The Strengths and Difficulties Questionnaire: A Research Note
The Strengths and Difficulties Questionnaire: A Research Note
The Strengths and Difficulties Questionnaire: A Research Note
581-586, 1997
Cambridge University Press
1997 Association for Child Psychology and Psychiatry
Printed in Great Britain, All rights reserved
0021-9630/97 $15.00 + 0,00
581
5S2 R. GOODMAN
SDQ-Rutter Correlations
Table 1 shows the correlations between SDQ and
0 0.2 0.4 0.6 0.8 Rutter scores. The correlations were only marginally
1-Specificity lower when the analyses were repeated for the dental and
Pigure 1. ROC curves for parent-rated questionnaires. psychiatric samples separately. No cross-measure corre-
584 R. GOODMAN
Table 1
Inter-measure Correlation for Each Type of Rater
SDQ-Rutter correlation
Table 2 used identical items for parents and teachers whereas the
Inter-rater Correlations for Each Type of Measure Rutter questionnaires were somewhat different for
Parent-Teacher parents and teachers. The ROC analyses showed that the
correlation two measures had equivalent predictive validity, as
(N == 128) judged by their ability to distinguish between psychiatric
and nonpsychiatric samples. Of course, discriminating
SDQ Rutter between psychiatric and dental clinic attenders is a
relatively easy task, but the high correlation between
Total Deviance/Difficulties score .62* .52
Conduct Problems score .65 .57
SDQ and Rutter scores within each clinic group suggests
Emotional Symptoms score .41 .47 that the two measures are also likely to be comparably
Hyperactivity score .54 .55 discriminating in more demanding screening tasks, such
Peer Problems score .59 as detecting nonreferred cases of child mental health
Prosocial Behaviour score .37 problems in the community; further empirical studies
would be needed to confirm this. Since previous studies
* Correlation significantly higher than the comparable
Rutter correlation (p < .02); all other comparisons non-
have shown that CBCL and Rutter parent questionnaire
significant. scores are highly correlated (Berg et al., 1992; Fombonne,
1989), and that these two sets of questionnaires are of
comparable predictive vahdity (Berg et al., 1992), it is
likely that the SDQ and CBCL will also be highly
correlated and have comparable validity; direct com-
lations can be presented on two SDQ scoresthe peer parisons are currently under way.
problems score and the prosocial behaviour scoresince
they have no Rutter counterpart. The SDQ and Rutter questionnaires can each be used
to generate separate scores for conduct problems,
emotional symptoms, and hyperactivity. For each of
these three scores, there was a high correlation between
Parent-Teacher Correlations the SDQ score and the Rutter score; and parent-teacher
correlations were comparable for the two sets of
Table 2 presents the correlation coefficients between measures. Despite its brevity, the SDQ also generated
teacher- and parent-derived scores when both are using two scores that have no Rutter counterparts; a peer
the SDQ or when both are using Rutter questionnaires. problems score and a prosocial behaviour score.
For comparable scores, the cross-situation correlations The performance of the SDQ could potentially have
of the SDQ and Rutter measures were generally similar, been undermined by three of its design features; inclusion
apart from the higher SDQ correlation for total score of strengths as well as difficulties; use of an identical
\X^ (1) = 5.90, p < .02]. Though the correlations were questionnaire for both parents and teachers; and a
lower when the analyses were repeated for the dental and compact presentation on just one side of paper. The
psychiatric samples separately, these correlations were equivalence of the SDQ and Rutter scores suggests that
generally comparable for the SDQ and Rutter measures, these three features have not had an adverse effect. This
apart from a higher SDQ correlation for total score in the should encourage researchers and clinicians who are
psychiatric sample [x^ (1) = 4.05, p < .05). contemplating incorporating similar features into other
questionnaires.
Rutter questionnaires are routinely used to categorise
Discussion children as likely psychiatric "cases" or "non-cases"
according to whether their total deviance score is equal to
Given the well-established validity and reliability ofthe or greater than a standard cut-off; 13 on the Rutter
Rutter questionnaires (Elander & Rutter, 1996), the high parent questionnaire and 9 on the Rutter teacher ques-
correlation between the total scores generated by the tionnaire (Rutter, 1967; Rutter et al., 1970). Using a
SDQ and Rutter questionnaires is evidence for the single cut-off for all studies has both advantages and
concurrent validity of the SDQ. Parent-teacher corre- disadvantages. The advantages are simplicity and equiv-
lations were either equivalent for the two measures or alence across studies. The main disadvantage is that
slightly favoured the SDQ, perhaps because the SDQ "caseness" does not have a comparable meaning in
STRENGTHS AND DIFEICULTIES QUESTIONNAIRE 585
different studies simply because those studies have the following additional advantages: a compact format;
employed the same cut-off. Comparability is particularly a focus on strengths as well as difficulties; better coverage
likely to be lost when high- and low-risk samples are of inattention, peer relationships, and prosocial behav-
contrasted. A worked example may be helpful. Study X iour; and a single form suitable for both parents and
involves 100 children from a high-risk population with a teachers, perhaps thereby increasing parent-teacher
true rate of psychiatric disorder of 50 %; if the screening correlations.
questionnaire has a sensitivity of .8 and a specificity of .8
when using the standard cut-off, the questionnaire will AcknowledgementsI am very grateful for the willing co-
identify 40 true positives and 10 false positives. Study Y operation of parents and teachers, and for the invaluable
involves 100 children from a low-risk population with a assistance of E)r. Hilary Richards and the staff of the three
true rate of disorder of 10%; even with the same London clinics that participated in the study: the Department
sensitivity and specificity, the questionnaire will identify 8 of Paediatric Dentistry of King's Dental Institute, Camberwell
Child Guidance Centre, and the Department of Child and
true positives and 18 false positives. Despite using the Adolescent Psychiatry, Hounslow.
same questionnaire and the same cut-off, a comparison of
"cases" from studies X and Y will primarily be a References
comparison of true positives from study X with false
positives from study Y. Achenbach, T. M. (1991a). Manual for the Child Behavior
Checklist 14-18 and 1991 Profile. Burlington, VT: University
Given these problems, the best strategy for researchers of Vermont Department of Psychiatry.
may be to choose cut-offs according to the likely disorder Achenbach, T. M. (1991b). Manual for the Youth Self-Report
rate in the sample being studied, and according to the and 1991 Profile. Burlington, v f : University of Vermont
relative importance for that study of false positives and Department of Psychiatry.
false negatives. It may also be appropriate to adjust cut- Berg, I., Lucas, C , & McGuire, R. (1992). Measurement of
offs for age and gender. Ongoing clinical and epidemio- behaviour difficulties in children using standard scales
logical studies using the SDQ should provide the basis for administered to mothers by computer: Reliability and val-
cut-offs adjusted for these sample characteristics. In idity. European Child and Adolescent Psychiatry, 1, 14-23.
addition, planned trials should establish if the predictive Elander, J., & Rutter, M. (1996). Use and development ofthe
validity of the SDQ can further be improved by an Rutter Parents' and Teachers' Scales. International Journal of
Methods in Psychiatric Research, 6, 63-78.
algorithm that combines SDQ scores with scores from an
Fombonne, E. (1989). The Child Behavior Checklist and the
additional and even briefer screening instrument that Rutter Parental Questionnaire: A comparison between two
elicits the respondent's view on whether the child has screening instruments. Psychological Medicine, 19, 111-1S5.
significant emotional or behavioural difficulties, and on Goodman, R. (1994). A modified version ofthe Rutter parent
the extent to which these difficulties result in social questionnaire including items on children's strengths: A
impairment or distress for the child, or burden for others. research note. Journal of Child Psychology and Psychiatry,
Until these various studies are completed, SDQ users can 35, 1483-1494.
use the provisional cut-off scores shown in Appendix B, Hanley, J. A., & McNeil, B. J. (1982). The meaning and use of
which are derived partly from the samples used for this the area under a receiver operating characteristic (ROC)
study and partly from other ongoing epidemiological curve. Radiology, 143, 29-36.
surveys using the SDQ. The "borderline" cut-offs can be Hanley, J. A., & McNeil, B. J. (1983). A method of comparing
used for studies of high-risk samples where false positives the areas under receiver operating characteristic curves
are not a major concern; the " abnormal" cut-offs can be derived from the same cases. Radiology, 148, 839-843.
Rutter, M. (1967). A children's behaviour questionnaire for
used for studies of low-risk samples where it is more completion by teachers: Preliminary findings. Journal oj
important to reduce the rate of false positives. Child Psychology and Psvehiatry, 8, 1-11.
Rutter, M., Tizard. J., & Whitmore, K. (1970). Education,
health and behaviour. London: Longman.
Schachar, R., Rutter, M., & Smith, A. (1981). The charac-
Conclusion teristics of situationally and pervasively hyperactive children:
Implications for syndrome definition. Journal of Child Psy-
These initial findings suggest that the SDQ may chology and Psychiatry, 22, 375-392.
function as well as the Rutter questionnaires (and, by
inference, the Achenbach questionnaires) while offering Accepted manuscript received 26 September 1996
586 R. GOODMAN
Date of Birth.
Not Somewhat Certainly
True True True
Considerate of other people's feelings D D
Restless, overactive. cannot stay still for long n D
Often complains of headaches, stomach-aches or sickness D D
Shares readily with other children (treats, toys, pencils etc) D D
Often has temper tantrums or hot tempers D D
Rather solitary, tends to play alone n
Generally obedient, usually does what adults request D
Many worries, often seems worried D D
Helpful if someone is hurt, upset or feeling ill n
Constantly fidgeting or squirming D D
Has at least one good friend D
Often fights with other children or bullies them
Often unhappy, down-hearted or tearful n D
Generally liked by other children D D
Easily distracted, concentration wanders D D
Nervous or clingy in new situations, easily loses confidence n
Kind to younger children n D D
Often lies or cheats D D
Picked on or bullied by other children D
Often volunteers to help others (parents, teachers, other children) n n D
Thinks things out before acting D D
Steals from home, school or elsewhere n
Gets on better with adults than with other children D n D
Many fears, easily scared D n
Sees tasks through to the end. good attention span D n
Signature Date