Validitas Konstruk Working Memory (Gwm) dan Learning Efficiency (Gl) Tes Kognitif AJT pada Anak dan Remaja

Muflihah Abustani, Rahmat Hidayat, S.Psi., M.Sc., Ph.D.

2025 | Tesis | S2 Magister Profesi Psikologi

Kemampuan kognitif merupakan penentu penting prestasi akademik dan fungsi adaptif sehingga validitas konstruk yang kuat menjadi krusial bagi asesmen kognitif. Tes Kognitif AJT dikembangkan berdasarkan teori Cattell-Horn-Carroll (CHC) untuk mengukur berbagai kemampuan kognitif, termasuk Working Memory (Gwm) dan Learning Efficiency (Gl). Namun, temuan empiris sebelumnya menunjukkan ketidakstabilan pada model pengukuran CHC standar (Model A), khususnya pada subtes Narration Recall (NARREC). Studi ini bertujuan mengevaluasi model pengukuran alternatif (Model B) dengan mereposisi NARREC sebagai indikator Gwm. Data sekunder dari 4.650 partisipan berusia 5–18 tahun dianalisis menggunakan Generalized Structured Component Analysis with Measurement Errors Incorporated (GSCAm) melalui GSCA Pro. Model B menunjukkan kecocokan model yang lebih baik (GFI = 0,985; SRMR = 0,060), serta pemisahan empiris Gwm dan Gl yang lebih jelas dibandingkan Model A. Temuan ini mengindikasikan bahwa kinerja NARREC lebih merefleksikan tuntutan pemrosesan aktif working memory dibandingkan efisiensi pembelajaran jangka panjang.

Cognitive ability is a central determinant of academic achievement and adaptive functioning, making robust construct validity essential for cognitive assessment. The AJT Cognitive Test was developed within the Cattell–Horn–Carroll (CHC) framework to assess multiple cognitive abilities, including Working Memory (Gwm) and Learning Efficiency (Gl). However, prior empirical findings have suggested instability in the standard CHC measurement model (Model A), particularly regarding the Narration Recall (NARREC) subtest. This study evaluated an alternative measurement model (Model B) in which NARREC was reassigned to Gwm. Secondary data from 4,650 individuals aged 5–18 years were analyzed using Generalized Structured Component Analysis with Measurement Errors Incorporated (GSCAm) implemented in GSCA Pro. Model B demonstrated superior overall model fit (GFI = 0.985; SRMR = 0.060) and a clearer empirical separation between Gwm and Gl compared to Model A. These findings suggest that within the AJT Cognitive Test, NARREC primarily reflect active processing characteristic of working memory rather than long-term learning efficiency.

Kata Kunci : validitas konstruk, working memory, learning efficiency, Tes Kognitif AJT, GSCAm