The Classification Accuracy of the MacArthur

The Classification Accuracy of the MacArthur Communicative Development Inventory-Level III
Wenonah Campbell and Elizabeth Skarakis-Doyle, Doctoral Program in Rehabilitation Sciences, University of Western Ontario
Lynn Dempsey, Department of Applied Linguistics, Brock University
INTRODUCTION
• Independent samples t-tests Æ LI group obtained
significantly lower percentage scores on all subtests of
CDI-III
Table 1. Characteristics of Children with
Language Impairment
• Caregiver involvement in assessment widely
regarded as best practice in early intervention1,2
• Indeed, IDEA takes caregiver involvement as one of
its central foci3
ªThus, considerable interest in developing valid
and reliable parent-report measures
• MacArthur-Bates Communicative Development
Inventories (CDI) are two such tools:
• Words & Gestures (CDI-WG) Æ 8-16 mos
• Words & Sentences (CDI-WS) Æ16-30 mos
• Validated for infants, toddlers, and children with
developmental disabilities4,5,6
• Appropriate for assessing developing language,
and with precautions, identifying risk for
language impairment (LI)4,7
• CDI-III Æ short-form upper extension of CDI-WS
intended for 30-43 month-olds8
Variable
M (SD)
[min-max]
PPVT-III SS
(M = 100; SD = 15)
85.78 (8.98)
[72-100]
(n = 9)
TACL-3 – Grammatical
Morphemes Subtest SS*
(M = 10; SD = 2)
9.67 (2.50)
[6-13]
(n = 6)
MLU (in morphemes)**
2.14 (0.74)
[1.04-3.02]
(n = 5)
McCarthy Performance SS
(M = 50; SD = 10)
• Assesses vocabulary, grammatical complexity,
and linguistic concepts
Vocabulary
Grammar
Use
→
→
→
t (10.59) = 8.01, p < .01, d = 2.88
t (11.62) = 5.65, p < .01, d = 1.89
t (9.12) = 4.76, p < .01, d = 1.93
Goal of Present Study
To determine the accuracy with which the
CDI-III can discriminate between children with
LI and those developing typically.
• Specificity Æ excellent
• 2/3 misclassified children with TLD fell
between age range for CDI-WS and
CDI-III
• Concern: not all children in LI group achieved
performance IQ score within normal limits
45 (8.35)
[36-57]
(n = 9)
• Conducted DA on children with LI who had
McCarthy Performance SS ≥ 40 (n = 5)
• Selected subset of TLD group (n = 45) to
maintain proportion of children with LI at ~
13%
METHODS
• 9 children with LI
• 31-45 months (M = 39.00, SD = 6.14 ; 7 males)
• Proportion (13%) = prevalence data for preschool
population, particularly for local area of London,
Ontario, Canada12,13
• Inclusion based upon status in treatment, not
upon tests scores of referring clinicians14
• All had expressive and receptive deficits per
parent report
• Parents completed CDI-III as part of a larger study on
language comprehension
ªThus , nonverbal IQ cannot account
for results of original DA
• Although average performance of the LI and TLD groups
differed significantly, still must determine if CDI-III
correctly identifies particular children known to have LI
Table 2. Descriptives for the CDI-III by Group
TLD
M (SD)
[min – max]
LI
M (SD)
[min – max]
Vocabulary
(max = 100)
73.17 (18.74)
[25-100]
19.89 (18.58)
[0-56]
Grammar
(max = 12)
8.78 (3.87)
[0-12]
2.00 (3.28)
[0-8]
Use
(max = 12)
9.17 (2.16)
[3-12]
3.89 (3.22)
[0-9]
Total Score
(max = 124)
91.12 (22.87)
[28-124]
25.78 (24.07)
[0-70]
• 60 children with typical language development (TLD)
• 30-45 months (M = 37.17, SD = 4.11; 23 males)
• Replicated results of original analysis
Initial analysis Æ LI group’s CDI-III scores were compared
to those obtained by the TLD group
CDI-III
Scores
• Comparable to many language tests and
based solely on parental report
• BUT…sample approximated prevalence of
LI in population rather than using equal
size samples15
RESULTS AND DISCUSSION
•
• Overall classification accuracy excellent at
92.8%
• Sensitivity Æ good
• Note large effect sizes for all three comparisons
• Norming and validity data are emerging8,9,10
• Yet validity is an ongoing process Æ need to
show that CDI-III is valid for its intended
purpose11
• CDI-III Total Score significantly predicted
language status (Wilks’ lambda = .514, X 2 (1)
= 44.26, p = .000)
•
Therefore, a Discriminant Analysis was conducted to
determine whether the CDI-III Total Score would
accurately discriminate children with LI from those with
TLD
CONCLUSIONS
• This preliminary study suggests CDI-III may be
valid for the purpose of identifying children with
LI who are between 30-45 months of age
• Could be used in conjunction with other
measures for a complete assessment of
language
• Clinical implications Æ CDI-III could be used as
a first step in identification of LI for SLPs who
must:
Table 3. Classification Table for the CDI-III
Predicted Language Status
• screen large numbers of children
• prioritize a waiting list
Known Language
Status
TLD
LI
TLD
LI
Specificity
57 (95%)
3 (5%)
2 (22%)
Sensitivity
7(78%)
• provide services to remote areas
• Research implications Æ CDI-III may be an
efficient means of selecting children for more
extensive testing in large-scale studies
ACKNOWLEDGEMENTS
This study was conducted with funding from The University of
Western Ontario, Toronto Hospital for Sick Kids Foundation, and the Social
Sciences and Humanities Research Council of Canada. The authors gratefully
acknowledge the participation of the children and their families, as well as the
assistance of Prof. Chris Lee, Jayna Amting, Melanie Beaudin, Joselynne
Jaques, Sarah Pifher, and Brooke Thornton.
REFERENCES
1. Crais, E.R. & Calculator, S.N. (1998). Role of caregivers in the assessment process. In A.M. Wetherby, S.F. Warren, & J. Reichle (Eds.), Transitions in prelinguistic communication (pp. 261-284). Baltimore,
MD: Paul H. Brookes.
2. Law, M., Hanna, S., Hurley, P., King, S., Kertoy, M., & Rosenbaum, P. (2003). Factors affecting family-centred service delivery for children with disabilities. Child: Care, Health and Development, 29(5), 357366.
3. Crais, E.R. (2000). Ecologically valid communication assessment of infants and toddlers. In L.R. Watson, E.R. Crais, and T.L. Layton (Eds.), Handbook of early language impairment in children (pp. 1-38).
Albany, NY: Delmar Thomson Learning.
4. Fenson, L., Bates, E., Dale, P., Goodman, J., Reznick, J.S., & Thal, D. (2000). Measuring variability in early child language: Don't shoot the messenger. Child Development, 71(2), 323-328.
5. Fenson, L.D., Dale, P.S., Reznick, J.S., Thal, D., Bates, E., Hartung, J.P., Pethick, S., et al. (1991). Technical manual for the MacArthur Communicative Development Inventories. San Diego, CA: San Diego
State University.
6. Miller, J.F., Sedey, A.L., & Miolo, G. (1995). Validity of parent report measures of vocabulary development for children with Down syndrome. Journal of Speech & Hearing Research, 38(5), 1037-1044.
7. Feldman, H. M., Dollaghan, C. A., Campbell, T. F., Kurs-Lasky, M., Janosky, J. E., & Paradise, J. L. (2000). Measurement properties of the MacArthur Communicative Development Inventories at ages one
and two years. Child Development, 71(2), 310-322.
8. Dale, P.S., Reznick, J.S., & Thal, D.J. (1998April 2-5). A parent report measure of language development for three-year-olds. Paper presented at the International Conference for Infant Studies, Atlanta, GA.
9. Feldman, H. M., Dale, P. S., Campbell, T. F., Colborn, D. K., Kurs-Lasky, M., Rockette, H. E., et al. (2005). Concurrent and predictive validity of parent reports of child language at ages 2 and 3 years. Child
Development, 76(4), 856-868.
10. Oliver, B., Dale, P.S., Saudino, K.J., Petrill, S.A., Pike, A., & Plomin, R. (2002). The validity of a parent-based assessment of cognitive abilities in three-year olds. Early Child Development and Care, 172(4),
337-348.
11. Messick, S. (1989). Validity. In R. L. Linn (Ed.), Educational measurement (3rd ed., pp. 13-103). Washington, DC: American Council on Education.
12. National Institute on Deafness and Other Communicative Disorders. (1995). National strategic research plan for language and language impairments, balance and balance disorders, and voice and voice
disorders (NIH Publication No. 97-3217). Bethesda, MD: Author.
13. Warr-Leeper, G., Smith, T., & Dunn, C.(2001). Report to the Thames Valley Region Ontario Preschool Speech and Language Initiative (Document No. 196-19-01). London, Ontario, Canada: Thames Valley
Regional Health District Council.
14. Dollaghan,C., & Campbell, T. (1998). Nonword repetition and child language impairment. Journal of Speech, Language, and Hearing Research, 41, 1136-1146.
15. Plante, E., & Vance, R. (1994). Selection of preschool language tests: A data-based approach. Language Speech and Hearing Services in Schools, 25, 15-24.