PENGEMBANGAN THREE-TIER MULTIPLE CHOICE WEB DIAGNOSTIC ASSESSMENT KESULITAN BELAJAR MATEMATIKA BERBASIS HOTS PADA SISWA SEKOLAH MENENGAH ATAS

KHAERUN NISA’, . (2026) PENGEMBANGAN THREE-TIER MULTIPLE CHOICE WEB DIAGNOSTIC ASSESSMENT KESULITAN BELAJAR MATEMATIKA BERBASIS HOTS PADA SISWA SEKOLAH MENENGAH ATAS. Doktor thesis, UNIVERSITAS NEGERI JAKARTA.

	Text COVER.pdf Download (1MB)
	Text BAB I.pdf Download (533kB)
	Text BAB II.pdf Restricted to Registered users only Download (1MB) | Request a copy
	Text BAB III.pdf Restricted to Registered users only Download (1MB) | Request a copy
	Text BAB IV.pdf Restricted to Registered users only Download (3MB) | Request a copy
	Text BAB V.pdf Restricted to Registered users only Download (506kB) | Request a copy
	Text DAFTAR PUSTAKA.pdf Download (358kB)
	Text LAMPIRAN_compressed.pdf Restricted to Registered users only Download (1MB) | Request a copy

Abstract

Rendahnya kompetensi numerasi dan Higher-Order Thinking Skills (HOTS) peserta didik SMA serta keterbatasan asesmen berbasis skor dalam mengidentifikasi atribut kognitif, pemahaman konseptual, miskonsepsi, dan kesulitan belajar menunjukkan perlunya asesmen diagnostik berbasis data. Penelitian ini bertujuan mengembangkan Three-Tier Multiple Choice Web-Based Diagnostic Assessment berbasis Computer-Based Testing (CBT) serta menghasilkan Integrated Multidimensional Cognitive Diagnostic Framework berbasis Generalized Deterministic Inputs, Noisy “And” Gate (G-DINA) pada materi fungsi eksponensial. Kerangka tersebut mengintegrasikan diagnosis atribut kognitif dan konseptual. Penelitian menggunakan pendekatan Research and Development dengan model 4D. Tahap Define mencakup analisis kebutuhan, kurikulum, domain dan tugas kognitif, penyusunan learning continuum dan hierarki konsep, serta identifikasi empat atribut HOTS dan pengembangan Attribute Hierarchy Model. Tahap Design meliputi penyusunan Q-matrix, instrumen Three-Tier Multiple Choice (3TMC), rancangan diagnosis integratif, dan desain sistem CBT berbasis web. Tahap Develop mencakup validasi pakar, validasi empiris Q-matrix menggunakan Proportion of Variance Accounted For (PVAF) dan Stepwise Wald Test, pengembangan sistem, serta dua tahap pengujian empiris terhadap 600 peserta didik SMA di Daerah Khusus Jakarta dan Sulawesi Selatan. Analisis meliputi perbandingan model, evaluasi kecocokan model dan butir, estimasi parameter, serta akurasi dan konsistensi klasifikasi. Tahap Disseminate mencakup finalisasi produk, penyusunan panduan, serta diseminasi melalui forum akademik dan publikasi. Instrumen menunjukkan kualitas konseptual, psikometrik, dan implementatif yang memadai. Validitas isi ditunjukkan oleh I-CVI sebesar 0,82–1,00 dan S-CVI/Ave sebesar 0,97. Q-matrix memperoleh dukungan teoretis dan empiris melalui konvergensi penilaian pakar, PVAF, dan Stepwise Wald Test. G-DINA merupakan model paling sesuai untuk merepresentasikan pola respons, dengan RMSEA sebesar 0,0569 dan SRMSR sebesar 0,0625. Kualitas keputusan diagnosis didukung oleh Overall Classification Accuracy sebesar 0,884, Attribute Classification Accuracy sebesar 0,921–0,990, Pattern Classification Accuracy sebesar 0,701–0,994, dan Classification Consistency Matrix sebesar 0,701–0,994. Sistem CBT berbasis web dinilai layak secara diagnostik, pedagogis, desain, dan fungsionalitas, dengan I-CVI 0,80–1,00 dan S-CVI/Ave di atas 0,90. Estimasi G-DINA menunjukkan bahwa penguasaan atribut HOTS belum merata. Atribut menganalisis grafik fungsi eksponensial memiliki penguasaan tertinggi sebesar 58,1%, sedangkan atribut membangun dan mengevaluasi model masing-masing sebesar 38,9% dan 39,5%. Profil 0100 merupakan profil paling dominan, yaitu 164 peserta didik (27,33%), yang menunjukkan penguasaan atribut analisis grafik. Diagnosis konseptual 3TMC didominasi Misconception sebesar 34,57%, diikuti Sound Understanding 24,51%, Procedural/Fragmented Understanding 18,52%, Lack of Knowledge/Guessing 12,14%, dan Partial Understanding 10,23%. Keunggulan integrasi tampak pada profil 0100: meskipun G-DINA mengklasifikasikan 164 peserta didik menguasai atribut analisis grafik, diagnosis 3TMC menunjukkan bahwa 132 peserta didik di antaranya (80,49%) mengalami miskonsepsi. Temuan ini menegaskan bahwa penguasaan atribut secara psikometrik belum selalu identik dengan pemahaman konseptual yang benar dan stabil. Integrasi G-DINA dan 3TMC menghasilkan enam profil kesulitan belajar. Fragmented Mastery merupakan profil paling dominan sebesar 48,67%, diikuti Mixed/Needs Qualitative Review 29,67%, Stable Mastery 15,00%, Consistent Non-Mastery 5,83%, Hidden Understanding 0,67%, dan Pseudo-Mastery/Fragile Mastery 0,17%. Mayoritas peserta didik memiliki penguasaan atribut parsial dan struktur pemahaman yang belum terintegrasi. Penelitian ini menghasilkan Integrated Multidimensional Cognitive Diagnostic Framework yang mengintegrasikan HOTS, G-DINA, 3TMC, dan CBT berbasis web untuk memetakan penguasaan atribut, pemahaman konseptual, miskonsepsi, dan profil kesulitan belajar sebagai dasar remediasi berbasis atribut, pembelajaran berdiferensiasi, perubahan konseptual, pembelajaran mendalam, dan pengambilan keputusan berbasis bukti. ***** The low levels of numeracy competence and higher-order thinking skills (HOTS) among senior high school students, together with the limitations of score-oriented assessments in identifying cognitive attribute mastery, conceptual understanding, misconceptions, and specific learning difficulties, indicate the need for data-driven diagnostic assessment. This study aimed to develop a Three-Tier Multiple Choice Web-Based Diagnostic Assessment delivered through computer-based testing (CBT) and to establish an Integrated Multidimensional Cognitive Diagnostic Framework based on the Generalized Deterministic Inputs, Noisy “And” Gate (G-DINA) model for exponential functions. The framework integrates cognitive attribute diagnosis with conceptual diagnosis. This study employed a research and development approach using the 4D model: Define, Design, Develop, and Disseminate. The Define stage comprised needs, curriculum, domain, and cognitive-task analyses; the construction of a learning continuum and conceptual hierarchy; the identification of four HOTS attributes; and the development of an Attribute Hierarchy Model. The Design stage included the construction of the Q-matrix, the Three-Tier Multiple Choice (3TMC) instrument, the integrated diagnostic scheme, and the web-based CBT system. The Develop stage involved expert validation, empirical Q-matrix validation using the Proportion of Variance Accounted For (PVAF) and Stepwise Wald Test, system development, and two phases of empirical testing involving 600 senior high school students in the Special Capital Region of Jakarta and South Sulawesi. Data analysis included model comparison, model- and item-fit evaluation, parameter estimation, and classification accuracy and consistency analyses. The Disseminate stage comprised product finalization, preparation of user manuals, and dissemination through academic forums and scholarly publications. The instrument demonstrated adequate conceptual, psychometric, and implementation quality. Content validity was supported by I-CVI values ranging from 0.82 to 1.00 and an S-CVI/Ave of 0.97. The Q-matrix received theoretical and empirical support through the convergence of expert judgment, PVAF, and the Stepwise Wald Test. G-DINA was the most appropriate model for representing students’ response patterns, with an RMSEA of 0.0569 and an SRMSR of 0.0625. Diagnostic decision quality was supported by an Overall Classification Accuracy of 0.884, Attribute Classification Accuracy values of 0.921–0.990, Pattern Classification Accuracy values of 0.701–0.994, and a Classification Consistency Matrix ranging from 0.701 to 0.994. The web-based CBT system was also judged feasible in terms of diagnostic, pedagogical, design, and functional aspects, with I-CVI values of 0.80–1.00 and an S-CVI/Ave above 0.90. G-DINA estimation showed uneven mastery of the HOTS attributes. The attribute related to analyzing graphs of exponential functions had the highest mastery rate at 58.1%, whereas constructing and evaluating exponential models had mastery rates of 38.9% and 39.5%, respectively. Profile 0100 was the most prevalent mastery profile, comprising 164 students (27.33%) who were psychometrically classified as having mastered graph analysis. The 3TMC conceptual diagnosis was dominated by Misconception at 34.57%, followed by Sound Understanding at 24.51%, Procedural/Fragmented Understanding at 18.52%, Lack of Knowledge/Guessing at 12.14%, and Partial Understanding at 10.23%. The advantage of integrating the two approaches was particularly evident in profile 0100: although G-DINA classified 164 students as having mastered graph analysis, the 3TMC diagnosis revealed that 132 of them (80.49%) exhibited misconceptions. This finding confirms that psychometrically estimated attribute mastery does not necessarily correspond to conceptually correct and stable understanding. The integration of G-DINA and 3TMC produced six learning-difficulty profiles. Fragmented Mastery was the most prevalent at 48.67%, followed by Mixed/Needs Qualitative Review at 29.67%, Stable Mastery at 15.00%, Consistent Non-Mastery at 5.83%, Hidden Understanding at 0.67%, and Pseudo-Mastery/Fragile Mastery at 0.17%. Most students therefore demonstrated partial attribute mastery and conceptually unintegrated knowledge structures. This study produced an Integrated Multidimensional Cognitive Diagnostic Framework that combines HOTS, G-DINA, 3TMC, and web-based CBT to map attribute mastery, conceptual understanding, misconceptions, and learning-difficulty profiles as a basis for attribute-based remediation, differentiated instruction, conceptual change, deep learning, and evidence-based instructional decision-making.

Item Type:	Thesis (Doktor)
Additional Information:	1). Prof. Dr. Iva Sarifah, M.Pd. ; 2). Dr. Riyadi, M.T.
Subjects:	Pendidikan > Evaluasi Pendidikan
Divisions:	PASCASARJANA > S3 Penelitian Dan Evaluasi Pendidikan
Depositing User:	Khaerun Nisa' .
Date Deposited:	31 Jul 2026 07:24
Last Modified:	31 Jul 2026 07:24
URI:	http://repository.unj.ac.id/id/eprint/67782

Actions (login required)

View Item