Revising Social Communication Questionnaire scoring procedures for

Autism Spectrum Disorder and potential Social Communication Disorder Lucy Barnard-Brak, David M. Richman, & Steven R. Chesnut

Texas Tech University

Abstract

In analyzing data from the National Database for Autism Research, we

examine revising the Social Communication Questionnaire, a commonly

used screening instrument for Autism Spectrum Disorder.. A combination of

Item Response Theory and Mokken scaling techniques were utilized to

achieve this and abbreviated scoring of the SCQ is suggested. The

psychometric sensitivity of this abbreviated SCQ was examined via

bootstrapped Receiver Operator Characteristic (ROC) curve analyses.

Additionally, we examined the sensitivity of the abbreviated and total scaled

SCQ as it relates to a potential diagnosis of Social (Pragmatic)

Communication Disorder (SCD). As SCD is a relatively new disorder

introduced with the fifth edition of the Diagnostic and Statistical Manual

(DSM-5), we derived potential diagnosis of SCD among individuals with

ASD via mixture modeling techniques using NDAR data. These analyses

revealed two classes or clusters of individuals when considering the two core

areas of impairment among individuals with ASD: social communication and

restricted, repetitive patterns of behavior.

Sample

From NDAR, the sample consisted of 1,031 individuals with SCQ and ADOS score information.

The average age of the sample was 103.45 months (SD = 70.19).

Findings

Conclusions

Results indicate evidence of an abbreviated scoring for the SCQ that appears to be psychometrically sufficient.

It appears that Mokken scaling in combination with Item Response Theory techniques provide a means of

abbreviating scales while maintaining psychometric properties of the total scale.

There is an unknown number of individuals without an ASD diagnosis who may qualify for an SCD diagnosis and

this procedure needs to be tested with a sample consisting of those individuals without ASD diagnoses as well.

Furthermore, the diagnostic assessment procedure for SCD is still very much in the development stage thus we are

careful to term this group of individuals as having potential SCD without the assistance of validated diagnostic

assessments or cutoffs.

Selected References:

Bolte, S., Holtmann, M., & Poustka, F. (2008). The Social Communication Questionnaire (SCQ) as a screener for ASDs. Journal

of the American Academy of Child & Adolescent Psychiatry, 47(6), 719-720.

Corsello, C., Hus, V., Pickles, A., Risi, S., Cook, E. H., Leventhal, B., & Lord, C. (2007). Between a ROC and a hard place:

decision making and making decisions about using the SCQ. Journal of Child Psychology and Psychiatry, 48(9), 932940.

Van der Ark, L. A. (2012). New developments in Mokken scale analysis in R. Journal of Statistical Software, 48(5), 1-27.

Wei, T., Chesnut, S. R., Barnard-Brak, L. & Richman, D.. (in press). Psychomertic analysis of the social communication

questionnaire using an item-response theory framework. Journal of Psychopathology and Behavioral Assessment.

Wiggins, L., Bakeman, R., Adamson, L., & Robins, D. (2007). The utility of the Social Communication Questionnaire in

screening for autism in children referred for early intervention. Focus on Autism and Other Developmental Disabilities, 22(1), 33-

38.

Introduction

Early screening is a prerequisite to early intervention, which may be

considered as especially important for Autism Spectrum Disorder (ASD).

The purpose of the current study was to develop a more psychometrically

efficient scoring procedure for the Social Communication Questionnaire

(SCQ).

Improving the integration of universal screening for ASD screening would

be served by developing an abbreviated and psychometrically valid

screening instrument.

Consideration should also be given to the new Social Communication

(Pragmatic Disorder) introduced with the DSM-5.

Social Communication Questionnaire

Commonly used screener for Autism Spectrum Disorder

Consists of 40 dichotomous items

1st item pertains to verbal ability

Typically completed by parents or caregivers

Normed for a sample aged 4 to 40 years old

SCQ has a clear psychometric provenance (e.g., Bolte, Holtmann, &

Poustka, 2008; Wiggins et al., 2007; Corsello et al., 2007).

Utilized SCQ lifetime form item scores over current form item scores

when both present based upon previous psychometric analyses (Wei,

Chesnut, Barnard-Brak, & Richman, in press)

The cutoff score of 15 for SCQ is typically used.

SCQ varies in sensitivity and specificity and this cutoff has been

questioned.

Subsequent meeting or exceeding the cutoff leads to referral for diagnostic

assessment.

Analyses

A combination of Item Response Theory and Mokken scaling techniques were utilized.

Mokken scaling analyses were performed in R (v. 3.1.2) using the Mokken package (van der Ark,

2012).

Item Response Theory analyses were performed in FlexMirt (v. 2.0).

Table 1 provides the seven items that have been identified along with factor loadings () and IRT

parameter estimates (a, b, and c).

Table 1. Factor loadings and IRT parameter estimates for seven-item SCQ Item a b c

Q4: Socially inappropriate questions/statements 0.61 0.92 0.04 -0.04 Q10: Used others hand like a tool 0.47 0.75 0.34 -0.25 Q11: Odd, preoccupying interests 0.77 2.49 -0.13 0.32 Q13: Unusual, intense special interests 0.80 2.24 -0.41 0.92 Q15: Odd ways or movements 0.62 1.02 -0.25 0.25 Q26: Look directly at you in communicating* 0.39 0.81 -0.41 0.33 Q39: Playing imaginative games* 0.40 0.77 -0.87 0.67

* Items reverse-recoded on SCQ

0

10

20

30

40

50

60

70

80

90

100

0 10 20 30 40 50 60 70 80 90 1001 Specificity

Sensitiv

ity

Sample

Whole Sample

Under 4

Over 4

Table 2. Model statistics for each class solution

Model # of free parameters BIC AIC

1-class 4 6,503.39 6,485.76

2-class 7 6,463.65 6,432.79

3-class 10 6,475.61 6,431.53

4-class 13 6,489.44 6,432.13

From the table above:

a refers to discrimination, b refers to difficulty c refers to the pseudo

guessing parameter

From the figure to the left:

Theta () refers to the continuum of the latent trait

The curves represented are the Item Characteristic

Curves for each item

The abbreviated SCQ appears to have similar levels of sensitivity and specificity

for the whole sample, under 4 years, and

over 4 years.

Mixture modeling analyses (see Table 2 for results) indicated the presence of two

classes in determining the prevalence of

potential SCD.

The graph below contains the distribution between ASD versus potential SCD.

Acknowledgements

This study was supported by grant, R40 MC27475, R40 MCH Autism Secondary Data Analysis Studies (SDAS)

Program, from the Maternal and Child Health Bureau, Health Resources and Services Administration, Department of

Health and Human Services.

Data used in the preparation of this poster resides in the NIH-supported NIMH Data Repositories, specifically from

the National Database for Autism Research. Information regarding collections and submitters is available upon request.

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

3 2 1 0 1 2 3Theta

Pro

bability

Question

4

10

11

13

15

26

39