Sensitivity and Specificity of the Words in Game Test: A contribution to the English validation study

Title: Sensitivity and Specificity of the Words in Game Test: A contribution to the English validation study Authors: Dora Bianchi, Emre Celik & Allegra Cattani Affiliation: Dipartimento di Psicologia dei processi di sviluppo e socializzazione, Sapienza Università di Roma Contacts: dora.bianchi@uniroma1.it; emre.celik@uniroma1.it; allegra.cattani@uniroma1.it Introduction: The Words in Game Test (Wing; Cattani et al., 2019) is an early screening tool to assess vocabulary skills in toddlers. We aim to provide new evidence for the Wing Test’s sensitivity and specificity, complementing previous validation studies (Bello et al., 2010). Method: English speaking children aged 19 to 36 months performed four Wing Test subscales (i.e., nouns and predicates comprehension and production), as well as other language scales. A large sample of 336 children (M age = 28.63 months; SD = 4.78) completed the two Wing Test comprehension scales, while 295 of them successfully completed the two production scales. Children’ subscales scores on the Wing Test were age-adjusted on 3-months standard scores, then two aggregate scores on Comprehension and Production scales were computed. A subsample of 86 participants also completed the PLS-4 Preschool Language Scale 4 (PLS-4; Zimmerman et al., 2011, M age = 27.76 months; SD = 4.79). The PLS-4 total score was considered as reference gold standard test (cut-off ≥ 85) and children were classified as typical (80.3%) and atypical (19.7%) language groups. The validity of Wing Test was then estimated by sensitivity, specificity, area under the ROC curve (AUC), accuracy and error rate. Reliability and external validity were also ascertained. Results: The optimal sensitivity/specificity of age-adjusted cut-off score was found at the 7th percentile for the Comprehension (sensitivity = 0.90, specificity = 0.75; AUC = .75, p = .002, accuracy = .78 and error rate = .22) and Production scales (sensitivity = 1.00, specificity = 0.73; AUC = .81, p < .001, accuracy = .86 and error rate = .14). The Wing Test dimensions showed good reliability (Cronbach’s alpha = .87 for Comprehension and = .91 for Production) and external validity (correlations with PLS-4 total score of .44 for Comprehension, and of .57 for Production). Conclusions: Findings provide new evidence for the optimal sensitivity/specificity age-adjusted cut-off scores of the Wing Test, supporting its adequacy for an early screening of typical/atypical vocabulary skills in toddlers. References: Bello, A. et al. (2010). Parole in gioco: una prova per la comprensione e produzione lessicale per la prima infanzia. Firenze: Giunti O.S. Cattani, A., et al. (2019). Words in Game Test: WING. Vocabulary assessment for preschooler children. St. Mabyn, UK: Stass Publications Zimmerman, I.L., et al. (2009). Preschool Language Scale. Fourth Edition UK. London: Pearson Assessment.