Content uploaded by Paul A Pilkonis
Author content
All content in this area was uploaded by Paul A Pilkonis
Content may be subject to copyright.
Stability of Borderline Personality Disorder Features in Girls
Stephanie D. Stepp, Paul A. Pilkonis, Alison E. Hipwell, Rolf Loeber, and Magda
Stouthamer-Loeber
University of Pittsburgh Medical Center
Abstract
Little empirical evidence exists regarding developmental antecedents of borderline personality
disorder (BPD) features in children and adolescents. As a first step in addressing this gap in our
knowledge, this study examined the factor structure and stability of putative underlying BPD
features, specifically impulsivity, negative affectivity and interpersonal aggression, in 6–12 year-
old girls. We report on results from exploratory and confirmatory factor analyses of underlying
BPD dimensions as rated by parents and teachers over six successive data waves in a large,
community sample of girls (N=2,451). Six factors were derived from parent ratings (i.e., Cognitive
Dyscontrol, (Lack of) Self-Control, Hostility, Depression/Anxiety, Hyperactivity, and Relational
Aggression) and five factors were derived from teacher reports (i.e., Cognitive Dyscontrol,
Hyperactivity, (Lack of) Self-Control, Relational Aggression, and Depression). The item
composition of similar parent and teacher factors was highly consistent. The year-to-year stability
from ages 6 to 12 was high for parent factor scores (r ranging from .71–.85) and moderately high
for teacher factor scores (r ranging from .49–.77). These findings suggest that underlying
dimensions of BPD features can be reliably measured and are stable in 6–12 year-old girls.
Borderline Personality Disorder (BPD) is a heterogeneous disorder characterized by
affective instability, cognitive disturbances, impulsive and self-damaging acts, and
dysfunctional interpersonal relationships (APA, 2000). Individuals with BPD features are
likely to experience poor outcomes in occupational, academic, and interpersonal functioning
(Bagge, et al., 2004; Zweig-Frank & Paris, 2002) and utilize more treatment services than
those without such features (Bagge, Stepp, & Trull, 2005). Furthermore, those with BPD are
well-represented in psychiatric settings, accounting for 10–20% of outpatients and 15% of
inpatients (Gunderson, 2001).
Given the extraordinary suffering endured by those afflicted and the strain this disorder
imposes upon individuals who come into contact with BPD, Lenzenweger and Cicchetti
(2005) discuss the importance of elucidating etiological pathways and the developmental
course of BPD to aid in early identification and prevention efforts. Most research to date on
BPD has been limited to studying adult samples. Thus, what is known about the
developmental history of individuals with BPD relies heavily on retrospective reporting.
Although formal diagnosis using the DSM criteria for BPD is usually postponed until age
18, the constructs that underlie this disorder have a developmental course and, thus, can be
measured in childhood and adolescence.
Three factors have reliably emerged from factor analytic work and are conceptualized as
core features of the disorder: impulsivity, negative affectivity, and interpersonal aggression
Corresponding Author: Correspondence concerning this article should be addressed to Stephanie D. Stepp, Ph.D., Western Psychiatric
Institute and Clinic, 3811 O’Hara St., Pittsburgh, PA 15213. You many also contact the author for an appendix of tables providing
additional information regarding the item selection process and further documentation of the results.
NIH Public Access
Author Manuscript
J Pers Disord. Author manuscript; available in PMC 2010 August 12.
Published in final edited form as:
J Pers Disord
. 2010 August ; 24(4): 460–472. doi:10.1521/pedi.2010.24.4.460.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript
(e.g., Gunderson, 2007; Skodol, Gunderson, Pfohl, Widiger, & Siever, 2002). These three
core features have also been implicated in the development of BPD. For example, Trull
(2001) found that personality traits assessing impulse control and negative affectivity
predicted BPD features in young adults over a 2-year time period. Additionally, Crick,
Murray-Close, and Woods (2005) found that interpersonal turmoil, characterized by
friendship exclusivity and relational aggression, predicted BPD over the course of one year
in 4th to 6th grade children. We hypothesize that similar associations exist between the
underlying constructs of impulsivity, negative affectivity, and interpersonal aggression and
later BPD features in children and adolescents. Explicating the nature of these relations has
implications for developmental pathways of risk for BPD.
Epidemiological studies estimate that BPD affects males and females equally (Torgersen,
Kringlen, & Cramer, 2001). Examining the stability of these constructs in girls is important
because this disorder appears to be particularly pernicious for females. In clinical settings,
75% of those with a BPD diagnosis are women (Skodol & Bender, 2003). Understanding the
precursors of BPD in girls will yield important information about the etiology and
developmental course for those who appear to be at particularly high risk for utilizing
treatment in adulthood.
Questions
1. Using exploratory factor analysis (EFA), what factor structure emerges when
sampling content from the domains of impulsivity, negative affectivity, and
interpersonal aggression in girls between ages 6 and 12? Does this factor structure
vary by informant (parent vs. teacher)?
2. Can the factor structure derived from EFA be validated in each of the age groups
using more stringent confirmatory factor analytic techniques?
3. How stable are the factors during this age period? Does stability increase with age?
Method
Sample Description
The participants of the Pittsburgh Girls Study (PGS) are 2,451 five to eight-year old girls
recruited from a sample of 103,238 households in the city of Pittsburgh. Participants were
identified by a stratified sampling of households in Pittsburgh neighborhoods where
households in low-income neighborhoods were over-sampled. For the purposes of this
study, neighborhoods were deemed low-income if at least 25% of the families were living at
or below the poverty level, using 1990 Census data. Enumeration was completed in 89 of the
90 City of Pittsburgh neighborhoods during 1999, when households in low-income
neighborhoods were fully enumerated. Half of the households in other neighborhoods were
randomly sampled. In total, 3,241 girls in the 5- to 8-year old age range – 83.7% of the girls
noted in the 2000 Census – were identified. Of those girls initially identified as meeting the
age criterion, 2,876 were asked to take part in the longitudinal study. From this pool, a total
of 2,451 (85.2%) girls agreed to participate (for further details, see Hipwell et al., 2002).
At the time of the first interview, the sample comprised 588 five-year olds, 630 six-year
olds, 611 seven-year olds, and 622 eight-year olds. African American girls made up slightly
more than half of the sample (52.8%), while 40.9% were Caucasian. Most of the remaining
6.3% of girls were described as multi-racial. In 92.7% of the interviews, the primary
caregiver was a biological parent and in 92.9% of the cases the caregiver interviewed was
female. The large majority of the parents (83.2%) had at least a high school education. In a
majority of households (58.8%), the parent was cohabiting with a spouse or domestic
Stepp et al. Page 2
J Pers Disord. Author manuscript; available in PMC 2010 August 12.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript
partner. Of the families surveyed, 38.9% reported receiving public assistance in the form of
Women, Infants and Children Program (WIC), food stamps, or welfare.
Data Collection
Separate in-home interviews for both the child and the parent were conducted annually by
trained interviewers using a laptop computer. Parents gave further feedback by completing
additional questionnaires. Teacher participation was obtained using questionnaire booklets,
distributed via a mix of mail and hand-delivery. All participants were reimbursed for their
involvement. Study procedures were approved by the University of Pittsburgh Institutional
Review Board and parental consent and child assent were obtained.
This paper covers the first six waves of parent and teacher data collected by the PGS. During
this time period, cohort 5 girls ranged in age from 5 to 10, cohort 6 were ages 6 to 11, cohort
7 girls were 7 to 12 years of age, and cohort 8 ranged from 8 to 13 years old. Because the
girls were not interviewed at age 5 and a full interview was not administered until age 7,
self-reported data from the girls were not used. Additionally, due to the relatively smaller
number of girls aged 5 (n=588) and 13 (n=565), this study is limited to girls aged 6–12.
All parents completed the interview during the first year. Valid teacher booklets were
obtained from 1,832 (74.8%) of the participants’ teachers during this wave. In year 2,
interviews were completed by 2,383 (97.2%) parents, while 2,145 (87.5%) teachers
completed and returned booklets. Parent participation was 95.4% (N=2,339) and teacher
participation was 84.8% (N=2,079) for the third interview year. At year 4, parent and teacher
participation rates were 94.3% (N=2,310) and 83.8% (N=2,054), respectively. For year 5,
parent and teacher participation rates were 92.9% (N=2,277) and 80.9% (N=1,982),
respectively. In year 6, parent and teacher participation rates were 92.2% (N=2,260) and
82.5% (N=2,021), respectively.
To assess the uniformity of the data across informants at each time point, an attrition
analysis was run. For each year, participants who had missing teacher data were compared
to those participants who received valid responses from both the parent and teacher on race
(African American, Caucasian, Other), single parent, public assistance (any participant
whose family received WIC, food stamps, or welfare), and low parental education (parent
with less than 12 years of formal education). During year 1, the only difference in rates of
those missing data concerned race: African Americans and Caucasians were very similar
(21.7% and 18.3% missing teacher data, respectively) while other minorities showed a much
lower attrition rate at 13.2%. In year 2, girls with missing teacher data differed on receipt of
public assistance (12.2% of girls whose family received public assistance were missing
teacher data, while only 7.7% of girls whose family did not receive public assistance were
missing) and race (12.7% of African Americans were missing, while 6.5% of both
Caucasians and other minorities were missing). There were no significant differences in year
3. The analysis of year 4 data showed a significant difference in race, as 13.3% of African
Americans, 11.7% of other minorities, and 8.0% of Caucasians had missing teacher data. In
year 5, girls with missing teacher data differed on receipt of public assistance (15.2%
receiving public assistance compared with 11.9% not receiving public assistance were
missing teacher data). There were no significant differences in year 6. Results from these
analyses do not suggest any systematic bias due to sample loss.
Selection of BPD-Related Items
We selected items that appeared to be valid indicators of impulsivity, negative affectivity,
and interpersonal aggression. To assess impulsivity, we mapped items onto three facets of
Whiteside and Lynam’s (2001) model of impulsivity: Urgency, (Lack of) Premeditation, and
Stepp et al. Page 3
J Pers Disord. Author manuscript; available in PMC 2010 August 12.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript
(Lack of) Perseverance. In their model, Urgency is defined as the tendency to engage in
impulsive behaviors when experiencing negative affect. Lack of Premeditation refers to
engaging in behavior with little planning or forethought about the consequences. Lack of
Perseverance is defined as the inability to follow-through with completing tasks, especially
when tasks require focused attention.
For the Urgency domain, we chose to examine items that measured (Lack of) Self-Control
(e.g., ‘Ends disagreements calmly’ and ‘Responds appropriately when hit’). For the domains
of (Lack of) Premeditation and (Lack of) Perseverance, we chose to sample from items
measuring Inattention (e.g., ‘Makes careless mistakes’ and ‘Has difficulty paying attention’)
and Hyperactivity (e.g., ‘Difficulty staying seated’ and ‘Acts as if driven by a motor’)
symptoms of Attention Deficit Hyperactivity Disorder, respectively.
Negative affectivity is often conceptualized as three emotional states: Depression, Anxiety,
and Anger (e.g., Watson & Clark, 1992). Items assessing Depressive, Anxiety, and
Oppositional Defiant Disorder symptoms in children were selected that measured the
experience of these affective states (e.g., ‘Feels worthless/guilty,’ ‘Is nervous,’ and ‘Is angry
and resentful’). We also chose items that measured intense emotional responses, (e.g.,
‘When frightened, she feels unreal’), negative emotions that were easily elicited, (e.g., ‘Is
easily annoyed, touchy’) and negative emotions with a long duration (e.g., ‘Depressed most
of the day’).
Lastly, we chose to focus on interpersonal aggression and defiant behaviors as clear markers
of interpersonal problems relevant to BPD features. These interpersonal problems are likely
due to difficulties with self-control (Geiger & Crick, 2001), and thus, overlap with the
Urgency domain of Impulsivity. These behaviors have been associated with BPD features in
adolescent female offenders (Burnette, South, & Reppucci, 2007). Overt forms of physical
aggression toward others clearly results in difficulties forming friendships and having
positive relationships with adults. We chose items that measured physical aggression (e.g.,
‘Starts physical fights’ and ‘Bullies’). We also chose to measure relational aggression,
which is defined as engaging in behaviors that cause harm to interpersonal relationships
(e.g., ‘When mad gets even by excluding others from the group’). Items assessing defiant
interpersonal behaviors were also selected, such as ‘Defies what you tell her to do.’
Measures
Children’s Peer Relationship Scale (CPR; Crick & Grotpeter, 1995; Crick, 1996)
—The CPRS measures child-peer relations through frequencies of behaviors. The PGS
administered adapted versions of the relational aggression subscale to the parent (5 items)
and teacher (7 items). This subscale was comprised of items such as: ‘When some kids are
mad at someone, they get back at the person by not letting the person in their group
anymore’. Both the parent and teacher versions comprise a 5-point answer format, which
ranges from never to almost always.
Child Symptom Inventory-4 (CSI-4; Gadow & Sprafkin, 1994)—The CSI-4
assesses the nature and severity of childhood emotional and behavioral disorder symptoms
using criteria found in the DSM-IV, including Conduct Disorder, Oppositional Defiant
Disorder, Attention Deficit Hyperactivity Disorder, and Major Depressive Disorder. Each
symptom was scored on 4-point scales of never, sometimes, a lot, and all the time. Items of
interest for the current study included 9 inattention items, 10 hyperactivity-impulsivity
items, 8 Oppositional Defiant Disorder items, 2 conduct disorder items assessing physical
aggression items, and 6 depression items for parent and teacher informants. In the first year
of data collection, symptoms were assessed for lifetime occurrence. In all ensuing years,
only past year occurrence was assessed.
Stepp et al. Page 4
J Pers Disord. Author manuscript; available in PMC 2010 August 12.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript
To measure girls’ social competencies, items were adapted from the Social Skills Rating
System (SSRS; Gresham & Elliott, 1990). Items of interest in the current analyses included
those adapted to assess self-control in girls for the parent (9 items) and teacher (8 items; e.g.,
‘Responds appropriately when hit’ and ‘Controls temper when in conflict with parents/
adults’). Both the parent and teacher versions comprise a 3-point scale of often, sometimes,
and never. Thus, higher ratings reflected poorer self-control.
Screen for Child Anxiety Related Emotional Disorders (SCARED; Birmaher et
al., 1997)—The SCARED is a screening instrument for childhood anxiety disorders. This
instrument was administered to the parent and child but not to teachers. For the purposes of
this study, 5 parent items were chosen that reflected trait anxiety (e.g., ‘Is nervous.’) and
reactive/intense anxiety (e.g., ‘Gets really frightened for no reason.’). Items were rated on 3-
point scale of not true or hardly ever, sometimes true, and very true.
Data Analysis
We examined parent and teacher reports separately using the same analytic strategy. Since
we are interested in the nature and stability of these constructs as girls mature, analyses were
also run separately by age (6–12; 7 years). Due to the sampling technique used at
recruitment, which oversampled girls in low income neighborhoods, a weighting variable
was applied to all analyses in order to obtain rates for the general population of girls in
Pittsburgh.
Prior to examining the factor structure, the sample was randomly divided into two groups of
about equal size: calibration (n=1210) and validation (n=1241) samples. An exploratory
factor analysis (EFA) was first performed on the calibration sample. The EFA utilized all of
the items reflecting content related to negative affectivity, impulsivity, and interpersonal
aggression as described previously. The EFA analyses were run using a mean and variance-
adjusted weighted least squares estimator (WLSMV) with an oblique rotation (promax) that
allowed for correlated factors in Mplus 5.0 (Muthén & Muthén, 2007). To determine the
most suitable number of factors, the number of eigenvalues greater than 1.0, visual
inspection of scree plots, and intrepetability of the solution were considered.
Next, the factor loading pattern matrix was examined to determine whether or not individual
items consistently loaded on a single factor across multiple years. The strength of item
loadings were considered poor if they did not reach a value of .35 in at least five of the
seven years examined. Items were considered to poorly discriminate between factors if they
exhibited loadings greater than or equal to .35 on more than one factor across three or more
years.
Items from the EFA that were found to consistently load on a single factor across time were
then submitted to a CFA using the validation sample. The CFA analyses were also
conducted using WLSMV in Mplus 5.0. We assessed absolute fit of the confirmatory
models using global fit indices, including the comparative fit index (CFI), the Tucker-Lewis
index (TLI), and the root mean square error of approximation (RMSEA). For the CFI and
TLI, we used the conventional cutoff values .90 or greater for acceptable fit, and .95 or
greater for good fit. RMSEA values between .05 and .08 represent an acceptable fit, while
values less than .05 indicate a good fit (McDonald & Ho, 2002).
Stepp et al. Page 5
J Pers Disord. Author manuscript; available in PMC 2010 August 12.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript
Results
Exploratory Factor Analysis Using Parent and Teacher Report
Separate EFA analyses were conducted with 53 parent and 50 teacher items using the
calibration sample. The EFA analyses for the parent data yielded between 9 and 12
eigenvalues greater than 1.0 (mode=10 across 7 years). The teacher EFA analyses yielded 5
and 6 eigenvalues greater than 1.0 (mode=6 across 7 years). Inspection of scree plots
suggested that a seven-factor solution for parent reports and a six-factor solution for teacher
reports was most appropriate. Since interpretability was of high importance, six, seven, and
eight rotated solutions were evaluated for the parent data while four, five, and six rotated
solutions were examined for the teacher data. We determined that the six factor solution
consistently yielded conceptually valid structure for the parent report. The resulting parent
factors were labeled: (1) Cognitive Dyscontrol, (2) (Lack of) Self-Control, (3) Hostility, (4)
Depression/Anxiety, (5) Hyperactivity, and (6) Relational Aggression. Next, we examined
the range of parent factor loadings for each item across the seven years of data and the
number of times each item had a significant loading (i.e., ≥ 0.35). Four items did not reach
significance in five of the seven years examined (i.e., ‘Does dangerous things,’ ‘Avoids
trouble situations,’ ‘Helps with tasks without prompts,’ and ‘Fidgets’). Two items
significantly cross-loaded on multiple factors (i.e., ‘Bullies,’ and ‘Starts physical fights’) and
were not retained for further confirmatory analyses.
Similarly, for the teacher data, the five factor solution was deemed most conceptually
compelling while a six factor solution failed to yield a meaningful extra factor. The five
parent factors were labeled: (1) Cognitive Dyscontrol, (2) Hyperactivity, (3) (Lack of) Self-
Control, (4) Relational Aggression, and (5) Depression. Teachers did not rate anxiety items;
thus, anxiety is not represented in the teacher solution. We then examined the range of
teacher factor loadings for each item and the number of times each item had a significant
loading across the seven years. All items reached significance in at least 5 years. Five items
significantly cross-loaded (i.e., ‘Fidgets,’ ‘Difficulty staying seated,’ ‘Appropriately
questions rules,’ ‘Angry and resentful,’ and ‘Bullies’) and were not retained for
confirmatory analyses.
Confirmatory Factor Analysis Using Parent and Teacher Report
Table 1 presents ranges for the estimated CFA standardized factor loadings for both parent
and teacher reports. All items consistently loaded above the .35 threshold for all ages
examined for both the parent and teacher reports with the exception of one item. For parent
data, ‘Receives criticism well’ fell just below the threshold (loading = .34) for one of the
years. Model fit statistics for the CFA models were examined. For parent models, TLI
values suggested good fit from ages 6–12 (>.95). RMSEA values suggested acceptable fit
for all years (<.08). CFI values suggested acceptable fit in one of the seven years. For
teacher models, CFI values suggested acceptable fit in three years and good fit for the
remaining four years. TLI values suggested good fit in all years (>.95). RMSEA values
suggested adequate fit in all but one year.
Generally, there was high consistency of items loading on equivalent factors derived from
the two informants. However, the teacher (Lack of) Self-Control factor includes items
relating to physical aggression and does not include items relating to the experience of
anger. The parent factors do not include physical aggression and the parent Hostility factor
contains items reflecting the propensity to experience anger. These differences may be due
to teacher’s ability to observe physical aggression in a school setting and parent’s superior
ability to notice girls experiencing anger.
Stepp et al. Page 6
J Pers Disord. Author manuscript; available in PMC 2010 August 12.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript
Among the parent factors, Hyperactivity was consistently highly correlated with Cognitive
Dyscontrol and Hostility, with coefficients ranging from .64 to .72 (p<.001) and .61 to .71
(p<.001), respectively. Additionally, Hostility strongly correlated with Cognitive
Dyscontrol, (Lack of) Self-Control, and Depression/Anxiety, with coefficients ranging
from .54 to .72 (p<.001), .60 to .67 (p<.001), and .51 to .67 (p<.001), respectively. All other
factors exhibited more varied magnitudes of associations, most at an intermediate level.
Among the teacher factors, Cognitive Dyscontrol, Hyperactivity, and (Lack of) Self-Control
were highly related, (r ranging from .63 to .82, p<.001). All other teacher factors were
related at an intermediate level.
Intercorrelation between parent and teacher factor scores—The Cognitive
Dyscontrol parent and teacher factors were the most inter-related, with coefficients ranging
from .35–.40 (p<.001) across the seven years. The Hyperactivity parent and teacher factors
were also related (r ranging from .28–.34, p<.001). The teacher (Lack of) Self-Control factor
was inter-related with both the parent (Lack of) Self-Control and Hostility factors (r ranging
from .23–.30 and .20–.29, respectively). The Relational Aggression parent and teacher
factors were more weakly associated at age 6 (r=.14, p<.001) than at ages 7–12 (r ranging
from .24–.31, p<.001). The parent and teacher Depression factors had the weakest
association, (r ranging from .13–.20, p<.001).
Examining stability—We calculated intra-class correlations (ICCs) for each parent- and
teacher-derived factor between the ages of 7 and 12. All correlations were statistically
significant (p<.001), and were moderate to very strong in magnitude. For parent-derived
factor scores, ICCs ranged from .77 to .84 for Cognitive Dyscontrol, from .76 to .82 for
(Lack of) Self-Control, from .80 to .85 for Hostility, from .75 to .83 for Depression/Anxiety,
from .80 to .85 for Hyperactivity, and from .71 to .81 for Interpersonal Aggression. For each
factor, the ICC was lowest between age 5 and age 6 (r = .77 for Cognitive Dyscontrol, .76
for (Lack of) Self-Control, .80 for Hostility, .75 for Depression/Anxiety, .85 for
Hyperactivity, and .71 for Relational Aggression) and appeared to gradually increase in
magnitude with age, with ICCs predicting age 11 to age 12 generally of the strongest
magnitude.
For teacher-derived factor scores, ICCs ranged from .69 to .77 for Cognitive Dyscontrol,
from .65 to .73 for Hyperactivity, from .68 to .77 for (Lack of) Self-Control, from .48 to .63
for Relational Aggression, and from .49 to .61 for Depression. Generally, the magnitude of
the ICCs was strongest at age 8, leveled off, and then decreased at age 11.
Discussion
These findings suggest that the putative underlying features of BPD, conceptualized as
impulsivity, negative affectivity, and interpersonal aggression, can be reliably measured in
6–12 year-old girls. Based on the exploratory and confirmatory factor analyses, these three
underlying features best fit a six factor solution for parent ratings and a five factor solution
for teacher ratings. Three factors were largely similar for both informants: Cognitive
Dyscontrol, Hyperactivity, and Relational Aggression. The parent Depression/Anxiety factor
and teacher Depression factor contained identical items regarding Depression. Anxiety items
did not appear on the teacher factor solution because teachers did not rate these items. The
(Lack of) Self-Control and Hostility factors were distinct factors in the parent solution, but
were combined into one factor in the teacher solution. Items from the teacher (Lack of) Self-
Control factor and the parent (Lack of) Self-Control and Hostility factors had a high degree
of overlap. The convergence between parent and teacher factor scores appeared to be
strongest for the cognitive dyscontrol, hyperactivity, and (lack of) self-control factors. The
convergence between these informants was weakest for the relational aggression and
Stepp et al. Page 7
J Pers Disord. Author manuscript; available in PMC 2010 August 12.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript
depression/anxiety factor scores. These results suggest that parents and teachers are more
similar in rating externalizing behaviors when compared to those behaviors that are more
covert. This lack of agreement might also reflect the slightly different items that parents and
teachers rated.
The year-to-year stability of each of the parent factor scores was high and appeared to
increase as girls matured. Although the stability of teacher factor scores was moderately
high across the age period under examination, it generally appeared to peak at age 8, level
off, and then decrease at age 11. The year-to-year stability of Relational Aggression and
Depression teacher factor scores was considerably lower than the stability of the Cognitive
Dyscontrol, Hyperactivity, and (Lack of) Self-Control teacher factor scores. This finding is
consistent with previous work suggesting that teachers are better informants regarding more
overt behavioral problems than internalizing problems when compared to parents and child
reports (e.g., Epkins, 1995). Overall, these findings support the notion that these features are
stable in 6–12 year-old girls. The stability of the teacher scores is particularly impressive
given that different teachers judged these behaviors each year.
To our knowledge, this is the first large-scale, prospective study regarding the factor
structure and stability of underlying features of BPD in girls. In a short-term longitudinal
study, Crick et al. (2005) found intermediate levels of stability for a measure of BPD in 4th
to 6th grade children across three assessment points over the course of one year. However,
only the stability for the overall measure, and not for each underlying feature, was reported.
Although not a study of BPD features, Vaillancourt and colleagues (2003) found that
relational aggression was distinct from physical aggression in 4–11 year-old children across
three time points.
This study is not without limitations. We did not set out to measure an exhaustive list of all
BPD criteria, such as paranoid ideation when under stress. Thus, we do not posit that other
items are unimportant when measuring BPD features in girls. For example, other
interpersonal processes related to themes of abandonment and sensitivity to interpersonal
rejection are also likely to be important constructs. We do believe that we thoughtfully and
adequately sampled the content domain for the constructs we set out to measure, namely
impulsivity, negative affectivity, and interpersonal aggression and that these are central
constructs to BPD.
As the girls in the PGS continue to be followed, we will re-examine the nature and stability
of these dimensions during adolescence. We will be able to examine the relation between
impulsivity, negative affectivity, and interpersonal aggression and a measure of BPD in
early adolescence. We are interested in examining the nature and specificity of these
pathways. Understanding the precursors of this disorder in childhood will yield important
information about the etiology of BPD and will allow for examining risk and protective
factors. Establishing the nature of these putative features of the disorder and their relation to
BPD will also allow for the development of screening tools and effective interventions for
girls who are at risk.
Acknowledgments
We would like to extend our deepest appreciation to the staff of the Pittsburgh Girls Study. This study would not be
possible without their commitment and hard work. This research was supported by grants from the Office of
Juvenile Justice and Delinquency Prevention (95-JD-FX-0018) and from the National Institute on Mental Health
(MH56630). The first author also received support from T32 MH18269 (Clinical Research Training for
Psychologists, PI: Paul A. Pilkonis).
Stepp et al. Page 8
J Pers Disord. Author manuscript; available in PMC 2010 August 12.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript
References
American Psychiatric Association. Diagnostic and statistical manual of mental disorders. 4.
Washington, DC: Author; 2000.
Bagge C, Nickell A, Stepp S, Durrett C, Jackson K, Trull TJ. Borderline personality disorder features
predict negative outcomes two years later. Journal of Abnormal Psychology 2004;113:279–288.
[PubMed: 15122948]
Bagge CL, Stepp SD, Trull TJ. Borderline personality disorder features and utilization of treatment
over two years. Journal of Personality Disorders 2005;19:420–439. [PubMed: 16178683]
Birmaher B, Khetarpal S, Brent D, Cully M, Balach L, Kaufman J, Neer SM. The Screen for Child
Anxiety Related Emotional Disorders (SCARED): Scale construction and psychometric
characteristics. Journal of the American Academy of Child and Adolescent Psychiatry
1997;36:545–553. [PubMed: 9100430]
Burnette ML, South SC, Reppucci ND. Cluster B personality pathology in incarcerated girls:
Structure, comorbidity, and aggression. Journal of Personality Disorders 2007;21:262–272.
[PubMed: 17536939]
Crick NR, Murray-Close D, Woods K. Borderline personality features in childhood: A short-term
longitudinal study. Development and Psychopathology 2005;17:1051–1070. [PubMed: 16613430]
Crick NR, Grotpeter JK. Relational aggression, gender, and social-psychological adjustment. Child
Development 1995;66:710–722. [PubMed: 7789197]
Crick NR, Ostrov JM, Werner ME. A longitudinal study of relational aggression, physical aggression,
and children’s social psychological adjustment. Journal of Abnormal Child Psychology
2006;34:127–138.
Epkins CC. Teachers’ ratings of inpatient children’s depression, anxiety, and aggression: A
preliminary comparison between inpatient-facility and community-based teachers’ ratings and their
correspondence with children’s self-reports. Journal of Clinical Child Psychology 1995;24:63–70.
Gadow, KD.; Sprafkin, J. Child Symptom Inventories Manual. Stony Brook, NY: Checkmate Plus;
1994.
Geiger, T.; Crick, NR. A developmental psychopathology perspective on vulnerability to personality
disorders. In: Ingram, R.; Price, JM., editors. Vulnerability to psychopathology: Risk across the
life span. New York: Guildford Press; 2001.
Gresham, FM.; Elliott, SN. Social Skills Rating System. Bloomington, MN: Pearson Assessments;
1990.
Gunderson, JG. Borderline personality disorder: A clinical guide. Washington, DC: American
Psychiatric Press; 2001.
Gunderson JG. Disturbed relationships as a phenotype for borderline personality disorder. American
Journal of Psychiatry 2007;164:1637–1640. [PubMed: 17974925]
Hipwell AE, Loeber R, Stouthamer-Loeber M, Keenan K, White HR, Kroneman L. Characteristics of
girls with early onset disruptive and antisocial behaviour. Criminal Behaviour and Mental Health
2002;12:99–118. [PubMed: 12357260]
Lenzenweger MF, Cicchetti D. Toward a developmental psychopathology approach to borderline
personality disorder. Development and Psychopathology 2005;17:893–898. [PubMed: 16613423]
McDonald RP, Ho MH. Principles and practice in reporting structural equation analyses.
Psychological Methods 2002;7:64–82. [PubMed: 11928891]
Muthén, LK.; Muthen, R. Mplus: The Comprehensive Modeling Program for Applied Researchers.
Los Angeles, CA: 2007. (version 5.0)
Skodol AE, Gunderson JG, Pfohl B, Widiger TA, Livesley WJ, Siever L. The borderline diagnosis I:
psychopathology, comorbidity, and personality structure. Biological Psychiatry 2002;51:936–950.
[PubMed: 12062877]
Skodol AE, Bender DS. Why are women diagnosed borderline more than men? Psychiatric Quarterly
2003;74:349–360. [PubMed: 14686459]
Torgersen S, Kringlen E, Cramer V. The prevalence of personality disorders in a community sample.
Arch Gen Psychiatry 2001;58:590–596. [PubMed: 11386989]
Stepp et al. Page 9
J Pers Disord. Author manuscript; available in PMC 2010 August 12.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript
Trull TJ. Structural relations between Borderline Personality Disorder features and putative etiological
correlates. Journal of Abnormal Psychology 2001;110:471–478. [PubMed: 11502090]
Vaillancourt T, Brendgen M, Boivin M, Tremblay RE. A longitudinal confirmatory factor analysis of
indirect and physical aggression: Evidence of two factors over time? Child Development
2003;74:1628–1638. [PubMed: 14669886]
Watson D, Clark LA. Affects separable and inseparable: On the hierarchical arrangement of negative
affects. Journal of Personality and Social Psychology 1992;62:489–505.
Whiteside SP, Lynam DR, Miller JD, Reynolds SK. Validation of the UPPS Impulsive Behavior Scale:
A four-factor model of impulsivity. European Journal of Personality 2005;19:559–574.
Zweig-Frank H, Paris J. Predictors of outcome in a 27-year follow-up of patients with borderline
personality disorder. Comprehensive Psychiatry 2002;43:103–107. [PubMed: 11893987]
Stepp et al. Page 10
J Pers Disord. Author manuscript; available in PMC 2010 August 12.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript
Stepp et al. Page 11
Table 1
Results from the confirmatory factor analysis
Parent Report Teacher Report
Item Stem Min Max Item Stem Min Max
Cognitive Dyscontrol Cognitive Dyscontrol
Careless mistakes 0.40 0.80 Careless mistakes 0.79 0.85
Diff. paying attention 0.58 0.85 Diff. paying attention 0.91 0.93
Not listening 0.74 0.85 Not listening 0.88 0.92
Diff. following directions 0.79 0.87 Diff. following directions 0.92 0.94
Diff. organizing 0.69 0.81 Diff. organizing 0.89 0.93
Avoids mental effort 0.50 0.82 Avoids mental effort 0.88 0.92
Loses things 0.40 0.72 Loses things 0.81 0.86
Easily distracted 0.77 0.84 Easily distracted 0.91 0.98
Forgetful 0.47 0.81 Forgetful 0.87 0.92
Hyperactivity Hyperactivity
Blurts out answers 0.42 0.76 Blurts out answers 0.79 0.89
Diff. waiting 0.74 0.84 Diff. waiting 0.91 0.95
Interrupts people 0.72 0.86 Interrupts people 0.95 0.96
Diff staying seated 0.67 0.85 Does dangerous things 0.82 0.94
Runs/Climbs 0.62 0.79 Runs/Climbs 0.80 0.89
Diff playing quietly 0.61 0.82 Diff playing quietly 0.84 0.90
Driven by motor 0.53 0.65 Driven by motor 0.69 0.83
Talks excessively 0.52 0.68 Talks excessively 0.81 0.90
(Lack of) Self-Control (Lack of) Self-Control
Appropriate response when hit 0.36 0.63 Appropriate response when hit 0.78 0.87
Receives criticism well 0.34 0.75 Receives criticism well 0.78 0.86
Ends disagreements calmly 0.60 0.84 Appropriate response when teased 0.78 0.86
Appropriate response when teased 0.53 0.71 Compromises in conflict situations 0.55 0.66
Appropriate tone when speaking 0.52 0.87 Appropriate peer pressure response 0.72 0.83
Controls temper with peers/friends 0.71 0.78 Loses temper 0.89 0.91
Controls temper with adults/parents 0.67 0.88 Easily annoyed, touchy 0.83 0.88
J Pers Disord. Author manuscript; available in PMC 2010 August 12.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript
Stepp et al. Page 12
Parent Report Teacher Report
Item Stem Min Max Item Stem Min Max
Hostility Controls temper with peers/friends 0.85 0.90
Loses temper 0.67 0.75 Controls temper with adults/parents 0.81 0.87
Easily annoyed, touchy 0.66 0.81 Argues with adults 0.80 0.91
Angry and resentful 0.67 0.87 Physically cruel to people 0.89 0.94
Argues with adults 0.61 0.78 Defies what you tell her to do 0.90 0.94
Defies what you tell her to do 0.61 0.80 Deliberately annoys others 0.91 0.95
Deliberately annoys others 0.69 0.76 Blames others for mistakes 0.91 0.93
Blames others for mistakes 0.58 0.87 Takes anger out on others 0.93 0.97
Takes anger out on others 0.63 0.82 Relational Aggression
Relational Aggression Excludes others when mad 0.89 0.92
Excludes others when mad 0.75 0.86 Excludes others from play 0.93 0.96
Spreads rumors 0.67 0.85 Spreads rumors 0.91 0.95
Gets others to exclude child when mad 0.84 0.96 Tells lies so that others won’t like 0.95 0.98
Threatens to stop being friends 0.75 0.93 Gets others to exclude child when mad 0.95 0.98
Ignores when mad 0.58 0.71 Threatens to stop being friends 0.90 0.94
Depression/Anxiety Ignores when mad 0.79 0.84
She is nervous 0.42 0.64 Depression
She is a worrier 0.40 0.66 Felt she hated herself 0.97 1.02
When frightened, feel unreal 0.42 0.66 Complains of loneliness 0.80 0.87
Get frightened for no reason 0.56 0.69 Depressed most of day 0.84 0.94
Felt she hated herself 0.73 1.01 Feels worthless/guilty 0.82 0.95
Complains of loneliness 0.56 0.78 Little confidence/self-conscious 0.71 0.81
Depressed most of the day 0.61 0.83 Feels things never work out 0.90 0.96
Feels worthless/guilty 0.75 0.91
Little confidence/self-conscious 0.55 0.77
Feels things never work out 0.69 0.91
Note: Minimum and maximum factor loadings from confirmatory factor analyses across ages (6–12 years) are provided.
J Pers Disord. Author manuscript; available in PMC 2010 August 12.
