self-Blame evidenced only a modest association with
Conflict Properties. These results indicated that higher levels of frequency and intensity of
marital conflict (as reported by the child) do not necessarily correspond with higher levels of
self-blame. Lastly, both Self-Blame and Triangulation/Stability appeared to both be positively
associated with ADHD and other externalizing behaviors, including ODD and CD. Yet, when

79

taking into account the covariance among ADHD, ODD, and CD, only Self-Blame remained a
significant and unique predictor of both ADHD and CD.
While self-blame might at first seem to be linked more to internalizing behaviors (e.g.,
depression, anxiety), the results of Study 1 are in line with past work indicating that self-blame
shows a generally stronger relationship with externalizing problems in children (Fosco & Grych,
2008; Grych, Harold, & Miles, 2003; Rhoades, 2008). While the mechanisms underlying this
relationship remain unknown, some have postulated that children who blame themselves for their
parents‟ conflict may be more likely to act out as a way of re-directing the focus of the conflict
back upon themselves, and thereby relieving distress associated with observing conflict (Fosco &
Grych, 2008). While more work is needed, particularly longitudinal studies of the relationships
between youth appraisals of self-blame and later behavior problems, the results of Study 1
indicate a significant relationship between self-blame and ADHD.
Tests of GxE Effects. Studies 2 and 3 then proceeded to examine self-blame as a potential
moderator of genetic (and environmental) influences on ADHD. Study 2 made use of biometric
twin models in order to examine moderation of etiological contributors at the latent level,
whereas Study 3 examined interactions with a specific candidate marker, the promoter
polymorphism of the serotonin transporter gene (5HTTLPR). Results of both studies provided
evidence of significant GxE effects for ADHD involving self-blame. Study 2 results indicated
significant moderation of latent genetic and environmental influences on ADHD. When levels of
self-blame were low, the contribution of genetic factors to ADHD was high, whereas unique
environmental influences were low and shared environmental effects were zero. Importantly, as
reports of self-blame increased, genetic effects significantly decreased whereas unique
environmental contributions increased substantially. A crucial component of the biometric
80

models implemented in Study 2 is that they allowed for a direct test of genetic overlap between
self-blame and ADHD and thus provided a measure of potential confounding effects of geneenvironment correlation (rGE). Indeed, the genetic covariance between self-blame and ADHD
was significant, indicating that rGE effects may be operating between self-blame and ADHD
(i.e., similar genes are contributing to both self-blame and ADHD). Yet, even after controlling
for this overlap, significant increases in unique environmental influences for ADHD were still
observed with higher levels of self-blame.
Whereas Study 2 examined potential shifts in latent genetic and environmental
contributions to ADHD with varying levels of self-blame, Study 3 investigated interactions
between 5HTTLPR and self-blame within a clinically-diagnosed sample of ADHD youth. Genes
from the dopamine and norephinephrine transmission systems have been more commonlyexamined candidates for association with ADHD based on prior work involving the effects of
psychostimulants on these systems. However, selection of 5HTTLPR was based upon an
integrated theoretical model of ADHD, which postulates that the behavioral symptoms of the
disorder are due to deficits in behavioral and emotion regulation abilities, underpinned by
frontal-limbic circuitry (Nigg & Casey, 2006). Furthermore, 5HTTLPR has been shown to be a
functional polymorphism that is sensitive to environmental context (Lesch et al., 1996; Manuck
et al., 2004) and prior meta-analytic work has shown evidence of association with ADHD
(Faraone et al., 2005; Gizer, Ficks, & Waldman, 2009).
Importantly, however, the relationship between serotonin functioning and ADHD
remains somewhat unclear as both high and low serotonin activity has been associated with
impulsive and aggressive behavior in children with ADHD (Castellanos et al., 1993; Halperin et
al., 1994, 1997). Thus, tests for GxE effects examined two potential models of association.
81

These included (1) linear GxE interactions, which presume that only low or high serotonergic
functioning is associated with ADHD, and (2) non-linear GxE interactions, which presume that
both low and high serotonergic functioning are associated with the disorder. The latter model is
consistent with the notion that optimal functioning may occur at the mid-range of serotonin
functioning associated with the heterozygote genotype, a phenomenon called heterozygote
advantage, which has been noted for other disease models.
Results of Study 3 indicated significant non-linear 5HTTLPR x self-blame interactions
predicting teacher reports of total ADHD symptoms. That is, for youth with the low functioning
and high activity 5HTTLPR genotypes, there was a significant and positive relationship between
self-blame and ADHD symptoms. Yet, for those with the intermediate functioning 5HTTLPR
genotype, there was no relationship between self-blame and ADHD symptoms. Of note, findings
for total ADHD symptoms were observed for two separate measures of ADHD symptomatology.
These included a DSM-based measure, the ADHD Rating Scale, as well as the Conners‟ Rating
Scale. Findings also remained significant after controlling for gender, age, and ethnicity, as well
as the main effects of 5HTTLPR and self-blame. Furthermore, interactions remained significant
after controlling for potential confounds of comorbid psychopathology, including ODD, CD, and
major depressive disorder.
The use of ratings of ADHD symptoms in Study 3 also allowed for separate examination
of the inattention and hyperactivity symptom dimensions. Meta-analytic work has indicated that
the genetic etiology of the two symptom dimensions may differ (Nikolas & Burt, in press).
Examination across two separate measures of each dimension however did not reveal a
consistent pattern of association. Significant interactions were observed for inattention and
hyperactivity symptoms on the ADHD Rating Scale, although the interaction predicting
82

hyperactivity did not survive correction for multiple tests. Conversely, the interaction predicting
Cognitive Problems (e.g., Inattention) on the Conners‟ Rating Scale was not significant, whereas
the interaction predicting Hyperactivity remained significant after correction. Thus, the
interactions appear to generalize to the combined symptom dimensions of inattention and
hyperactivity, indicating that 5HTTLPR may represent a general genetic risk factor for ADHD.
Integration of GxE Results
Studies 2 and 3 both provided evidence of significant GxE effects on ADHD utilizing
two complimentary methodologies across two different samples (community-based twin sample
and clinically-diagnosed ADHD sample). Positive findings from both studies thus provide good
initial evidence that self-blame may have an important etiological role in ADHD. However,
integration of the results from each line of inquiry remains necessary for understanding how GxE
processes may be operating more generally for the disorder.
Study 2 indicated that genetic influences on ADHD were highest at low levels of selfblame. In other words, reports of higher levels of self-blame may be reducing genetic effects for
the disorder. At first, this may appear to stand in contrast to Study 3, which demonstrated a
positive association between self-blame and ADHD for those with low and high activity
5HTTLPR genotypes. Importantly, however, results from Study 2 indicate that while genetic
influences on ADHD decline with increases in self-blame, they were still significantly greater
than zero at the highest levels of self-blame. That is, while overall genetic influences decreased
with higher self-blame, there were still significant genetic contributions to ADHD within this
high risk environmental context. Thus, while genetic influences may be smaller in magnitude at
higher levels of self-blame, 5HTTLPR may be one of the genetic factors that exert influence on
ADHD within the context of high levels of environmental risk.
83

While a decline in genetic influences on ADHD was observed with higher levels of selfblame, the increase in unique environmental contributions to ADHD was substantial and
independent of rGE effects. This increase in e2 may reflect an increase in contributions from the
main effects of child-specific environmental risk factors. That is, reports of self-blame may
signal the presence of other child-specific environmental risk factors exerting main effects on
ADHD. Additionally, the observed increase in unique environmental contributors to ADHD
may also reflect an increase in contributions from gene x non-shared, or AxE interactions, which
have been shown to contribute to the non-shared environmental variance in behavioral genetic
models (Purcell, 2002). Increases in self-blame may be signaling the presence of other childspecific environmental risk processes (e.g., family conflict, parental discipline), which are then,
in turn, also interacting with genetic influences on ADHD. That is, the high-risk environmental
context associated with higher reports of self-blame may reflect increases in the contribution
from exchanges between genetic and child-specific environmental factors. In other words, high
levels of self-blame may serve to signal the presence of multiple AxE interactions that contribute
to ADHD. Indeed, Study 2 revealed that self-blame, while partially influenced by genetic factors
(a2=23%), was primarily influenced by child-specific environmental risk factors (e2=77%).
Thus, the interactions observed in Study 3 between 5HTTLPR genotype and self-blame may be
characterized as AxE interactions, which would then correspond with the increase in E observed
with higher levels of self-blame in Study 2.
Implications
Understanding the Genetic Etiology of ADHD. Previous work examining GxE
interactions for ADHD has primarily relied on molecular genetic approaches, which have
examined the statistical interactions between specific DNA polymorphisms and environmental
84

risk factors that have been previously associated with ADHD (e.g., DAT1 VNTR genotype and
prenatal cigarette exposure). Yet, findings of positive GxE effects on ADHD involving selfblame in the current project emerged using both behavioral genetic methodology, which
examined shifts in latent genetic and environmental contributions to the disorder, as well as with
molecular genetic approaches. Thus, both approaches provide complimentary lines of evidence
suggesting that self-blame may indeed play an etiological role in ADHD. This work thus
extends beyond prior reports of correlational relationships between self-blame and child behavior
problems and provides some initial evidence that self-blame may serve to moderate genetic
influences on ADHD.
Findings from the current project are in line with a recent review of initial work
examining GxE effects on ADHD, which has indicated mostly consistent moderating effects of
post-natal environmental risk factors on genetic risk for ADHD (Nigg, Nikolas, & Burt, under
review). This stands in contrast to GxE studies examining pre-or perinatal risk factors, such as
prenatal cigarette and alcohol exposure and low birth weight, which have shown less consistent
positive findings. In particular, variables relating to the family context, including parental
discipline, parental education, marital stability, and psychosocial adversity (conceptualized
broadly via measures of family stress, SES, and parental psychopathology) appear to moderate
the association between a variety of candidate genes and ADHD. Results from the current set of
studies provide additional evidence of the importance of the family environment for shaping the
development of behavioral and emotional regulation capabilities in children (Nigg, Hinshaw, &
Huang-Pollock, 2006). Further, this work, combined with previous reports, indicates that the
quality of the family environment may in fact play an important role in shaping the genetic
etiology of ADHD by moderating the influence of a variety of candidate genes. Thus, the family
85

environment may effectively turn “on and off” the association between candidate genes from the
catecholamine and serotonin neurotransmission systems and ADHD as children develop.
Additionally, the emergence of positive GxE effects in both Studies 2 and 3 may have
crucial implications for future work aiming to identify the numerous genetic markers that are
likely involved in the etiology of ADHD. First, positive GxE effects indicate differential
association between candidate genes and ADHD at various levels of environmental risk. That
said, the presence of GxE effects may be partially responsible for the numerous non-replications
in ADHD molecular genetic research (as well as for other phenotypes). Therefore, the search for
genetic factors operating for ADHD may be improved by considering the heterogeneity of the
environment, particularly the family environment, in order to obtain a clearer signal of
association. Understanding the ways in which particular genetic risk factors operate may thus be
enhanced by examination of association among probands that have common environmental
characteristics (in addition to a more homogenous phenotype as suggested by Todd et al., 2001).
Enhanced Versus Reduced Genetic Influences on ADHD. The results of Study 2 suggest
that genetic effects for ADHD may be attenuated as opposed to enhanced within the high-risk
environmental context, which stands in contrast to previous GxE work for ADHD. While this
phenomenon of reduced genetic influence within a risk environment has been observed for other
phenotypes, including IQ, personality, and reading disorder (Burt, 2008; Pennington et al., 2009;
Turkheimer et al., 2003), the notion of reduced genetic expression within a high-risk
environment has not been advanced for ADHD. It may be the case that certain types of
environments “override” genetic influences for ADHD. That is, perhaps certain family
environments serve to enhance ADHD symptoms for a wide-range of genetic vulnerabilities,
causing the environment to appear to exert a type of “main effect.” Alternatively, an increase in
86

the variance due to the non-shared environment may also be due to additional AxE interactional
processes which are subsumed in the variance due to this „main environmental effect.‟ Thus,
while the models may suggest different processes at first (i.e., enhanced versus attenuated
genetic risk in the high-risk environment), consideration of the impact of AxE interactions may
serve to integrate the two process models.
Pennington and colleagues (2009) described these processes as bioecological GxE effects
and diathesis-stress GxE effects. They argue that for some disorders, the high-risk environment
“turns on” genetic influences, which would be consistent with the traditional diathesis-stress
framework of the development of psychopathology. In contrast, the bioecological model of GxE
effects posits that genetic influences express their wide range of influence with the low-risk
environmental context. Increases in environmental risk thus serve to depress genetic influences.
Importantly, the authors argue that disorders may be defined by one type of process and suggest
that GxE effects for ADHD are operating via a diathesis-stress process, whereas GxE effects for
reading disorder are operating via a bioecological process.
While Pennington et al. (2009) initial analyses support this type of model (i.e., a specific
GxE process for each disorder), they also examined only one type of moderator (e.g., parental
education). Results from the current study suggest that different environmental moderators (as
opposed to overall disorders) may operate via both diathesis-stress and bioecological processes.
In other words, it may be the case that some environmental risk factors serve to activate some
genetic effects (and reduce others) and other environmental risk factors may serve to reduce
some genetic influences (but enhance others). While this idea is only a hypothesis and remains
to be tested, it allows for the possibility that different environmental risk factors may operate
differentially over time, either serving to enhance or reduce genetic influence for ADHD.
87

Furthermore, it allows for the possibility of interactional processes with the environment both
when overall estimates of genetic influence are high (e.g., a2 contains variance due to genetic
main effects and due to AxC interactions), as well as when genetic influences are lower and
unique environmental effects are high (e.g., e2 contains variance due to non-shared
environmental main effects and AxE interactions). Importantly however, these ideas remain
speculative and further examination of latent G x measured E interactions will be crucial for
gaining a greater understanding as to the type of interactional processes that are occurring for
ADHD as well as for other types of psychopathology.
Appraisals of the Environment as an Etiological Mechanism. The current projects
provided some evidence suggesting a role for appraisals of self-blame in the causal processes
that give rise to ADHD. These findings add support to Grych and Fincham‟s (1990) argument
for the cognitive-contextual framework, which posits that children‟s understanding of conflict
and the ways in which they attribute meaning to those experiences are what determines the
impact of conflict on behavior. That is, the objective presence of conflict does not seem to be as
important to the etiology of behavior problems as is the interpretation and attribution of the
conflict. Results of the current study indicated that while appraisals of self-blame were partially
influenced by genetic factors, the remaining variance was due to non-shared environmental
factors. In other words, appraisals of self-blame appear to be a child-specific as opposed to a
family-wide environmental risk factor that serves to create differences among siblings, even
when they are likely exposed to similar environmental contexts.
Similar findings have also emerged for different types of “environmental” processes,
such as stressful life events. Caspi and colleagues (2003) influential GxE investigation found
that genetic risk for depression was enhanced by the presence of multiple stressful life events.
88

Yet, further examination of these events determined that, like self-blame, they are partially
influenced by genetic factors (Kendler & Baker, 2007). That is, stressful life events are not
random but tend to occur more commonly in people that are already genetically vulnerable for
developing depression (Kendler et al., 1995; Jaffee & Price, 2007; Silberg et al., 1999). Recent
work has also indicated that the number of events may not be as important for understanding the
potential etiological role as is the perception of distress associated with the events (Monroe &
Reid, 2008). Thus, appraisals of the situation (e.g., how stressful is the event, how much does
one blame themselves for their parents‟ marital conflict) may be the key mechanisms by which
environmental contexts exert their risk on the etiology of psychopathology (e.g., by serving to
turn genetic risk on and off). Furthermore, these processes may serve as an entry point for
interventions. Cognitive-behavioral interventions that may serve to reduce the negative
attributions and resultant negative affect (e.g, via cognitive restructuring) associated with these
risk factors may aid in serving to reduce genetic vulnerability for the development and perhaps
the maintenance of psychopathology.
Limitations
There are some limitations to the current work that are important to note. First, both
samples were cross-sectional and thus the longitudinal relationships between conflict, appraisals
of self-blame, and ADHD behaviors could not be determined. Secondly, the size of the twin
sample did not permit the examination of separate estimates by sex or by age. As prior work
indicates that ADHD behaviors vary in both mean levels and variance by sex and by age, these
will be important considerations in future GxE studies of ADHD. Furthermore, the twin sample
relied on the CBCL and thus separate examination of moderation for inattention and
hyperactivity was not possible. Future work investigating GxE effects for each of the symptom
89

dimensions will likely be valuable for considering what risk processes are common to both as to
which processes may be unique to either inattention or hyperactivity.
The molecular genetic GxE study relied on case-control data from a single marker
(5HTTLPR). While results were robust to measure and survived correction for multiple tests, the
chance of false positive data in molecular studies remains high, as each marker likely only
accounts for a small proportion of the variance in ADHD. However, steps were taken to guard
against false positive results including the use of two dimensional measures for replication
effects as well as correction for multiple tests. Additionally, because the data were case-control,
population stratification effects, while unlikely, cannot be fully ruled out.
Overall Conclusion
Overall, the current project replicated prior work demonstrating a significant link
between appraisals of self-blame and ADHD. Yet, examination of GxE effects using both
behavioral and molecular genetic approaches took this link one step further by investigating a
potential etiological role of self-blame in ADHD. The results indicated that self-blame may
moderate genetic influences on ADHD. Examination of GxE effects using a behavioral genetics
approach indicated a substantial increase in non-shared environmental influences on ADHD with
increases in self-blame, a finding that remained when accounting for potential gene-environment
correlation. Furthermore, investigation of specific interactions involving 5HTTLPR and selfblame revealed that self-blame and ADHD were only significantly related for individuals with
low and high 5HTTLPR activity genotypes.

Both GxE investigations taken together suggest an

increase in the potential importance of genetic x non-shared (or AxE) interactional processes on
ADHD within a high-risk environmental context. These results are also consistent with a
developmental psychopathology conceptualization of the etiology of ADHD and point to the
90

potential utility of future investigations of the exchanges between genetic and environmental risk
processes in order to understand the mechanisms that give rise to psychopathology.

91

APPENDICES: TABLES AND FIGURES
Table 1. Internal Consistency Estimates of Four CPIC Factors across the Age Range.
____________________________________________________________________________
Scale
Overall α
5-7
8-10
11-13
14-16
Conflict Properties

.82

.66

.79

.88

.90

Triangulation/Stability

.88

.84

.87

.87

.91

Self-Blame

.85

.81

.83

.86

.87

Threat

.84

.77

.80

.87

.85

Note. Conflict Properties scale (n=11 items); Triangulation/Stability scale (n=13 items); SelfBlame scale (n=9 items); Threat scale (n=6 items).

92

Table 2. Bivariate Correlations among CPIC Factors and Externalizing Behaviors – Full
Sample
CP

TS

SB

TH

ADHD

ODD

CP

1.0

TS

.59**

1.0

SB

.21**

.36**

1.0

TH

.40**

.47**

.16*

1.0

ADHD .17**

.23**

.25**

.04

1.0

ODD

.10*

.15**

.16**

-.04

.51**

1.0

CD

.10*

.19**

.28**

-.06

.54**

.69**

CD

1.0

Note. CP=Conflict Properties, Triangulation/Stability, SB=Self-Blame, TH=Threat,
ADHD=DSM oriented CBCL raw score, ODD=DSM oriented CBCL raw score, CD=DSM
oriented CBCL raw score.
* p<.05, **p<.01

93

Table 3. Bivariate Correlations among CPIC Factors and Externalizing Behaviors – Ages 5-7
(n=33 twin pairs).
____________________________________________________________________________
CP
TS
SB
TH
ADHD
ODD
CD
CP

1.0

TS

.41**

1.0

SB

.03

.57**

1.0

TH

.45**

.48**

.16*

1.0

ADHD .09

.29**

.29**

- .04

1.0

ODD

-.28*

.05

.25**

-.21

.50**

1.0

CD

-.33 *

.06

.32**

-.13

.42**

.74**

1.0

Note. CP=Conflict Properties, Triangulation/Stability, SB=Self-Blame, TH=Threat,
ADHD=DSM oriented CBCL raw score, ODD=DSM oriented CBCL raw score, CD=DSM
oriented CBCL raw score.
* p<.05, **p<.01

94

Table 4. Bivariate Correlations among CPIC Factors and Externalizing Behaviors – Ages 8-10
(n=81 twin pairs).
CP
TS
SB
TH
ADHD
ODD
CD
CP

1.0

TS

.61**

1.0

SB

.27**

.32**

1.0

TH

.34**

.41**

.21**

1.0

ADHD .18*

.13

.20**

.08

1.0

ODD

.17*

.12

.08

.04

.46**

1.0

CD

.13

.07

.20**

-.08

.57**

.65**

1.0

Note. CP=Conflict Properties, Triangulation/Stability, SB=Self-Blame, TH=Threat,
ADHD=DSM oriented CBCL raw score, ODD=DSM oriented CBCL raw score, CD=DSM
oriented CBCL raw score.
* p<.05, **p<.01

95

Table 5. Bivariate Correlations among CPIC Factors and Externalizing Behaviors – Ages 11-13
(n=68 twin pairs).
CP
TS
SB
TH
ADHD
ODD
CD
CP

1.0

TS

.77**

1.0

SB

.21*

.23**

1.0

TH

.52**

.59**

.14

1.0

ADHD .16

.23*

.26**

.03

1.0

ODD

.17*

.17*

.29**

.01

.50**

1.0

CD

.22*

.25*

.33**

.06

.51**

.64**

1.0

Note. CP=Conflict Properties, Triangulation/Stability, SB=Self-Blame, TH=Threat,
ADHD=DSM oriented CBCL raw score, ODD=DSM oriented CBCL raw score, CD=DSM
oriented CBCL raw score.
* p<.05, **p<.01

96

Table 6. Bivariate Correlations among CPIC Factors and Externalizing Behaviors – Ages 14-16
(n=32 twin pairs).
CP
TS
SB
TH
ADHD
ODD
CD
CP

1.0

TS

.47**

1.0

SB

.30*

.35**

1.0

TH

.23

.31*

.12

1.0

ADHD .25*

.33**

.32**

- .16

1.0

ODD

.16

.32**

.08

- .12

.69**

1.0

CD

.24

.47**

.27*

-.09

.73**

.82**

1.0

Note. CP=Conflict Properties, Triangulation/Stability, SB=Self-Blame, TH=Threat,
ADHD=DSM oriented CBCL raw score, ODD=DSM oriented CBCL raw score, CD=DSM
oriented CBCL raw score.
* p<.05, **p<.01

97

Table 7. Hierarchical linear models examining CPIC factors as predictors of ADHD, ODD, and
CD behaviors among 5-17 year old twins.
Scale

Estimate

Std Error

t

p

Conflict Properties

.05

.04

1.4

.16

Triangulation/Stability

.08

.03

2.4

.02*

Self-Blame

.19

.05

3.8

<.001*

Threat

-.10

.06

-1.8

.073

Conflict Properties

.02

.02

.88

.38

Triangulation/Stability

.06

.02

2.3

.03*

Self-Blame

.08

.03

2.4

.02*

Threat

-.12

.04

-2.7

.007*

Conflict Properties

.01

.03

.32

.75

Triangulation/Stability

.10

.03

2.9

.004*

Self-Blame

.23

.05

4.9

<.001*

Outcome: ADHD Scale

Outcome: ODD Scale

Outcome: CD Scale

Threat
-.20
.06
-3.5
Note. Age and gender covaried in all models. * indicates significance at p<.05.

98

.001*

Table 8. Hierarchical linear models examining CPIC factors as predictors of ADHD, ODD, and
CD behaviors among 5-17 year old twins: Specificity of relationships.
Scale
Estimate
Std Error
t
p
Outcome: ADHD Scale
Conflict Properties

.04

.03

1.4

.17

Triangulation/Stability

.03

.03

.89

.37

Self-Blame

.08

.04

1.9

.05*

Threat

-.02

.05

-.17

.72

ODD

.38

.08

4.8

<.001*

CD

.32

.06

5.6

<.001*

Conflict Properties

.02

.04

.51

.61

Triangulation/Stability

.01

.05

.16

.88

Self-Blame

-.06

.03

-1.5

.13

Threat

-.01

.04

-.28

.78

ADHD

.20

.04

4.8

<.001*

CD

.60

.04

14.1

<.001*

Conflict Properties

-.02

.04

.61

.54

Triangulation/Stability

.07

.04

1.6

.11

Self-Blame

.12

.04

3.2

.001*

Threat

-.09

.04

-2.2

.03*

ADHD

.22

.04

5.6

<.001*

Outcome: ODD Scale

Outcome: CD Scale

ODD
.54
.04
14.2
Note. Age and gender covaried in all models. * indicates significance at p<.05.

99

<.001*

Table 9. Twin Intraclass Correlations for Mother-Rated CBCL ADHD Score: Overall and by
level of CPIC Self-Blame.
MZ
DZ
ADHD Score

.55**

.15*

CPIC Self-Blame: LOW (N=66)

.80**

.21*

CPIC Self-Blame: MODERATE (N=24)

.52*

.04

CPIC Self-Blame: HIGH (N=36)

.10

.01

Note. * p<.05, **p<.01

100

Table 10. Fit statistics for nested “straight” GxE models.
2

-2lnL
df
Δχ
Δdf
p
BIC
______________________________________________________________________________
Non-Linear Model
1143.019
413
NA
NA
NA
-534.626
Linear Model

1143.604

416

.585

3

.90

-550.374

Main Effects Model
1154.187
419
10.58
3
.014 -542.368
Note. In the main effects model, genetic, shared, and non-shared environmental parameter
estimates do not vary by self-blame. In the linear model, genetic and environmental parameter
estimates vary linearly. In the non-linear model, parameter estimates vary linearly and
quadratically. Each model is compared with the preceding model when calculating the change in
2
χ and degrees of freedom. Non-significant changes in chi-square indicate that the more
restrictive model (i.e., that model with fewer estimated parameters and therefore more degrees of
freedom) provides a better fit to the data. Lower or more negative values of BIC also indicate
the best-fitting model. By these criteria, the linear moderation model fit the data best.

101

Table 11.Unstandardized path and moderator estimates in the best-fitting “Straight” GxE model.

PATHS

LINEAR

QUADRATIC

a

c

e

A1

C1

E1

A2

C2

E2

.890*

.000

.468*

-.188*

.000

.184*

----

----

----

(.634, 1.07)

(-.546, .546)

(.323, .673)

(-.369, -.013) (-.354, .354)

ADHD

(.043, .301)

Note. Paths and moderators are presented; their 95% confidence intervals are presented below them in brackets. A, C, and E (both
upper and lower case) represent genetic, shared, and non-shared environmental parameters, respectively. In the left portion of the
table, the path estimates (i.e., a, c, and e) are presented. Because self-blame was divided into tertiles and those with scores within the
lowest third of the distribution were coded as zero, these estimates function as intercepts. Accordingly, the genetic and environmental
variance components at low levels of self-blame can be obtained simply by squaring these path estimates. At each subsequent level of
self-blame, significant linear (i.e., A1, C1, E1) moderators were added to these genetic and environmental paths using the following
2 2

2 2

equation: Unstandardized VarianceTotal = (a + A1(self-blame) + A2(self-blame )) + (c + C1(self-blame) + C2(self-blame )) + (e +
2 2

E1(self-blame) + E2(self-blame )) . The variance component estimates calculated this way are presented in Figure 1. The quadratic
moderation terms (e.g., A2, C2, E2) are not included as the linear model provided the best fit to the data (and are thus zero in the above
equation). * indicates that the estimate is statistically significant at p<.05.

102

Table 12.Unstandardized path and moderator estimates for the GxE in the presence of rGE model.
PATHS
a

c

e

A1

LINEAR MODERATION
C1
E1

Covariance (or overlap) between ADHD and self-blame
.489*

.325

.014

.003

-.243

.101

(.024, .798)

(-.715, .715)

(-.230, .205)

(-.289, .353)

(-.451, .451)

(.-.067, .273)

Variance unique to ADHD
.636*

.000

.426*

-.115

.000

.200*

(.287, .886)

(-.663, .663)

(.303, .598)

(-.347, .347)

(-.335, .335)

(.079, .303)

Note. Paths and moderators are presented; their 95% confidence intervals are presented below them in brackets. A, C, and E (both
upper and lower case) represent genetic, shared, and non-shared environmental parameters, respectively. In the left portion of the
table, the path estimates (i.e., a, c, and e) are presented. Because self-blame was divided into tertiles and those with scores within the
lowest third of the distribution were coded as zero, these estimates function as intercepts. Accordingly, the genetic and environmental
variance components at low levels of self-blame can be obtained simply by squaring these path estimates. At each subsequent level of
self-blame, significant linear (i.e., A1, C1, E1) moderators are added to these genetic and environmental paths using the following
2 2

2 2

equation: Unstandardized VarianceTotal = (a + A1(self-blame) + A2(self-blame )) + (c + C1(self-blame) + C2(self-blame )) + (e +
2 2

E1(self-blame) + E2(self-blame )) . The variance component estimates calculated this way are presented in Figure 2. * indicates that
the estimate is statistically significant at p<.05.

103

Variance

Figure 1. Unstandardized Genetic (A), Shared Environmental (C), and Non-Shared
Environmental Variance Contributions to ADHD by Level of Self-Blame: “Straight” GxE Model.

0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0

A
C
E

Low

Medium

High

CPIC Self-Blame
Note. Moderation analyses revealed that the increase in unique environmental contributions to
the variance in ADHD (E) was significant. In addition, the decrease in genetic contributions to
ADHD with increasing levels of self-blame was also significant.
For interpretation of the references to color in this and all other figures, the reader is referred to
the electronic version of this dissertation.

104

Variance

Figure 2. Unstandardized Genetic (A), Shared Environmental (C), and Non-Shared
Environmental Variance Contributions to ADHD by Level of Self-Blame: GxE in the Presence of
rGE Model.
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0

A
C
E

Low

Medium

High

CPIC Self-Blame
Note. Moderation analyses revealed that the increase in unique environmental contributions to
the variance in ADHD (E) was significant. The decrease in genetic contributions to ADHD with
increasing levels of self-blame was only significant when all shared environmental paths (C)
were constrained to be zero.

105

Table 13. Demographics and Descriptive Statistics for ADHD Cases and non-ADHD Controls.
Control
(N=137)
ADHD (N=151)
p

% Male

48.2

64.2

.006

% Caucasian

74.5

78.1

.46

% African American

14.6

11.3

.40

% Latino

6.6

3.3

.20

% Other

4.4

7.3

.30

Age (SD)

14.7 (2.4)

13.5 (2.8)

<.001

74.1

62.2

.03

74 (36)

63 (36)

.02

Inattentive Symptoms (SD)

1.2 (2.0)

7.3 (1.8)

<.001

Hyperactive Symptoms (SD)

.57 (1.1)

4.1 (3.2)

<.001

% ODD

8.8

27.2

<.001

% CD

1.5

9.9

.002

% MDD

10.2

21.9

.008

Cognitive Problems T Score

51.8 (12.3)

61.7 (14.6)

<.001

Hyperactivity T score

49.8 (9.3)

60.3 (11.8)

<.001

Inattentive Symptoms

.62 (1.7)

4.0 (3.2)

<.001

Hyperactive Symptoms

.32 (.10)

2.1 (2.8)

<.001

% High Activity (La/La)

.28

.25

.53

% Moderate Activity (La/Lg, La/s)

.42

.46

.49

% Low Activity (Lg/Lg, Lg/s, or s/s) .30

.29

.88

% Two Parent Households
Yearly Household Income (SD)

+

KSAD Diagnostics

Conners’ Teacher Report

ADHD Rating Scale – Teacher Report

5HTTLPR Genotype

Note. SD=standard deviation; ODD=Oppositional Defiant Disorder; CD=Conduct Disorder;
MDD=Major Depressive Disorder-Lifetime. KSADS-E Diagnostics based upon parent report.
+ Income in thousands of dollars.

106

Table 14. 5HTTLPR Allele Frequencies and Genotypes by Self-Reported Ethnic Group.
La

Lg

Short

.49

.06

.45

African-American (N=42)* .61

.19

.20

Latino (N=15)

.33

.07

.60

Other (n=18)

.28

.17

.55

Total

.48

.09

.43

Caucasian (N=229)

Note. *La and Lg allele frequency significantly greater in African-American participants
compared to non-African American participants, (La, p=.048; Lg, p=.007).

107

Table 15. ADHD and Externalizing Symptoms: Main Effect Tests of High, Moderate and Low Activity 5HTTLPR Genotypes.
High (La/La)

Moderate (La/Lg, La/s)

Low (Lg/Lg, Lg/s, s/s)

78

137

5.5 (3.5)

6.0 (3.2)

Hyperactive Symptoms

3.7 (3.3)

ODD Symptoms
CD Symptoms

eas

89

Inattentive Symptoms

e

p

p

p

5.5 (3.7)

.49

.51

.22

3.8 (3.3)

4.2 (3.3)

.60

.47

.52

1.7 (2.3)

1.6 (2.2)

1.8 (2.4)

.81

.80

.83

.33 (.65)

.57 (1.3)

.30 (.87)

.10

.09

.09

Cognitive

3.0 (3.5)

3.4 (4.0)

3.8 (4.7)

.47

.45

.44

Hyperactivity

4.3 (4.8)

3.6 (3.8)

3.1 (3.8)

.20

.24

.13

N
KSAD Diagnostics

Conners’ Teacher Report

ADHD Rating Scale Teacher Report
Inattention

4.3 (6.1)

4.9 (6.5)

.95

.96

.93

Hyperactivity
e

4.6 (6.7)
3.1 (5.9)

2.4 (4.5)

2.3 (4.6)

.49

.65

.46

eas

p = ethnicity corrected p value, p = ethnicity age, and sex corrected p value. Inattentive, Hyperactive, ODD, and CD symptoms are
parent report on the KSADS-E. Conners‟ Scores and ADHD Rating Scale scores are raw total scores on each measure (higher scores
signify symptoms/problems).

108

Figure 3. Relationship between CPIC Self-Blame and ADHD Rating Scale Total Symptoms
Score (Teacher Report) by 5HTTLPR Genotype Group.

Note. Scatter plot data and best-fitting regression line are color-coded by genotype group.
ADHD Rating Scale Total Symptom score is total raw score on likert scale (0-3) for all18 DSMIV ADHD items.

109

Figure 4. Relationship between CPIC Self-Blame and ADHD Rating Scale Inattention Symptom
Scores (Teacher Report) by 5HTTLPR Genotype Group.

Note. Scatter plot data and best-fitting regression line are color-coded by genotype group.
ADHD Rating Scale symptom dimension scores are the total raw scores on likert scale (0-3) for
the 9 inattention symptoms and 9 hyperactive-impulsive symptoms, respectively.

110

Figure 5. Relationship between CPIC Self-Blame and ADHD Rating Scale Hyperactivity
Symptom Scores (Teacher Report) by 5HTTLPR Genotype Group.

Note. Scatter plot data and best-fitting regression line are color-coded by genotype group.
ADHD Rating Scale symptom dimension scores are the total raw scores on likert scale (0-3) for
the 9 inattention symptoms and 9 hyperactive-impulsive symptoms, respectively.

111

Figure 6. Relationship Between CPIC Self-Blame and Conners’ ADHD Index (Teacher Report)
by Genotype Group.

Note. Scatter plot data and best-fitting regression line are color-coded by genotype group.
ADHD Rating Scale Total Symptom score is total raw score on likert scale (0-3) for ADHD
Index items.

112

Figure 7. Relationship Between CPIC Self-Blame and Conners’ Cognitive Problems (Teacher
Report) by Genotype Group.

Note. Scatter plot data and best-fitting regression line are color-coded by genotype group.
ADHD Rating Scale Total Symptom score is total raw score on likert scale (0-3) for ADHD
Index items.

113

Figure 8. Relationship Between CPIC Self-Blame and Conners’ Hyperactivity (Teacher Report)
by Genotype Group.

Note. Scatter plot data and best-fitting regression line are color-coded by genotype group.
ADHD Rating Scale Total Symptom score is total raw score on likert scale (0-3) for ADHD
Index items.

114

REFERENCES

115

REFERENCES
Achenbach, T., & Rescorla, L.A. (2001). Manual for the Child Behavior Checklist and 2001
Profile. Burlington, VT:University of Vermont, Department of Psychiatry.
American Psychiatric Association. (2000). Diagnostic and Statistical Manual of Mental
Disorders (4th ed., text revision). Washington, DC: Author.
Arnsten, A.F., Goldman-Rakic, P.S. (1998). Noise impairs prefrontal cortical cognitive function
in monkeys: Evidence for a hyperdopaminergic mechanism. Archives of General
Psychiatry 55, 362-368.
Banerjee, T.D., Middleton, F., & Faraone, S.V. (2007). Environmental risk factors for attention
deficit hyperactivity disorder. Acta Paediatrica, 96, 1269-1274.
Barkley, R.A. (2006). Attention-deficit hyperactivity disorder: A handbook for diagnosis and
treatment, 3rd edition. New York: The Guilford Press.
Barry, R.A., Kochanska, G., & Philibert, R.A. (2008). GxE interaction in the organization of
attachment: Mother‟s responsiveness as a moderator of children‟s genotypes. Journal of
Child Psychology and Psychiatry, 49, 1313-1320.
Beitchman, J.H., Baldassarra, L., Mik, H., DeLuca, V., King, N., Bender, D,. Ehtesham, S.,
& Kennedy, J.L. (2006). Serotonin transporter polymorphisms and persistent, pervasive
childhood aggression. American Journal of Psychiatry, 163, 1103-1105.
Bellgrove, M. A., Hawi, Z., Kirley, A., Gill, M.,& Robertson, I.H. (2005). Dissecting the
attention deficit hyperactivity disorder phenotype: Sustained attention, response
variability, and spatial attentional asymmetries in relation to dopamine transporter
(DAT1) genotype. Neuropsychologia, 43, 1847-1857.
Bergen, S.E., Gardner, C.O., & Kendler, K.S. (2007). Age-related changes in heritability of
behavioral phenotypes over adolescence and young adulthood: A meta-analysis. Twin
Research and Human Genetics, 10, 423-433.
Bennet, A., Lesch, K.P., Heils, A., Long, J.C., Lorenz, J.G., Schoaf, S.E., Champoux, M.,
Suomi, S.J., Linnoila, M.V., & Higley, J.D. (2002). Early experience and serotonin
transporter gene interact to influence primate CNS function. Molecular Psychiatry, 7,
118-122.
Biederman, J., Faraone, S.V., Keenan, K., Benjamin, J., Krifcher, B., Moore, C. et al. (1992).
Further evidence for family-genetic risk factors in attention deficit hyperactivity disorder:
Patterns of comorbidity in probands and relatives in psychiatrically and pediatrically
referred samples. Archives of General Psychiatry, 49, 728-738.
Biederman, J., Faraone, S.V., & Monuteaux, M. (2002). Differential effect of environmental
116

adversity by gender: Rutter‟s index of adversity in a group of boys and girls with and
without attention-deficit hyperactivity disorder. American Journal of Psychiatry, 159,
1556-1662.
Biederman, J., Milberger, S., Faraone, S.V., Kiely, K., Guite, J., Mick, E., Ablon, J.S.,
Warburton,R., & Reed, E. (1995). Family environment risk factors for attention-deficit
hyperactivity disorder: A test of Rutter‟s indicators of adversity. Archives of General
Psychiatry, 52, 464-470.
Bloom, D.R., Levin, H.S., Ewing-Cobbs, L., Saunders, A.E., Song, J., Fletcher, J.M., et al.
(2001). Lifetime and novel psychiatric disorders after pediatric traumatic brain injury.
Journal of the American Academy of Child and Adolescent Psychiatry, 40, 572-579.
Breslau, N., Breslau, J., Peterson, E., Miller, E., Lucia, V.C., Bonhert, K. &, Nigg, J. (2009.
Shifts in academic achievement from age 11 to age 17: The role of early development of
attention problems. Pediatrics, 123, 1472-1476.
Breslau, J., Lane, M., Sampson, N., & Kessler, R.C. (2008). Mental disorders and subsequent
educational attainment in a US National Sample. Journal of Psychiatry Research, 42,
708-716.
Brookes, K., Xu, X., Chen, W., Zhou, K., Neale, B., Lowe, N., et al. (2006). The analysis of 51
genes in DSM-IV combined type attention deficit hyperactivity disorder: Association
signals in DRD4, DAT1, and 16 other genes. Molecular Psychiatry, 11, 934-953.
Buehler, C., Anthony, C., Krishnakumar, A., Stone, G., Gerard, J., & Pemberton, S. (1997).
Interparental conflict and youth behavior problems: A meta-analysis. Journal of Child
and Family Studies, 6, 233-247.
Buehler, C., Lange, G., & Franck, K.L. (2007). Adolescents‟ cognitive and emotional responses
to marital hostility. Child Development, 78, 775-789.
Buhrmester, D., Comparo, L., Christiansen, A., Gonzales, L.S., & Hinshaw, S.P. (1992).
Mothers and fathers interacting in dyads and triads with normal and hyperactive sons.
Developmental Psychology, 28, 500-509.
Burt, S.A. (2008). Gene-environment interactions and their impact on the development of
personality traits. Psychiatry, 7, 507-510.
Burt, S.A. (2009). Rethinking environmental contributions to child and adolescent
psychopathology: A meta-analysis of shared environmental influences. Psychological
Bulletin, 135, 608-637.
Campbell, S.B. (1994). Hard-to-manage preschool boys: Externalizing behavior, social
competence, and family context at two-year follow-up. Journal of Abnormal Child
Psychology, 22, 147-166.
117

Campbell, S.B., Breaux, A.M., Ewing, L.J., & Szumowski, E.K. (1986). Correlates and
predictors of hyperactivity and aggression: A longitudinal referred study of parentreferred problem preschoolers. Journal of Abnormal Child Psychology, 14, 217-234.
Campbell, S.B., Pierce, E.W., March, C.L., Ewing, L.J., & Szumowski, E.K. (1994). Hard-to
manage preschool boys: Symptomatic behavior across contexts and time. Child
Development, 65, 836-851.
Carlson, E.A., Jacobvits, D.J., & Sroufe, L.A. (1995). A developmental investigation of
inattentiveness and hyperactivity. Child Development, 66, 37-54.
Caspi, A., Moffitt, T.E., Morgan, J., Rutter, M., Taylor, A., Arsenault, L., Tully, L., Jacobs, C.,
Kim-Cohen, J., & Polo-Tomas, M. (2004). Maternal expressed emotion predicts
children‟s antisocial behavior problems: Using monozygotic-twin differences to identify
environmental effects on behavioral development. Developmental Psychology, 40, 149161.
Caspi, A., Sugden, K., Moffitt, T.E., Taylor, A., Craig, I.W., Harrington, H., McClay, J., Mill, J.,
Martin, J., Braithwaite, A., & Poulton, R. (2003). Influence of life stress on depression:
Moderation by a polymorphism in the 5HTT gene. Science, 301, 291-293.
Castellanos, F.X., Elia, J., Kruesi, M.J., Gulotta, C.S., Mefford, I.N., Potter, W.Z., Ritchie, G.F.,
& Rapoport, J.L. (1994). Cerebrospinal fluid monoamine metabolites in boys with
attention-deficit hyperactivity disorder. Psychiatry Research 52, 305-316.
Castellanos, F.X., Sharp, W.S., Gottesman, R.F., Greenstein, D.K., Giedd, J.N., & Rapoport,
J.L. (2003). Anatomic brain abnormalities in monozygotic twins discordant for attention
deficit hyperactivity disorder. American Journal of Psychiatry, 160, 1693-1696.
Clarke, R.A., Murphy, D.L., & Constantino, J.N. (1999). Serotonin and externalizing behavior in
young children. Psychiatry Research 86, 29-40.
Coccaro, E.F., Siever, L.J., Klar, H.M., Maurer, G., Cochrane, K., Cooper, T.B., Mohs, R.C.,
Davis, K.L. (1989). Serotonergic studies in patients with affective and personality
disorders: Correlates with suicidal and impulsive aggressive behavior. Archives of
General Psychiatry 46, 587-599.
Connors, C.K. (1997). Connors’ Rating Scales – Revised. Toronto, Ontario, Canada: Multi
Health Systems, Inc.
Cornish. K.M., M., T., Savage, R., Swanson, J., Morisano, D., Butler, N, et al. (2005).
Association of the dopamine transporter (DAT1) 10/10 repeat genotype with ADHD
symptoms and response inhibition in a general population sample. Molecular Psychiatry,
10, 686-698.
118

Counts, C.A., Nigg, J.T., Stawicki, J.A., Rappley, M.D., & Von Eye, A. (2005). Family adversity
in DSM-IV ADHD Combined and Inattentive subtypes as associated disruptive behavior
problems. Journal of the American Academy of Child and Adolescent Psychiatry, 44,
690-698.
Culbert, K.C., Breedlove, S.M., Burt, S.A., & Klump, K.L. (2008). Prenatal hormone exposure
and risk for eating disorders: A comparison of opposite-sex and same-sex twins. Archives
of General Psychiatry, 65, 329-336.
Cummings, E.M., & Davies, P.T. (1994). Children and marital conflict: The impact of marital
dispute and resolution. New York: The Guilford Press.
Cummings, E.M., Davies, P.T., & Simpson, K.S. (1994). Marital conflict, gender, children‟s
appraisals, and coping efficacy as mediators of child adjustment. Journal of Family
Psychology, 8, 141-149.
Cummings, E.M., Goeke-Morey, M.C., & Papp, L.M. (2004). Everyday marital conflict and
child aggression. Journal of Abnormal Child Psychology, 32, 191-202.
Cummings, E.M., Goeke-Morey, M.C., Papp, L.M., & Dukewich, T. (2002). Children‟s
responses to mother‟s and father‟s emotionality and tactics in marital conflict in the
home. Journal of Family Psychology, 16, 478-492.
Curran, S., Purcell, S., Craig, I., Asherson, P., & Sham, P. (2005). The serotonin transporter
gene as a QTL for ADHD. American Journal of Medical Genetics Part B:
Neuropsychiatric Genetics 134B, 47-47.
Davies, P.T., Harold, G.T., Goeke-Morey, M.C., & Cummings, E.M. (2002). Child emotional
security and interparental conflict. Monographs of the Society for Research in Child
Development, 67, 116-127.
Davies, P.T., Sturge-Apple, M.L., Cicchetti, D., & Cummings, E.M. (2008). Adrenocortical
underpinnings of children‟s psychological reactivity to interparental conflict. Child
Development, 79, 1693-1706.
Duncan, G.J., Dowsett, C.J., Claessens, A., Magnuson, J., Huston, A.C., Klebanoc, P., Pagani,
L., Feinstein, L., Engel, M., Brooks-Gunn, J., Sexton, H., Duckworth, K., & Japel, C.
(2007). School readiness and later achievement. Developmental Psychology, 43, 14281446.
Du Paul, G.J., McGoey, K.E., Eckert, T.L., & VanBrackle, J. (2001). Preschool children with
attention-deficit hyperactivity disorder: Impairments in behavioral, social, and school
functioning. Journal of the American Academy of Child and Adolescent Psychiatry, 40,
508-515.
119

Du Paul, G.J., Power, T.J., Anastopoulos, A.D., & Reid, R. (1998): The ADHD rating scale –
IV:Checklists, norms, and clinical interpretation. New York: Guilford Press.
Eaves, L.J. (2006). Genotype x environment interaction in psychopathology: Fact or artifact?
Twin Research and Human Genetics 9, 1-8.
El-Sheikh, M., & Harger, J. (2001). Appraisals of marital conflict and children‟s adjustment,
health, and physiological reactivity. Developmental Psychology, 37, 875-885.
El-Sheikh, M., Kouros, C.D., Erath, S., Cummings, E.M., Keller, P., & Staton, L. (2009).
Interactions among marital conflict, sympathetic, and parasympathetic nervous system
activity in the prediction of children‟s externalizing problems. Monographs of the Society
for Research in Child Development, 74, 19-34.
Evans, J., Platts, S., Lightman, S., & Nutt, D. (2000). Impulsiveness and the prolactin response
to d-flenfluramine. Psychopharmacology, 149, 147-152.
Faraone, S.V., Biederman, J., Mich, E., Williamson, S., Wilens, T., Spencer, T., et al. (2000).
Family study of girls with attention deficit hyperactivity disorder. American Journal of
Psychiatry, 157, 1077-1083.
Faraone, S.V., Perlis, R.H., Doyle, A.E., Smoller, J.W., Goralnick, J.J., Holmgren, M.A. &
Sklar, P. (2005). Molecular genetics of ADHD. Biological Psychiatry, 57,1313-1323.
Friedel, S., Saar, K., Sauer, S., Dempfle, A., Walitza, S., Renner, T., et al. (2007). Association
and linkage of allelic variants of the dopamine transporter gene and ADHD. Molecular
Psychiatry, 12, 923-933.
Flory, J.D., Newcorn, J.H., Miller, C., Harty, S., & Halperin, J.M. (2007). Serotonergic function
in children with attention-deficit hyperactivity disorder: Relationship to later antisocial
personality disorder. British Journal of Psychiatry, 190, 410-414.
Fosco, G.M., & Grych, J.H. (2008). Emotional, cognitive, and family systems mediators of
children‟s adjustment to interparental conflict. Journal of Family Psychology, 22, 843854.
Frankle, W.G., Huang, Y., Hwang, D.R., Talbot, P.S., Slifstein, M., Van Heertum, R., Abu
Dargham, A., & Laurelle, M. (2004). Comparative evaluation of serotonin transporter
radioligands 11C-DASB and 11C-McN 5652 in healthy humans. Journal of Nuclear
Medicine, 45, 682-694.
Gerard, J.M., Beuhler, C., Franck, K., & Anderson, O. (2005). In the eyes of the beholder:
Cognitive appraisals as mediators of the association between interparental conflict and
youth maladjustment. Journal of Family Psychology, 19, 376-384.

120

Gizer, I.R., Ficks, C., & Waldman, I.D. (2009). Candidate gene studies of ADHD: A metaanalytic review. Human Genetics, 126, 51-90.
Greenberg, B.D., Tolliver, T.J., Huang, S.J., Li, Q., Bengel, D., Murphy, D.L. (1999). Genetic
variation in the serotonin transporter promoter region affects serotonin uptake in human
blood platelets. American Journal of Medical Genetics, 88, 83-87.
Grych, J.H., Fincham, F.D., Jouriles, E.N., & McDonald, R.N. (2000). Interparental conflict and
child adjustment: Testing the mediational role of appraisals in the cognitive-contextual
framework. Child Development, 71, 1648-1661.
Grych, J.H., & Fincham, F.D. (1990). Marital conflict and children‟s adjustment: A cognitive
contextual framework. Psychological Bulletin, 108, 267-290.
Grych, J.H., Harold, G., & Miles, C. (2003). A prospective investigation of appraisals as
mediators of the link between interparental conflict and child adjustment. Child
Development, 74, 1176-1193.
Grych, J.H., Seid, M., & Fincham, F.D. (1992). Assessing marital conflict from the child‟s
perspective: The Children‟s Perception of Interparental Conflict Scale. Child
Development, 63, 558-572.
Haberstick, B.C., Smolen, A., & Hewitt, J.K. (2006). Family-based association test of 5HTTLPR
and aggressive behavior in a general population sample of children. Biological
Psychiatry, 59, 836-843.
Hallikainen, T., Saito, T., Lachman, H.M., Volavka, J., Pohjalainen, T., Ryynanen, O.P.,
Kauhanen, J., Syvalahti, E., Hietala, J., & Tiihonen, J. (1999). Association between low
activity serotonin transporter promoter genotype and early-onset alcoholism with habitual
violent impulsive behavior. Molecular Psychiatry, 4, 385-388.
Halperin, J.M., Sharma, V., Siever, L.J., Schwartz, S.T., Matier, K., Wornell, G., & Newcorn,
J.H. (1994). Serotonergic function in aggressive and nonaggressive boys with attentiondeficit hyperactivity disorder. American Journal of Psychiatry, 151, 243-248.
Harold, G.T., Fincham, F.D., Osborne, L.N., & Conger, R.D. (1997). Mom and dad are at it
again: Adolescent perceptions of marital conflict and adolescent psychological distress.
Developmental Psychology, 33, 333-350.
Harold, G.T., Shelton, K.H., Goeke-Morey, M.C., & Cummings, E.M. (2004). Marital conflict,
child emotional security about family relationships, and child adjustment. Social
Development, 13, 350-376.
Haslam, N., Williams, B., Prior, M., Haslam, R., Graetz, B., & Sawyer, M. (2006). The latent
121

structure of attention-deficit hyperactivity disorder: A taxometric analysis. Australian
and New Zealand Journal of Psychiatry, 40, 639-647.
Hawi, Z., Dring, M., Kirley, A., Foley, D., Kent, L., Craddock, N., et al. (2002). Serotonergic
system and attention deficit hyperactivity disorder (ADHD): a potential susceptibility
locus at the 5-HT (1B) receptor gene in 273 nuclear families from a multi-center sample.
Molecular Psychiatry, 7, 718-725.
Hu, X.Z., Lipsky, R.H., Zhu, G., Akhtar, L.A., Taubman, J., Greenberg, B.D., Xu, K., Arnold,
P.D., Richter, M.A., Kennedy, J.L., Murphy, D.L., & Goldman, D. (2006). Serotonin
transporter promoter gain-of-function genotypes are linked obsessive-compulsive
disorder. American Journal of Human Genetics, 78, 815–826.
Hudziak, J.J., Copeland, W., Stanger, C., & Wadsworth, M. (2004). Screening for DSM-IV
externalizing disorders with the Child Behavior Checklist: A Receiver-Operating
characteristic analysis. Journal of Child Psychology and Psychiatry and Allied
Disciplines, 45, 1299-1307.
Jaffee, S.R., & Price, T.S. (2007). Gene-environment correlations: A review of the evidence and
implications for prevention of mental illness. Molecular Psychiatry, 12, 432-442.
Johnston, C., & Mash, E.J. (2001). Families of children with attention-deficit/hyperactivity
disorder: Review and recommendations for future research. Clinical Child and Family
Psychology Review, 4, 183-207.
Jouriles, E.N., Collazos, S. Spiller, L., Stephens, N., McDonald, R., & Swank, P. (2000).
Variability in adjustment of child of battered women: The Role of child appraisals of
interparental conflict. Cognitive Therapy and Research, 24, 233-249.
Jouriles, E.N., Murphy, C.M., & O‟Leary, K.D. (1989). Interspousal aggression, marital discord,
and child problems. Journal of Consulting and Clinical Psychology, 57, 453-455.
Keller, M.C., & Miller, G. (2006). Resolving the paradox of common, harmful, heritable mental
disorders: Which evolutionary model works best? Behavioral and Brain Sciences, 29,
385-404.
Kendler, K.S., & Baker, J.H. (2007). Genetic influences on measures of the environment: A
systematic review. Psychological Medicine, 37, 615-626.
Kendler, K.S., Kessler, R.C., Walters, E.E., MacLean, C.J., Neale, M.C., Heath, A.C., & Eaves,
L.J. (1995). Stressful life events, genetic liability, and the onset of an episode of major
depression in women: Evidence of genetic control of sensitivity to the environment.
American Journal of Psychiatry, 152, 833-842.

122

Kenny, D.A., Kashy, D.A., & Cook, D.L. (2006). Dyadic data anaylsis. New York: The Guilford
Press.
Kent, L., Doerry, U., Hardy, E., Parmar, R., Gingell, K., Hawi, Z., Kirley, A., Lowe, N.,
Fitzgerald, M., Gill, M., & Craddock, N. (2002). Evidence that variation at the serotonin
transporter
gene influences susceptibility to attention deficit hyperactivity disorder
(ADHD): Analysis and pooled analysis. Molecular Psychiatry, 7, 908-912.
Keppel, G., & Wickens, T.D. (2004). Design and analysis: A researcher’s handbook – 4th
edition. Upper Saddle River, New Jersey: Pearson/Prentice Hall.
Klump, K.L., & Burt, S.A. (2006). The Michigan State University Twin Registry: Genetic,
environmental, and neurobiological influences on behavior across development. Twin
Research and Human Genetics, 9, 971-979.
Knopik, V.S., Sparrow, E.P., Madden, P.A., Bucholz, K.K., Hudziak, J.J., Reich, W., et al.
(2005). Contributions of parental alcoholism, parental substance exposure, and genetic
transmission to child ADHD risk: A female twin study. Psychological Medicine, 35, 625635.
Kraft, J.M., Slager, S.L., McGrath, P.J., & Hamilton, S.P. (2005). Sequence analysis of the
serotonin transporter and associations with antidepressant response. Biological
Psychiatry, 58, 374-381.
Kruesi, M.J., Hibbs, E.D., Zahn, T.P., Keysor, C.S., Hamburger, S.D., Bartko, J.J., & Rapoport,
J.L. (1992). A 2-year prospective follow-up study of children and adolescents with
disruptive behavior disorders: Prediction by cerebrospinal fluid 5-hydroxyinoleacetic
acid, homovanillic acid, and autonomic measures? Archives of General Psychiatry, 49,
429-435.
Kruesi, M.J., Rapoport, J.L., Hamburger, S., Hibbs, E., Potter, W.Z., Lenane, M., & Brown, G.L.
(1990). Cerebrospinal fluid monamine metabolites, aggression, and impulsivity in
disruptive behavior disorders of children and adolescents. Archives of General
Psychiatry, 47, 419-426.
Langley, K., Turic, D., Rice, F., Holmans, P., van den Bree, M.B., Craddock, N., Kent, L.,
Owen, M.J., O‟Donovan, M.C., & Thapar, A. (2008). Testing for gene x environment
effects in attention deficit hyperactivity disorder and associated antisocial behavior.
American Journal of Medical Genetics: Part B Neuropsychiatric Genetics, 147B, 49-53.
Laucht, M., Treutlein, J., Blomeyer, D., Buchmann, A.F., Schmid, B., Becker, K., et al (2009).
Interaction between the 5HTTLPR serotonin transporter polymorphism and
environmental adversity for mood and anxiety psychopathology: Evidence from a highrisk community sample of young adults. International Journal of
Neuropsychopharmacology, 20, 1-11.
123

Lehn, H., Derks, E.M., Hudziak, J.J., Heutink, P., van Beijsterveldt, T.C., & Boomsma, D.I.
(2007). Attention problems and attention-deficit/hyperactivity disorder in discordant and
concordant monozygotic twins: Evidence of environmental mediators. Journal of the
American Academy of Child and Adolescent Psychiatry, 46, 83-91.
Lesch, K.P., Bengel, D. Heils, A., Sabol, S.Z., Greenberg, B.D., Petri, S., Benjamin, J., Muller,
C.R., Hamer, D.H., Murphy, D.L. (1996). Association of anxiety-related traits with a
polymorphism in the serotonin transporter gene regulatory region. Science, 274, 15271531.
Levy, F., Hay, D., McStephen, M, Wood, C., & Waldman, I. (1997). Attention deficit
hyperactivity disorder: A category of a continuum? Genetic analysis of a large-scale twin
study. Journal of the American Academy of Child and Adolescent Psychiatry, 36, 737744.
Manor, I., Eisenberg, J., Tyano, S., Sever, Y., Cohen, H., Ebstein, R.P., & Kotler, M. (2001).
Family-based association study of the serotonin transporter promoter polymorphism (5HTTLPR) in attention-deficit hyperactivity disorder. American Journal of Medical
Genetics, 105, 91-95.
Manuck, S.B., Flory, J.D., Muldoon, M.F., & Ferrell, R.E. (2002). Central nervous system
serotonergic responsivity and aggressive disposition in men. Physiology and Behavior,
77, 705-709.
Martel, M.M., & Nigg, J.T. (2006). Child ADHD and personality/temperament traits of reactive
and effortful control, resiliency, and emotionality. Journal of Child Psychology and
Psychiatry, 47, 1175-1183.
Markon, K.E., & Krueger, R.F. (2004). An empirical comparison of information-theoretic model
selection criteria. Behavior Genetics, 34, 593-610.
McGue, M., & Bouchard, T.J. (1984). Adjustment of twin data for the effects of age and sex.
Behavior Genetics, 14, 325-343.

Meulenbelt, I., Droog, S., Trommelen, G.J., Boomsma, D.I., & Slagboom, P. (1995). High-yield
non-invasive human genomic DNA isolation method for genetic studies in geographically
dispersed families and populations. American Journal of Human Genetics, 57, 12521254.
Miller, T., Nikolas, M., & Nigg, J.T. (in preparation). Relationships among children‟s appraisals
of marital conflict and clinically diagnosed attention-deficit hyperactivity disorder.

124

Moffitt, T.E., Brammer, G.L., Caspi, A., Fawcett, J.P., Raleigh, M., Yuwiler, A., & Silva, P.
(1998). While blood serotonin relates to violence in an epidemiological study. Biological
Psychiatry, 43, 446-457.
Moffitt, T.E., Caspi, A., & Rutter, M. (2005). Strategy for investigating interactions between
measured genes and measured environments. Archives of General Psychiatry 62, 473481.
Monroe, S.M., & Reid, M.W. (2008). Gene-environment interactions in depression research:
Genetic polymorphisms and life stress polyprocedures. Psychological Science, 19, 947956.
Nadder, T.S., Rutter, M., Silberg, J.L., Maes, H.H., & Eaves, L.J. (2002). Genetic effects on the
variation and covariation of attention-deficit hyperactivity disorder (ADHD) and
oppositional-defiant/conduct disorder symptomatologies across informant and occasion
of measurement. Psychological Medicine, 32, 39-53.
Nakamura, M., Ueno, S., Sano, A., & Tanabe, H. (2000). The human serotonin transporter gene
linked polymorphism (5HTTLPR) shows ten novel allelic variants. Molecular
Psychiatry, 5, 32-38.
Neale, M.C. (1999). Mx: Statisical modeling (5th ed.) Richmond, VA: Department of Psychiatry.
Neuman, R.J., Heath, A., Reich, W., Bucholz, K.K., Madden, P.A.F., Sun, L., et al (2001).
Latent class analysis of ADHD and comorbid symptoms in a population sample of
adolescent female twins. Journal of Child Psychology and Psychiatry, 42, 933-942.
Neuman, R.J., Lobos, E., Reich, W., Henderson, C.A., Sun, L.W., & Todd, R.D. (2007). Prenatal
smoking exposure and dopaminergic genes interact to cause a severe ADHD subtype.
Biological Psychiatry, 61, 1320-1328.
Nigg, J.T. (2006). What causes ADHD? Understanding what goes wrong and why. New York:
The Guilford Press.
Nigg, J.T., & Casey, B.J., (2006). An integrative theory of attention-deficit hyperactivity
disorder based on the cognitive and affective neurosciences. Development and
Psychopathology, 17, 785-806.
Nigg, J.T., Hinshaw, S.P., & Huang-Pollock, C. (2006). Disorders of attention and impulse
regulation. In Developmental Psychopathology, Volume 3: Risk, disorder, and adaptation
(2nd edition). D Cicchetti & D.J. Cohen (Eds.) pp. 458-403. Hoboken, New Jersey: John
Wiley & Sons.

125

Nigg, J.T., Nikolas, M.A., Miller, T., Burt, S.A., Klump, K.L., & von Eye, A. (2009). Factor
structure of the child perception of marital conflict scale for studies of youth with
externalizing behavior problems. Psychological Assessment, 21, 450-456.
Nikolas, M., & Burt, S.A. (in press). Genetic and environmental influences on ADHD
symptom dimensions of inattention and hyperactivity: A meta-analysis. Journal of
Abnormal Psychology.
O‟Keane, V., Maloney, E., O‟Neil, H., O‟Connor, A., Smith, C., Dinan, T.G. (1992). Blunted
prolactin response to d-fenfluramine in sociopathy: Evidence for subsensitivity of central
serotonergic function. British Journal of Psychiatry, 160, 643-646.
Oades, R.D., Lasky-Su, J., Christiansen, H., Faraone, S.V., Sonuga-Barke, E.J.S.,
Banaschewski, T., Chen, W., Anney, R.J.L., Buitelaar, J.K., Ebstein, R.P., Franke, B.,
Gill, M., Miranda, A., Roeyers, H., Rothenberger, A., Sergeant, J.A., Steinhausen, H.,
Taylor, E.A., Thompson, M., & Asherson, P. (2008). The influence of serotonin-and
other genes on impulsive behavioral aggression and cognitive impulsivity in children
with attention-deficit hyperactivity disorder (ADHD): Findings from a family-based
association test (FBAT) analysis. Behavioral and Brain Functions, 4:48.
Pauli-Pott, U., Friedl, S., Hinney, A., & Hebebrand, J. (2009). Serotonin transporter gene
polymorphism (5HTTLPR), environmental conditions, and developing negative
emotionality and fear in early childhood. Journal of Neural Transmission, 116, 503-512.
Peeters, H., Van Gestel, S., Vlietinck, R., Derom, C., & Derom, R. (1998). Validation of a
telephone zygosity questionnaire in twins of known zygosity. Behavior Genetics, 28,
159-161.
Pennington, B.F., McGrath, L.M., Rosenberg, J., Barnard, H., Smtih, S.D., Willcutt, E.G., &
Friend, A. (2009). Gene x environment interactions for reading disability and attentiondeficit hyperactivity disorder. Developmental Psychology, 45, 77-89.
Pressman, L.J., Loo, S.K., Carpenter, E.M., Asarnow, J.R., Lynn, D., McCracken, J.T.,
McGough, J.J., Lubke, G.H., Hang, M.H., & Smalley, S.L. (2006). Relationship of family
environment and parental psychiatric diagnosis to impairment in ADHD. Journal of the
American Academy of Child and Adolescent Psychiatry, 45, 346-354.
Puig-Antich, J., & Ryan, N. (1986). The Schedule for Affective Disorders and Schizophrenia for
School-Age Children (Kiddie-SADS). Pittsburgh, Pennsylvania. Western Psychiatric
Institute and Clinic.
Purcell, S. (2002). Variance component models for gene-environment interaction in twin
analysis. Twin Research, 5, 554-571.

126

Rathouz, P.J., Van Hulle, C.A., Rodgers, J.L., Waldman, I.D., & Lahey, B.B. (2008).
Specification, testing, and interpretation of gene-by-measured-environment interaction
models in the presence of gene-environment correlation. Behavior Genetics, 38, 301-315.
Retz, W., Freitag, C.W., Retz-Jungunger, P., Wenzler, D., Schneider, M., Kissling, C., Thome,
J., & Rosler, M. (2008). A functional serotonin transporter gene polymorphism increases
ADHD symptoms in delinquents: Interactions with adverse child environment.
Psychiatry Research, 158, 123-131.
Retz, W., Retz-Jungunger, P., Supprian, T., Thome, J., & Rosler, M. (2004). Association with
serotonin transporter gene polymorphism with violence: Relation with personality
disorders, impulsivity, and childhood ADHD psychopathology. Behavioral Science and
the Law, 22, 415-425.
Reynolds, C.R., & Kamphaus, R.W. (1992). Behavioral Assessment System for Children
manual. Circle Pines, MN: American Guidance Service.
Rhoades, K.A. (2008). Children‟s responses to interparental conflict: A meta-analysis of their
associations with child adjustment. Child Development, 79, 1942-1956.
Rutter, M. (1999). Psychosocial adversity and child psychopathology. British Journal of
Psychiatry, 174, 480-493.
Rutter, M., Moffitt, T.E., & Caspi, A. (2006). Gene-environment interplay and psychopathology:
Multiple varieties but real effects. Journal of Child Psychology and Psychiatry, 47, 226261.
Rutter, M., & Silberg, J. (2002). Gene-environment interplay in relation to emotional and
behavioral disturbance. Annual Review of Psychology, 53, 463-490.
Rutter, M., & Sroufe, L.A. (2000). Developmental psychopathology: Concepts and challenges.
Development and Psychopathology, 12, 265-296.
Sakai, J.T., Young, S.E., Stallings, M.C., Timberlake, D., Smolen, A., Stelter, G.L., & Crowley,
T.J. (2006). Case-control and within-family tests for an association between conduct
disorder and 5HTTLPR. American Journal of Medical Genetics Part B: Neuropsychiatric
Genetics, 141B, 825-832.
Samuel, V., George, P., Thornell, A., Curtis, S., Taylor, A., Brome, D., et al. (1999). A pilot
controlled family study of DSM-III-R and DSM-IV ADHD in African-American
children. Journal of the American Academy of Child and Adolescent Psychiatry, 38, 3439.

127

Schulz, K.P., Newcorn, J.H., McKay, K.E., Himelstein, J., Koda, V.H., Siever, L.J., Sharma, V.,
& Halperin, J.M. (2001). Relationship between central serotonergic function and
aggression in prepubertal boys: Effect of age and attention-deficit/hyperactivity disorder.
Psychiatry Research, 101, 1-10.
Seeger, G., Schloss, P., & Schmidt, M.H. (2001). Functional polymorphism within the promoter
of the serotonin transporter gene is associated with severe hyperkinetic disorders.
Molecular Psychiatry, 6, 235-238.
Sjoberg, R.L., Nilsson, K.W., Nordquist, N., Ohrvik, J., Leppert, J., Lindstrom, L., & Oreland, L.
(2006). Development of depression: sex and the interaction between environment and a
promoter polymorphism of the serotonin transporter gene. International Journal of
Neuropsychopharmacology, 9, 443-449.
Sherman, D.K., McGue, M.K., & Iacono, W.G. (1997). Twin concordance for attention deficit
hyperactivity disorder: A comparison of teachers‟ and mothers‟ reports. American
Journal of Psychiatry, 154, 532-535.
Silberg, J., Pickles, A., Rutter, M., Hewitt, J., Simonoff, E., Maes, H., Carbonneau, R., Murrelle,
L., Foley, D., & Eaves, L. (1999). The influence of genetic factors and life stress on
depression among adolescent girls. Archives of General Psychiatry, 56, 225-232.
Skopp, N.A., McDonald, R., Manke, B., & Jouriles, E.N. (2005). Siblings in domestically violent
families: Experiences of interparent conflict and adjustment problems. Journal of Family
Psychology, 19, 324-333.
Smalley, S.L., Kustanovich, V., Minassian, S.L., Stone, J.L., Ogdie, M.N., McGough, J.J., et al.
(2002). Genetic linkage of attention deficit/hyperactivity disorder on chromosome 16p13,
in a region implicated in autism. American Journal of Human Genetics, 71, 959-963.
Sonuga-Barke, E.J.S. (2005). Causal models of attention-deficit/hyperactivity disorder: From
common simple deficits to multiple developmental pathways. Biological Psychiatry, 57,
1231-1238.
Sonuga-Barke, E.J.S., Auerbach, J., Campbell, S.B., Daley, D., & Thompson, M. (2005).
Varieties of preschool hyperactivity: Multiple pathways from risk to disorder.
Developmental Science, 8, 141-150.
Sonuga-Barke, E. J., Lasky-Su, J., Neale, B. M., Oades, R., Chen, W., Franke, B., Buitelaar, J.,
Banaschewski, T., Ebstein, R., Gill, M., Anney, R., Miranda, A., Mulas, F., Roeyers, H.,
Rothenberger, A., Sergeant, J., Steinhausen, H. C., Thompson, M., Asherson, P. &
Faraone, S. V. (2008). Does parental expressed emotion moderate genetic effects in
ADHD? An exploration using a genome wide association scan. American Journal of
Medical Genetics Part B: Neuropsychiatric Genetics, 147B, 1359-68.
128

Spencer, T., Biederman, J. & Wilens, T. (2000). Pharmacoptherapy of attention deficit
hyperactivity disorder. Child and Adolescent Psychiatry Clinics of North America, 9, 7797.
Sprich, S., Biederman, J., Crawford, M.H., Mundy, E., & Faraone, S.V. (2000). Adoptive and
biological families of children and adolescents with ADHD. Journal of the American
Academy of Child and Adolescent Psychiatry, 39, 1432-1437.
Stoff, D.M., Pasatiempo, A.P., Yeung, J., Cooper, T.B., Bridger, W.H., & Rabinovich, H.
(1992). Neuroendocrine responses to challenge with dl-fenfluramine and aggression in
disruptive behavior disorders of children and adolescents. Psychiatry Research, 43, 263276.
Sturge-Apple, M.L., Davies, P.T., Winter, M.A., Cummings, E.M., & Schermerhorn, A. (2008).
Interparental conflict and children‟s school adjustment: The explanatory role of
children‟s representations of interparental and parent-child relationships. Developmental
Psychology, 44, 1678-1690.
Taylor, S.E., Way, B.M., Welch, W.T., Hilmert, C.J., Lehman, B.J., & Eisenberger, N.I. (2006).
Early family environment, current adversity, the serotonin transporter promoter
polymorphism, and depressive symptomatology. Biological Psychiatry, 60, 671-676.
Thapar, A., Fowler, T., Rice, F., Scourfield, J., van den Bree, M., Thomas, H., Harold, G., &
Hay, D. (2003). Maternal smoking during pregnancy and attention-deficit hyperactivity
disorder in offspring. American Journal of Psychiatry, 160, 1985-1989.
Thapar, A., O‟Donovan, M., & Owen, M.J. (2005). The genetics of attention-deficit
hyperactivity disorder. Human Molecular Genetics, 14, 275-282.
Todd, R.D., & Neuman, R.J. (2007). Gene-environment interactions in the development of
combined type ADHD: Development of a synapse-based model. American Journal of
Medical Genetics Part B: Neuropsychiatric Genetics, 144B, 971-975.
van den Oord, E. J., Boomsma, D.I., & Verhulst, F.C. (1994). A study of problem behaviors in
10- to 15-year-old biologically related and unrelated international adoptees. Behavioral
Genetics, 24, 193-205.
Volk, H.E., Neuman, R.J., & Todd, R.D. (2005). A systematic evaluation of ADHD and
comorbid psychopathology in a population twin sample. Journal of the American
Academy of Child and Adolescent Psychiatry, 44, 768-775.
Waldman, I.D. (2007). Gene-environment interactions re-examined: Does mother‟s marital
stability interact with the dopamine receptor D2 gene in the etiology of childhood
attention-deficit hyperactivity disorder? Development and Psychopathology, 19, 11171128.
129

Waldman, I. D., Nigg, J.T., Gizer, I.R., Park, L., Rappley, M.D., & Friderici, K. (2006). The
adrenergic receptorl alpha-2A gene (ADRA2A) and neuropsychological executive
functions as putative endophenotypes for childhood ADHD. Cognitive, Affective, and
Behavioral Neuroscience, 6, 18-30.
Whalen, C.K., & Henker, B. (1999). The child with attention-deficit hyperactivity disorder in
family contexts. In H.C. Quay and A.E. Hogan (Eds.) Handbook of Disruptive Behavior
Disorders (pp. 139-155). New York: Kluwer Academic/Plenum.
Wigg, K.G., Takhar, A., Ickowicz, A., Tannock, R., Kennedy, J.L., Pathare, T., Malone, M.,
Schachar, R., & Barr, C. (2006). Gene for the serotonin transporter and ADHD: No
association with two functional polymorphisms. American Journal of Medical Genetics,
Part B: Neuropsychiatric Genetics, 141B, 566-570.
Willcutt, E.G., Pennington, B.F., & DeFries, J.C. (2000). Etiology of inattention and
hyperactivity/impulsivity in a community sample of twins with learning difficulties.
Journal of Abnormal Child Pyschology, 28, 149-159.
Wymbs, B.T., Pelham, W.E., Molina, B.S.G., & Gnagy, E.M. (2008). Mother and adolescent
reports of interparental discord among parents of adolescents with and without attentiondeficit/hyperactivity disorder. Journal of Emotional and Behavioral Disorders, 16, 29-41.

130