The early development of self-injurious behavior: An
Hall, S., Oliver, C. & Murphy, G.
Cerebra Centre for Neurodevelopmental Disorders,
School of Psychology,
University of Birmingham
Please use this reference when citing this work:
Hall, S., Oliver, C. & Murphy, G. (2001). The early development of self-injurious behavior: An
empirical study. American Journal on Mental Retardation, 106, 189-199.
The Cerebra Centre for Neurodevelopmental Disorders,
School of Psychology, University of Birmingham, Edgbaston, Birmingham, B15 2TT
Website: www.cndd.Bham.ac.uk E-mail: email@example.com
The Early Development of Self-injurious Behavior:
An Empirical Study
Scott Hall and Chris Oliver
University of Birmingham (Birmingham, UK)
University of Kent (Canterbury, UK)
Running Head: Early SIB
The early development of self-injurious behavior (SIB) in young children with developmental
disabilities was examined by tracking 16 school-age children who had recently started to show
‘early’ SIB over an 18 month period. Naturalistic observations were conducted in each child’s
classroom every 3 months and the association between early SIB and environmental events was
examined. Results showed that for the 4 children whose early SIB had escalated over the 18
month period, there was a significant association between early SIB and low levels of social
contact across observation points, supporting models of the development of SIB. This association
might be considered as a risk marker for the exacerbation of SIB. The implications of this
finding for targeting early interventions for SIB are discussed.
The Early Development of Self-injurious Behavior:
An Empirical Study
Self-injurious behavior (SIB) shown by people with developmental disabilities has long been
recognized as an intransigent and stigmatizing behavior, as evidenced by increasing research
attention over the last few decades (Carr, 1977; Iwata et al., 1994; Oliver, 1995). Most research
into SIB has examined its occurrence at a single point in time, either through studies of
prevalence (Oliver, Murphy & Corbett, 1987; Rojahn, 1986; Schroeder, Schroeder, Smith &
Dalldorf, 1978), functional analysis (Carr & Durand, 1985; Iwata, Dorsey, Slifer, Bauman &
Richman, 1982), or interventions (Carr & Durand, 1985; Durand & Carr, 1991; Howlin, 1993).
Less attention however, has been paid to the early development of SIB and its chronicity
(Murphy, Hall, Oliver, & Kissi-Debra, 1999; Schroeder, Bickel & Richmond, 1986; Windahl,
1988). Such research could have implications for models of the development of SIB and thus for
the early intervention and prevention of SIB.
Data from prevalence studies have indicated that SIB begins in childhood and progresses into the
teenage years with a corresponding increase in incidence (Kebbon & Windahl, 1986; Oliver et
al., 1987). Prevalence and cohort studies suggest that individuals most at risk for developing SIB
are children with a severe or profound degree of developmental disability, a diagnosis of autism,
and/or sensory and physical disabilities (Oliver, 1995; Schroeder et al., 1978). Once established
however, SIB is notoriously difficult to treat and presents significant problems to the individuals
concerned, their carers and service providers (Murphy et al., 1993). It would appear that a
possible strategy would be to introduce early interventions for children whose SIB is just
beginning, prior to it becoming established in the child’s repertoire (Oliver, 1995). Before such a
strategy can be implemented however, it is important to determine why SIB develops in these
children and whether individualized treatments could then be implemented to eliminate SIB at
this early stage. Additionally, predictive risk markers might facilitate specific targeting within an
early intervention strategy.
Several theories have been advanced to explain the early development of SIB. It is thought that
SIB may emerge from stereotypical behaviors commonly seen in childhood (e.g., Lovaas,
Newsom & Hickman, 1987), from accidental motor responses (Murphy & Wilson, 1985), as a
result of a minor illness (Carr & McDowell, 1980), as a respondent behavior (Romancyck,
1986), or from disrupted neurotransmitter systems (Harris, 1992). In an attempt to integrate some
of these accounts, Guess and Carr (1991) have proposed a 3-level model for the emergence and
maintenance of SIB. At level 1, internally regulated rhythmic patterns emerge in the child’s
repertoire in order to regulate maturation and development. As such, these behaviors are unlikely
to be influenced by environmental factors. At level 2, the behaviors begin to modulate arousal
levels in response to environmental stimulation. Low levels of arousal result in an increase in
rhythmic behaviors, high levels of arousal result in a decrease. At level 3, the behaviors develop
into stereotypy and self-injury that has an effect on the behaviors of other people. In this way,
SIB is maintained by contingent environmental events through the processes of positive and
negative reinforcement (see Oliver, 1995). As yet however, the model remains untested.
The next step is to move from the hypothetical to empirical, and to provide some data on the
early development of SIB. In a previous paper, we examined the early development of SIB in 17
children, aged under 15 years, who had recently started to show particular forms of behavior
which could be considered potentially self-injurious (see Murphy et al., 1999). Although the
children’s parents were carefully interviewed to determine precisely why and when their ‘early’
SIB had first started, it was difficult to determine the reason for the onset of the behavior.
Sixteen of the children were observed in their classrooms at school over a period of up to 2 years
at approximately 3-monthly intervals. For four of the children, early SIB had clearly escalated
whilst for the other children it had not done so. Whilst the reasons for the onset of early SIB
remained unclear in these children, the degree of concern expressed about the children's behavior
at time 1 was predictive of an increase in early SIB. Indeed, the critical component of SIB may
have been its aversiveness to others, acting as an establishing operation to produce escape
behavior which is socially reinforcing (see Oliver, 1993). No relationship was found between the
children’s early SIB and other stereotypical behaviors (e.g., body rocking and hand flapping) that
were also present in the child’s repertoire. For example, the children whose early SIB had
increased were no more likely to show other stereotypical behaviors than those children whose
SIB did not increase. Overall, these findings would tend to support level 3 of Guess and Carr's
(1991) model of the development of SIB i.e., the social responses of others to the child's SIB
could primarily account for the further development and exacerbation of SIB. A descriptive
analysis of these data, necessary to examine the relationship between early SIB and
environmental events, was not conducted however.
The present paper has two broad aims. Firstly, to determine whether the early SIB shown by
these children was associated with socially mediated environmental events. Secondly, to
determine whether an association between early SIB and socially mediated environmental events
across successive observation points would predict an increase in early SIB over time. No
attempt was made in this study to deliberately manipulate environmental events. Although
experimental analyses are generally more powerful than correlational analyses, they carry with
them their own threats to validity. The present study therefore provides an opportunity to press
the correlational approach to its limits.
All 16 of the children observed in the previous study were selected from 22 schools for children
with developmental disabilities and/or autism. According to their class teachers, the children had
begun to show behaviors within the previous 3 month period which could be considered
topographically similar to those shown by individuals with more established SIB (cf. Rojahn,
1986). Table 1 shows the participant characteristics.
+++ Insert Table 1 here +++
Five of the children showed hand biting, five hand-to-head hitting, three head-to-object banging,
two hair-pulling, two body hitting, one self-scratching and one showed eye-poking (3 of the 16
showed more than one topography). 11 were boys with the mean age of the children being 5.27
years (SD = 1.94 years). The mean developmental equivalent age of the children (from the
Vineland Adaptive Behavior Scales) was 1.31 years (SD = 0.66 years). 4 children were non-
ambulant, 4 had poor ambulation and 8 were ambulant. 6 children had been diagnosed with a
specific syndrome i.e., Sturge-Weber (2), Trisomy 6Q (1), Seckel (1), Fragile X (1) and Cornelia
de Lange (1) and 6 children had a diagnosis of autism. 4 children suffered from epilepsy, 5 had a
visual deficit and 3 had a hearing deficit. On two occasions at the end of the study, one of the
children (case 2) was required to wear straight arm splints to prevent her hand biting.
Observational response definitions, recording, and interobserver agreement. Child responses
recorded included body banging, head banging, body hitting, head hitting, eye poking, eye
pressing, hand mouthing, hand biting, scratching or rubbing of the skin and hair pulling.
Operational definitions for each of these responses can be found in Murphy et al., (1999).
Five teacher behaviors were also recorded: demands (defined as any verbal or physical direction
by the teacher in order for the child to complete an action or task); attention (defined as any other
physical or verbal contact made by the teacher to the child e.g., touching, response blocking,
offering drinks or food, reprimanding and commenting); demand removal (defined as the
discontinuation of demands for 10-s following the occurrence of a demand, or the time between
successive occurrences of demands, if demands recurred within 10-s); attention removal (defined
as the discontinuation of attention for 10-s following the occurrence of attention, or the time
between successive occurrences of attention, if attention recurred within 10-s) and no interaction,
defined as the absence of demands, attention, demand removal and attention removal. The
categories no interaction, attention removal and demand removal were automatically coded by
the computer at analysis. All teacher behaviors were therefore exhaustive, but not necessarily
mutually exclusive i.e., attention could co-occur with demand removal and demands could co-
occur with attention removal.
All responses were recorded on a laptop computer (Olivetti Quaderno, Model PT-XT-20) using
software that allowed continuous documentation of the frequency and/or duration of each
behavior and their interrelations (see Repp, Harman, Felce, VanAcker, & Karsh, 1989). Two
observers (one standard observer) independently scored responses during 30 of the total of 304
hours (i.e., 10%) of observational data collected. Observer records were compared on a 10-s
interval-by-interval basis, scoring agreements and disagreements on occurrence and non-
occurrence for each recorded behavior (Murphy, 1987). For example, for each 10-s interval, an
agreement on the occurrence of a behavior would be scored if both observers had scored the
behavior. The mean total percentage agreement across behaviors was 88.87% (range, 84.95% to
91.61%). The corresponding mean Kappa statistic was 0.68 (range, 0.59 to 0.74), suggesting that
agreement between observers was good.
All children were observed in their regular classroom at school at the beginning of the study
(time 1) and on a number of subsequent occasions (see below). Classrooms usually contained the
child's teacher, one or more teacher assistants and at least 4 but no more than 10 other children.
Other children in the class were not included in this study, though some were comparison
children (controls) for the first study (see Murphy et al., 1999). Each child in the study was
observed for a 3 to 4 hour period at each observation point. Throughout each observation period,
the observer followed the child as unobtrusively as possible and did not interact with the child at
any time. Teachers were reminded prior to each observation period to ignore the presence of the
observer and to interact normally with the child. Observations included a representative sample
of activities: meals, group activities, individual work, and leisure time. The observer stood in the
corner of the room and out of the child's line of sight.
The aim of the study was to conduct direct observations in classrooms for each child, once every
three months for 18 months. However, the mean length of interval between follow-ups for each
child was 3.67 months (range 3.2 months to 6 months). Long follow-up interval lengths (i.e., 6
months) occurred when 1 child (case 8) was admitted to hospital for a treatment unrelated to
SIB. The mean total length of follow-up period for each child was 17.69 months (range 12 to 24
months). Longer follow-up periods occurred for those children who were enrolled early in the
study, while shorter follow-up periods occurred for those children enrolled later in the study and
for no other reason.
Descriptive analysis. In order to examine the association between a child’s early SIB and socially
mediated environmental events, the conditional probability of SIB given the occurrence of a
particular teacher response, Ti, was calculated at each observation point for each child; p(SIB|Ti).
This was done by imposing 10-s intervals on the data and then determining the number of times
that SIB had occurred given that a particular teacher response had occurred in each 10-s interval.
This number was then divided by the number of 10-s intervals during which the particular
teacher response occurred. If the base rate of the teacher response was low however, a high
conditional probability could occur simply by chance. To control for this possibility, the
conditional probability of each teacher response, Ti, given SIB was also calculated (see Lerman
& Iwata, 1993); p(Ti|SIB). This was done by determining the number of times that a particular
teacher response occurred given that SIB had occurred in each 10-s interval and dividing this
number by the number of 10-s intervals during which SIB occurred. To illustrate the analysis,
Figure 1 shows occurrences of early SIB and a teacher behavior, T, represented in 10-s interval
format. Here, the conditional probability of SIB given the occurrence of T would be 0.75, and the
conditional probability of T given SIB would be 0.38.
+++ Insert Figure 1 here +++
The information in these two conditional probabilities can be combined using the z statistic
(Bakeman & Gottman, 1997; Bakeman & Quera, 1995). The formula for z is given as follows:
)1 )(1( / )
where xij = observed joint frequency of SIB and Ti, mij = expected joint frequency of SIB and Ti,
pi = unconditional probability of SIB, and pj = unconditional probability of Ti. A high value for z
indicates a significant association between SIB and the teacher response (i.e., higher than would
be expected by chance). All conditional probabilities and z statistics were calculated using the
SDIS-GSEQ software (Bakeman & Quera, 1995). Only the z statistics will be reported here.
Given that some of the children were followed up for differing lengths of time and that the
periods between measurement occasions inevitably varied slightly because of scheduling
constraints (see above), a linear growth model was adopted to characterize change in early SIB
over time (see Willett, 1988). Linear regression lines were fitted to each individual’s data using
Ordinary Least Squares with the slope parameter, β, representing the degree of change in SIB
over time (see Murphy et al., 1999). A positive value for β indicates an increase in early SIB, a
negative value indicates a decrease in early SIB.
Z-scores indexing the degree of association between early SIB and teacher responses for each of
the children at each observation point are shown in Figure 21. The numbers shown in bold above
each graph are the slope parameters of the regression line fitted to the duration of early SIB
observed across observations for each child (see above).
+++ Insert Figure 2 here +++
The figure shows that in 3 cases (i.e., data displays for cases 1, 2 and 8, shown at the top of
Figure 2) there was a significant association between SIB and environmental events as evidenced
by z-scores greater than 5.0 across 4 or more consecutive observation points. For one additional
case, z-scores were greater than 3.0 across 4 of the observation points. For cases 1, 5 & 8 the
association was consistently high between SIB and ‘no interaction’. For case 2, SIB was
associated with ‘no interaction’ over the first 4 observation points and then with ‘attention’ at the
last observation point. It can be seen from the numbers shown in bold above each graph that
these 4 children also obtained the highest slope parameters, suggesting that an association
between SIB and environmental events predicted the subsequent exacerbation of early SIB.
This consistent pattern of association between SIB and environmental events observed in cases 1,
2, 5 & 8 was not observed for the remaining children. These children also obtained lower slope
parameters, indicating that their SIB did not increase. However, associations with environmental
events (i.e., z-scores greater than 3.0) did occur for some of the children at some observation
points. For cases 6 and 12 for example, SIB was associated with demand removal at the third and
sixth observation points respectively; for case 11, SIB was associated with teacher interaction at
1 For case 6 at observation 2, case 12 at observation 1 and case 15 at observation 4, the number of intervals
containing SIB was less than 10. Z-scores for these individuals should therefore be interpreted with caution (see
Bakeman & Gottman, 1997, p. 145).
observation 5, and with no teacher interaction at observation 7; for case 3, SIB was associated
with no teacher interaction at observations 1 and 2; for case 9, SIB was associated with no
teacher interaction at observations 1 and 4 and with attention at observation 2; for case 16, SIB
was associated with no teacher interaction at observations 1 and 4 and with demands at
These data suggest that if an association with an environmental event is not present across
several observation points, SIB does not increase in the child’s repertoire. In order to examine
whether this was indeed the case, the observational data were pooled across observation points
for each child and a Yule’s Q statistic was computed for the association between SIB and no
interaction for each child. A high value for Q indicates a strong association. The Yule’s Q
statistic was employed because, unlike z, it is unaffected by the number of tallies in the data
(follow-up time) and so is comparable as an index of association across children (see Bakeman
& Gottman, 1997). Figure 3 shows change in early SIB (i.e., the slope parameter) plotted as a
function of the Yule’s Q statistic obtained between SIB and no interaction when the data were
pooled across observation points for each child.
+++ Insert Figure 3 here +++
The data show that when the data were pooled across observation points, high Yule’s Q values
were obtained for those children whose SIB had increased, suggesting that the association
between SIB and no interaction was predictive of an increase in SIB.
In this paper, we have attempted to trace the socially mediated correlates of the development of
SIB in young children with developmental disabilities whose SIB had recently started at school.
These children were followed-up over a period of up to two years and at each observation point,
the association between SIB and socially mediated environmental events (i.e., teacher behaviors)
was examined. In order to aid the interpretation of the resultant data and to correct for the
problem of "chance" in the descriptive analysis, a statistical measure of association, the z-score,
was employed for the appraisal of conditional probabilities. Most z-scores were low for most
children, suggesting that the associations between early SIB and environmental events were at
chance levels. However, for 4 of the children, z-scores indexing the association between early
SIB and ‘no interaction’ were consistently high across 4 or more observation points, indicating
that SIB was more likely to occur when these children had been left alone. In a study conducted
by Iwata et al., (1982), 3 of the 9 children who had been referred to an inpatient treatment center,
showed the highest rates of SIB when they had been left alone in a room for a period of 15
minutes. Although it was not stated when the self-injury had begun in the children in the Iwata et
al., (1982) study, all subjects showed SIB that produced tissue damage, suggesting that the
behaviors were well established. The data from the present study suggests that environmental
correlates can also be identified in young children showing very early SIB (i.e., SIB which had
begun less than a year ago but did not yet produce tissue damage).
It should be noted that only correlations between socially mediated environmental events and
early SIB could be ascertained in the present study. Previous descriptive analyses have attempted
to identify both the antecedents and consequences of SIB via the calculation of conditional
probabilities (e.g., Lalli, Browder, Mace & Brown, 1993; Lerman & Iwata, 1993). That is, these
authors attempted to identify three-chain sequences of events in the observational data. For
example, Lerman & Iwata (1993) identified a ‘no interaction-SIB-no interaction’ sequence in
one individual who lived in a group home whilst Lalli et al., (1993) identified ‘instructions-SIB-
instruction removal’ and ‘no interaction-SIB-attention’ sequences in two children aged 10 and
14. However, none of the conditional probabilities were corrected for chance in these studies.
The results of the present study therefore need to be set in this context.
In the present study, all four children whose early SIB was found to be associated with socially
mediated environmental events across successive observation points had shown increases in
early SIB. Inspection of Table 1 indicates that these children were no different from the other
children in terms of age, developmental age, clinical condition, or topography of early SIB. In
addition, the children whose early SIB had escalated were no more likely to show other
stereotypical behaviors than those whose SIB did not increase. For example, whilst child 1
engaged in body rocking and arm waving, in addition to his headbanging and headhitting, and
child 2 engaged in head rolling, body rocking and arm waving, in addition to her hand biting and
hair pulling, cases 5 and 8 did not show any other stereotypical behaviors. In addition, case 3,
whose SIB did not increase, also showed body rocking, head rolling and arm waving. It appears
therefore that stereotypcial behaviors already present in the children’s repertoires were not
predictive of an increase in SIB and that it was an association between SIB and no interaction
that best predicted the development of SIB in these children. These findings suggest that a
descriptive analysis of SIB similar to the one conducted here could be employed as a useful
screening device for the assessment of risk in children showing early SIB.
For cases 1, 5, and 8, the association between SIB and no interaction appeared to develop across
successive 3 monthly intervals. These data provide support for the transition of early SIB
between levels 1 and 2 of Guess & Carr’s (1991) theoretical model i.e., early SIB was unrelated
to socially mediated environmental events but subsequently occurred during periods of low
interaction. In these children then, SIB may have served to modulate levels of arousal - in this
case arousal in the absence (as opposed to an overload) of environmental stimuli. This
interpretation is highly speculative however, given that other sources of environmental
stimulation e.g., vibrating objects, light sources, loud noises etc. would have been available in
the classroom which could also help to modulate an adequate homeostatic level in these children.
A second interpretation to the arousal theory has been put forward by Lovaas et al., (1987) who
have suggested that self-stimulatory behaviors such as SIB serve to provide sensory
consequences which arise “automatically” when the behaviors occur. Evidence to support this
contention comes from the studies conducted by Rincover and his colleagues (Rincover, 1978;
Rincover, Cook, Peoples & Packard, 1979; Rincover & Devany, 1982) who showed that when
the sensory consequences of stereotypic and self-injurious behaviors were masked or blocked,
they reduced in frequency. Given that no attempt was made in this study to deliberately mask or
block the sensory consequences of these behaviors, we were unable to determine directly
whether the early SIB of these children served as automatic reinforcement. Further research
should be conducted to determine whether blocking the sensory consequences of early SIB (i.e.,
providing sensory extinction) could provide an effective early intervention strategy for early SIB.
The data for case 2 showed a different pattern of results. Here, an association between SIB and
no interaction was already well-established. Over successive observation points, an association
between SIB and teacher attention emerged. These data would appear to support the transition
between levels 2 and 3 of Guess & Carr’s (1991) model i.e., early SIB now served to control the
behavior of others, being maintained by either positive and/or negative reinforcement. However,
this hypothesis is based on a single case and should therefore be interpreted with caution. In the
Iwata et al., (1982) study, 2 of the children showed the highest rates of SIB when demands were
removed contingent on SIB, indicating that SIB was maintained by negative reinforcement. One
participant showed the highest rates of SIB when attention was provided contingent on SIB,
indicating that SIB was maintained by positive reinforcement. In the present study, few
associations between early SIB and socially-mediated environmental events could be detected,
suggesting that level 3 self-injury had not been established in these children.
The fact that an association between an exacerbation of early SIB and low levels of interaction
could be consistently detected in 4 of the children in the present study suggests that these
children are at high risk for developing SIB and should therefore be targeted for intervention.
Individual reports were compiled for each of these children and the reports were subsequently
passed on to the child’s parents and teachers. In only one case (i.e., child 5) was help actually
sought from services. The problem for preventative or early intervention strategies is that mild or
‘early’ SIB is unlikely to provoke teachers or parents to seek help from services. On many
occasions during participant selection, for example, we were directed away from the children
who subsequently became participants in the study and toward those children whose SIB was
already a prominent feature of their behavioral repertoire. For this reason, it appears that services
will have to be proactive and seek to actively identify children showing the early signs of SIB.
In addition, a lack of behavioral knowledge and understanding of the causes of SIB in schools
may contribute to the emergence and development of the behavior. Hall and Oliver (1992) have
suggested that the critical component of SIB may be its aversiveness to others. For instance,
parents and/or teachers of children engaging in SIB adopt strategies to escape the aversive
quality of the child’s SIB. The behaviors adopted by parents and/or teachers (e.g., providing
attention, removing demands) may serve to reinforce the child’s SIB which may subsequently
increase in frequency and/or intensity. Parents and/or teachers may therefore be engaged in a
behavioral trap. Further research should attempt to delineate this sequencing of events in early
Finally, Oliver, Hall, Hales and Head (1996) found that those working in close contact with
individuals who showed SIB were more likely to choose a strategy to stop SIB which would
ultimately reinforce the SIB. These data suggest that the dissemination of research on SIB is not
at present filtering down into practice. Because of this, it would appear that any early
identification strategy should be supplemented with training and support for teachers and parents
in the management of behavior problems.
Bakeman, R. & Gottman, J.M. (1997). Observing Interaction: An Introduction to Sequential
Analysis. Second Edition. Cambridge, Cambridge University Press.
Bakeman, R. & Quera, V. (1995). Analyzing Interaction: Sequential Analysis with SDIS and
GSEQ. Cambridge, Cambridge University Press.
Carr, E.G. (1977). The motivation of self-injurious behavior. Psychological Bulletin, 84, 800-
Carr, E.G. & Durand, V.M. (1985). Reducing behavior problems through functional
communication training. Journal of Applied Behavior Analysis, 18, 111-126.
Carr, E. G. & McDowell, J.J. (1980). Social control of self-injurious behavior of organic
etiology. Behavior Therapy, 11, 402-409.
Durand, V.M. & Carr, E.G. (1991). Functional communication training to reduce challenging
behavior: Maintenance and application in new settings. Journal of Applied Behavior Analysis,
Guess, D. & Carr, E.G. (1991). Emergence and maintenance of stereotypy and self-injury.
American Journal on Mental Retardation, 96, 299-319.
Hall, S. & Oliver C. (1992). Differential effects of severe self-injurious behaviour on the
behaviour of others. Behavioral Psychotherapy, 20, 355-365.
Harris, J.C. (1992). Neurobiological factors in self-injurious behavior. In J.K. Luiselli, J.L.
Matson & N.N. Singh (Eds.), Self-injurious behavior: Analysis, assessment, and treatment (pp.
93-121). New York: Springer-Verlag.
Howlin, P. (1993). Behavioral techniques to reduce self-injurious behaviour in children with
autism. Acta Paedopsychiatrica, 56, 75-84.
Iwata, B.A., Dorsey, M.F., Slifer, K.J., Bauman, K.E. and Richman, G.S. (1982). Toward a
functional analysis of self-injury. Analysis and Intervention in Developmental Disabilities, 2, 3-
Iwata, B.A., Pace, G.M., Dorsey, M.F., Zarcone, J.R., Vollmer, T.R., Smith, R.G., Rodgers,
T.A., Lerman, D.C., Shore, B.A., Mazaleski, J.L., Goh, H., Cowdery, G.E., Kalsher, M.J.,
McCosh, K.C. & Willis, K.D. (1994). The functions of self-injurious behavior: An experimental
epidemiological analysis. Journal of Applied Behavior Analysis, 27, 215-240.
Kebbon, L. & Windahl, S.I. (1986). Self-injurious behavior: Results of a nationwide survey
among mentally retarded persons in Sweden. In J.M. Berg & J.M. Dejong (Eds.), Science and
Service in Mental Retardation (pp. 142-148). London, Methuen.
Lalli, J.S., Browder, D.M., Mace, F.C. & Brown, D.K. (1993). Teacher use of descriptive
analysis data to implement interventions to decrease students' problem behaviors. Journal of
Applied Behavior Analysis, 26, 227-238.
Lerman, D.C. & Iwata, B.A. (1993). Descriptive and experimental analyses of variables
maintaining self-injurious behavior. Journal of Applied Behavior Analysis, 26, 293-319.
Lovaas, O.I., Newsom, C. & Hickman, C. (1987). Self-stimulatory behavior and perceptual
reinforcement. Journal of Applied Behavior Analysis, 20, 45-68.
Murphy, G. (1987). Direct observation as an assessment tool in functional analysis and
treatment. In J. Hogg, & N.H. Raynes (Eds.), Assessment in Mental Handicap: A Guide to
Assessment Practice, Tests and Checklists (pp. 190-238). London: Croom Helm.
Murphy, G., Oliver, C., Corbett, J., Crayton, L., Hales, J., Head, D., & Hall, S. (1993).
Epidemiology of self injury, characteristics of people with severe self-injury and initial treatment
outcome. In C. Kiernan (Ed.), Research to Practice? Implications of Research on the
Challenging Behavior of People with Learning Disability (pp. 1-35). Kidderminster: BILD
Murphy, G., Hall, S., Oliver, C. & Kissi-Debra, R. (1999). Identification of early self-injurious
behaviour in young children with intellectual disability. Journal of Intellectual Disability
Research, 43, 149-163.
Murphy, G. & Wilson, B. (1985). Self-injurious behavior. Kidderminster: BIMH Publications.
Oliver, C. (1993). Self-injurious behavior: from response to strategy. In C. Kiernan (Ed.),
Research to Practice? Implications of Research on the Challenging Behavior of People with
Learning Disability (pp. 136-188). Kidderminster: BILD Publications.
Oliver, C. (1995). Self-injurious behavior in children with learning disabilities: Recent advances
in assessment and intervention. Journal of Child Psychology and Psychiatry, 30, 909-927.
Oliver, C., Murphy, G.H. & Corbett, J.A. (1987). Self-injurious behavior in people with mental
handicap: A total population study. Journal of Mental Deficiency Research, 31, 147-162.
Oliver, C., Hall, S. Hales, J. & Head, D. (1996). Self-injurious behaviour and people with
intellectual disabilities: Assessing the behavioral knowledge and causal explanations of care
staff. Journal of Applied Research in Intellectual Disabilities, 9, 229-239.
Repp, A.C., Harman, M.L., Felce, D. Van Acker, R. & Karsh, K.G. (1989). Conducting
behavioral assessments on computer collected data. Behavioral Assessment, 11, 249-268.
Rincover, A. (1978). Sensory extinction: A procedure for eliminating self-stimulatory behavior in
developmentally disabled children. Journal of Abnormal Child Psychology, 6, 299-310.
Rincover, A., Cook, R., Peoples, A., & Packard, D. (1979). Sensory extinction and sensory
reinforcement principles for programming multiple adaptive behavior change. Journal of Applied
Behavior Analysis, 12, 221-233.
Rincover, A. & Devany, J. (1982). The application of sensory extinction procedures to self-injury.
Analysis and Intervention in Developmental Disabilities, 2, 67-81.
Rojahn, J. (1986). Self-injurious and stereotypic behavior of non-institutionalized mentally
retarded people: Prevalence and classification. American Journal of Mental Deficiency, 91, 268-
Romancyck, R. (1986). Self-injurious behavior: Conceptualization, assessment, and treatment. In
K.D. Gadow (Ed.), Advances in Learning and Behavioral Disabilities (Vol. 5). Greenwich, CT:
Schroeder, S.R., Bickel, W.K. & Richmond, G. (1986). Primary and secondary prevention of
self-injurious behaviors: a life-long problem. In K.D. Gadow (Ed.), Advances in Learning and
Behavioral Disabilities (Vol. 5). Greenwich, CT: JAI Press.
Schroeder, S.R., Schroeder, C.S., Smith, B. & Dalldorf, J. (1978). Prevalence of self-injurious
behavior in a large state facility for the retarded: A three-year follow-up study. Journal of Autism
and Childhood Schizophrenia, 8, 261-269.
Willett, J.B. (1988). Questions and answers in the measurement of change. In E.Z. Rothkopf
(Ed.), Review of Research in Education (Vol. 15, pp. 345-422). Washington, DC: American
Educational Research Association.
Windahl, S.I. (1988). SIB in a time perspective: A follow-up of SIB prevalence and frequency
after 10-13 years. Paper presented at the 8th IASSMD conference, Dublin, UK.
We are very grateful to the children, their parents and teachers who helped in this study. This
research was funded by the Mental Health Foundation, UK. Requests for reprints should be sent
to Scott Hall, School of Psychology, University of Birmingham, Edgbaston, Birmingham B15
Figure 1. Hypothetical data showing occurrences of SIB and a teacher behavior T in 10-s
Figure 2. Z-scores indexing the degree of association between early SIB and environmental
events. The numbers in bold above each graph indicate the degree of change in observed SIB
over the duration of the study where positive values indicate an increase in SIB, negative values,
Figure 3. Change in early SIB plotted as a function of the Yule’s Q value for the association
between early SIB and no interaction when observations were pooled for each child.
Characteristics of the children
Physical disability Clinical condition Topography of ‘early’
epilepsy, vision and
Head banging, head
epilepsy, vision deficit
Hand biting, hair
Hand biting, hair
Profound MR, infantile
Severe MR, infantile
Severe MR, Trisomy
Severe MR, infantile
Profound MR, Seckel
Profound MR, Fragile
Severe MR, infantile
Profound MR, Sturge-
epilepsy, vision and
Severe MR, infantile
Severe MR, infantile
Severe MR, infantile
Profound MR, Cornelia
de Lange syndrome
1 adaptive behaviour composite of Vineland Adaptive Behavior Scales
134679 10 1112 1314 15
Association between SIB and:
Time of Observation (months)
-0.20 0.2 0.4 0.60.8
Yule's Q (no interaction and SIB)
Change in SIB