Literature Review

Description

There are two articles that need a literature review so that means you have too answer these questions for both articles there are the articles when answering these question do one article at a time so I will provide you with the work sheet that has the questions. So the literature review template should be filled out twice.

PSY 215 Literature Review Template
Please note: Keep in mind that the following questions are practice and preparation for the more
detailed literature review elements to come. When you complete your literature review, you will
be addressing more specific elements. These questions are the first thing to think about when
beginning a literature review.
What is the title of the article? Provide a citation for the article in APA format.
What is the purpose of the article?
What is the hypothesis of the study? In other words, what claims do the authors make in
the article? What was the outcome(s) of the study, that is, what conclusions did the authors
make as a result of the study?
What variables (factors) are being looked at as an influence on abnormal behavior?
If these variables or the relationship between these variables have been studied before,
what have other studies found? This shows historical significance.
What type of research design is used in the study?
Do you think the research in this article was conducted in an ethical manner? Why or why
not?
Journal of Abnormal Psychology
2012, Vol. 121, No. 2, 458 – 466
© 2011 American Psychological Association
0021-843X/11/$12.00 DOI: 10.1037/a0026393
The Covariation of Trait Anger and Borderline Personality:
A Bivariate Twin-Siblings Study
Marijn A. Distel, Mark Patrick Roeling,
Jorim J. Tielbeek, and Désie van Toor
Catherine A. Derom
University Hospital Gasthuisberg, Katholieke Universiteit
Leuven, Belgium
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
VU University Amsterdam, Amsterdam, the Netherlands
Timothy J. Trull
Dorret I. Boomsma
University of Missouri–Columbia
VU University Amsterdam, Amsterdam, the Netherlands
Anger can be defined as an emotion consisting of feelings of variable intensity, from mild irritation
or annoyance to intense fury and rage. Borderline personality disorder (BPD) is characterized by
impulsivity and instability of interpersonal relationships, of self-image, and of negative affects.
Borderline personality and trait anger are often observed together. The present study examined the extent
to which a genetic association explains the covariation between a trait measure of borderline personality
and trait anger. To this end, self-report data of 5,457 twins and 1,487 of their siblings registered with the
Netherlands Twin Register and the East Flanders Prospective Twin Survey were analyzed using genetic
structural equation modeling. A significant phenotypic correlation was observed between the two traits (rP ⫽
.52). This correlation was explained by genetic (54%) and by environmental influences (46%). A shared
genetic risk factor is thus one of the explanations for the covariation of borderline personality and trait anger.
Keywords: borderline personality, trait anger, twin study, genetic factors
(“anger-in”) or expressed outwardly in some form of aggressive
behavior (“anger-out”). Anger has been related to several important constructs in behavioral medicine and psychological research.
For example, high levels of internal expression of anger and trait
anger have been associated with increased blood pressure and
induced hypertension (Markovitz, Matthews, Wing, Kuller, &
Meilahn, 1991; Schneider, Egan, Johnson, Drobny, & Julius,
1986), increased risk for coronary heart diseases (Atchison &
Condon, 1993; Williams et al., 2000; Leon, 1992; Kawachi, Sparrow, Spiro, Vokonas, & Weiss, 1996; Eaker, Sullivan, Kelly–
Hayes, D’Agonstino, & Benjamin, 2004; Chang, Ford, Meoni,
Wang, & Klag, 2002), and mental disorders such as anorexia and
bulimia nervosa (Fassino, Daga, Piero, Leombruni, & Rovera,
2001) bipolar disorder (Posternak & Zimmerman, 2002), and
borderline personality disorder (Morse et al., 2009).
Borderline personality disorder (BPD) is characterized by a
pervasive pattern of instability of interpersonal relationships, of
self-image, and of negative affects, and marked impulsivity that
begins by early adulthood and is present in a variety of contexts
(American Psychiatric Association [APA], 2000). BPD is diagnosed in approximately 1% to 2% of the general population
(Lenzenweger, Lane, Loranger, & Kessler, 2007; Torgersen, Kringlen, & Cramer, 2001) and is associated with a variety of negative
outcomes such as self-harm behavior, suicidal behavior, impaired
occupational and interpersonal functioning, delinquent behavior,
and substance abuse (Skodol et al., 2002). Inappropriate, intense
anger or difficulty controlling anger is the most prevalent BPD
criterion in clinical samples (Zanarini, Frankenburg, Hennen,
Reich, & Silk, 2005), nonclinical samples (Trull, 1995), and in first
degree relatives of BPD patients (Zanarini et al., 2004). Further,
Anger can be defined as an emotion that consists of feelings of
variable intensity, from mild irritation or annoyance to intense fury
and rage (Spielberger, Jacobs, Russell, & Crane, 1983). It can be
conceptualized as state anger, referring to an episode of anger
occurring at a specified time, or as trait anger, referring to an
aspect of personality (Eckhardt, Norlander, & Deffenbacher,
2004). Feelings of anger may be suppressed or directed inward
This article was published Online First December 12, 2011.
Marijn A. Distel, Department of Biological Psychology, VU University
Amsterdam, Amsterdam, The Netherlands, and EMGO⫹ Institute for
Health and Care Research, VU University Medical Center, Amsterdam, the
Netherlands; Mark Patrick Roeling, Jorim Tielbeek, and Désie van Toor,
Department of Biological Psychology, VU University Amsterdam, Amsterdam, The Netherlands; Catherine A. Derom, Department of Human
Genetics, University Hospital Gasthuisberg, Katholieke Universiteit Leuven, Belgium; Timothy J. Trull, Department of Psychological Sciences,
University of Missouri–Columbia, Columbia, Missouri; and Dorret I.
Boomsma, Department of Biological Psychology, VU University Amsterdam, Amsterdam, The Netherlands, EMGO⫹ Institute for Health and Care
Research, and Neuroscience Campus Amsterdam, VU University Medical
Center, Amsterdam, the Netherlands.
The present study was supported by the Borderline Personality Disorder
Research Foundation, Spinozapremie (NWO/SPI 56 – 464-14192), and
Twin-Family Database for behavior genetics and genomics studies (NWO
480 – 04-004). The authors declare no financial or other conflicts of interest.
Correspondence concerning this article should be addressed to Marijn A.
Distel, Department of Biological Psychology, VU University Amsterdam,
Van der Boechorststraat 1, 1081 BT Amsterdam, the Netherlands. E-mail:
ma.distel@psy.vu.nl/m.distel@ggzIngeest.nl
458
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
TRAIT ANGER AND BORDERLINE PERSONALITY
459
this feature was found to be among the slowest symptoms to remit
in BPD patients across 8- to 10-year follow up (Zanarini et al.,
2007).
Both trait anger and the trait of borderline personality are
heritable. Rebollo and Boomsma (2006) conducted a longitudinal
twin family study into the genetics of trait anger. The genetic
architecture of the trait differed in men and women. In males 23%
of the variance was explained by additive genetic effects and 26%
by dominant genetic effects, leading to a total heritability of 49%.
In women, 34% of the variance was explained by additive genetic
effects and no dominant genetic effects were found. Other studies
only focused indirectly on the heritability of anger or violence, for
example in genetic studies on antisocial personality disorder for
which the heritability was estimated at 38% (Cadoret & Stewart,
1991; Torgersen et al., 2008). Large-scale twin and twin family
studies of BPD and the trait of borderline personality report
heritability estimates around 40% (Distel et al., 2008a; Bornovalova et al., 2009; Kendler et al., 2008; Torgersen et al., 2008;
Distel et al., 2009). The study by Distel et al. (2009), included
twins as well as their parents and nontwin siblings and provided
evidence for the influence of nonadditive genetic effects, while
suggesting no effects of cultural transmission from parents to
offspring.
Some overlap between trait anger and the trait of borderline
personality is expected, considering that one of the nine criteria for
BPD concerns inappropriate expressions of anger and intense,
chronic feeling of anger (APA, 2000). However, most of the BPD
criteria do not directly tap trait anger. On the other hand, there are
many cognitive processes associated with trait anger (Owen, 2011;
Wilkowski & Robinson, 2010) that may at least partially explain
some of the remaining symptoms and features of borderline personality. For example, selective attention and reasoning biases
associated with trait anger may lead individuals to be hypervigilant
to possible threat or aggression and to attribute hostile intentions to
others so as to arouse anger more frequently (Wilkowski & Robinson, 2010). These cognitive processes associated with anger may
lead to rejection sensitivity and to interpersonal conflict and disruption often seen in those with BPD, for example (Romero–
Canyas et al., 2010). Therefore, an examination of the phenotypic
and genotypic association between trait anger and borderline personality can help to index the degree to which the two traits may
share a common underlying cause as well as point to potential
shared mechanisms (e.g., cognitive processes and biases) that may
inform theories of the etiology of BPD.
In the present study, we explored shared genetic risk factors as
a possible explanation for the covariation of borderline personality
and trait anger in the population. Data from twins and their siblings
were available from the Netherlands Twin Register (NTR;
Boomsma et al., 2006) and the East Flanders Prospective Twin
Survey (EFPTS; Derom et al., 2006) to disentangle genetic and
environmental influences on the covariance between borderline
personality and trait anger.
the NTR established in 1978 (Boomsma et al., 2006). Every 2
years, surveys on health and lifestyle were sent to the twin families. For the present study, data from the seventh survey were used
which was sent in 2004 –2005. Dutch-speaking twins in Belgium
were also asked to take part in the Dutch health, lifestyle, and
personality study. Belgian participants were recruited through the
EFPTS, a population-based register of multiple births in the Belgian province of East Flanders which was started in 1964 (Derom
et al., 2006). Young adult twins were contacted by mail and invited
to complete a survey which was enclosed with the letter. A
nonresponse study (Distel et al., 2007) showed that for a substantial group of targeted participants the addresses were incorrect.
This group thus never received the questionnaire. After correcting
for this, the response rate was estimated at 52.2% in the group of
targeted participants who participated before and 13.6% in the
group of targeted participants who were already registered, but
never completed a questionnaire.
For the Dutch sample zygosity was determined either from
DNA typing or from self-report answers to eight survey questions
on physical twin resemblance and confusion of the twins by family
members and strangers. Zygosity agreement reached 97% (Willemsen, Posthuma, & Boomsa, 2005). For the Belgian sample,
twin zygosity was determined through sequential analysis based on
sex, fetal membranes, umbilical cord blood groups, and placental
alkaline phosphatase until 1985. After that time, DNA fingerprinting was used. In case of missing or insufficient DNA information,
the zygosity of the same-sex DZ twins was based on survey items
on physical twin resemblance and confusion of the twins (see
Derom & Derom, 2005).
Data from 7,261 twins and siblings with valid scores on the trait
measures of borderline personality and anger were available. A
total of 928 twins were registered with the EFPTS. Twins with
unknown zygosity (N ⫽ 94), individuals with an unknown age
(N ⫽ 148) or sex (N ⫽ 12) and individuals aged below 18 (N ⫽
14) were excluded. A maximum of two brothers and two sisters
were included in the analyses, remaining siblings were excluded
(N ⫽ 49). This resulted in a total sample of 5,457 twins and 1,487
siblings from 3,946 families. The twin sample consisted of 813
monozygotic males (MZM), 416 dizygotic males (DZM), 2,095
monozygotic females (MZF), 1008 dizygotic females (DZF),
1,125 dizygotic opposite sex (DOS), and 528 brothers and 959
sisters. Table 1 shows the complete family configuration of the
sample. There were 1,866 families in which both members of a
twin pair completed the questionnaire, 1,725 families in which
only one member of the twin pair completed the questionnaire and
355 families in which only nontwin siblings completed the questionnaire. The mean (M) age of the twins was 34.46 years (standard deviation [SD] ⫽ 10.98, range ⫽ 18 – 87 years). The mean
age for the siblings was 39.89 years (SD ⫽ 12.55, range ⫽ 18 –91
years).
Methods
Trait anger was measured with the Dutch adaptation of the State
Trait Anger Scale (STAS, Spielberger et al., 1983; van der Ploeg,
Defanes, & Spielberger, 1982). The scale is designed to assess the
frequency of which an individual experiences the state anger over
time and in response to a variety of situations. The STASadaptation was scored on a 4-point Likert scale (1– 4; almost never,
Participants
The present study is part of an ongoing study on health, lifestyle,
and personality in twins and their family members registered with
Measures
DISTEL ET AL.
460
Table 1
Family Configuration in the Sample According to Zygosity, Cohort, and Number of Additional Nontwin Siblings
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Families yielding
MZM
Families yielding a twin pair
Families yielding a single twin
DZM
Families yielding a twin pair
Families yielding a single twin
MZF
Families yielding a twin pair
Families yielding a single twin
DZF
Families yielding a twin pair
Families yielding a single twin
DOS
Families yielding a twin pair
Families yielding a single twin
Families yielding no twins
Total
Note.
No siblings
1 sibling
2 siblings
3 siblings
4 siblings
Total
202
207
64
27
12
9
6
0
1
0
285
243
67
157
35
35
3
6
3
0
1
0
109
198
613
373
164
56
39
12
9
0
2
0
827
441
233
290
70
42
24
14
2
4
0
0
329
350
223
404
—
2769
79
67
282
921
11
17
62
209
2
4
10
40
1
1
1
7
316
493
355
3,946
MZM ⫽ monozygotic males; DZM ⫽ dizygotic males; MZF ⫽ monozygotic females; DZF ⫽ dizygotic females; DOS ⫽ dizygotic opposite sex.
sometimes, often, almost always) and consists of 10 items concerning for example getting easily annoyed and irritated, being a
hotheaded person, flying off the handle, and having a fiery temper.
Participants were asked to indicate the extent to which an item
occurred in their everyday lives. The items were scored according
to the test manual, which states that at least 80% of the items must
be answered to calculate a sum score and that missing values or
ambiguous answers should be substituted by a score of 2 (sometimes true). Several studies have reported excellent psychometric
qualities, reliability (␣) of 0.86 and good discriminant and convergent validity (Eckhardt et al., 2004). In the present sample, the
internal consistency (Cronbach’s alpha) was 0.85. Borderline personality was measured by the Dutch translation of the Personality
Assessment Inventory – Borderline features scale (PAI-BOR;
Morey, 1991). The PAI-BOR consists of 24 items rated on a
4-point scale (0 –3; false, slightly true, mainly true, very true) that
tap features of severe personality pathology clinically associated
with BPD, such as stability of mood and affects, self-image,
feelings of emptiness, intense and unstable relationships, impulsivity, and self-harm. The items were scored according to Morey’s
test manual, which states that at least 80% of the items must be
answered to calculate a sum score and that missing values or
ambiguous answers should be substituted by a zero score. Several
studies in clinical, as well as nonclinical samples, have supported
the reliability and validity of the PAI-BOR total score in indexing
the degree to which borderline features are present (Distel et al.,
2008b; Trull, 1995; Trull, 2001; Morey, 1991). Bell–Pringle et al.
(1997) and Stein, Pinkster–Aspen, and Hilsenroth (2007), for
example, showed that the PAI-BOR differentiates between patients
diagnosed with BPD and patients without borderline personality
pathology or unscreened controls with 75% to 80% accuracy.
Jacobo, Blais, Baity, and Harley (2007) administered the PAIBOR to patients diagnosed with BPD and found a significant
correlation of .58 between the total number of BPD SCID-II
criteria and the PAI-BOR scale. Finally, the 6-month test–retest
correlation of the Dutch version of the PAI-BOR assessed on 200
unrelated individuals was 0.78 (Distel et al., 2008a) and multigroup confirmatory factor analysis showed that the PAI-BOR is
measurement invariant across sex and age (De Moor, Distel, Trull,
& Boomsa, 2009).
The PAI-BOR does include two items that directly assess anger,
one asking whether the respondent has control over his or her
anger, and the other asking whether the respondent gets so mad,
s/he has trouble controlling feelings of anger. Therefore, as detailed below, we repeated all phenotypic and genotypic analyses
after deleting these two PAI-BOR items from the total PAI-BOR
score. In the present sample, the internal consistency (Cronbach’s
alpha) was 0.83 for the full PAI-BOR scale and 0.82 for the
22-item scale.
Twin Studies
In twin-family studies, the different degree of genetic relatedness of monozygotic (MZ) and dizygotic (DZ) twins and sibling
pairs is used to identify the relative contribution of genes and
environment to the phenotypic variation in a trait, or to the covariation between traits. MZ twins are (almost) genetically identical
and DZ twin and sibling pairs share on average half of their
segregating genes (Boomsma, Busjahn, & Peltonen, 2002). The
total phenotypic variance is decomposed into a part due to the
additive effects of alleles at all genomic loci (A), the nonadditive
genetic effects of alleles (D; dominance), the effects of the environment that is shared by individuals growing up in the same
family (C), and the effects of nonshared environment (which also
includes measurement error, E). The expectation for the phenotypic variance may be written as: V(P) ⫽ V(A) ⫹ V(D) ⫹ V(C) ⫹
V(E). Broad-sense heritability (h2) is the proportion of phenotypic
variance that is attributable to genotypic variance, h2 ⫽ [V(A) ⫹
V(D)]/V(P); narrow-sense heritability is the proportion of variation
explained by additive genetic factors, h2n ⫽ V(A)/V(P). Based on
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
TRAIT ANGER AND BORDERLINE PERSONALITY
data from only MZ and DZ twins and siblings this full model (i.e.,
ACDE) is not identified, and either an ADE or an ACE model can
be fitted to the data. The choice between these latter two models
may be based on prior knowledge or on the pattern of correlations
in MZ and DZ twins. When the DZ correlation is more than half
the MZ correlation, there is evidence for environmental effects
shared by twins from the same family (C) and when the DZ
correlation is less than half the MZ correlation, there is evidence
for nonadditive genetic effects (D). In the present study an ADE
model was fitted to the data. The ADE model is identified because
of the difference in correlations among the latent factors influencing the phenotype in MZ and DZ twin pairs. For MZ twin pairs
correlations between the A and the D factor score in twin 1 and
twin 2 are both one. For DZ and sibling pairs, these correlations are
0.5 and 0.25, respectively. Correlations between unique environmental factor scores in twin 1 and twin 2 are zero in MZ and DZ pairs
(e.g., Falconer & Mackay, 1996; Boomsma & Molenaar, 1986).
Bivariate genetic analyses can be applied to determine to what
extent the covariation between two traits can be explained by
genetic and environmental factors. The comparison of MZ and DZ
cross-twin cross-trait correlations provides a first indication about
the shared etiology between traits. If a significant cross-twin
cross-trait correlation is present it suggests that there is a familial
influence on the etiology of the correlation between the two traits.
If the MZ cross-twin cross-trait correlation exceeds the DZ crosstwin cross-trait correlation it suggests that the familial influence on
the correlation is at least partly genetic in origin. A graphical
representation of the bivariate genetic model is shown in Figure 1.
Statistical Analyses
We first fitted a saturated model in which variances, covariances
(among family members and among traits) and means were estimated. Mean borderline personality and trait anger scores were
461
estimated separately for twins and siblings. An effect of sex (coded
as 0 for males and 1 for females) and age (in years) were included
as fixed effects (regression coefficients) on each trait.
We tested for the significance of differences in mean scores of
twins and siblings and for the effect of sex and age on borderline
personality and trait anger scores. Significant effects of sex and
age were retained in subsequent genetic analyses. All correlations
between MZ and DZ twin and sibling pairs within and between
traits were estimated as a function of zygosity and sex. By constraining within-trait and cross-trait correlations to be equal for DZ
twins and nontwin siblings and between men and women within
the zygosity groups we tested for a specific twin environment and
for qualitative and quantitative sex differences. Qualitative sex
differences (i.e., different genes influence the trait in males and
females) are suggested if correlations in DZ twins of opposite sex
(DOS) cannot be predicted based on the pattern of correlations in
same-sex twin pairs. Quantitative sex differences (differences in
the magnitude of A, D, and E between males and females) are
suggested when the correlations in male–male and female–female
pairs within zygosity cannot be constrained to be equal without a
significant deterioration in the fit of the model.
To assess to what extent borderline personality and trait
anger share genetic liability, a bivariate genetic model was
fitted to the data in which the variance in borderline personality
and trait anger and the covariance between them was decomposed into sources of A, D, and E. In this model the first
variable loads only on the first factor and the second variable
loads on the first two factors. Constraining the contributions of the
latent factors of A or D at zero provides a test of whether these factors
significantly contribute to the total variance in the traits. The significance of the genetic and environmental covariance structure was
tested by constraining subsequent pathways (a21, d21, and e21 in
Figure 1) in the model at zero.
Figure 1. Bivariate genetic model; A1 and A2 ⫽ additive genetic factors; D1 and D2 ⫽ dominant genetic
factors; E1 and E2 ⫽ unique environmental factors; a ⫽ factor loading of A; d ⫽ factor loading of D; e ⫽ factor
loading of E. All latent A, D, and E factors have unit variance. For clarity reasons the nontwin sibling is not
drawn.
DISTEL ET AL.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
462
All analyses were conducted using structural equation modeling
in MX (Neale, Boker, Xie, & Maes, 2006). Testing of submodels
was done by means of likelihood ratio tests, by subtracting the
negative log-likelihood (-2LL) for the more general model from
the ⫺2LL of the more restricted model. This gives a ␹2 test with
the degrees of freedom (df) equal to the difference in the number
of estimated parameters in the two models. A significant ␹2(p ⬍
.05) indicates that the constrained model is significantly worse
than the previous model and is therefore rejected. As a result, the
previous model is kept as the most parsimonious model, to which
a new model can be compared. In line with previous publication
based on these data, a square root data transformation was performed for the borderline personality data but not for the trait
anger data. To make sure this approach did not influence our
results we reran all analyses with transformations for both measures but this did not change the results.
Results
Tests of Fixed Effects on the Means and Variances
The borderline personality and the trait anger scores were not
2
significantly different for twins and siblings, ␹(1)
⫽ 0.008, p ⫽
2
.929 for borderline and ␹(1)
⫽ 0.053, p ⫽ .818 for anger, and were
dependent on age (all p ⬍ .01). Both age regression coefficients
were negative indicating that the borderline and anger scores
decrease with age. A sex effect was only significant for borderline
2
personality trait, ␹(1)
⫽ 6.349, p ⫽ .012. Women had significantly
higher borderline personality scores than men. In subsequent analyses the significant effects of sex and age were retained in the
means model. Standard deviations were equal in males and fe2
males for both measures, ␹(1)
⫽ 0.072, p ⫽ .788 for borderline
2
personality and ␹(1) ⫽ 2.092, p ⫽ .148 for trait anger.
Correlation Structure
Table 2 shows the twin and sibling correlations from the saturated model (upper part) and the correlations from the most constrained model (lower part). Table 3 gives the results of the tests
performed on the correlation structure. The phenotypic correlation
between borderline personality and trait anger scores did not differ
for men and women and was estimated at .52 (Model 1). Correlations were similar for DZ twins and siblings for males and
females for both variables (Model 2). For both variables, the
correlations were equal for DZ males and females and same sex
siblings and for MZ males and females suggesting that the heritability is the same for men and women (Model 3). Additionally,
the DZ and sibling same sex correlations were equal to the DZ and
sibling opposite sex correlations indicating that the same genes
influence the borderline personality and trait anger in men and
women (Model 4).
Genetic Model
Based on the correlation structure, which does not provide
evidence for the influence of C, and the results from prior studies
(Distel et al., 2009; Torgersen et al., 2008; Kendler et al., 2008),
we fitted an ADE model to the data. Genetic model-fitting results
are summarized in Table 4. Removal of the dominant genetic
effects (Model 1) resulted in a significant worsening of the goodness of fit (p ⫽ .027). Dropping path d21 from the model did not
result in a significant deterioration in model fit (Model 2; p ⫽
.237), but dropping the path a21 (Model 3; p ⬍ .001) and the path
e21 (Model 4; p ⬍ .001) did result in a significant decrease in
model fit. This means that there are additive genetic and unique
environmental factors that contribute to the covariance between
the traits. Table 5 shows the estimates A, D, and E on variance in
borderline personality and trait anger, the additive genetic and
environmental correlations and the percentage of the phenotypic
correlation explained by A and E. Figure 1 is a graphical representation of the bivariate model and gives the path coefficients
from the best fitting model in the right part of the graph. The total
variance in borderline personality can be written as (a211) ⫹ (d211) ⫹
(e211). The broad-sense heritability of borderline personality [calculated as (a211) ⫹ (d211)/(a211) ⫹ (d211) ⫹ (e211)] was estimated at
46%. The influence of E on individual differences in borderline
personality [calculated as (e211)/(a211) ⫹(d211) ⫹ (e211)] was estimated at 54%. Total variance in trait anger can be written as
(a221) ⫹ (a222) ⫹ (d221) ⫹ (d222) ⫹ (e221) ⫹ (e222). The broad-sense
heritability of trait anger [calculated as (a221) ⫹ (a222) ⫹ (d221) ⫹
(d222)/(a221) ⫹ (a222) ⫹ (d221) ⫹ (d222) ⫹ (e221) ⫹ (e222)] was estimated
at 40%. The influence of E on trait anger [calculated as (e221) ⫹
(e222)/(a221) ⫹ (a222) ⫹ (d221) ⫹ (d222) ⫹ (e221) ⫹ (e222)] was estimated
at 60%. The additive genetic and environmental covariance may be
written as (a11 â«» a21) and (e11 â«» e21), respectively. The dominant
genetic covariance did not significantly explain covariance be-
Table 2
Twin and Sibling Correlations for Trait Anger and Borderline Personality
Monozygotic males twin pairs
Dizygotic male twin pairs
Monozygotic female twin pairs
Dizygotic female twin pairs
Dizygotic opposite sex twin pairs
Brothers
Sisters
Brother–sister pairs
All monozygotic twins
All dizygotic twins/siblings
Twin correlation
trait anger
Twin correlation
borderline personality
Cross-twin cross-trait
correlation
.31
.14
.44
.12
.14
.15
.17
.11
.41
.14
.46
.23
.46
.15
.20
.14
.24
.13
.47
.18
.27
.11
.29
.09
.12
.10
.15
.08
.29
.12
TRAIT ANGER AND BORDERLINE PERSONALITY
463
Table 3
Model Fit Results for the Saturated Bivariate Model
Model
Test
Versus
⫺2LL
df
␹2
⌬df
p
0
1
Full model
Phenotypic correlation males ⫽ Phenotypic
correlation females
Twin correlations ⫽ sibling correlations
Male correlations ⫽ female correlations
Same sex DZ/sibling correlations ⴝ opposite
sex DZ/sibling correlations
—
0
56053.875
56053.885
13855
13856
.01
1
.920
1
2
3
56059.194
56068.141
56069.810
13865
13871
13874
5.309
8.947
1.669
9
6
3
.807
.177
.644
2
3
4
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Note.
⫺2LL ⫽ ⫺2 log likelihood; df ⫽ degrees of freedom; p ⫽ p-value; DZ ⫽ dizygotic. The best fitting model is printed in bold.
tween the traits. The additive genetic correlation (rGa) [calculated
as a11 ⫻ a21/(公(a211) ⫻ 公(a221 ⫹ a222))] was estimated at .93 and
the environmental correlation [calculated as e11 â«» e21/(å…¬(e211) â«»
公(e221 ⫹ e222))] at .42. The percentage of the phenotypic correlation explained by A may be calculated as 公(h2bpd) ⫻ rG ⫻
2
å…¬(hanger
) divided by the phenotypic correlation. Likewise, the
percentage of the phenotypic correlation explained by E may be
2
calculated as å…¬(e2bpd) â«» rE â«» å…¬(eanger
) divided by the phenotypic
correlation. Based on these calculations the phenotypic correlation
can be explained for 54% by A and 46% by E.
Replication Using the 22-Item PAI-BOR Scale
We repeated all analyses after deleting the two PAI-BOR items
that directly tapped anger from the total PAI-BOR score for each
participant. The same pattern of results was obtained. Specifically,
the phenotypic correlation was estimated at 0.50 as compared with
.52 in the original analyses. The correlation pattern was similar to
the original analyses. The MZ twin correlations and DZ twin/sib
correlations were estimated at 0.41 and 0.14 for trait anger, and at
0.46 and 0.18 for borderline personality, respectively. The cross
correlation was estimated at 0.28 for MZ twins and at 0.11 for DZ
twins and siblings. The heritability estimated found in the genetic
analyses were equal to those found in the original analyses. Only
the genetic correlation decreased from 0.93 to .89 and the unique
environmental decreased from 0.42 to 0.40.
Discussion
Both trait anger and borderline personality are influenced by
additive genetic, dominant genetic, and unique environmental factors. These findings are in line with previous studies reporting
heritability estimates for BPD, borderline personality scores, and
trait anger scores. Previous research showed that there is a moderate to high phenotypic correlation between trait anger and BPD
(Dolan, Anderson, & Deakin, 2001; Morse et al., 2009; Newhill,
Eack, & Mulvey, 2009). The basis of the overlap however remained unclear in these previous studies. In the present study,
structural equation modeling was applied to disentangle the relative influence of the genes and the environment on the covariance
between the trait of borderline personality and trait anger. Significant correlations between the latent additive genetic and unique
environmental factors that influence the two traits were found.
Results showed that the phenotypic association (r ⫽ .52) could be
explained by additive genetic factors that are shared between the
traits (54%) and by shared unique environmental influences
(which also includes some measurement error; 46%). A similar
level of correlation (r ⫽ .50) was found even after the two
PAI-BOR items that directly indexed anger were deleted from the
total PAI-BOR score. This result suggests that the moderate correlation between anger and borderline personality is not solely due
to shared item content.
Shared genetic risk is thus one of the possible explanations for
the association between trait anger and borderline personality. As
mentioned earlier, one possible shared set of mechanisms that
appear to characterize both those high in trait anger and those with
significant borderline personality features is a cognitive style that
includes selective attention to hostile social cues, a tendency to
interpret the actions of others as potentially hostile or aggressive,
and a propensity to ruminate over past transgressions by others or
anger-provoking experiences in general (Owen, 2011). Attentional
biases observed in BPD may help explain the tendency for those
with this disorder to experience chronic negative affect that is
interrupted only by intense episodes of fear, anxiety, or hostility
(Carpenter, Bagby–Stone, & Trull, 2011). Whether the intense
Table 4
Genetic Model Fitting Results for Borderline Personality and Trait Anger
Model
0.
1.
2.
3.
4.
ADE model
AE model
Drop path d21
Drop path a21
Drop path e21
Test
⫺2LL
df
␹2
⌬df
p
—
1 vs. 0
2 vs. 0
3 vs. 2
4 vs. 2
52603.736
52612.911
52605.132
52813.758
52915.985
12994
12997
12995
12996
12996
—
9.175
1.396
208.626
310.853
—
3
1
1
1
—
.027
.237
⬍.001
⬍.001
Note. v ⫽ versus; ⫺2LL ⫽ ⫺2 log likelihood; df ⫽ degrees of freedom; p ⫽ p-value; A ⫽ additive genetic factors; D ⫽ nonadditive genetic factors
(dominance); E ⫽ unique environmental factors. The best fitting model is printed in bold.
DISTEL ET AL.
464
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Table 5
Estimates of the Contribution of the Additive and Dominant
Genetic Factor and the Unique Environmental Factor to
Variance in Borderline Personality Trait and Trait Anger, the
Phenotypic Correlation, the Genetic Correlation, the
Environmental Correlation and the Percentage of Covariance
Between Borderline Personality and Trait Anger Explained
by A and E
Additive genetic factor (A)
Dominant genetic factor (D)
Unique environmental factor (E)
Phenotypic correlation (rP)
Genetic correlation
Environmental correlation
% rP explained by A
% rP explained by E
Borderline
personality
Trait
anger
.36
.10
.54
.25
.15
.60
.52
.93
.42
54%
46%
negative affect is experienced and expressed as anger or as fear, for
example, may depend on nature of the situation that arouses the
negative affect; anger is more likely to be aroused if approach
tendencies (goals) are blocked whereas fear or anxiety arise from
avoidance motivation (i.e., experience of threat; Carver &
Harmon–Jones, 2009). It is also worth noting that deficits in the
recruitment of effortful control resources (e.g., Rothbart, 1989) in
hostility-related contexts may play a role in the frequency, duration, and intensity of anger expression (Wilkowski & Robinson,
2010; Wilkowski & Robinson, 2010). Such a cognitive set characterized by these attentional biases, biased appraisals, ruminative
tendencies, and deficits in effortful control may contribute to BPD
features such as conflicted interpersonal relationships, impulsive
behavior, paranoid ideation, and even self-harm behaviors, in
addition to intense negative affects.
These cognitive processes described above, are activated by
environmental circumstances that are likely unique to the individual (i.e., nonshared environmental influence [e.g., rejection]). This
is in line with the moderate degree of correlation between unique
environmental influences on trait anger and on borderline personality found in this study. As there are no known specific environmental events that are unique to the development of BPD or trait
anger, another possibility to consider is the concept of an “invalidating environment,” which is a major component of Linehan’s
biopsychosocial model of BPD (Linehan, 1993). Briefly, Linehan’s model highlights the transactional process between emotional hypersensitivity and the experience of an invalidating environment, a process starting in childhood, which in turn leads to
major features of BPD. In her theory, an invalidating environment
is one that communicates to the individual that her or his emotional/internal experience or behavior is inappropriate or wrong.
Thus, it is not a specific environmental event per se, but rather an
encounter with others in the person’s unique environment context
that contributes to the “invalidating” experience. Therefore, invalidation may take the form of abuse, neglect, or excessive criticism
for example. Further, it is important to note that not all (e.g.,
siblings) are equally vulnerable to invalidation. Based on the
standard quantitative genetic model (Falconer & Mackay, 1996)
we know that siblings within the same family have correlated, but
different genotypes. Linehan proposes that those who are temperamentally emotionally sensitive are most vulnerable to the effects
of invalidation. It is important Linehan (1993) notes that the
consequences of invalidation include increased emotional arousal,
and negative affects such as anger. Therefore, we speculate that
this experience of an invalidating environment may be at least
partially responsible for the finding of a correlation between environmental influences on trait anger and on borderline personality
scores. More research will be necessary to directly test this proposal.
When interpreting the outcomes of this study, some limitations
should be kept in mind. First, as we noted above, the PAI-BOR
questionnaire includes two items (items 10 and 18) that explicitly
deal with anger. Although trait anger and anger as part of BPD are
two related but different concepts, this might have influenced the
results. Therefore, we reran the analyses excluding the two anger
related items from the PAI-BOR. The genetic architecture however
did not change, nor did the general pattern of results. Second,
nonresponse may limit the validity of questionnaire studies when
nonresponse is associated with the traits under study. Distel et al.
(2007) suggest that nonresponse may be higher among subjects
with more BPD features because the participating members of less
cooperative families showed somewhat higher scores on the PAIBOR scale compared to member of highly cooperative families.
However, the difference in borderline personality scores between
less and highly cooperative families was quite small so the practical importance of this difference should not be overestimated. For
trait anger, no significant association with nonresponse was found
(Distel et al., 2007). Third, self-report questionnaires were used to
assess borderline personality and trait anger. Although interview
data and questionnaire data for BPD are highly correlated (Kurtz &
Morey, 2001), results should be generalized to clinical populations
and to a BPD diagnosis specifically with caution. Finally, the
present study did not take possible gene– environment (GE) correlation or interaction into account (Livesley, 2008). If GE interaction plays an important role, its effects would be included in the
“E” component of the model, and thus the role of genetics might
be larger than indicated by the current results. If GE correlation
(the nonrandom distribution of genotypes across environments) is
present, part of the genetic influences derive from the effect of GE
correlation (Rutter, 2007). GE correlation was suggested for borderline personality and certain life events (Distel et al., 2011).
Future studies focusing on identifying the genes and the environmental factors that influence both trait anger and borderline personality will help us better understand the covariance of these
traits.
References
American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders. (Fourth ed., text revision) Washington, DC:
American Psychiatric Publishing.
Atchison, M., & Condon, J. (1993). Hostility and anger measures in
coronary heart disease. Australian and New Zealand Journal of Psychiatry, 27, 436 – 442. doi:10.3109/00048679309075800
Bell–Pringle, V. J., Pate, J. L., & Brown, R. C. (1997). Assessment of
borderline personality disorder using the MMPI-2 and the personality
assessment inventory. Assessment, 4, 131–139.
Boomsma, D., Busjahn, A., & Peltonen, L. (2002). Classical twin studies
and beyond. Nature Reviews Genetics, 3, 872– 882. doi:10.1038/nrg932
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
TRAIT ANGER AND BORDERLINE PERSONALITY
Boomsma, D. I., de Geus, E. J. C., Vink, J. M., Stubbe, J. H., Distel, M. A.,
Hottenga, J. J., . . . Willemsen, G. (2006). Netherlands Twin Register:
From twins to twin families. Twin Research and Human Genetics, 9,
849 – 857. doi:10.1375/twin.9.6.849
Boomsma, D. I., & Molenaar, P. C. M. (1986). Using lisrel to analyze
genetic and environmental covariance structure. Behavior Genetics, 16,
237–250. doi:10.1007/BF01070799
Bornovalova, M. A., Hicks, B. M., Iacono, W. G., & Mcgue, M. (2009).
Stability, change, and heritability of borderline personality disorder traits
from adolescence to adulthood: A longitudinal twin study. Development and
Psychopathology, 21, 1335–1353. doi:10.1017/S0954579409990186
Cadoret, R. J., & Stewart, M. A. (1991). An adoption study of attention–
deficit hyperactivity aggression and their relationship to adult antisocial
personality. Comprehensive Psychiatry, 32, 73– 82. doi:10.1016/0010440X(91)90072-K
Carpenter, R., Bagby–Stone, S., & Trull, T. J. (2011). Emotional awareness: Attention dysregulation in borderline personality disorder. In M. D.
Robinson, E. Harmon–Jones, & E. R. Watkins (Eds.), Handbook of
cognition and emotion. New York, NY: Guilford Press.
Carver, C. S., & Harmon–Jones, E. (2009). Anger is an approach-related
affect: Evidence and implications. Psychological Bulletin, 135, 183–
204. doi:10.1037/a0013965
Chang, P. P., Ford, D. E., Meoni, L. A., Wang, N. Y., & Klag, M. J. (2002).
Anger in young men and subsequent premature cardiovascular disease—
The precursors study. Archives of Internal Medicine, 162, 901–906.
doi:10.1001/archinte.162.8.901
De Moor, M. H. M., Distel, M. A., Trull, T. J., & Boomsma, D. I. (2009).
Assessment of borderline personality disorder features in population
samples: Is the Personality Assessment Inventory-Borderline Scale measurement invariant across sex and age? Psychological Assessment, 21,
125–130. doi:10.1037/a0014502
Derom, C., & Derom, R. (2005). The East Flanders Prospective Twin
Survey. In I. Blickstein & L. G. Keith (Eds.), Multiple pregnancy:
Epidemiology, gestation and perinatal outcome (Second ed., pp. 39 –
47). Oxford, U.K.: Taylor and Francis.
Derom, C. A., Vlietinck, R. F., Thiery, E. W., Leroy, F. O. G., Fryns, J. P.,
& Derom, R. M. (2006). The East Flanders Prospective Twin Survey
(EFPTS). Twin Research and Human Genetics, 9, 733–738. doi:
10.1375/twin.9.6.733
Distel, M. A., Hottenga, J. J., Trull, T. J., & Boomsma, D. I. (2008b).
Chromosome 9: Linkage for borderline personality disorder features. Psychiatric Genetics, 18, 302–307. doi:10.1097/YPG.0b013e3283118468
Distel, M. A., Ligthart, L., Willemsen, G., Nyholt, D. R., Trull, T. J., &
Boomsma, D. I. (2007). Personality, health and lifestyle in a questionnaire family study: A comparison between highly cooperative and less
cooperative families. Twin Research and Human Genetics, 10, 348 –353.
doi:10.1375/twin.10.2.348
Distel, M. A., Middeldorp, C. M., Willemsen, G., Trull, T. J., Derom,
C. A., & Boomsma, D. I. (2011). Life events and borderline personality:
Gene-environment correlation or gene-environment interaction? Psychological Medicine: A Journal of Research in Psychiatry and the Allied
Sciences, 41, 849 – 860. doi:10.1017/S0033291710001297
Distel, M. A., Rebollo–Mesa, I., Willemsen, G., Derom, C. A., Trull, T. J.,
Martin, N. G., & Boomsma, D. I. (2009). Familial resemblance of
borderline personality disorder features: Genetic or cultural transmission? PLoS ONE, 4, e5334. doi:10.1371/journal.pone.0005334
Distel, M. A., Trull, T. J., Derom, C. A., Thiery, E. W., Grimmer, M. A.,
Martin, N. G., et al. (2008a). Heritability of borderline personality
disorder features is similar across three countries. Psychological Medicine: A Journal of Research in Psychiatry and the Allied Sciences, 38,
1219 –1229. doi:10.1017/S0033291707002024
Dolan, M., Anderson, I. M., & Deakin, J. F. W. (2001). Relationship
between 5-HT function and impulsivity and aggression in male offend-
465
ers with personality disorders. British Journal of Psychiatry, 178, 352–
359. doi:10.1192/bjp.178.4.352
Eaker, E. D., Sullivan, L. M., Kelly–Hayes, M., D’Agostino, R. B., &
Benjamin, E. J. (2004). Anger and hostility predict the development of
atrial fibrillation in men in the Framingham Offspring Study. Circulation, 109, 1267–1271. doi:10.1161/01.CIR.0000118535.15205.8F
Eckhardt, C., Norlander, B., & Deffenbacher, J. (2004). The assessment of
anger and hostility: A critical review. Aggression and Violent Behavior,
9, 17– 43. doi:10.1016/S1359-1789(02)00116-7
Falconer, D. S., & Mackay, T. F. C. (1996). Introduction to quantitative
genetics. (Fourth ed.) Essex, England, U.K.: Longman Group, Ltd.
Fassino, S., Daga, G. A., Piero, A., Leombruni, P., & Rovera, G. G. (2001).
Anger and personality in eating disorders. Journal of Psychosomatic
Research, 51, 757–764. doi:10.1016/S0022-3999(01)00280-X
Jacobo, M. C., Blais, M. A., Baity, M. R., & Harley, R. (2007). Concurrent
validity of personality assessment inventory scales in patients seeking
dialectical behaviour therapy. Journal of Personality Assessment, 88,
74 – 80.
Kawachi, I., Sparrow, D., Spiro, A., Vokonas, P., & Weiss, S. T. (1996).
A prospective study of anger and coronary heart disease—The normative aging study. Circulation, 94, 2090 –2095.
Kendler, K. S., Aggen, S. H., Czajkowski, N., Roysamb, E., Tambs, K.,
Torgersen, S., . . . Reichborn-Kjennerud, T. (2008). The structure of
genetic and environmental risk factors for DSM-IV personality disorders
a multivariate twin study. Archives of General Psychiatry, 65, 1438 –
1446. doi:10.1001/archpsyc.65.12.1438
Kurtz, J. E., & Morey, L. C. (2001). Use of structured self-report assessment
to diagnose borderline personality disorder during major depressive episodes. Assessment, 8, 291–300. doi:10.1177/107319110100800305
Lenzenweger, M. F., Lane, M. C., Loranger, A. W., & Kessler, R. C.
(2007). DSM-IV personality disorders in the National Comorbidity Survey Replication. Biological Psychiatry, 62, 553–564. doi:10.1016/
j.biopsych.2006.09.019
Leon, C. F. (1992). Anger and impatience/irritability in patients of low
socioeconomic status with acute coronary heart disease. Journal of
Behavioral Medicine, 15, 273–284. doi:10.1007/BF00845356
Linehan, M. M. (1993). Cognitive-behavioral treatment of borderline
personality disorder. New York, NY: Guilford Press.
Livesley, J. (2008). Toward a genetically-informed model of borderline
personality disorder. Journal of Personality Disorders, 22, 42–71. doi:
10.1521/pedi.2008.22.1.42
Markovitz, J. H., Matthews, K. A., Wing, R. R., Kuller, L. H., & Meilahn,
E. N. (1991). Psychological, biological and health behavior predictors of
blood-pressure changes in middle-aged women. Journal of Hypertension, 9, 399 – 406. doi:10.1097/00004872-199105000-00003
Morey, L. C. (1991). The Personality Assessment Inventory: Professional
manual. Odessa, FL: Psychological Assessment Resources.
Morse, J. Q., Hill, J., Pilkonis, P. A., Yaggi, K., Broyden, N., Stepp,
S., . . . Feske, U. (2009). Anger, preoccupied attachment, and domain
disorganization in borderline personality disorder. Journal of Personality Disorders, 23, 240 –257. doi:10.1521/pedi.2009.23.3.240
Neale, M. C., Boker, S. M., Xie, G., & Maes, H. H. (2006). Mx: Statistical
modeling. (Sixth ed.) Richmond, VA: Virginia Commonwealth University Department of Psychiatry.
Newhill, C. E., Eack, S. M., & Mulvey, E. P. (2009). Violent behavior in
borderline personality. Journal of Personality Disorders, 23, 541–554.
doi:10.1521/pedi.2009.23.6.541
Owen, J. M. (2011). Transdiagnostic cognitive processes in high trait
anger. Clinical Psychology Review, 31, 193–202. doi:10.1016/
j.cpr.2010.10.003
Posternak, M. A., & Zimmerman, M. (2002). Anger and aggression in
psychiatric outpatients. Journal of Clinical Psychiatry, 63, 665– 672.
doi:10.4088/JCP.v63n0803
Rebollo, I., & Boomsma, D. I. (2006). Genetic analysis of anger: Genetic
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
466
DISTEL ET AL.
dominance or competitive sibling interaction. Behavior Genetics, 36,
216 –228. doi:10.1007/s10519-005-9025-8
Romero–Canyas, R., Downey, G., Berenson, K., Ayduk, O., & Kang, N. J.
(2010). Rejection sensitivity and the rejection-hostility link in romantic
relationships. Journal of Personality, 78, 119 –148. doi:10.1111/j.14676494.2009.00611.x
Rothbart, M. K. (1989). Temperament in development. In G. A. Kohnstamm, J. E. Bates, & M. K. Rothbart (Eds.), Temperament in childhood
(pp. 187–247). New York, NY: Wiley.
Rutter, M. (2007). Gene-environment interdependence. Developmental
Science, 10, 12–18. doi:10.1111/j.1467-7687.2007.00557.x
Schneider, R. H., Egan, B. M., Johnson, E. H., Drobny, H., & Julius, S.
(1986). Anger and anxiety in borderline hypertension. Psychosomatic
Medicine, 48, 242–248.
Skodol, A. E., Gunderson, J. G., Pfohl, B., Widiger, T. A., Livesley, W. J.,
& Siever, L. J. (2002). The borderline diagnosis I: Psychopathology,
comorbidity, and personality structure. Biological Psychiatry, 51, 936 –
950. doi:10.1016/S0006-3223(02)01324-0
Spielberger, C. D., Jacobs, G., Russell, J. S., & Crane, R. S. (1983).
Assessment of anger: The state-trait anger scale. In J. N. Butcher & C. D.
Spielberger (Eds.), Advances in personality assessment (Vol. 2). Hillside, NJ: Erlbaum.
Stein, M. B., Pinkster–Aspen, J. H., & Hilsenroth, M. J. (2007). Borderline
pathology and the Personality Assessment Inventory (PAI): An evaluation of criterion and concurrent validity. Journal of Personality Assessment, 88, 81– 89.
Torgersen, S., Czajkowski, N., Jacobson, K., Reichborn–Kjennerud, T.,
Roysamb, E., Neale, M. C., & Kendler, K. S. (2008). Dimensional representations of DSM-IV cluster B personality disorders in a population-based
sample of Norwegian twins: A multivariate study. Psychological Medicine,
38, 1617–1625. doi:10.1017/S0033291708002924
Torgersen, S., Kringlen, E., & Cramer, V. (2001). The prevalence of
personality disorders in a community sample. Archives of General
Psychiatry, 58, 590 –596. doi:10.1001/archpsyc.58.6.590
Trull, T. J. (1995). Borderline personality disorder features in nonclinical
young adults: 1. Identification and validation. Psychological Assessment, 7, 33– 41. doi:10.1037/1040-3590.7.1.33
Trull, T. J. (2001). Structural relations between borderline personality
disorder features and putative etiological correlates. Journal of Abnormal Psychology, 110, 471– 481. doi:10.1037/0021-843X.110.3.471
van der Ploeg, H. M., Defares, P. B., & Spielberger, C. D. (1982).
Handleiding bij de zelf-analyse vragenlijst ZAV. Een vragenlijst voor het
meten van boosheid en woede, als toestand en als episode. Een Nederlandse bewerking van de Spielberger State-Trait Anger Scale. Lisse, the
Netherlands: Swets & Zeitlinger.
Wilkowski, B. M., & Robinson, M. D. (2010). The anatomy of anger: An
integrative cognitive model of trait anger and reactive aggression. Journal of Personality, 78, 9 –38. doi:10.1111/j.1467-6494.2009.00607.x
Willemsen, G., Posthuma, D., & Boomsma, D. I. (2005). Environmental
factors determine where the Dutch live: Results from the Netherlands
Twin Register. Twin Research and Human Genetics, 8, 312–317. doi:
10.1375/twin.8.4.312
Williams, J. E., Paton, C. C., Siegler, I. C., Eigenbrodt, M. L., Nieto, F. J.,
& Tyroler, H. A. (2000). Anger proneness predicts coronary heart
disease risk—Prospective analysis from the Atherosclerosis Risk in
Communities (ARIC) study. Circulation, 101, 2034 –2039.
Zanarini, M. C., Frankenburg, F. R., Hennen, J., Reich, D. B., & Silk, K. R.
(2005). The McLean Study of Adult Development (MSAD): Overview
and implications of the first six years of prospective follow-up. Journal
of Personality Disorders, 19, 505–523. doi:10.1521/pedi.2005.19.5.505
Zanarini, M. C., Frankenburg, F. R., Reich, D. B., Silk, K. R., Hudson, J. I.,
& McSweeney, L. B. (2007). The subsyndromal phenomenology of
borderline personality disorder: A 10-year follow-up study. The American Journal of Psychiatry, 164, 929 –935. doi:10.1176/appi.ajp
.164.6.929
Zanarini, M. C., Frankenburg, F. R., Yong, L., Raviola, G., Reich, D. B.,
Hennen, J., . . . Gunderson, J. G. (2004). Borderline psychopathology in
the first-degree relatives of borderline and axis II comparison probands.
Journal of Personality Disorders, 18, 439 – 447. doi:10.1521/
pedi.18.5.439.51327
Received February 15, 2011
Revision received September 26, 2011
Accepted October 4, 2011 䡲
E-Mail Notification of Your Latest Issue Online!
Would you like to know when the next issue of your favorite APA journal will be available
online? This service is now available to you. Sign up at http://notify.apa.org/ and you will be
notified by e-mail when issues of interest to you become available!
Psychology of Addictive Behaviors
2015, Vol. 29, No. 3, 590 – 602
© 2015 American Psychological Association
0893-164X/15/$12.00 http://dx.doi.org/10.1037/adb0000109
Can Marijuana Make It Better? Prospective Effects of Marijuana and
Temperament on Risk for Anxiety and Depression
Victoria A. Grunberg, Kismet A. Cordova, L. Cinnamon Bidwell, and Tiffany A. Ito
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
University of Colorado Boulder
Increases in marijuana use in recent years highlight the importance of understanding how marijuana
affects mental health. Of particular relevance is the effect of marijuana use on anxiety and depression
given that marijuana use is highest among late adolescents/early adults, the same age range in which risk
for anxiety and depression is the highest. Here we examine how marijuana use moderates the effects of
temperament on level of anxiety and depression in a prospective design in which baseline marijuana use
and temperament predict anxiety and depression 1 year later. We found that harm avoidance (HA) is
associated with higher anxiety and depression a year later, but only among those low in marijuana use.
Those higher in marijuana use show no relation between HA and symptoms of anxiety and depression.
Marijuana use also moderated the effect of novelty seeking (NS), with symptoms of anxiety and
depression increasing with NS only among those with high marijuana use. NS was unrelated to symptoms
of anxiety and depression among those low in marijuana use. The temperament dimension of reward
dependence was unrelated to anxiety and depression symptoms. Our results suggest that marijuana use
does not have an invariant relationship with anxiety and depression, and that the effects of relatively
stable temperament dimensions can be moderated by other contextual factors.
Keywords: marijuana, harm avoidance, novelty seeking, anxiety, depression
Supplemental materials: http://dx.doi.org/10.1037/adb0000109.supp
Health, 2014b, 2014c), making understanding the factors that
affect them of particular clinical significance. Adding to the clinical relevance, the risk of developing anxiety and depression is
highest within the same age range in which marijuana use is the
highest. That is, 75% of all lifetime cases of anxiety and depression start by age 24 (Kessler et al., 2005) and among adolescents,
roughly 32% have had lifetime prevalence of anxiety disorders and
roughly 14% have had lifetime prevalence of mood disorders
(Merikangas et al., 2010). At the same time, marijuana use is
typically the highest in the teens through early twenties as compared with all other age ranges (Degenhardt et al., 2008; Kessler et
al., 2005), and about 52% of 18- to 25-year-olds have used
marijuana in their lifetime (National Institute of Drug Abuse,
2014).
Here, we take the approach that understanding anxiety and
depression within this population at heightened risk can be improved by examining whether behaviors that are frequent within
this same age range relate to symptoms of anxiety and depression.
That is, given the relatively high rate of marijuana use within late
adolescence/early adulthood and the possibility that it may increase in the face of increasing legalization, there is public health
relevance in knowing the relation of marijuana use to the risk of
anxiety and depression within this age range. This can improve our
understanding of whether increases in marijuana legalization
might affect rates of anxiety and depression, and whether anxiety
and depression prevention and treatment strategies could benefit
by targeting marijuana use.
There is relatively little relevant comorbidity data speaking to
the relation between marijuana and anxiety/depression, as most
large epidemiological studies collapse marijuana use into a broader
Marijuana is the third most commonly used drug in the United
States (after alcohol and tobacco), and the leading illicit drug in
states where its recreational use is currently illegal (CNN Gallup,
2013; National Institute of Drug Abuse, 2014). It is estimated that
more than a third of the American population has used marijuana
and that roughly 7% of Americans currently are regular users
(CNN Gallup, 2013). The perception that marijuana is dangerous
has been decreasing since 2007, corresponding with increasing use
among young people (National Institute of Drug Abuse, 2014) and
increasing legalization for recreational and medical purposes (Colorado Amendment 64, 2012; Washington Initiative 502, 2012).
In the face of such high levels of use and rapid changes to laws
and perceptions, it is critically important to better understand the
consequences of marijuana use. One issue in particular need of
further exploration is the relation of marijuana use to mental
health. Anxiety and depression are the most common mental health
conditions in the United States (National Institute of Mental
Victoria A. Grunberg and Kismet A. Cordova, Department of Psychology and Neuroscience, University of Colorado Boulder; L. Cinnamon
Bidwell, Institute of Cognitive Science, University of Colorado Boulder;
Tiffany A. Ito, Department of Psychology and Neuroscience, University of
Colorado Boulder.
This work was supported by National Institutes of Health (NIH)
DA024002 to Tiffany A. Ito and NIH K23DA033302 to L. Cinnamon
Bidwell.
Correspondence concerning this article should be addressed to Tiffany
A. Ito, Department of Psychology and Neuroscience, University of Colorado, 345 UCB, Boulder, CO 80309-0345. E-mail: tiffany.ito@colorado
.edu
590
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
MARIJUANA, TEMPERAMENT, ANXIETY, AND DEPRESSION
substance use disorder category (for review, see Degenhardt, Hall,
& Lynskey, 2003). Those studies that do separately examine
marijuana use focus only on marijuana dependence and/or examine a wide age range (Chen, Wagner, & Anthony, 2002; Degenhardt, Hall, & Lynskey, 2001). Results from studies that have
focused on recreational users and/or young adults are quite variable; some show a negative association between marijuana use and
anxiety/depression (e.g., Denson & Earleywine, 2006; Sethi et al.,
1986; Stewart, Karp, Pihl, & Peterson, 1997), others a positive
association (e.g., Bonn-Miller, Zvolensky, Leen-Feldner, Feldner,
& Yartz, 2005; Hayatbakhsh et al., 2007; Scholes-Balog, Hemphill, Patton, & Toumbourou, 2013), and still others no association
(e.g., Green & Ritter, 2000; Musty & Kaback, 1995). Such a
diverse pattern of results suggests that other factors may also
interact with marijuana use to affect anxiety and depression. Unfortunately, there has been a great deal of diversity in the extant
research along multiple dimensions (e.g., community vs. college
samples, samples unselected vs. selected for marijuana use, different types of marijuana, anxiety, and depression measures),
making it difficult to identify variables that explain the different
patterns of associations obtained. Here we begin the process of
identifying factors that affect the relation between marijuana use
and anxiety and depression by examining a variable that is itself
known to relate to anxiety and depression. We specifically examine relatively stable aspects of temperament whose relation to
anxiety and depression have been frequently studied and ask
whether marijuana use interacts with temperament in its relationship with anxiety and depression.
591
depression. Specifically, to the degree that marijuana produces
anxiolytic and/or euphoriant effects— either directly through its
biochemical effects on neurotransmitters and receptors or indirectly through expectations and/or the facilitation of mood and
beneficial social interactions—marijuana use may buffer individuals high in HA from increased risk for anxiety and depression.
The other major temperament dimensions in the biosocial model
have shown no consistent associations with anxiety or depression.
Novelty seeking (NS) is thought to bias individuals toward impulsivity and exploration in response to novelty; reward dependence
(RD) reflects a tendency to maintain previously rewarded behaviors. Although these dimensions have been associated with anxiety
and depression in some samples (with occasional negative associations of NS and RD with depression, Farmer et al., 2003; Hansenne et al., 1998), they most often show no association with
anxiety and depression (Copeland, Landry, Stanger, & Hudziak,
2004; Starcevic, Uhlenhuth, Fallon, & Pathak, 1996; Strakowski,
Dunayevich, Keck, & McElroy, 1995; Young et al., 1995). Given
the behavioral biases linked with the temperament dimensions, the
lack of associations with anxiety and depression are theoretically
sensible (i.e., the biases associated with these temperament dimensions would not seem to increase risk for anxiety and depression).
At the same time, these relations have most often been examined
in studies with relatively small samples (fewer than 100 participants), making small effects difficult to detect. More importantly
for the present analyses, the moderating effect of marijuana use has
never to our knowledge been tested.
Current Study
Temperament, Anxiety, and Depression
According to the biosocial model (Cloninger, Svrakic, & Przybeck, 1993), temperament affects mental health via genetically
determined biases that influence automatic responses to novelty,
punishment, and reward. The temperament dimension of harm
avoidance (HA) is particularly relevant for understanding anxiety
and depression as it is characterized by heightened apprehension,
shyness, pessimism, and inhibition of behaviors. Given these biases, it is not surprising that HA is positively associated with both
anxiety and depression (Hansenne et al., 1998; Jiang et al., 2003;
Manfredi et al., 2011; Matsudaira & Kitamura, 2006).
While HA likely increases anxiety and depression, marijuana
can have anxiolytic and euphoriant effects. Such positive mood
effects are reported among the top motives for marijuana use (Lee
et al., 2009; Newcomb, Chou, Bentler, & Huba, 1988; Simons,
Correia, Carey, & Borsari, 1998). Animal research suggests a
direct anxiolytic effect of cannabis administration (e.g.,
Guimarães, Chiaretti, Graeff, & Zuardi, 1990; de Paula Soares et
al., 2010; for a review, see Mechoulam, Parker, & Gallily, 2002).
The exact mechanism of these effects has not been determined,
although they seem to be restricted to the effects of cannabidiol
and not ⌬9-tetrahydrocannabinol (e.g., Zuardi, Crippa, Hallak,
Moreira, & Guimarães, 2006) and likely involve serotonergic
receptors in the dorsal periaqueductal gray matter as the basis for
anxiolytic effects (de Paula Soares et al., 2010). Marijuana use
may also facilitate social contact (Green & Ritter, 2000), which
could, in turn, improve mood and ultimately mental health. The
potential for marijuana use to affect mood suggests a possible
moderating role of marijuana on the relation of HA to anxiety and
The present study seeks to better understand how marijuana use
relates to anxiety and depression within late adolescents/early
adults by examining how it might moderate the effects of temperament on symptoms of anxiety and depression. We also examine,
in a larger sample than past studies, the relation of temperament to
anxiety and depression. We did this in a prospective design in
which marijuana use and temperament assessed at baseline were
used to predict anxiety and depression symptoms assessed 1 year
later in a relatively large (n ⫽ 338) sample of 18- to 21-year-old
male and female college students. Roughly equal numbers of men
and women allow us to test whether relations among temperament,
marijuana use, and anxiety/depression differ for men and women.
Hypotheses
Given past research and the nature of the behavioral biases
associated with HA, we predict that baseline HA will positively
predict both anxiety and depression symptoms assessed 1 year
later. However, given potential anxiolytic and euphoriant effects,
we expect marijuana to moderate this relationship, such that the
positive association of HA with anxiety and depression symptoms
will be most evident when marijuana use is low. Marijuana may
itself show a simple relation to anxiety and depression, with fewer
symptoms of anxiety and depression among those who use marijuana more frequently. We assess these relations while also controlling for baseline anxiety and depression. If HA and its interaction with marijuana use have effects independent of current
anxiety and depression, we expect these relations to be evident
even after controlling for baseline levels of psychopathology.
GRUNBERG, CORDOVA, BIDWELL, AND ITO
592
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Given the lack of consistent relations of NS and RD with
anxiety and depression, we had no specific predictions for these
analyses, but we nevertheless tested them to provide a comprehensive assessment of the relations among temperament dimensions, marijuana use, and risk for anxiety and depression.
We have no a priori expectations that these relations will differ
for men and women, but given gender differences in rates of
anxiety and depression (Kessler, McGonagle, Swartz, Blazer, &
Nelson, 1993; McLean, Asnaani, Litz, & Hofmann, 2011), it is
important to assess whether factors that relate to anxiety and
depression differ for men and women.
Method
Participants
Potential participants were recruited via e-mail invitations to
their university account and advertisements on campus to take part
in a 3-year longitudinal study of marijuana use. Those who were
interested in the study were initially interviewed on the phone by
study personnel to determine whether their marijuana use fit into
one of three categories: never users (i.e., never tried marijuana),
relatively infrequent marijuana users (i.e., used marijuana 4 times
or less per month for less than 3 years), and regular frequent
marijuana users (i.e., used marijuana an average of 5 days a week
or more for at least the past year). Both quantity and frequency
criteria were implemented to ensure that variability in marijuana
use reflected relatively stable tendencies. Because the full protocol
also included electroencephalography measures, individuals who
reported a history of head trauma, neurological disorder, or the use
of prescription medication (with the exception of oral contraceptives or medical marijuana) were excluded from the study. One of
the interests in the larger study was on change in marijuana use
over time, so we oversampled participants with lower levels of use,
whom we expected to be more likely to change their use over time.
We continued sampling within each use category until we had
roughly equal numbers of men and women. Participants who met
criteria for inclusion were invited to participate in two sessions a
year for three total years. Data in the present analyses come from
the first sessions in years 1 and 2.
Our final sample consisted of 375 University of Colorado primarily freshman (see Table 1 for sample characteristics). Of the
337 participants who provided racial information, 1 identified as
Black, 12 as Asian, 11 as Hispanic, 1 as Pacific Islander, 2 as East
Indian, 1 as Middle Eastern, 63 as multiracial, and 246 as White.
Four additional participants were initially enrolled but later found
to have provided inaccurate information at the time of recruitment
and so were dropped from the study. Thirty-seven individuals did
not return to complete Year 2, so analyses are based on the 338
with complete data. Those who failed to return in Year 2 did not
differ from those who did in gender, age, race, marijuana use
group, temperament, or psychopathology, all p values ⬎ .13. Only
the measures of interest to our current hypotheses will be described
Table 1
Sample Characteristics by Marijuana Use Group
Never
Demographics
N (Year 1/Year 2)
Gender (% female)
Age
Ethnicity (% White)
Substance use Year 1
Total days of marijuana use (0–30)
Total grams of marijuana use
Average grams per use day
Temperament Year 1
Harm avoidance
Novelty seeking
Reward dependence
Psychopathology Year 1
Anxiety symptoms
Anxiety % at risk
Depression symptoms
Depression % at risk
Psychopathology Year 2
Anxiety symptoms
Anxiety % subclinical or greater
Depression symptoms
Depression % subclinical or greater
126/114
57 (50%)
18.30 (0.46)
81 (71.1%)
Infrequent
Frequent
F or ␹2 value
146/133
72 (54.1%)
18.38 (0.52)
93 (69.9%)
103/91
47 (51.6%)
18.34 (0.50)
72 (80%)
1.65 (1.88)b
0.87 (1.43)a
0.32 (0.44)b
26.07 (3.52)c
24.45 (16.57)b
0.92 (0.56)c
2.49 (0.66)
2.91 (0.42)a
3.55 (0.58)
2.58 (0.64)
3.08 (0.37)b
3.57 (0.53)
2.52 (0.51)
3.38 (0.40)c
3.58 (0.46)
0.67
36.59ⴱⴱⴱ
0.10
4.40 (2.60)
4.39a
3.64 (2.67)a
0.88a
4.67 (3.39)
15.04b
4.73 (3.90)b
8.27b
4.91 (3.08)
13.19b
5.29 (4.00)b
14.29b
0.71
7.79â´±
5.81ⴱⴱ
13.53ⴱⴱⴱ
3.90 (2.79)
7.02
3.80 (3.28)a
5.26
4.66 (3.45)
11.28
4.99 (4.28)b
12.03
4.40 (2.86)
6.59
4.84 (4.00)ab
10.99
1.88
2.05
3.23â´±
3.62
0.00a
0.00a
0.00a
0.43
0.76
3.10
4,475.48ⴱⴱⴱ
256.76ⴱⴱⴱ
134.33ⴱⴱⴱ
Note. Gender shows number of females. Ethnicity shows number of Whites. Numbers in parentheses are
standard deviations. Possible ranges are 1–5 for harm avoidance, novelty seeking, and reward dependence; 0 –14
for ASR anxiety symptoms; and 0 –24 for ASR depression symptoms. Higher values indicate greater marijuana
use, HA, NS, RD, anxiety, and depression. Anxiety and depression symptoms reflect total number of symptoms
endorsed. % at risk shows percentage of participants who scored at or above the ASR “at-risk” threshold for clinical
levels of anxiety or depression (T score ⱖ 65). F and ␹2 values reflect the test of the omnibus marijuana use group
main effect, df ⫽ 2, 335, and 2, respectively. Marijuana use group means within the same row with different subscripts
differ at p ⬍ .05. For omnibus marijuana use group effects, ⴱ p ⱕ .05. ⴱⴱ p ⱕ .01. ⴱⴱⴱ p ⱕ .001.
MARIJUANA, TEMPERAMENT, ANXIETY, AND DEPRESSION
in detail, but where appropriate (e.g., when they preceded the
measures of interest), other measures collected will be noted.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Self-Report Measures
Temperament and Character Inventory (Year 1 and Year 2).
HA, NS, and RD were measured with the Temperament and
Character Inventory (TCI; Cloninger et al., 1993). HA was assessed with 33 items that assess anticipatory worry, fear of uncertainty, shyness with strangers, and fatigability (e.g., “Usually I am
more worried than most people that something might go wrong in
the future,” “I usually feel tense and worried when I have to do
something new and unfamiliar,” “When I meet a group of strangers, I am more shy than most people,” “I have less energy and get
tired more quickly than most people,” ␣ ⫽ .93). NS was assessed
with 35 items that assess exploratory excitability, impulsivity,
extravagance, and disorderliness (e.g., “When nothing new is
happening, I usually start looking for something that is thrilling or
exciting,” “I often do things based on how I feel at the moment
without thinking about how they were done in the past,” “I often
spend money until I run out of cash or get into debt from using too
much credit,” “I like when people can do whatever they want
without strict rules and regulations,” ␣ ⫽ .85). RD was assessed
with 30 items that assess sentimentality, openness to warm communication versus aloofness, attachment, and dependence (e.g., “I
am strongly moved by sentimental appeals, like when asked to
help crippled children,” “I like other people to know that I really
care about them,” “I like to discuss my experiences and feelings
openly with friends instead of keeping them to myself,” “I don’t
care very much whether other people like me or the way I do
things,” reverse-coded, ␣ ⫽ .88).
All items were answered with respect to how the participants
usually or generally act and feel using a 5-point scale (1 ⫽
definitely false to 5 ⫽ definitely true). Separate mean scores were
created for overall HA, NS, and RD, with higher scores reflecting
greater HA, NS, and RD. The biosocial model currently includes
a fourth temperament dimension of persistence (P) that was previously included as part of RD (Cloninger et al., 1993). P is
associated with determination and industriousness. It has been
much less frequently measured in association with anxiety and
depression, and when it has been, shows inconsistent relations
(Cloninger, Svrakic, & Przybeck, 2006; Hansenne et al., 1999).
Measures of P were omitted from the present study out of space
considerations.
Marijuana use (Year 1). Self-reported marijuana use during
the past 30 days was assessed using the Time-Line Follow Back
(TLFB; Sobell & Sobell, 1992), a calendar-assisted structured
interview in which participants were asked to indicate over the past
30 days the quantity of marijuana used on each day. Frequency and
quantity reports were highly correlated, r ⫽ .82, p ⬍ .0001.
Relative to other substances such as alcohol and nicotine, where
individuals might consume an entire beer or cigarette, marijuana
users might just take a few hits. There are also many different ways
to consume marijuana (joints, vaporizers, edibles). Because of this
potential for variability, our main analyses used marijuana use
frequency as our measure of marijuana use. However, secondary
analyses were also conducted with marijuana quantity measures
and yielded identical results (see Supplement Tables S1, S2, and
S3 in the online supplemental material).
593
Adult Self-Report (ASR; Year 1 and Year 2). Symptoms of
anxiety and depression were measured with the Achenbach System
of Empirically Based Assessment Adult Self-Report (ASR;
Achenbach & Rescorla, 2003), a self-report measure of current
internalizing and externalizing psychopathology that is the adult
parallel to the Child Behavior Checklist. These internalizing and
externalizing scales on the ASR have been well validated and have
adequate psychometric characteristics (Achenbach & Rescorla,
2003). Of interest were anxiety problems and depressive problems.
Participants were asked how well each item described them over
the past 6 months, with responses ranging from 0 ⫽ not true, 1 ⫽
somewhat or sometimes true, and 2 ⫽ very true or often true.
Anxiety problems were assessed with 7 items (e.g., “I am nervous
or tense,” Year 1: ␣ ⫽ .76; Year 2: ␣ ⫽ .77). Depressive problems
were assessed with 14 items (e.g., “I am unhappy, sad, or depressed,” Year 1: ␣ ⫽ .77; Year 2: ␣ ⫽ .80). A total score was
created for each construct with higher scores reflecting a continuous measure of greater endorsement of anxiety and depressive
problems. Because of the conceptual similarity between HA and
anxiety and depression, we conducted factor analyses to confirm
that HA was distinct from anxiety and depression (see online
supplemental material).
Procedure
Participants who met criteria for inclusion were invited to participate in a total of six laboratory sessions over 3 years. Data in
the present analyses come from the first sessions in Years 1 and 2
at which marijuana use, temperament, and psychopathology were
assessed. The assessments occurred approximately 12 months
apart (M ⫽ 356.98 days, SD ⫽ 19.78 days).
Participants were instructed to abstain from alcohol for 24 hr,
recreational drugs (including marijuana) for 6 hr, and caffeine and
cigarettes for 1 hr prior to each laboratory session. In both sessions, participants were breathalyzed to ensure a breath alcohol
concentration of zero. Adherence to other abstinence requirements
was verified verbally. Although it would have been preferable to
assess abstinence biochemically, it was prohibitively expensive.
While failure to meet the requirements could add variability to the
responses, none of the participants were visibly impaired, and we
have no reason to think failure to conform to the abstinence
requirements introduced any systematic artifact (i.e., failure to
meet the abstinence requirement seems unlikely to have created
the pattern of relationships among the variables that we observed).
Participants next completed the TLFB followed by a questionnaire
including demographics, the ASR, and the TCI. Prior to the ASR,
participants completed measures of handedness, ADHD symptoms
(Barkley & Murphy, 1998), the Beck Depression Index (Beck,
Steer, & Carbin, 1988), and the Beck Anxiety Index (Beck, Epstein, Brown, & Steer, 1988). Prior to completing the TCI, participants completed the Shortened Self-Regulation Questionnaire
(Carey, Neal, & Collins, 2004). Participants received $25 at each
session.
Analysis Strategy
We first performed preliminary descriptive analyses to assess
the relation between marijuana use, temperament, and anxiety and
depression symptoms, with separate analyses representing mari-
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
594
GRUNBERG, CORDOVA, BIDWELL, AND ITO
juana use either categorically or continuously. Categorical analyses were done with one-way analyses of variance (ANOVAs)
using a 3-level marijuana use group variable based on the participant’s use at time of study enrollment (i.e., never, infrequent,
frequent). Continuous analyses consisted of bivariate correlations
between number of days marijuana was used in past 30 days from
the baseline Year 1 TLFB and temperament and psychopathology
variables.
Our primary analyses assessed whether Year 1 HA, NS, and RD
predict Year 2 anxiety and depression symptoms, and whether this
relationship is moderated by marijuana use. This was tested using
a cross-lag structural regression approach (Rogosa, 1980). Under
this approach, two multiple regression equations are used to test
the relations shown in Figure 1, illustrated using HA and anxiety
symptoms as an example. Model 1 tests our hypothesized relations
that Year 2 anxiety is predicted by Year 1 HA, and that the relation
between HA and anxiety is moderated by marijuana use. More
specifically, ␤1 assesses the autoregressive or lagged effect of Year
1 anxiety predicting Year 2 anxiety. Because we assume that initial
anxiety and depression symptoms will be a strong predictor of
subsequent anxiety and depression, including this lagged effect
provides a strong test of the degree to which temperament and
marijuana use predict anxiety and depression over and above
baseline anxiety and depression. Of primary theoretical interest are
␤2 and ␤3. ␤2 tests the effects of Year 1 HA on Year 2 anxiety,
reflecting a simple crossed effect, whereas ␤3 assesses whether the
simple crossed effect of HA on Year 2 anxiety is moderated by
Year 1 marijuana use. As described by Baron and Kenny (1986),
this predicted moderation is tested by assessing the interaction
between Year 1 HA and Year 1 marijuana use. Although our data
are correlational, given the temporal precedence of the variables
(i.e., that HA and marijuana use in Year 1 are predicting anxiety in
Year 2), significant coefficients for ␤2 and ␤3 are consistent with
the possibility that initial HA affects subsequent anxiety, and that
the relation of initial HA on subsequent HA is moderated by initial
marijuana use, respectively (Rogosa, 1980).
To further test our hypothesized relations, additional autoregressive, cross-lagged, and cross-lagged moderation effects are tested
in Model 2 in which Year 2 HA is the outcome. Specifically, ␥1
Figure 1. Sample cross-lag structural regression model. Hypothesized
relations were tested with two multiple regression models. Heavier lines
indicate the two paths of primary theoretical interest assessing the crosslagged effect of initial temperament on subsequent psychopathology (␤2)
and the degree to which initial marijuana use moderates the cross-lagged
effect of initial temperament on subsequent psychopathology (␤3).
assesses the autoregressive or lagged effect of Year 1 HA predicting Year 2 HA. These relations are not of particular theoretical
interest here, but because this tests the temporal stability of temperament, we expect the autoregressive effects in these second
models to be significant. Of primary theoretical interest are ␥2 and
␥3. ␥2 tests the effect of Year 1 anxiety symptoms on Year 2 HA.
Because we expect the relations between temperament and psychopathology to reflect the effect of the more stable temperament
variables affecting subsequent psychopathology rather than initial
symptoms of psychopathology affecting subsequent temperament,
we do not expect initial anxiety symptoms to predict subsequent
HA. Thus, we expect ␥2 to be nonsignificant. Similarly, we have
no theoretical expectation that the impact of initial anxiety symptoms on subsequent temperament will be moderated by marijuana
use, so we do not expect ␥3 to be significant. In this way, nonsignificant coefficients for ␥2 and ␥3, coupled with significant coefficients for ␤2 and/or ␤3, provide additional evidence for our
hypothesized relations. In sum, to the degree that relations between
temperament and anxiety reflect the effect of initial temperament
on subsequent anxiety and not the effect of initial anxiety on
subsequent temperament, we expect significant effects in ␤2 and/or
␤3, but not ␥2 and ␥3.
This framework just described was repeated 6 times to test the
relation of each aspect of temperament (HA, NS, and RD) on each
outcome variable (symptoms of anxiety and depression). Figure 1
shows a simplified model highlighting the autoregressive, crosslagged, and moderated cross-lagged relations. In addition to these
three variables, each model also contained five additional predictors. First, given the presence of marijuana use in the interaction
term to test for moderation (e.g., Year 1 HA â«» Year 1 marijuana
use), all models also included the simple effect of Year 1 marijuana use as a predictor. This variable is also of theoretical interest
because it assesses the simple effect of marijuana on anxiety and
depression symptoms. We also included gender and its interactions
with temperament and marijuana use (e.g., HA â«» Gender, Marijuana Use â«» Gender, HA â«» Marijuana Use â«» Gender) to test
whether interrelations among temperament, marijuana use, and
symptoms of anxiety and depression differ for males and females.
All continuous variables were mean-centered before analyses,
and gender was coded as 1 ⫽ male and ⫺1 ⫽ female. All model
assumptions (e.g., homoscedasticity, normality of distributions)
were met. When the predicted interaction between temperament
and marijuana use reflecting our primary test of moderation was
significant, we explored the form of the interaction by plotting and
testing the simple effects of temperament on anxiety or depression
symptoms at lower and higher levels of marijuana use following
Aiken and West (1991). The values of lower and higher marijuana
use selected to test the simple effects were based on examination
of the distribution of marijuana use reported in the TLFB in Year
1 with the goal of assessing effects at values that reflect actual
levels of low and high use in our sample. Based on use within our
sample, we plot and statistically test the effects of temperament on
anxiety and depression at 0 days (low use) and 25.80 days (high
use) of marijuana use, with the latter reflecting the mean level of
use reported by frequent users on the TLFB in Year 1. The low
marijuana use group included all participants recruited as never
users (i.e., they all reported 0 days of use in the past 30 days on the
TLFB) as well as 48 infrequent users who also happened to have
no days of use in the 30 days prior to completion of the Year 1
MARIJUANA, TEMPERAMENT, ANXIETY, AND DEPRESSION
TLFB. We also conducted ancillary simple effects tests using ⫾.5
standard deviations to represent low (M ⫽ 2.07 days) and high (M ⫽
13.47 days) levels of marijuana use. Results were identical to those
reported here.
Results
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Preliminary Analyses
Table 1 presents descriptive statistics from our sample by categorical marijuana use group classification at time of study enrollment (never, infrequent, frequent). Age, ethnicity, and gender measured in Year 1 did not differ across groups. One-way
ANOVAs on the temperament variables revealed a significant
effect of marijuana use group only on NS, with more frequent
marijuana use associated with higher novelty seeking. One-way
ANOVAs also revealed significant differences in depression
among the marijuana use groups in both Year 1 and Year 2, with
those who use marijuana reporting more depression symptoms.
Marijuana users were also more likely to meet or exceed the
“at-risk” threshold for clinical levels of depression in Year 1 (ASR
T score greater than or equal to 65). While there was no marijuana
use group effect on anxiety symptoms, those who use marijuana
were more likely to meet or exceed the “at-risk” threshold for
clinical levels of anxiety in Year 1.
In addition to examining marijuana use categorically based on
use at study enrollment, we can also examine marijuana use
continuously based on TLFB via bivariate correlations (see Table
2). When examined continuously, marijuana use was only weakly
associated with Year 1 depression symptoms, but not with Year 2
depression symptoms. It was unrelated to anxiety. More frequent
marijuana use was associated with higher NS. Of theoretical interest, HA was positively correlated with Year 1 and Year 2
anxiety and depression symptoms. Neither marijuana use frequency, NS, nor RD were correlated with anxiety or depression
symptoms.
Main Analyses
The preliminary correlational analyses (see Table 2) show consistent simple relations between HA and anxiety and depression
symptoms in both Years 1 and 2. To more specifically test our
hypotheses about the relation of temperament to subsequent anxiety and depression symptoms, as well as the moderating effect of
marijuana on this relation, we conducted a series of cross-lag
595
regression models, as described in the Analysis Strategy. To facilitate interpretation, Figure 2 presents the coefficients of greatest
interest in testing our hypotheses (cf. Figure 1) while Table 3
presents full model output including all predictors.
Harm avoidance and anxiety. The first model we ran tests
our primary hypotheses that initial temperament predicts subsequent psychopathology, and that this relation may be moderated by marijuana use (i.e., Model 1 in Figure 1). This was done
by regressing Year 2 anxiety on Year 1 HA, Year 1 marijuana
use, Year 1 anxiety, the HA â«» Marijuana Use interaction term,
gender, and all interaction terms involving gender. This model
revealed three significant effects (see Table 3). Not surprisingly, Year 1 anxiety symptoms significantly predicted Year 2
anxiety symptoms. There was also a significant gender effect,
with women reporting more Year 2 anxiety symptoms than
men. Of interest, when these other variables were included in
the model, HA was not an independent predictor of Year 2
anxiety symptoms. However, consistent with hypotheses, HA
did interact with marijuana use in predicting anxiety symptoms
(Figure 2, Panel A). To understand this interaction, we conducted simple slope analyses separately assessing the relation
between HA and Year 2 anxiety for those low and high in
marijuana use (Aiken & West, 1991). As can be seen in Figure
3, Panel A, when frequency of marijuana use was low, increases
in Year 1 HA were associated with greater anxiety in Year 2,
␤ ⫽ .15, t(329) ⫽ 2.69, p ⬍ .01. By contrast, as predicted,
marijuana use had a buffering effect as reflected in a nonsignificant relation between HA and anxiety when marijuana use
frequency was high, ␤ ⫽ ⫺.14, t(329) ⫽ ⫺1.40, p ⫽ .16.
To further evaluate our hypotheses, we also tested a second
model assessing the other possible cross-lagged effect—that initial
psychopathology predicts subsequent temperament (Model 2 in
Figure 1). This was tested by regressing Year 2 HA on Year 1 HA,
Year 1 anxiety, Year 1 marijuana use, the Anxiety â«» Marijuana
Use interaction term, gender, and all interaction terms involving
gender. Not surprisingly, Year 1 HA predicted Year 2 HA (see
Table 3). Of primary theoretical relevance, neither the simple
cross-lagged effect of Year 1 anxiety nor the moderated crosslagged effect of Year 1 Anxiety â«» Marijuana Use were significant
(Figure 2, Panel A). The only other significant effect in this model
was the Anxiety â«» Gender interaction. Tests of simple slopes
showed that Year 1 anxiety was associated with greater Year 2 HA
for women, ␤ ⫽ .11, t(329) ⫽ 2.15, p ⫽ .03, but not men,
␤ ⫽ ⫺.04, t(329) ⫽ ⫺.79, p ⫽ .43.
Table 2
Bivariate Correlations Among Variables
Variable
1.
2.
3.
4.
5.
6.
7.
8.
⫹
Year 1 marijuana use
Harm avoidance
Novelty seeking
Reward dependence
Year 1 anxiety
Year 1 depression
Year 2 anxiety
Year 2 depression
p ⬍ .10.
â´±
p ⬍ .05.
ⴱⴱ
1
2
3
4
5
6
7
—
⫺.01
.40ⴱⴱ
.01
.05
.13â´±
.01
.05
—
⫺.32ⴱⴱ
⫺.14ⴱ
.57ⴱⴱ
.55ⴱⴱ
.46ⴱⴱ
.42ⴱⴱ
—
.12â´±
⫺.06
.04
⫺.04
.05
—
.03
.02
.00
⫺.06
—
.68ⴱⴱ
.68ⴱⴱ
.51ⴱⴱ
—
.50ⴱⴱ
.63ⴱⴱ
—
.69ⴱⴱ
p ⬍ .01.
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
596
GRUNBERG, CORDOVA, BIDWELL, AND ITO
Figure 2. Cross-lag structural regression models assessing the relation of harm avoidance and marijuana use
to levels of anxiety (Panel A) and depression (Panel B), novelty seeking and marijuana use to levels of anxiety
(Panel C) and depression (Panel D), and reward dependence and marijuana use to levels of anxiety (Panel E) and
depression (Panel F). Heavier lines indicate the cross-lagged and moderated cross-lagged relations of primary
theoretical significance (see Table 3 for exact significance levels). ⴱ p ⬍ .05.
Harm avoidance and depression. Figure 2, Panel B shows
the relations of primary theoretical interest in the cross-lag
model assessing the relation between HA and depression symptoms. Considering the first regression model that tests whether
initial temperament predicts subsequent psychopathology, Year
1 depression symptoms were a significant predictor of Year 2
depression symptoms (see Table 3). The only other significant
effect was the predicted HA â«» Marijuana Use interaction.
Simple effects tests revealed a pattern of effects very similar to
that obtained for anxiety (Figure 3, Panel B): HA was significantly positively associated with depression symptoms when
marijuana use frequency was low, ␤ ⫽ .15, t(329) ⫽ 2.69, p ⬍
.01, but marijuana use appeared to have a buffering effect as reflected
in a nonsignificant negative relation between HA and depression
symptoms at high levels of marijuana use frequency, ␤ ⫽ ⫺.09,
t(329) ⫽ ⫺.92, p ⫽ .36.
In the second model predicting Year 2 HA, the only significant
predictor was Year 1 HA (see Table 3). Of importance, Year 1
depression symptoms did not predict Year 2 HA, nor did the Year
1 Depression â«» Marijuana Use interaction.
Novelty seeking and anxiety. The cross-lag model in Figure
2, Panel C shows the coefficients of primary theoretical interest in
assessing the relation between NS and anxiety. Considering the
first regression model that tests whether initial temperament predicts subsequent psychopathology, three effects were significant.
Year 1 anxiety symptoms significantly predicted Year 2 anxiety
symptoms, and greater marijuana use frequency in Year 1 was
associated with less anxiety in Year 2 (see Table 3). While NS did
not have a direct effect on anxiety levels, its effect was moderated
by marijuana use, as reflected in the NS â«» Marijuana Use interaction. Simple effects displayed in Figure 3, Panel C show that the
relation between NS and anxiety occurs among those with high
marijuana use frequency. That is, when marijuana use frequency
was high, Year 1 NS was positively associated with anxiety
symptoms, ␤ ⫽ .28, t(329) ⫽ 3.46, p ⫽ .001. There was no
relation between NS and anxiety symptoms when marijuana use
frequency was low, ␤ ⫽ ⫺.08, t(329) ⫽ ⫺1.61, p ⫽ .11.
In the second model predicting Year 2 NS, the only significant
predictor was Year 1 NS (see Table 3). Year 1 anxiety did not
predict Year 2 NS, nor did the Year 1 Anxiety â«» Marijuana Use
interaction.
Novelty seeking and depression. The regression models predicting Year 2 depression from NS revealed effects very similar to
those in the models predicting anxiety from NS (See Figure 2,
Panel D). In the first regression model that tests whether initial
temperament predicts subsequent psychopathology, there were
MARIJUANA, TEMPERAMENT, ANXIETY, AND DEPRESSION
597
Table 3
Full Regression Models
Harm avoidance
HA and anxiety
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Model 1: DV ⫽ Anxiety
HA and depression
␤
t
p
0.62
0.06
⫺0.02
⫺0.09
⫺0.11
0.01
0.00
⫺0.02
12.76
1.24
⫺0.54
⫺2.31
⫺2.69
0.20
⫺0.09
⫺0.35
⬍.001
.22
.59
.02
.01
.84
.93
.73
Model 2: DV ⫽ HA
␤
t
p
Year 1 HA
Year 1 anxiety
Marijuana use frequency
Gender
Anxiety â«» Marijuana Use
Marijuana Use â«» Gender
Anxiety â«» Gender
Anxiety â«» Marijuana Use â«» Gender
0.76
0.03
⫺0.05
⫺0.05
⫺0.04
0.01
⫺0.07
⫺0.01
18.88
0.79
⫺1.59
⫺1.39
⫺1.19
0.17
⫺2.24
⫺0.26
⬍.001
.43
.11
.17
.23
.86
.03
.80
Year 1 anxiety
Year 1 HA
Marijuana use frequency
Gender
HA â«» Marijuana Use
Marijuana Use â«» Gender
HA â«» Gender
HA â«» Marijuana Use â«» Gender
Model 1: DV ⫽ Depression
␤
t
p
0.59
0.08
⫺0.03
0.01
⫺0.10
⫺0.02
0.00
0.02
11.30
1.52
⫺0.65
0.17
⫺2.22
⫺0.50
0.07
0.49
⬍.001
.13
.52
.87
.03
.62
.94
.63
Model 2: DV ⫽ HA
␤
t
p
Year 1 HA
Year 1 depression
Marijuana use frequency
Gender
Depression â«» Marijuana Use
Marijuana Use â«» Gender
Depression â«» Gender
Depression â«» Marijuana Use â«» Gender
0.77
0.01
⫺0.05
⫺0.05
⫺0.02
0.01
⫺0.06
0.03
19.24
0.28
⫺1.38
⫺1.53
⫺0.43
0.24
⫺1.82
0.89
⬍.001
.78
.17
.13
.67
.81
.07
.37
␤
t
p
0.65
0.05
⫺0.12
0.04
0.13
0.07
⫺0.05
⫺0.07
14.84
1.18
⫺2.46
0.76
2.93
1.41
⫺1.06
⫺1.51
⬍.001
.24
.02
.45
.00
.16
.29
.13
Year 1 depression
Year 1 HA
Marijuana use frequency
Gender
HA â«» Marijuana Use
Marijuana Use â«» Gender
HA â«» Gender
HA â«» Marijuana Use â«» Gender
Novelty seeking
NS and anxiety
Model 1: DV ⫽ Anxiety
NS and depression
␤
t
p
0.67
0.03
⫺0.10
⫺0.08
0.16
0.09
⫺0.05
⫺0.04
16.67
0.58
⫺2.22
⫺1.78
3.80
1.87
⫺1.26
⫺0.91
⬍.001
.56
.03
.08
⬍.001
.06
.21
.36
Model 2: DV ⫽ NS
␤
t
p
Model 2: DV ⫽ NS
␤
t
p
Year 1 NS
Year 1 anxiety
Marijuana use frequency
Gender
Anxiety â«» Marijuana Use
Marijuana Use â«» Gender
Anxiety â«» Gender
Anxiety â«» M…
Purchase answer to see full
attachment

We offer the bestcustom writing paper services. We have done this question before, we can also do it for you.

Why Choose Us

  • 100% non-plagiarized Papers
  • 24/7 /365 Service Available
  • Affordable Prices
  • Any Paper, Urgency, and Subject
  • Will complete your papers in 6 hours
  • On-time Delivery
  • Money-back and Privacy guarantees
  • Unlimited Amendments upon request
  • Satisfaction guarantee

How it Works

  • Click on the “Place Order” tab at the top menu or “Order Now” icon at the bottom and a new page will appear with an order form to be filled.
  • Fill in your paper’s requirements in the "PAPER DETAILS" section.
  • Fill in your paper’s academic level, deadline, and the required number of pages from the drop-down menus.
  • Click “CREATE ACCOUNT & SIGN IN” to enter your registration details and get an account with us for record-keeping and then, click on “PROCEED TO CHECKOUT” at the bottom of the page.
  • From there, the payment sections will show, follow the guided payment process and your order will be available for our writing team to work on it.