Relative Weight Analysis of the Western Aphasia Battery
Relative Weight Analysis of the Western Aphasia Battery Charles Ellis Ph. D CCC-SLP, Richard K. Peach, Ph. D CCC-SLP & Kathrin Rothermich, Ph. D INTRODUCTION • The Western Aphasia Battery (WAB and WAB-R) has been criticized due to its psychometric properties [1 -3]. • WAB items are weighted and the percentages by which the 10 subtests contribute to the aphasia quotient (AQ) are different [4]. DISCUSSION RESULTS Table 1. Summary of Relative Weight Analysis of Four Components of WAB AQ • How the four major components and 10 subtests of the WAB contribute to the AQ is questionable [4]. • The AQ is believed to be primarily impacted by expressive skills [3, 4]. • • Crary & Gonzalez found 98% of the AQ is derived from information content, repetition, sequential commands and fluency [4]. Hula et al. reported spontaneous speech accounts for ~40% of the AQ [3]. Raw Relative Weight 95% CI Relative Contribution (%) Spontaneous Speech Auditory Verbal Comprehension . 306. 201 . 296 -. 319. 182 -. 215 30. 6* 20. 1+ Repetition Naming/Word Finding . 247. 246 . 236 -. 256. 237 -. 254 24. 7† 24. 6# *The RC for this AQ component differs from all major components +The RC for this AQ component differs from all major components • Accurate accounting of the contributions of each WAB subtest is important for characterizing patients’ aphasia. †The RC for this AQ component differs from spontaneous speech & auditory verbal comprehension #The RC for the AQ component differs from spontaneous speech & auditory verbal comprehension AIM Table 2. Summary of Relative Weight Analysis of Ten Subtests of WAB AQ • To determine the relative importance of: a) the four major components and b) the 10 subtests of the WAB to the AQ. METHODS Data Source • WAB scores from 288 persons with aphasia (PWA) were obtained from the Aphasia. Bank [8]. Statistical Approach • Relative weight analysis (RWA) was performed to determine the “relative importance” of each component and each subtest to the AQ • RWA calculates the percentage contribution of each variable to the prediction and totals them to 100% [5 -7]. • RWA provides a more accurate partitioning of variance when multicollinearity is present among the predictor variables • The RWA was completed with R relaimpo package. • Significance tests were based on bootstrapping with 10000 repetitions. Information Content Fluency, Grammatical Competence, Paraphasias Yes/No Auditory Word Recognition Sequential Commands Repetition Object Naming Word Fluency Sentence Completion Responsive Speech Raw Relative Weight 95% CI. 130. 124 -. 139. 144. 134 -. 154 . 053. 079. 084 . 140 . 105. 085. 091. 086 . 014 -. 063. 069 -. 086. 073 -. 092 . 132 -. 150 . 099 -. 110. 076 -. 092. 082 -. 097. 078 -. 092 Relative Contribution (%) 13. 1* 14. 4+ 5. 4† 7. 9^ 8. 4~ 14. 1! 10. 5# 8. 5@ 9. 1% 8. 7& * The RC for this subtest differs from all subtests except repetition +The RC for this subtest differs from all subtests • RWA offers clinicians and researchers unique information regarding the way in which WAB subtests contribute to AQ scores. • RWA suggests that Spontaneous Speech contributes 30%, Auditory Verbal Comprehension contributes 20% and Repetition and Naming/Word Finding contribute 25% respectively to the AQ. • Individual subtest contributions to the AQ do not sum in the same manner as the four components of the AQ. • Similar to previous studies, spontaneous speech is the major contributor to the AQ. Expressive tasks are the largest contributors to the AQ. • This relative weight analyses designed to elucidate the relative contributions of WAB 4 major areas and 10 subtests to the AQ highlights the impact that advancements in statistical analysis approaches can have on the interpretation of established measurement tools such as the WAB. REFERENCES 1. Kertesz, A. (1982). The Western Aphasia Battery. New York: Grune and Stratton. 2. Kertesz, A. (2006). The Western Aphasia Battery – Revised. Texas: Harcourt Assessments. 3. Hula, W. , Donovan, N. , Kendall, D. L. & Gonzalez-Rothi, L. J. (2010). Item response theory analysis of the Western Aphasia Battery. Aphasiology, 24, 1326 -1341. 4. Crary, M. A. & Gonzalez-Rothi, L. J. (1989). Predicting the Western Aphasia Battery Quotient. Journal of Speech and Hearing Disorders, 54, 163 -166. 5. Stadler, M. , Cooper-Thomas, H. D. , & Greiff, S. (2017). A primer on relative importance analysis: illustrations of its utility for psychological research. Psychological Test and Assessment Modeling, 59, 381403. 6. Tonidandel, S. , Le. Breton, J. M. , & Johnson, J. W. (2009). Determining the statistical significance of relative weights. Psychological Methods, 14, 387 -399. 7. Tonidandel, S. & Le. Breton, J. M. (2011). Relative importance analysis: A useful supplement to regression analysis. Journal of Business & Psychology, 26, 1 -9. 8. Forbes, M. , Fromm, D. , & Mac. Whinney, B. (2012). Aphasia. Bank: A resource for clinicians. Semin Speech Lang, 17; 33; (3), 217 -222. CONTACT †The RC for this subtest differs from all subtests ^The RC for this subtest differs from info content, fluency, yes/no, repetition, object naming ~ The RC for this subtest differs from info content, fluency, yes/no repetition, object naming ! The RC for this subtest differs from all subtests except info content #The RC for this subtest differs from all subtests except sentence comprehension @The RC for this subtest differs from info content, fluency, yes/no, repetition, object naming %The RC for this subtest differs from info content, fluency, yes/no, repetition &The RC for this subtest differs from info content, fluency, yes/no, repetition, object naming Charles Ellis, Ph. D CCC-SLP Communication Equity and Outcomes Lab Department of Communication Sciences & Disorders East Carolina University Greenville, NC 27834 ellisc 14@ecu. edu
- Slides: 1