An Investigation of the Biosocial Model of Borderline

Personality Disorder
Duncan Gill and Wayne Warburton
Macquarie University
Objectives: We sought to test the Biosocial Theory of borderline personality disorder (BPD) that
posits that borderline traits are due to emotional dysregulation, caused by the interaction between child-
hood emotional vulnerability and invaliding parenting. Method: A total of 250 adults (76% female,
median age = 32.06 years) from a nonclinical population completed self-report measures assessing
current levels of borderline traits and emotional dysregulation. They also completed retrospective mea-
sures of childhood emotional vulnerability and parental invalidation. Results: Invalidating parenting
and emotional vulnerability independently predicted emotion dysregulation, but an interaction effect
was not found. Having experienced validating parenting was found to be a protective factor for develop-
ing borderline traits but was not significantly related to emotional dysregulation. Conclusion: Data
in this sample did not support the underlying genesis of BPD proposed by the Biosocial Theory and
a model that more parsimoniously explains the development of BPD is proposed.  C 2014 Wiley

Periodicals, Inc. J. Clin. Psychol. 70:866–873, 2014.

Keywords: borderline personality disorder; biosocial theory; emotional dysregulation; invalidating

parenting; DBT

Dialectical behavior therapy is a common treatment for borderline personality disorder (BPD). It
is based upon the Biosocial Theory that posits that borderline traits are due to extensive emotion
dysregulation, caused in turn by an interaction between an individual’s innate emotional vulner-
ability and the experience of being raised in an emotionally invalidating environment (Linehan,
1993). While recent revisions of the model have hypothesized that childhood impulsivity may
also play a role in the development of borderline traits (Crowell, Beauchaine, & Linehan, 2009),
the main emphasis of the Biosocial Theory rests upon a strong interaction between childhood
emotional vulnerability and invalidating parenting.
The Biosocial model has received some empirical support (Arens, Grabe, Spitzer, & Barnow,
2011), but the posited interaction between childhood emotional vulnerability and invaliding
parenting has failed to predict emotion dysregulation in at least one study (Reeves, 2007) and
borderline traits in another (Sauer & Baer, 2010). Unfortunately, both of these studies had
methodologies that make it hard to draw firm conclusions about the effect of the proposed
interaction – in the Reeves (2007) study, the measure used for childhood emotional vulnerability
(General Emotional Dysregulation Measure; Newhill, Mulvey, & Pilkonis, 2007) does not focus
upon the respondent’s childhood and contains some items that are almost identical to the those
used in the measure of current emotional dysregulation (Difficulties in Emotion Regulation
Scale [DERS]; Gratz & Roemer, 2004), thus confusing construct validity. The Sauer and Baer
(2010) study did not incorporate a measure of emotion dysregulation, and so it was not possible
to determine whether this construct mediated the effect of the childhood measures on current
borderline traits.

We thank Dr. Alan Taylor for his assistance with the data analysis for this study.
Please address correspondence to: Duncan Gill, Department of Psychology, Macquarie University, NSW,
2109, AUSTRALIA. E-mail: [email protected]

JOURNAL OF CLINICAL PSYCHOLOGY, Vol. 70(9), 866–873 (2014) 

C 2014 Wiley Periodicals, Inc.

Published online in Wiley Online Library (wileyonlinelibrary.com/journal/jclp). DOI: 10.1002/jclp.22074

 Investigating the Biosocial Model of BPD 867

The Present Study

The current study sought to test whether the best fitting model of the precursors to borderline
traits would include a significant interaction effect between childhood emotional vulnerability
and parental invalidation, leading to current emotional dysregulation and resulting in borderline
traits.
Anecdotally, this prediction is contrary to our clinical experience, where clients may possess
borderline traits while reporting only a history of emotional vulnerability or of invalidating
parenting. These anecdotal reports from clinicians, considered in conjunction with the findings
of Reeves (2007) and Sauer and Baer (2010), lead to the hypothesis that an interaction effect
would not play an important role in predicting either emotional dysregulation or borderline
traits.
An argument could be mounted that the interaction that the Biosocial Theory posits is not
a statistical interaction but rather a functional one, whereby an emotionally vulnerable child is
likely to elicit invalidating responses from their environment, leading to further emotionality. The
most recent incarnation of the Biosocial Theory suggests that this functional interaction occurs
(Crowell et al., 2009). Consequently, if this interpretation of the Biosocial Theory is correct, then
it would be expected that measures of emotional vulnerability and parental invalidation would
be highly correlated and would share substantial covariance (as a causal cycle is posited to exist
between the two factors). Further, it is possible that both functional and statistical interactions
may be present. Although we sought to test both forms of interactions in the current study,
based on emerging evidence, we hypothesized that neither would be present.

Method
Participants
The study was conducted online, with 150 subjects recruited from the general population. The
sample was further supplemented by 100 first-year psychology students completing the study
for course credit.
The final sample comprised of 250 participants (60 males, 190 females) with a mean age of
32.06 years (standard deviation [SD] = 15.80) and a mean number of years of education of
16.14 (SD = 3.25). Of the participants, 60% identified with an Australian background and the
sample was predominately middle class (24% with a household income of $37,000 or less; 24%
$37,001–80,000; and 52% over $80,000).

Measures
Emotion dysregulation. The issue of what constitutes emotional dysregulation was con-
sidered before the commencement of the study, given that the definition varies across theorists.
For the purposes of this study, emotional dysregulation was conceptualized as having a mal-
adaptive reaction to an emotion rather than having a high level of intensity of the emotion itself,
a distinction made by previous researchers examining emotional dysregulation (e.g., Gratz &
Roemer, 2008; Mennin, Heimberg, Turk, & Fresco, 2005). With regard to the Biosocial model,
while this distinction differs from the definition utilized in the most recent description of the
model (Crowell et al., 2009), a similar construct is identified further down the model, before the
emergence of borderline traits (albeit with the label of “Reactions to emotional situations”).

DERS (Gratz & Roemer, 2004). The DERS was selected with these considerations
in mind, with its items focusing upon the second-order responses to emotions (rather than
the intensity of emotions themselves). DERS was developed with consideration of Linehan’s
theory and comprises six facets of emotion dysregulation. It has demonstrated good internal
consistency (α = .93 – .94), test-retest reliability (ρI = .88, p < .01) and predictive power in
relation to the anticipated behavioral outcomes of emotional dysregulation (Gratz & Roemer,
2004, 2008). Further, it is a significant predictor of BPD (Gratz, Tull, Baruch, Bornovalova, &
Lejuez, 2008).
868 Journal of Clinical Psychology, September 2014

Borderline Traits
Borderline Personality Questionnaire (BPQ; Poreh et al., 2006). The BPQ is a self-
report measure that has been previously used with nonclinical populations as a dimensional
measure of borderline traits (Fonseca-Pedrero et al., 2011). It contains nine subscales, one for
each of the facets of BPD as per the Diagnostic and Statistical Manual of Mental Disorders
Fourth Edition Text Revision (DSM-IV-TR; American Psychiatric Association [APA], 2000).
The subscales demonstrate adequate-excellent internal consistency (α = .78 − .93; Fonseca-
Pedrero et al., 2011), and the measure has performed favourably when compared with other
measures of BPD in the screening of outpatient youth for the disorder (Chanen et al., 2008).

Childhood Invalidation
Recalled Childhood Socialization of Emotion Scale (RCSES; Krause, Mendelson, &
Lynch, 2003). The RCSES is a self-report measure retrospectively rating the levels of validation
and invalidation experienced as a child, by assessing the manner in which the respondent’s
primary caregiver would respond to a series of common scenarios experienced during childhood
(e.g., a child becoming upset at losing a prized possession). It has previously been used in a
number of studies to assess perceived levels of parental invalidation (e.g., Sauer & Baer, 2010;
Thomas, DiLillo, Walsh, & Polusny, 2011), having significantly predicted borderline symptoms
and having shown a modest relationship to parent’s self-report of their parenting (Sauer & Baer,
2010).
In this study, the measure used did not include all the questions, but instead removed the
scenarios identified as being redundant by Sauer and Baer (2010), with such refinements having
previously resulted in highly internally consistent measures of parental validation and invalida-
tion (α = .88 − .95). In this study, the RCSES scales were reversed (cascading from left to right)
to keep the meaning of a high score consistent with the other measures used. Both invalidation
and validation scales were used in this study to determine (a) whether they differ as risk and
protective factors and (b) whether one construct is merely the inverse of the other.

Childhood Emotional Vulnerability

Emotional Vulnerability–Child Scale (EV-Child; Sauer & Baer, 2010). The EV-
Child is a self-report measure that not only retrospectively rates emotional vulnerability, based
upon the Affective Intensity Measure (AIM; Bryant, Yarnold, & Grimm, 1996), but also incor-
porates some items relating to Linehan’s conceptualization of emotional vulnerability involving
a slow return to baseline from emotional arousal. It has high levels of internal consistency
(α = .91) and enjoys a modest convergence with retrospective parental ratings of childhood
vulnerability (Sauer & Baer, 2010).

Results
Calculating Scores
The standard manner of establishing the total score on the BPQ was considered inappropriate for
this study, as it involves summing all the scores despite each subscale having a variable number of
items that may be endorsed at differing rates. This creates the potential for some of the subscales
to have a greater influence on the overall score, whereas the DSM-IV-TR (APA, 2000) does
not give primacy to any of the traits. Consequently, the scoring of the BPQ was completed by
transforming the mean level of endorsement of each of the subscores into a z-score before the
creation of an overall mean score, thus not favoring one facet of BPD over the others. A similar
issue existed for the DERS and so the same solution was applied (i.e., the mean of each subscale
was transformed into a z-score before an average score was determined).
Both the RCSES-Invalidation and the RCSES-Validation were calculated with regard to the
refinements suggested in previous research (Sauer & Baer, 2010) to deliver a two-factor solution.
 Investigating the Biosocial Model of BPD 869

Table 1
Descriptive Statistics and Internal Reliability (N = 250)

Mean S.D. Crombach’s alpha

EV-Child 3.11 .83 .94

RCSES-Validation 3.99 1.39 .95
RCSES-Invalidation 2.79 1.31 .93
(prior to transformation)
DERS (mean of z-scores) 0 .71 .81
DERS NA 2.28 .96 .91
DERS GDB 2.96 .98 .88
DERS ICD 1.95 .82 .88
DERS LEA 2.40 .76 .81
DERS LS 2.15 .90 .92
DERS LEC 2.18 .74 .84
BPQ (mean of z-scores) 0 .72 .89
BPQ Impulsivity .18 .20 .68
BPQ Affective Instability .35 .31 .87
BPQ Abandonment .18 .20 .74
BPQ Relationships .32 .31 .82
BPQ Self Image .30 .29 .82
BPQ Suicide/Self- .16 .23 .78
Mutilation
BPQ Emptiness .26 .28 .84
BPQ Intense Anger .26 .28 .86
BPQ Quasi-Psychotic States .19 .21 .62

Note. EV-Child = Emotional Vulnerability–Child Scale; RCSES = Recalled Childhood Socialization of

Emotion Scale; DERS = Difficulties in Emotion Regulation Scale; BPQ = Borderline Personality Ques-
tionnaire; NA = Nonacceptance of Emotional Responses; GDB = Difficulties in Goal-Directed Behaviour;
ICD = Impulse Control Difficulties; LEA = Lack of Emotional of Emotional Awareness; LS = Limited
Access to Emotion Regulation Strategies; LEC = Lack of Emotional Clarity.

Data Transformation and Internal Consistency

Because data normality was found to be an issue for RCSES-Invalidation scores, a logarith-
mic transformation was used to resolve this issue (the transformed variable was used for all
subsequent analysis).
Internal consistency for each of the measures was found to be adequate to excellent for all
the scales, other than the Quasi-Psychotic States subscale of the BPQ, which was marginal (see
Table 1).

Model Comparisons
The EV-Child, RCSES-Invalidation, and RCSES-Validation scores were converted to z-scores to
allow for the calculation of interaction effects (obtained by multiplying the standardized scores
together and generating the variables EV*Invalid and EV*Valid). The demographics variables
were recoded into dichotomous or ordinal variables for analysis. The model used composite
rather than latent variables because of the large number of parameters.
A model was designed using AMOS 7.0, whereby all of the demographic variables, EV-
Child, RCSES-Invalidation, RCSES-Validation, EV*Invalid, and EV*Valid were entered as
independent variables with direct effects upon emotional dysregulation (DERS; the mediating
variable). All possible covariances between the independent variables were estimated. Borderline
traits (BPQ total score) was then entered as the dependent variable, with all variables (including
DERS) identified as having a direct effect upon this measure.
Because the model had many paths and covariances, a large number of which were not
significantly different from zero, the model was simplified by removing nonsignificant paths
870 Journal of Clinical Psychology, September 2014

Figure 1. Model with all nonsignificant direct effects removed (p < .05). Note. The integers represent
standardized estimates for each direct path or covariance, with each of these being significant (p < .05). For
simplicity of presentation, not all the demographics measures and significant covariances are displayed.

and covariances. The initial approach was based on chi-squared difference tests. However,
because of the large sample size, the removal of nonsignificant paths and covariances often
led to significant changes in the value of chi-squared, even though, by other fit indices less
affected by sample size, the goodness of fit of the model had not been reduced (and in some
cases even improved). Consequently, it was determined to prune the model on the basis of the
significance of the covariances and direct effects. The model was pruned of all weak covariances
and rerun, and further covariances were pruned. This process was repeated until only significant
(p < .05) covariances remained. The same pruning technique was then utilized for the direct
effects.
The resulting model (see Figure 1) was a very good fit to the data (goodness of fit index =
.96, comparative fit index = .98, root mean square residual = .04, standarised root mean square
residual = .07, normed fit index = .93). The model was also parsimonious (root mean square
error of approximation = .04, probability of close fit = .87) and had an acceptable chi-squared
statistic, χ2 (58, N = 250) = 76, p = .056 (χ2 /df = 1.31).
The final model was compared to the model where all the direct paths between the variables
were retained, yielding a nonsignificant difference, χ2 (17, N = 250) = 15.3, p = .57.

Correlational Data
Significant but weak correlations existed between childhood emotional vulnerability and per-
ceived parenting (see Table 2).
 Investigating the Biosocial Model of BPD 871

Table 2
Correlation Between Measures of Emotional Vulnerability and Parenting Environment

EV-Child RCSES-Validation RCSES-Invalidation

EV-Child 1
RCSES-Validation −.15* 1
RCSES-Invalidation .26** −.53** 1

Note. EV-Child = Emotional Vulnerability–Child Scale; RCSES = Recalled Childhood Socialization of

Emotion Scale.
*p < .05. **p < .01.

Discussion
In line with the Biosocial Theory, and as expected, emotional dysregulation significantly pre-
dicted borderline personality traits. However, contrary to the Biosocial Theory (Crowell et al.,
2009), the hypothesis that emotional vulnerability and invalidating parenting would individually
predict emotional dysregulation but have a limited interaction with each other was supported.
In addition, both the correlation and covariance between invalidating parenting and emotional
vulnerability, although significant, was not strong, a finding at odds with the contention that the
interaction between these two constructs is of a functional, rather than statistical, nature. These
findings suggested that both emotional vulnerability and invalidating parenting independently
exert their effects upon borderline traits. This has clinical implications, particularly in the case
of patients with emerging borderline traits, in which both aspects should be assessed and may
be targets for intervention.
It is noteworthy that there was a strong relationship between borderline traits and emotional
dysregulation, which was operationalized in this study as an individual’s reaction to the experi-
ence of emotion, rather than the strength of the emotion itself. Despite this strong relationship,
the items contained in the emotion dysregulation measure do not appear to encompass the same
constructs as the measure of borderline traits that are more behaviorally based. Together these
findings support the notion that an individual’s reaction to emotions, rather than the intensity
of the emotions they feel, is more fundamental to the development of borderline traits.
An unanticipated finding was that perceived parental validation correlated significantly and
negatively with borderline traits, but not with emotional dysregulation, suggesting that it might
exert a protective effect against the development of borderline traits. This is a significant finding,
as it suggests that emotion validation is not merely the opposite of emotion invalidation, given
the differing roles they occupy in the model. Rather, based on the items in the measure, it involves
being taught how to respond to emotional challenges in a constructive and helpful manner. This
finding is also of note as it highlights that not all borderline aetiology is mediated via emotion
dysregulation.

Further Research
It should be noted that this study used retrospective self-report measures, cross-sectional rather
than longitudinal data to explore the dynamics between the various factors, and a nonclinical
sample to examine a clinical construct. Future studies would benefit from replicating this study
in a clinical sample. Longitudinal studies are also indicated; however, to follow a sufficiently
large sample from early childhood to the development of BPD would be costly.
It would also be of interest to establish whether the narrow focus on invalidating parenting
may be misguided. It is plausible that invalidating parenting per se may not be of particular
importance, but rather the measure of invalidating parenting may be tapping into a broader
factor of generally harmful parenting (which may include a wide range of behaviors). It is
possible that other types of poor parenting beyond emotionally invalidating parenting may also
influence the development of borderline traits. Consequently, including a measure assessing
various types of poor parenting would assist to clarify this situation.
872 Journal of Clinical Psychology, September 2014

Similarly, it could be argued that the construct of an emotionally vulnerable child may be
indistinguishable from the broader construct of neuroticism, a construct that correlates with not
only borderline traits (Distel et al., 2009) but also a range of other physical and mental health
conditions (Claridge & Davis, 2001). Consequently, if the measure of emotional vulnerability
substantially measures neuroticism, its inclusion in a model for BPD, while accounting for
variation in the associated measures, may not possess the capacity to explain why an individual
develops BPD (as opposed to another disorder).
Finally, it would also be of interest to ascertain whether poor parenting exerts an effect in a
nonlinear fashion. The classification of BPD as a form of complex posttraumatic stress disorder
has previously been postulated (e.g., Driessen et al., 2002). This would suggest that negative
childhood events may have to reach a threshold of stress to induce a trauma response, with poor
parenting (up until a point) not exerting a significant effect upon longer term behavior.

References
American Psychiatric Association. (2000). Diagnostic and statistics manual of mental disorders (4th ed.,
text rev.). Washington, DC: Author.
Arens, E. A., Grabe, H. J., Spitzer, C., & Barnow, S. (2011). Testing the biosocial model of borderline
personality disorder: Results of a prospective 5-year longitudinal study. Personality and Mental Health,
5(1), 29–42. doi:10.1002/pmh.143
Bryant, F. B., Yarnold, P. R., & Grimm, L. M. (1996). Toward a measurement model of the af-
fect intensity measure: A three factor structure. Journal of Research in Personality, 30(2), 223–247.
doi:10.1006/jrpe.1996.0015
Chanen, A. M., Jovev, M., Djaja, D., McDougall, E., Yuen, H. P., Rawlings, D., & Jackson, H. J. (2008).
Screening for Borderline Personality Disorder in outpatient youth. Journal of Personality Disorders,
22(4), 353–364. doi:10.1521/pedi.2008.22.4.353
Claridge, G., & Davis, C. (2001). What’s the use of neuroticism? Personality and Individual Differences,
31(3), 383–400. doi:10.1016/S0191-8869(00)00144-6
Crowell, S. E., Beauchaine, T. P., & Linehan, M. M. (2009). A biosocial developmental model of border-
line personality: Elaborating and extending Linehan’s theory. Psychological Bulletin, 135(3), 495–510.
doi:10.1037/a0015616
Distel, M. A., Trull, T. J., Willemsen, G., Vink, J. M., Derom, C. A., Lynskey, M., . . . Boomsma, D. I.
(2009). The five-factor model of personality and borderline personality disorder: A genetic analysis of
comorbidity. Biological Psychiatry, 66(12), 1131–1138. doi:10.1016/j.biopsych.2009.07.017
Driessen, M., Beblo, T., Reddemann, L., Rau, H., Lange, W., Silva, A., . . . Ratzka, S. (2002). Is the borderline
personality disorder a complex post-traumatic stress disorder? The state of research. Nervenarzt, 73(9),
820–829. doi:10.1007/s00115-002-1296-1
Fonseca-Pedrero, E., Paino, M., Lemos-Giráldez, S., Sierra-Baigrie, S., González, M. P., Bobes, J., & Muniz,
J. (2011). Borderline personality traits in nonclinical young adults. Journal of Personality Disorders,
25(4), 542–556. doi:10.1521/pedi.2011.25.4.542
Gratz, K. L., & Roemer, L. (2004). Multidimensional assessment of emotion regulation and dysregulation:
Development, factor structure, and initial validation of the Difficulties in Emotion Regulation Scale.
Journal of Psychopathology and Behavioral Assessment, 26(1), 41–54. doi:0882-2689/04/0300-0041/0
Gratz, K. L., & Roemer, L. (2008). The relationship between emotion dysregulation and deliberate self-harm
among female undergraduate students at an urban commuter university. Cognitive Behaviour Therapy,
37(1), 14–25. doi:10.1080/16506070701819524
Gratz, K. L., Tull, M. T., Baruch, D. E., Bornovalova, M. A., & Lejuez, C. W. (2008). Factors associ-
ated with co-occurring borderline personality disorder among inner-city substance users: The roles of
childhood maltreatment, negative affect intensity/reactivity, and emotion dysregulation. Comprehensive
Psychiatry, 49(6), 603–615. doi:10.1016/j.comppsych.2008.04.005
Krause, E., Mendelson, T., & Lynch, T. (2003). Childhood emotional invalidation and adult psycho-
logical distress: The mediating role of emotional inhibition. Child Abuse & Neglect, 27(2), 199–213.
doi:10.1016/S0145-2134(02)00536-7
Linehan, M. M. (1993). Cognitive-behavioral treatment of borderline personality disorder. New York:
Guilford.
 Investigating the Biosocial Model of BPD 873

Mennin, D. S., Heimberg, R. G., Turk, C. L., & Fresco, D. M. (2005). Preliminary evidence for an emotion
dysregulation model of generalized anxiety disorder. Behaviour Research and Therapy, 43(10), 1281–
1310. doi:10.1016/j.brat.2004.08.008
Newhill, C. E., Mulvey, E. P., & Pilkonis, P. A. (2007). The general emotional dysregulation measure (4th
ed., Vol. 2). New York: Oxford University Press.
Poreh, A. M., Rawlings, D., Claridge, G., Freeman, J. L., Faulkner, C., & Shelton, C. (2006). The BPQ:
A scale for the assessment of borderline personality based on DSM-IV criteria. Journal of Personality
Disorders, 20(3), 247–260. doi:10.1521/pedi.2006.20.3.247
Reeves, M. D. (2007). The effect of invalidation on emotion regulation: An empirical investigation of
Linehan’s biosocial theory (Unpublished doctoral dissertation). Florida State University. Retrieved
from http://diginole.lib.fsu.edu/cgi/viewcontent.cgi?article=4411&context=etd
Sauer, S. E., & Baer, R. A. (2010). Validation of measures of biosocial precursors to borderline personality
disorder: Childhood emotional vulnerability and environmental invalidation. Assessment, 17(4), 454–
466. doi:10.1177/1073191110373226
Thomas, R., DiLillo, D., Walsh, K., & Polusny, M. A. (2011). Pathways from child sexual abuse to adult
depression: The role of parental socialization of emotions and alexithymia. Psychology of Violence, 1(2),
121–135. doi:10.1037/a0022469
Copyright of Journal of Clinical Psychology is the property of John Wiley & Sons, Inc. and
its content may not be copied or emailed to multiple sites or posted to a listserv without the
copyright holder's express written permission. However, users may print, download, or email
articles for individual use.