Brain Injury, March 2006; 20(3): 307–319
The overt behaviour scale (OBS): A tool for measuring challenging
behaviours following ABI in community settings
, JENNY TODD
, GRAHAME SIMPSON
, PETER KREMER
ABI Behaviour Consultancy, Epworth Hospital, Victoria, Australia,
Brain Injury Rehabilitation Unit, Liverpool
Hospital, New South Wales, Australia,
School of Exercise & Nutrition Science, Deakin University, Victoria,
Brain Injury Rehabilitation Unit, Liverpool Hospital, New South Wales, Australia
(Received 31 March 2005; accepted 20 September 2005)
Objectives: The Overt Behaviour Scale (OBS) was designed as a comprehensive measure of common challenging behaviours
observed after acquired brain injury (ABI) in community settings. The OBS comprises 34 items in nine categories that
measure aggression, inappropriate sexual behaviour, perseveration, wandering, inappropriate social behaviour and lack of
initiation. The aim of the current study was to determine the reliability, validity and responsiveness of the OBS.
Method: Two adult community-based samples of people with ABI were recruited. Sample 1 (n¼30) were concurrently
evaluated on the OBS by two raters and again 1 week later to test stability. Other validating scales were also administered.
Sample 2 (n¼28) were clients of the ABI Behaviour Consultancy who were treated for challenging behaviours and were
administered the OBS before treatment commenced and then again 4 months later.
Results: Inter-rater reliability and stability coefficients for the OBS total score was strong (0.97 and 0.77, respectively). Initial
evidence of convergent and divergent validity was shown by the differential pattern of correlations with other measures.
Moderate-to-strong coefficients (range 0.37–0.66) were observed between the OBS and other measures that had
behavioural content (i.e. Mayo-Portland Adaptability Inventory, Current Behaviour Scale, Neurobehavioural Rating
Scale–Revised). Divergent validity was shown by the lack of correlation between the OBS and the sub-scales of these tools
that do not measure challenging behaviour. Finally, responsiveness was demonstrated with a significant decrease in OBS
scores in the expected direction over the 4-month period. This improvement was confirmed by corroborating evidence from
Conclusion: The OBS shows promise as a reliable, valid and responsive measure that can be used for the systematic
assessment of challenging behaviours in community settings.
Keywords: Acquired brain injury, challenging behaviour, community settings, outcome measures
People with acquired brain injury (ABI) can display
various types of behavioural disturbance including
excesses of behaviour such as aggression, inappro-
priate social behaviours and inappropriate sexual
behaviours and deficiencies of behaviour such as
adynamia [1–3]. These behaviours can endure and
worsen over time, particularly in unstructured set-
tings where there is often little control over the envi-
ronmental contingencies that govern behaviour .
Negative consequences that flow from such behav-
iours can include exclusion from needed services,
increased staffing costs for agencies managing such
clients, criminal charges being laid against the
person with ABI, unwanted admissions to inappro-
priate institutional care, and significant distress
caused to family members and staff as well as the
person with ABI .
The ABI behaviour management literature can
give a reader the impression that clients typically live
Correspondence: Dr Glenn Kelly, ABI Behaviour Consultancy, PO Box 1228, North Fitzroy, 3068, Australia. Tel: þ61 3 9383 7176. Fax: þ61 3 9383 7278.
ISSN 0269–9052 print/ISSN 1362–301X online ß2006 Taylor & Francis
in specialized accommodation and engage in struc-
tured activities supported by well-trained staff [5–8].
However, most people with ABI spend most of their
time in community settings. Community settings can
be defined as the living situation of the person with
ABI (e.g. own home, parents’ home, supported
residential services, hostels, private rental or public
housing) and the associated social environment
(e.g. typically public, diverse and largely unstruc-
tured environments such as supermarkets, shopping
centres, gymnasiums, hotels and train stations).
Given the expressed preference of people with ABI
to live in such settings , the emphasis within
the International Classification of Functioning on
people with disabilities participating to the fullest
extent possible in community life  and the
growth of community-based rehabilitation , the
challenge for brain injury services is to develop ways
of managing behaviours in these environments.
Within this context, there is a need for comprehen-
sive, reliable and valid measurement instruments.
The absence of such tools has contributed to the lack
of clarity and consensus in the definition and study
of challenging behaviours [12–14].
Current scales that measure challenging behaviour
have not been designed to capture data on the
breadth of behaviours that occur in community
settings; this includes scales that focus solely on
behaviour, as well as global inventories of the
sequelae of ABI. First, generic measures of chal-
lenging behaviour have often been developed to
address issues specific to one type of setting (e.g.
Nursing Home Behaviour Problem Scale ) or
for clients from specific diagnostic groups, such as
intellectual disability (e.g. Behaviour Disorder
Scale , ICAP ) or psychiatric disability
(e.g. Overt Aggression Scale ). Secondly, the
behavioural scales that have been developed for
ABI populations or which appear more suited to
them often measure one or only a few behavioural
domains, such as aggression , agitation (Agitated
Behavior Scale ), apathy (Apathy Evaluation
Scale ) or emotional control and motivation (e.g.
Current Behaviour Scale ). Such scales are of
limited usefulness as clinical experience has found
that behavioural disturbances after ABI typically
occur across a number of domains . Thirdly,
some measures of challenging behaviours have
been purpose-designed for a particular study with
no reported reliability or validity data [4, 22].
Global measures of neurobehavioural impair-
ments after ABI typically include individual items
or sub-scales that record overt behaviours as a part
of the broader range of motor-sensory, cognitive
and affective sequelae [23–26]; however, these
measures lack the specificity needed for a closer
analysis of challenging behaviours. In these cases,
items recording behaviours are either aggregated into
a small number of ‘behavioural’ sub-scales [24, 26]
or else single items are used to summarize whole
domains of behaviour, for example an item recording
the presence of inappropriate sexual behaviour
may be required to cover such diverse behaviours
as sexual innuendo, frotteurism, child molestation,
exhibitionism and rape .
Finally, a number of the scales cited only measure
the frequency with which behaviours occur .
Although frequency is an important indicator that
has been used for many decades [27, 28], there are
other behavioural indices, such as the severity of
a behaviour or its impact on others that can also be
measured. Although these indices are less frequently
documented, they can provide critical clinical infor-
mation for effective assessment and management.
One scale that addresses some of these outlined
difficulties is the Overt Aggression Scale (OAS ),
a rating scale with reported reliability and validity
designed to measure the frequency of four categories
of overt aggressive behaviours in adults and children
in psychiatric settings. Each category has four levels
of severity, defined by behavioural descriptors.
In addition to severity, the scale can measure behav-
iour frequency by being completed each time an
aggressive incident occurs or by using a summary
behaviour frequency measure (such as rating from
‘never’ through to ‘always’ ). The OAS provides
a degree of objectivity to behaviour measurement
and a consistency in clinical descriptions across time
and settings. Furthermore, a version of the scale
modified for people with ABI has been developed
[14, 30]. Importantly, a summary frequency mea-
sure on the OAS meets the needs for an alternative,
practical way of systematically recording levels of
behavioural disturbance, because it is often not
logistically possible to comprehensively chart single
behaviours in community settings [3, 8]. However,
one limitation of the OAS is that it only measures
aggression and, as previously noted, clients often
display a range of challenging behaviours.
The ABI Behaviour Consultancy (the
‘Consultancy’) is a state-wide, community-based,
government-funded service, operating in Victoria,
Australia . The Consultancy receives 200 refer-
rals each year to assist with the management of
challenging behaviours, with most referrals for
people with ABI living in the general community.
Based on this experience, the Consultancy devel-
oped the Overt Behaviour Scale (OBS ) as a
clinical rating scale, extending the OAS by devising
additional sub-scales that incorporate a wider range
of challenging behaviours that are commonly
encountered after ABI. The additional sub-scales
were modelled on the structure of the OAS and
record data on the severity and frequency of the
308 G. Kelly et al.
challenging behaviours. However, an additional
index has been introduced with the OBS, making
provision to record the impact that challenging
behaviours have on the people (such as staff or
family) exposed to them. Finally, it is often difficult
to measure behaviour in community settings by
direct observation because people rapidly move
between many different environments and so the
OBS was designed principally to measure behaviour
based on informant’s reports, although it can also
be completed through direct observation. The aim
of this study is to report on the reliability, validity
and responsiveness of the OBS.
Two samples were used: Sample 1 was recruited for
the reliability and validity trials; Sample 2 was used
to test the sensitivity of the OBS.
Sample 1. Participants were drawn from the Brain
Injury Rehabilitation Unit (BIRU) at Liverpool
Hospital, Sydney, Australia, which provides special-
ist inpatient and community-based rehabilitation to
adults with traumatic brain injury (TBI) who are
resident in South Western and Southern Sydney
. Thirty clients who sustained a TBI between the
ages of 16–65 years and exhibited challenging
behaviours were identified from the BIRUs commu-
nity outreach programme. Mean age was 31.5 years
(SD ¼13.2) and mean time post-injury 8.6 years
(SD ¼8.4). The group sustained extremely severe
injuries (mean duration of PTA 77.2 days,
SD ¼59.4). Current living circumstances included
own home (11/30, 36.7%), parents’ home (9/30,
30.0%) or other environments (e.g. rental accom-
modation, living with friends; 10/30, 33.3%).
Informants were eight allied health staff (i.e. social
workers, psychologists, case managers, occupational
therapists) of the BIRU community team who were
actively involved in working with the clients and
Sample 2. The sample comprised 28 clients of the
Consultancy who were part of a larger consecutive
series of 112 clients referred over an 8-month period
for the assessment and treatment of challenging
behaviours. The group of 28 comprised the clients
for whom it had been possible to administer an
OBS on two occasions, with a median interval of 4.0
months, thereby providing an opportunity to deter-
mine whether the OBS detected change following
behaviour management intervention. Twenty-four
participants were male (85.7%). Mean age at time
of injury was 39.0 years (SD ¼13.0), and mean
time post-injury 8.1 years (SD ¼12.0). There
were diverse causes of the acquired brain injuries
(TBI 11/28, 39.3%; CerebroVascular Accident 5/28,
17.9%; Hypoxia 3/28, 10.7%; Other causes 9/28,
32.1%). Although data on initial severity of injury
was not available, the disabling consequences of
the brain injuries were clearly evident, with only
14.3% of the sample in paid employment and a
separate 14.3% of the sample engaged in an educa-
tional programme or volunteer work. Current living
circumstances included own home (11/28, 39.3%),
parents’ home (6/28, 21.4%) or other environments
(supported residential services, nursing homes or
hospitals; 11/28, 39.3%).
Measures: Overt Behaviour Scale
The Consultancy is the only agency within Victoria
to specialize exclusively in the management of
challenging behaviours after ABI. Referral criteria
are that clients have an ABI and challenging
behaviour, are aged between 18–65 years at
the time of referral and are not compensible
(e.g. through Victoria’s Transport Accident
Commission) and thereby eligible for private ser-
vices. Knowledge of the service is widespread within
the state, with referrals received from every region
(metropolitan and rural) and from a broad range of
rehabilitation (inpatient and outpatient), community
brain injury (e.g. case management services) and
generic service providers (e.g. nursing homes).
Hence, the Consultancy has a broad referral base
and is very experienced in the range of challenging
behaviours that occur in community settings.
The following steps were taken to develop the
.Development of behaviour categories. A review was
conducted by Consultancy staff of hundreds of
examples of overt challenging behaviours drawn
from 543 referrals to the service over a 5-year
period. This review found that, in addition to
the four original OAS verbal and physical aggres-
sion sub-scales (i.e. Verbal Aggression, VA;
Physical Aggression against objects, PA objects;
Physical Aggression against self, PA self; Physical
Aggression against other people, PA people),
a number of other behaviours could be sorted
into categories that accounted for 10% or more
of referred behaviours. Hence, a further five
categories were selected for development as sub-
scales: inappropriate sexual behaviour (SEX),
perseveration/repetition (PER/REP), wandering/
absconding (WAN/ABS), inappropriate social
behaviour (SOC), and lack of initiation (INI),
bringing the total number of categories to nine.
Approximately 90% of all behaviours referred to
A tool for measuring challenging behaviours 309
the Consultancy could be classified into one of
these sub-scales, with the remaining behaviours
(e.g. voyeurism) proving either too idiosyncratic
to be classified in an existing sub-scale or too
heterogenous as a collection to combine into an
.Aggression categories. The anglicized behavioural
descriptors published in the Overt Aggression
Scale-Modified for Neurorehabilitation  were
adopted as the four OBS aggression scales.
.New categories. Turning to the five new categories,
the next step involved reviewing the pooled
behaviours to develop a series of severity levels.
Consultancy staff used their clinical judgement
to sort the behaviours into levels of lesser
to greater severity (see Table I). However,
INI was qualitatively different to the other
four categories because it involved an absence of
overt behaviours. Experience in trialling a number
of pilot versions found difficulties in creating
a multi-level sub-scale. As a result, INI was
treated as a dichotomous sub-scale (present vs.
absent) in which variations were measured by
the frequency data that recorded differing levels
of prompting required throughout a day (range:
1¼less than once/day, 5 ¼all tasks, everyday).
.Inappropriate social behaviour. Having completed
this preliminary stage of scale development,
eight experienced ABI-specialist staff working at
a community-based ABI case management service
then reviewed the categories and the severity level
descriptors. The consultation found that most
sub-scales were well-received, but also highlighted
difficulties with the severity levels in the SOC
category. SOC encompassed a diverse set of
behaviours reflecting many different dimensions
Table I. The overt behaviour scale.
Category Severity levels CWS
1. Verbal aggression (VA)
1. Makes loud noises, shouts angrily ... 1
2. Makes mild personal insults ... 2
3. Swearing, use of foul language ... 3
4. Makes clear threats of violence ... 4
2. Physical aggression against objects (PA objects)
1. Slams doors, scatters clothing ... 1
2. Throws objects down ... 2
3. Breaks objects ... 3
4. Sets fire, throws objects dangerously ... 4
3. Physical aggression against self (PA self)
1. Picks or scratches skin ... 1
2. Bangs head ... 2
3. Inflicts small cuts or bruises ... 3
4. Mutilates self, causes deep cuts ... 4
4. Physical aggression against other people (PA people)
1. Makes threatening gestures ... 1
2. Strikes, kicks, pushes ... 2
3. Attacks others ... 3
4. Causes severe physical injury ... 4
5. Inappropriate sexual behaviour (SEX) 1. a. Sexual talk 1
b. Touching (non-genital)
2. a. Exhibitionism 2
3. Touching (genital) 3
4. Coercive sexual behaviour, rape 4
6. Perseveration/Repetition (PER/REP) Prolonged repetition of behaviour resulting
1. ... in no physical harm (e.g. questions) 1
2. ... in minor physical harm 2
3. ... in serious physical harm 3
7. Wandering/Abscond (WAN/ABS) 1. Go into prohibited areas (e.g. staff office) 1
2. Leaving the familiar, ‘safe’ environment ... 2
3. Escapes secure premises 3
8. Inappropriate Social Behaviour (SOC) 1. Socially awkward 1
2. Nuisance/annoyance 2
3. Non-compliant/oppositional 3
4. a. Petty crime/unlawful behaviour 4
b. Presents a danger or risk to self/others
9. Lack of initiation
(INI) 1. Present vs. absent 1–5
Cluster score ___/9 Total levels score ___/34 __/77
CWS ¼Clinical Weighting Score.
Scales taken from OAS-MNR;
Lack of initiation (INI) has only one severity level (Proxy severity descriptor: 1 less than monthly – 5
multiple times daily).
310 G. Kelly et al.
of inappropriate social behaviour, including poor
conversational turn-taking, poor inter-personal
distance, personal hygiene problems, urinating in
public, hoarding and non-compliance. The diver-
sity and quantity of behaviours created difficulties
for development of appropriate severity levels. To
address this issue, a list of 63 recorded behaviours
was compiled; these behaviours had been sorted
to the SOC category; did not overlap with other
categories; and were not primarily manifestations
of psychiatric problems (e.g. hallucinations),
discrete cognitive functions (e.g. memory prob-
lem) or mood-related conditions (e.g. anxiety).
A group of 282 staff volunteers from across the
state of Victoria reviewed the 63 behaviours,
rating each behaviour on a 4-point scale ranging
from 1 (least severe) to 4 (most severe), based on
their clinical experience. All respondents special-
ized in working with people with ABI in rehabil-
itation or community settings and had a median
of 6 years specialist experience (range 1–37 years).
Severity was defined in terms of the extent to
which the behaviour might present a problem or
concern, cause distress to staff and/or family,
disrupt service delivery or interfere with social and
A Principal Components Analysis was then con-
ducted on the data, which indicated a two factor
structure. However, the result was uninterpretable
due to the considerable overlap of items between
factors. Consequently, a descriptive data reduction
strategy was used. Behaviours were allocated to
particular severity levels when (a) there was a clear
modal response for that level (i.e. >40%), (b) the
most common rating was (preferably) 20% more
frequent than the second most common response
level, (c) no more than two response options had
more than 20% of responses (i.e. there was not
a rectangular response distribution across multiple
severity levels) and (d) Consultancy staff agreed
that the behavioural example well represented
only one level of severity. This strategy resulted in
the identification of the five severity levels displayed
in Table I. At the end of this development process,
the OBS had nine sub-scales with 34 levels of
Agreement tasks. To conduct a final check on the
scale construction, 20 allied health staff volunteers
with a median of 6 years experience in the ABI field
and who had not previously seen the OBS were
recruited. Each of the severity level behavioural
descriptors for eight of the nine behaviour categories
were printed onto separate cards (INI having only
one level was not included). Staff were first asked
to sort the 33 cards into the eight categories, with
the only guidelines being the number of severity
level descriptors needed for each category (Category
Agreement). Following this task, any ‘incorrectly’
assigned cards were placed into the ‘correct’ category
before the second task was undertaken. Staff were
then asked to arrange the descriptors within each
category from the least severe to the most severe
(Severity Level Agreement).
In terms of the Category Agreement task, the
overall number of descriptors ‘correctly’ allocated
to the eight proposed categories by the BIRU staff
were totalled and then expressed as a percentage.
The first row of Table II shows there was high
overall agreement between the proposed OBS struc-
ture and the categories to which the allied health
staff assigned these same descriptors (¼0.94,
p< 0.001). Secondly, the agreement between the
clinician raters with respect to assignment of the
descriptors to categories was also analysed.
Responses of the 20 participants were divided into
10 pairs of data to test inter-rater agreement for
each category, with strong overall agreement found
(¼0.88, p< 0.001).
Turning to the results of the Severity Level
Agreement task, the second row of Table II shows
the level of agreement between the OBS proposed
hierarchy of behavioural severity (lowest to highest
levels) and the raters’ ordering of the behaviour
descriptors within each category. For most cate-
gories agreement was significant and high. However,
to reach adequate levels of agreement (greater than
0.60 ) for SOC it was necessary to combine
levels 4 (petty crime) and 5 (risky behaviours).
Table II. Agreement (%) between authors’ proposed categories and raters’ assigned categories (row 1), and agreement () between
proposed hierarchy of behavioural severity and raters’ severity structure (row 2) (n¼20 raters).
Task Measure VA PA object PA self PA people SEX PER/REP WAN/ABS SOC
Category agreement % agree 94 90 95 95 99 97 100 88
Severity agreement (calibration) Kappa () 0.67* 0.87* 0.63* 0.90* 0.65* 0.88* 0.75* 0.75*
VA ¼verbal aggression, PA ¼physical aggression, SEX ¼inappropriate sexual behaviour, WAN/ABS ¼wandering/absconding,
PER/REP ¼perseveration/repetitive behaviour, SOC ¼inappropriate social behaviour.
A tool for measuring challenging behaviours 311
Similarly, it was necessary to combine levels 1 and 2
(sexual talk; non-genital touching) and levels 3 and 4
(exhibitionism; masturbating in public) of SEX as
these also were not consistently discriminated by
raters. These findings were used in the determina-
tion of scores for the clinical weighting of the severity
items (see next section).
OBS indices and scoring
The final structure of the OBS is presented in
Table I. The OBS produces three key indices. The
first, ‘Cluster’ (range 0–9), comprises the sum of the
number of categories for which challenging behav-
iours have been observed (present ¼1, absent ¼0).
Similarly, the second, ‘Total Levels’ (range 0–34),
comprises the sum of the number of individual
severity levels endorsed (behaviour present ¼1,
absent ¼0). The final score represents the ‘Total
Clinical Weighted Severity’ score (range 0–77).
In contrast to the Total Levels score in which every
behaviour that is observed scores the same value,
the weighted severity score reflects clinical opinion
that some behaviours within each category are more
severe than others. However, the behaviour levels
in SEX and SOC about which staff disagreed as to
the relative level of severity (see previous section
‘agreement’) were assigned the same weighted value.
The following example illustrates the scoring of
these three indices. A client displayed three different
types of verbally aggressive behaviour (VA level 1
‘shouting’, VA level 2 ‘swearing’ and VA level 4
‘verbal threats’), but no other type of challenging
behaviour. In this case, the person would be rated as
1 on the Cluster score (1/9), 3 on the Total Levels
(3/34) and 7 (1 þ2þ4) for the Total Weighted
Severity (7/77). The two other indices, frequency of
behaviour and the impact on others (each rated on a
5-point Likert scale), are not reported in the current
psychometric analyses as they do not form the
structure of the scale, but rather provide additional
clinical data. In the case of the INI sub-scale,
because there was only one severity level, the
frequency measure was used as a proxy for Severity
To test the validity of the OBS, a number of
other measures with reported reliability and valid-
ity data were administered with Sample 1. The
Neurobehavioural Rating Scale-Revised (NRS-R
[34, 35]) is a well-established measure, comprising
29 items rated on a 4-point scale for evaluating the
potential impact of neurobehavioural sequelae (e.g.
disorientation, emotional withdrawal, poor plan-
ning) on social and occupational independence.
Analysis by Vanier et al.  produced a five-factor
structure including Intentional Behaviour,
Emotional State, Survival Oriented Behaviour,
Arousal State and Language. Only 26 of the original
items are employed in this structure, so a global
score ranges from 26–104, with higher values signi-
fying greater levels of impact. The Mayo Portland
Adaptability Inventory (MPAI ) is a 36-item
scale that measures adaptive functioning after brain
injury. It measures six domains: Physical/Medical,
Cognition, Emotion, Everyday Activities, Social
Behaviour and Behaviour. The scale produces a
total score (range 0–90), with lower scores repre-
senting higher levels of independence and sub-scale
scores can also be calculated. The Current
Behaviour Scale (CBS [21, 36] is a 25-item scale
that represents two factors: Loss of Emotional
Control (e.g. impulsive, short tempered) and Loss
of Motivation (e.g. amount of initiative, degree of
spontaneity). Mean score ranges for each of the
two factors are 1–7, with higher scores representing
greater degrees of behavioural disturbance.
The Sydney Psychosocial Reintegration Scale
(SPRS ) is a 12-item scale that measures
psychosocial outcome and provides a score for each
of three domains (range 0–24), Work and Leisure,
Relationships and Living Skills as well as a Total
score (0–72), with higher scores representing better
levels of reintegration.
For Sample 1, permission for the study was provided
by the South Western Sydney Area Health Service
Human Research Ethics Committee. When a client
was identified as having met the inclusion criteria
and consent had been obtained, inter-rater reliability
was tested by GS and CM independently completing
an OBS based on behavioural information provided
by a staff informant during an interview. In addition,
the other measures for testing the validity of the OBS
were also completed (i.e. NRS-R, MPAI, CBS and
SPRS). To examine test–re-test reliability, GS then
readministered the OBS in a follow-up interview
with the informant 1 week later.
For Sample 2, OBS data were collected on 28
clients of the Consultancy both prior to commence-
ment of an intervention (‘pre-intervention’) and
4 months after intervention had commenced
(‘4 month intervention’) as part of a quality assur-
ance project. OBS data was obtained from infor-
mants (primarily family members and service
providers) and at the second measuring point infor-
mants were not made aware of their initial ratings.
The 4-month interval did not necessarily signal
the completion of an intervention, nor a successful
outcome, but clinical experience had found that this
312 G. Kelly et al.
was a reasonable period of time for an initial
intervention to have had a measurable effect.
Descriptive statistics were generated for all variables
of interest. Given the measurement characteristics of
the data (i.e. the sums of counts), non-parametric
statistical procedures were used. For Sample 1, the
reliability data were analysed using Spearman corre-
) to evaluate the level of agreement
between the ratings of GS and CM (inter-rater) as
well as the agreement between the time 1 and time 2
ratings of GS (test–re-test). Convergent and diver-
gent validity was assessed by using correlations to
examine the level of association between the OBS
indices and the global and sub-scale scores of the
NRS-R, MPAI, CBS and SPRS. It was hypothesized
that support for convergent and divergent validity
would be found by the differential pattern of
correlations between the OBS and the other mea-
sures: Convergent validity would be demonstrated
by the presence of significant associations between
the OBS indices and the total/sub-scale scores of the
other measures that contained behavioural items and
divergent validity would be demonstrated by the
absence of significant correlations between the OBS
indices and the other measures that did not contain
such items. For Sample 2, change between the
‘pre-intervention’ and ‘4 month intervention’ scores
was analysed using Wilcoxon Signed Rank Tests.
Descriptive statistics for the OBS indices as well as
the total or domain scores for the other measures are
displayed in Table III.
Inter-rater reliability was examined using the data
from Sample 1. This was accomplished by correlat-
ing OBS indices (Cluster and Total Levels) from
rater 1 (GS) with rater 2 (CM) at time 1. Correlation
coefficients were very strong for both the OBS
¼0.99, p< 0.001) and OBS Total Levels
¼0.97, p< 0.001), indicating that the clinical
descriptors can be used by different raters with a
high degree of consistency.
Test–re-test reliability was evaluated by correlat-
ing the OBS indices obtained by GS at time 1 and
time 2 (a period of 1 week). Correlation coefficients
were again strong for OBS Cluster (r
p< 0.001) and OBS Total Levels (r
p< 0.001), indicating good stability of the OBS
across a period of 1 week.
Convergent and divergent validity
Total scores. Convergent and divergent validity of
the OBS was initially assessed by correlating the
overall indices of the OBS (Cluster, Total Levels and
Total Clinical Weighted Severity) with the total
scores of the other scales. Results are displayed
in Table IV. All possible correlation coefficients
were calculated, but, for ease of viewing, the non-
significant coefficients are not displayed. Initial
evidence for convergent and divergent validity was
found at this broad level of analysis. As expected, the
OBS was related to scales that incorporated mea-
sures of behavioural disturbance including measures
of neurobehavioural sequelae (NRS-R), adaptive
functioning (MPAI) and behaviour (CBS), but not
broader psychosocial reintegration (SPRS).
Specifically, the OBS indices correlated significantly
with the MPAI and NRS-R Total scores and with
the Loss of Emotional Control factor of the CBS,
but not with the Loss of Motivation factor of the
CBS or the SPRS Total score.
Domain scores. To further test the validity of the
OBS, the three indices were then correlated with the
domain or factor scores from the other scales (see
Table IV). Once again, the results provided provi-
sional evidence for both convergent and divergent
validity. In terms of the MPAI, the OBS indices did
not have significant associations with the Physical/
Medical or Everyday Activities domains, but were
significantly related to the three domains that con-
tained behavioural items, namely Emotion (includes
an item on aggression), Social Behaviour (e.g.
initiation, appropriate social interaction) and
Behavior (e.g. initiation, law violation, drug use).
Turning to the NRS-R, the correlation coefficients
between the OBS indices and three factors were non-
significant (i.e. Intentional Behaviour, Emotional
State and Language), but the OBS indices were
significantly related to two other factors, namely
Survival Oriented Behaviour (e.g. irritability, hostil-
ity, disinhibition) and Arousal State (e.g. alertness,
mental fatigue, attention). Finally, as expected,
Table III. Descriptive statistics for global indices (n¼30).
Mean (SD) Range
OBS Cluster 4.87 (1.59) 2–8
OBS Total levels 9.53 (4.49) 3–23
OBS Total clinical weighted severity 19.93 (9.79) 3–47
MPAI Total 37.97 (9.89) 14–58
NRS-R Total 62.30 (10.50) 37–86
CBS Loss of emotional control 5.58 (0.78) 4–7
CBS Loss of motivation 3.94 (0.76) 2–5
SPRS Total 21.17 (9.44) 7–36
A tool for measuring challenging behaviours 313
none of the SPRS domains were significantly corre-
lated with the OBS indices.
In a third level of analysis each of the OBS sub-
scale scores was correlated with domain scores of the
other scales. Table V shows all statistically signifi-
cant correlations. Taking the OBS sub-scales in turn,
the aggression sub-scales VA and PA people corre-
lated significantly with other domains with aggres-
sion content (e.g. irritability, hostility and short
temper) such as MPAI Emotion, NRS-R Survival
Oriented Behaviour and CBS Loss of Emotional
Control, but not with sub-scales lacking aggression
content. PA object had one strong significant corre-
lation with MPAI Behaviour—a domain represent-
ing a collection of issues such as psychiatric
symptomatology plus behaviours such as law viola-
tions and drug use. PA self represents acts of
self-harm; apart from a small positive association
with NRS-R Arousal, PA self did not correlate
significantly with any other domains.
SEX, PER/REP and WAN/ABS collectively
showed few significant relationships with domains
from other scales. This can be expected because
those domains either do not have closely related
items or, if they do, their contribution to the domain
score is small. SEX showed one significant result
with MPAI Social Behaviour possibly due to one
of only three items being about socially inappro-
priate behaviour (which may include sexualized
behaviours). PER/REP showed one significant
relationship with NRS-R Emotional State, which
has items relating to anxiety, depression and emo-
tional withdrawal. WAN/ABS showed no significant
relationships with other domains.
SOC is a broad sub-scale that reflects a range of
behaviours from awkward inter-personal behaviour
to risky behaviours and law violations. It was found
to correlate significantly with a number of other
domains including MPAI Social Behaviour and
MPAI Behaviour, which represent socially inappro-
priate behaviours. In addition, SOC correlated
significantly with domains containing emotion and
aggression content. Of interest is the fact that it is
the only OBS sub-scale to have multiple significant
correlations with the SPRS, indicating that increas-
ing scores on inappropriate social behaviours go
hand-in-hand with decreasing social reintegration.
The last OBS sub-scale, INI, was found to be related
to domains that measure impairments in initiation,
planning, organization and drive (i.e. MPAI
Cognition, NRS-R Intentional Behaviour, NRS-R
Emotional State and CBS Loss of Motivation).
Responsiveness to change
Responsiveness, namely the ability to measure
changes in client functioning resulting from inter-
ventions, is an important characteristic of a scale.
To evaluate this, OBS scores prior to behaviour
management intervention and 4 months into
Table IV. Significant correlations between OBS indices and domain scores from other neurobehavioural
Cluster Total levels Total clinical weighted severity
Mayo Portland Adaptability Inventory
Total score – 0.43* 0.45*
Physical/medical – – –
Cognition – 0.39* –
TEmotion – 0.51** 0.59**
Everyday activities – – –
TSocial behaviour – 0.42* 0.44*
TBehavior 0.43* 0.49** 0.56**
Neurobehavioural Rating Scale – R
Total score 0.40* 0.37* 0.42*
Intentional behaviour – – –
Emotional state – – –
TSurvival oriented behaviour – 0.43* 0.45*
Arousal state 0.38* – 0.39*
Language – – –
Current Behaviour Scale
TLoss of emotional control 0.51** 0.66** 0.63**
Loss of motivation – – –
Sydney Psychosocial Reintegration Scale
Total score – – –
Work and leisure – – –
Relationships – – –
Living skills – – –
*p< 0.05; **p< 0.01.
Trepresents sub-scales with overt challenging behaviour content.
314 G. Kelly et al.
intervention were compared. Specifically, the
Cluster, Total Levels and Total Clinical Weighted
Severity scores for the Sample 2 participants (n¼28)
at ‘pre-intervention’ and ‘4 month intervention’ were
computed and these are displayed in Table VI.
A decrease in scores represents a reduction in
the range and severity of challenging behaviours.
As expected, the ‘4 month intervention’ scores were
lower for each of the global measures, with all
improvements being statistically significant, suggest-
ing that the OBS is sensitive to real changes
occurring in challenging behaviours over time.
Clinically, the improvements occasioned by a
reduction in behavioural disturbance were corrobo-
rated by the data collected directly from the 28
informants. The informants were asked to make
a subjective evaluation of the extent to which the
challenging situation had changed. At the 4 month
follow-up, although one informant reported no
improvement (3.6%; 1/28), the most common
response was that the situation had ‘somewhat’
improved (53.6%; 15/28), with the remaining 42.9%
(12/28) of informants reporting the most positive
responses (i.e. ‘quite’, ‘very’ or ‘extremely well’).
The OBS shows initial promise of having good
reliability, validity and responsiveness in measuring
challenging behaviours among people with ABI
living in community settings. The coefficients for
the inter-rater and test–re-test reliability analyses
are within the ideal range for measures (i.e. 0.75)
as outlined by Andresen . This may be due to
the emphasis within the OBS on providing clear
operational definitions of the behaviours (see the
Appendix for an example), as well as the structured
Table V. Correlations between OBS sub-scales
and domains of other neurobehavioural scales
people SEX PER/REP WAN/ABS SOC INI
Mayo Portland Adaptability Inventory
Physical/medical (mobility, vision) – – – – – – – – –
Cognition (problem solve, communication, memory) – – – – – – – 0.45* 0.41*
Emotion (anxiety, depression, aggression) 0.39* – – 0.40* – – – 0.43* –
Everyday activities (live independent, self-care) – – – – – – – – –
Social behaviour (relationships, socially appropriate) – – – 0.36* 0.38* – – 0.56** –
Behaviour (initiation, law violation, drug use) – 0.61** – – – – – 0.40* 0.53**
Neurobehavioural rating scale – R
Intentional behaviour (initiative, affect, planning) – – – – – – – – 0.38*
Emotional state (mood, anxiety) – – – – – 0.44* – – 0.40*
Survival oriented behaviour (irritability, hostility) 0.45* – – 0.44* – – – 0.45* –
Arousal state (alertness, mental fatigue) – – 0.39* – – – – – –
Language (expression, comprehension) – – – – – – – – –
Current behaviour scale
Loss of emotional control (impulsive, short temper) 0.55** – – 0.63** – – – 0.60** –
Loss of motivation (initiative, spontaneous) 0.42* – – – – – – – 0.46*
Sydney psychosocial reintegration scale
Work and leisure (work and organizational skills) – – – – – – – 0.37* –
Relationships (spouse and family interactions) – – – 0.46* – – – 0.38* –
Living skills (social skills, transport, accommodation) – – – – – – – – –
The OBS_Levels score has been used for correlations;
Sample items are shown next to domain title. All statistically significant results
have been included.
*p< 0.05; **p< 0.01. PA self and WAN/ABS had restricted ranges; checks using Eta correlations indicated similar relationships to those
Table VI. Sensitivity data for clients 4 months into behavioural intervention (n¼28).
Pre-intervention 4 months intervention
Measure (range) Median (IQR) Median (IQR) Z-statistic p-value
Cluster (1–8) 3.5 (2.0) 3.0 (2.0) 2.49 0.013*
Total levels (1–29) 5.0 (7.8) 3.0 (4.8) 2.41 0.016*
Total clinical weighted 11.0 (13.0) 7.5 (10.0) 2.24 0.025*
Data obtained from pilot version of OBS that did not have the SOC sub-scale, hence score ranges are reduced.
A tool for measuring challenging behaviours 315
approach to collecting data that is promoted by the
Support was also found for the construct validity
of the scale, with evidence for both convergent
and divergent validity being identified. In terms
of convergent validity, the correlation between the
CBS Loss of Emotional Control factor score, the
scale providing the most ‘pure’ measure of behav-
ioural excesses after ABI and two of the three OBS
indices falls within the ‘excellent’ range of clinical
agreement (coefficients r
> 0.60) . The correla-
tion coefficients for the total scores of the MPAI and
NRS-R were in the moderate range for convergent
validity , which is not surprising as both these
scales provide a multi-dimensional measure of neuro-
behavioural impairment and functioning, within
which behavioural disturbance is only one of a
number of domains. However, more fine-grained
analysis found a pattern of significant correlations
between the MPAI and NRS-R sub-scales within
which behavioural items were typically grouped,
with the strength of the correlation coefficients
falling within the moderate range with the NRS-R
sub-scales, but approaching the benchmark for
‘excellent’ (i.e. 0.60) with the Emotion and
Behaviour sub-scales of the MPAI.
Initial evidence was also found for divergent
validity, well illustrated by the differential pat-
tern of correlation coefficients between the OBS
Total indices and the MPAI sub-scales, with only
one significant correlation with the sub-scales mea-
suring non-behavioural impairment and functional
domains. Divergent validity is also well demon-
strated by the lack of significant relationship
between OBS Total indices and any of the SPRS
Total or domain scores.
Further analysis, at the level of the OBS sub-
scales, showed that a somewhat predictable pattern
of significant and non-significant correlations
occurred between the sub-scales of the OBS and
the other sub-scales with similar content. Significant
correlations ranged from ‘moderate’ to ‘excellent’
(0.36–0.63). Importantly, the patterns of correla-
tions occurred both for measures of behavioural
excess (e.g. PA people) and deficiencies of
behaviours (i.e. INI), thereby furnishing further
evidence of the validity of the OBS. It is possible
that some correlation values presented in Table V
were an artefact of restricted response range, but
additional eta correlation analysis discounted this
Little data on the responsiveness of scales mea-
suring behavioural disturbance after ABI have
been published to date, a critical property for any
scale that aims to evaluate the effectiveness of
management interventions. The OBS demonstrated
responsiveness over a median period of 4 months,
with the median score changing in the expected
direction. Furthermore, the ‘real world’ significance
of these improvements was corroborated by the
In terms of the ICF , the OBS indices
primarily measured behavioural disturbance at the
impairment level, as evidenced by its correlation
with the other measures of behavioural excesses at
the impairment or, at most, activity levels (i.e.
MPAI, NRS-R, CBS). If the OBS indices had also
tapped into the broader psychosocial consequences
of behavioural disturbance (e.g. problems in the
workplace, with relationships, with moving around
in the community), some level of association could
have been expected with the SPRS, which measures
psychosocial outcome at the level of participation
[37, 39]. Therefore, the OBS appears to be a good
measure of behavioural disturbance per se, but not of
the indirect effects or consequences that arise as the
result of such impairments.
The OBS has a number of potential clinical
applications. First it provides a common language
that clinicians, family members, other service pro-
viders and the person with ABI can use in seeking
to address challenging behaviours. It helps to clarify
nebulous ‘problems’ and provides a clear focus for
management. Furthermore, it can elicit information
that might otherwise not be obtained because
administration of the OBS provides a structured
format that promotes reporting of all challenging
behaviours, not only those most salient to an
informant. It also promotes consideration of the
severity, frequency and impact of behaviours, indices
that are not all considered in many instruments. In
addition to identifying and acknowledging challeng-
ing behaviours, the OBS can facilitate the prioritiza-
tion of behaviours to be targeted, which is necessary
for goal-setting. It can be used as a form of reality
check, helping to clarify and test the perceptions
of staff and/or family members, who may either
understate or overstate the seriousness of behav-
iours. It also can also be used to evaluate efficacy of
There are a number of limitations and outstanding
issues that still need to be addressed in relation to
the OBS. In terms of content validity, although the
Consultancy referrals come from a very broad base,
there is still some chance that the pattern of referrals
reflects some service-system related bias and that
this has influenced the structure of the scale
categories and severity levels to some degree. The
scale would benefit from further reliability and
validity assessment with larger samples and other
behavioural measures. In addition, issues of the
scale’s measurement characteristics and internal
consistency are still to be addressed. It will also be
important to conduct further reliability testing using
316 G. Kelly et al.
family members as prime informants. Finally, it is
important to recognize that the weighting of the scale
items represents a clinical consensus about the
relative severity of these behaviours and is not
intended to suggest that the scale has the properties
of interval or ratio levels of measurement.
In terms of further development of the OBS,
Pender and Fleminger  have highlighted the
importance of developing scales that are reliable and
valid across more than one section of the rehabili-
tation continuum. Although the OBS has been
primarily developed and used within community
settings, it has also been applied in a small number of
cases in acute medical wards and non-specialized
inpatient rehabilitation settings, with staff in those
settings providing positive feedback about the utility
of the measure. One aim of the authors is to explore
the generalizability of the OBS to other settings on
the continuum of brain injury recovery.
In summary, the OBS appears to fill a niche by
providing a tool for measuring challenging behaviour
following ABI in community settings. It promotes
effective communication among community-based
clinicians and provides a means of clarifying the
types of challenging behaviours that clinicians may
not see first hand. Too often, the presence of
challenging behaviour has provided grounds for
people with ABI to be excluded from services and
restricted in their choices. In contrast, the develop-
ment of the OBS represents a step in the ongoing
effort to develop new rehabilitation models and ways
of working with people with ABI that are inclusive
and which promote their participation in the com-
munity to the maximum degree possible .
Special thanks to Suzanne Brown, Samantha Burns,
Kathryn Hoskin, Jan Loewy and Ann Parry for the
underlying clinical work that formed the basis of this
research. Thanks to Melbourne Citymission’s
Statewide ABI Case Management Service for con-
tributions to an early version of the scale. Thanks to
Diane Martine, Irene Ko, Marianne Bush, Thelma
Osoteo, Marcella Forman, Maggie McFadyen,
Barbara Strettles and Kate Hopman for assistance
in data collection.
1. Wood RL, McMillan TM. Neurobehavioural disability and
social handicap following traumatic brain injury. Hove, East
Sussex: Psychology Press; 2001.
2. Ponsford J, Sloan S, Snow P. Traumatic brain injury:
Rehabilitation for everyday adaptive living. Hove, UK:
Lawrence Erlbaum Associates; 1995.
3. Todd J, Loewy J, Kelly G, Simpson G. Managing challenging
behaviours: Getting interventions to work in nonspecialised
community settings. Brain Impairment 2004;5:42–52.
4. Johnson R, Balleny H. Behaviour problems after brain injury:
Incidence and need for treatment. Clinical Rehabilitation
5. Gardner RM, Bird FL, Maguire H, Carreiro R, Abenaim N.
Intensive positive behavior supports for adolescents with
acquired brain injury: Long-term outcomes in community
settings. Journal of Head Trauma Rehabilitation
6. Willis TJ, Lavigna GW. The safe management of physical
aggression using multi-element positive practices in com-
munity settings. Journal of Head Trauma Rehabilitation
7. Yody BB, Schaub C, Conway J, Peters S, Strauss D,
Helsinger S. Applied behavior management and acquired
brain injury: Approaches and assessment. Journal of Head
Trauma Rehabilitation 2000;15:1041–1060.
8. Peters MD, Gluck M, McCormick M. Behaviour rehabili-
tation of the challenging client in less restrictive settings.
Brain Injury 1992;6:299–314.
9. Strettles B, Bush M, Simpson G, Hodgkinson A. Forced
choices: Accommodation in New South Wales for adults with
high care needs after traumatic brain injury. Sydney: New
South Wales Motor Accidents; 2005.
10. World Health Organisation. International classification of
functioning, disability and health (ICF). Geneva: World
Health Organisation; 2001.
11. Barnes P, Radermacher H. Neurological rehabilitation
in the community. Journal of Rehabilitation Medicine
12. Simpson G, Blaszczynski A, Hodgkinson A. Sex offending
as a psychosocial sequela of traumatic brain injury. Journal of
Head Trauma Rehabilitation 1999;14:567–580.
13. Kay SR, Wolkenfeld F, Murrill LM. Profiles of aggression
among psychiatric patients: I. Nature and prevalence. Journal
of Nervous and Mental Disease 1988;176:539–546.
14. Alderman N, Knight C, Morgan C. Use of a modified version
of the Overt Aggression Scale in the measurement and
assessment of aggressive behaviours following brain injury.
Brain Injury 1997;11:503–523.
15. Ray WA, Taylor JA, Lichtenstein MJ, Meador KG. The
Nursing Home Behavior Problem Scale. Journal of
16. Tustin RD, Kent PA, Haskell S, Bond MJ. Measuring
severity of challenging behaviours: A Behaviour Disorder
Scale. Australia and New Zealand Journal of Developmental
17. Bruininks RH, Hill B, Weatherman R, et al. Inventory for
Client and Agency Planning (ICAP). Itasca: The Riverside
Publishing Company; 1986.
18. Yudofsky SM, Silver JM, Jackson W, Endicott J, Williams D.
The Overt Aggression Scale for the objective rating of verbal
and physical aggression. American Journal of Psychiatry
19. Corrigan JD. Development of a scale for assessment of
agitation following traumatic brain injury. Journal of
Clinical and Experimental Neuropsychology 1989;11:
20. Glenn MB, Burke DT, O’Neil-Pirozzi T, Goldstein R,
Jacob L, Kettell J. Cutoff score on the apathy evaluation
scale in subjects with traumatic brain injury. Brain Injury
21. Kinsella G, Packer S, Olver J. Maternal reporting of
behaviour following very severe closed blunt injury. Journal
of Neurology, Neurosurgery, and Psychiatry 1991;
A tool for measuring challenging behaviours 317
22. Frosch S, Gruber A, Jones C, Myers S, Noel E,
Westerlund A, Zavisin T. The long term effects of traumatic
brain injury on the roles of caregivers. Brain Injury
23. Levin HS, High WM, Goethe KE, Sisson RA, Overall JE,
Rhoades HM, Eisenberg HM, Kalisky Z, Gary HE. The
neurobehavioural rating scale: Assessment of the behavioural
sequelae of head injury by the clinician. Journal of Neurology,
Neurosurgery, and Psychiatry 1987;50:183–193.
24. Kreutzer JS, Marwitz JH, Seel R, Serio CD. Validation of
a neurobehavioral functioning inventory for adults with
traumatic brain injury. Archives of Physical Medicine and
25. Godfrey HPD, Harnett MA, Knight RG, Marsh NV,
Kesel DA, Partridge FM, Robertson RH. Assessing distress
in caregivers of people with a traumatic brain injury (TBI): A
psychometric study of the Head Injury Behaviour Scale.
Brain Injury 2003;17:427–435.
26. Malec JF. Thompson JM. Relationship of the Mayo-
Portland Adaptability Inventory to functional outcome and
cognitive performance measures. Journal of Head Trauma
27. Hulse SH, Deese J, Egeth H. The psychology of learning.
4th ed. Kogakusha, Tokyo: McGraw-Hill; 1958.
28. Swan L, Alderman N. Measuring the relationship between
overt aggression and expectations: A methodology for deter-
mining clinical outcomes. Brain Injury 2004;18:143–160.
29. Sorgi P, Ratey J, Knoedler DW, Markert RJ, Reichman M.
Rating aggression in the clinical setting. A retrospective
adaptation of the Overt Aggression Scale: Preliminary results.
Journal of Neuropsychiatry 1991;3:52–56.
30. Alderman N, Davies JA, Jones C, McDonnel P. Reduction of
severe aggressive behaviour in acquired brain injury: Case
studies illustrating clinical use of the OAS-MNR in the
management of challenging behaviours. Brain Injury
31. Todd J, Kelly G, Simpson GK. Measuring challenging
behaviour: Development of the Overt Behaviour Scale
(OBS) [Abstract]. Australian Journal of Psychology
32. Tate RL, Strettles B, Osoteo T. Enhancing outcomes after
traumatic brain injury: A social rehabilitation approach.
In: Wilson BA, editor. Neuropsychological rehabilitation:
Theory and practice. Lisse, The Netherlands: Swets &
Zeitlinger; 2003. pp 139–169.
33. Landis J, Koch GG. The measurement of observer
agreement for categorical data. Biometrics 1977;33:
34. McCauley SR, Levin HS, Vanier M, Mazaux JM, Boake C,
Goldfader PR, Rockers D, Butters M, Kareken DA,
Lambert J, Clifton GL. The neurobehavioural rating scale-
revised: Sensitivity and validity in closed head injury assess-
ment. Journal of Neurology, Neurosurgery, and Psychiatry
35. Vanier M, Mazaux JM, Lambert J, Dassa C, Levin HS.
Assessment of neuropsychologic impairments after head
injury: Interrater reliability and factorial and criterion
validity of the Neurobehavioral Rating Scale-Revised.
Archives of Physical Medicine and Rehabilitation 2000;81:
36. Elsass L, Kinsella G. Development of a scale for
measuring behaviour change following closed head injury.
In: Anderson V, Bailey M, editors. Proceedings of
the fourteenth Annual Brain Impairment Conference.
Melbourne: Australian Society for the Study of Brain
Impairment; 1989. pp 124–131.
37. Tate R, Hodgkinson A, Veerabangsa A, Maggiotto S.
Measuring psychosocial recovery after traumatic brain
injury: Psychometric properties of a new scale. Journal of
Head Trauma Rehabilitation 1999;14:543–557.
38. Andresen EM. Criteria for assessing the tools of disability
outcomes research. Archives of Physical Medicine and
39. Tate RL, Strettles B, Osoteo T. The clinical practice of
a community rehabilitation team for people with acquired
brain injury. Brain Impairment 2004;5:81–92.
40. Pender N, Fleminger S. Outcome measures on inpatient
cognitive and behavioural units: An overview.
Neuropsychological Rehabilitation 1999;9:345–361.
318 G. Kelly et al.
Appendix: Items for the inappropriate sexual behaviour (SEX) category
Severity Item descriptions
1 Sexual talk
.Comments of a sexual nature (e.g. ‘I’ve got a big dick’, ‘I want to make babies with you’, ‘You’ve got nice tits’,
‘I could give you a good time’) where comments may be face-to-face or in the form of phone calls or letters.
.Explicit accounts of sexual activities (e.g. ‘When I am with a woman I like to ...’).
.Touching other people who do not want to be touched (but contact does not involve genitals). For example kissing hand
or arm, putting arm around shoulder, patting someone’s knee, rubbing or caressing arm or leg or back.
.Also includes touching clothing (e.g. lifting skirts).
.‘Flashing’, exhibiting genitals, undressing in public.
.Failing to dress (e.g. Walking about house without clothes on when co-residents could be or are present.
Answering door when naked).
.Masturbation in a public or shared setting when other people are in the area (e.g. masturbating in a car in a public car park
where passers by may see; masturbating in a common area in a supported residential setting).
3 Touching (genital)
.Touching (or making attempts to touch) other people’s breasts, buttocks or genitals (e.g. groping staff who walk by,
fondling breasts of support workers, pulling other’s hands toward own groin).
4 Sexual assault, rape
.Attempt to forcibly undress another person.
.Use of threat to obtain sex.
.Sexual penetration of another person who has not consented.
Note: Child molestation is not included as a behaviour severity level per se, as the act of molestation may involve any of the behaviours
outlined above. Rather the victim’s age is also recorded and, if under the age of 16, then the term child molestation is applied as a second
descriptor in addition to the specified behaviours.
A tool for measuring challenging behaviours 319