Maitsa, Nabila (2019) Studi Simulasi: Missing Data Treatment Technique Zero Imputation Dan Non-Imputation Pada Ability Estimation Dengan Model Satu Parameter Logisti. Sarjana thesis, Universitas Brawijaya.
Abstract
Penelitian ini bertujuan untuk melihat estimasi ability ketika menggunakan zero imputation dan non-imputation menggunakan model Satu Parameter Logistik (1PL). Pendekatan yang digunakan pada penelitian ini adalah kuantitatif psikometri serta studi simulasi dengan dua kondisi imputasi zero imputation dan non imputation, dua jumlah butir (20, 40) dan tiga persentase missing data (10%, 20%, dan 50%). Format respons yang digunakan adalah tes dikotomus dimana jika benar diberi skor 1 dan salah diberi skor 0. Teknik analisis yang digunakan adalah Item Response Theory (IRT) model 1PL dengan metode Expected A Posteriori (EAP) untuk mengestimasi ability. Hasilnya, penanganan missing data zero imputation memberikan perubahan pada akurasi estimasi ability meskipun perubahan tersebut tidak terlalu berarti, sehingga estimasi ability akan sedikit lebih baik dan akurat jika data dengan missing data ditangani oleh zero imputation pada kondisi missing data yang besar serta pada jumlah butir yang lebih sedikit. Selain itu hasil dari studi simulasi dan real data menunjukkan kesamaan dimana pada keduanya, estimasi ability akan lebih baik jika missing data ditangani dengan zero imputation.
English Abstract
This study aims to see the difference of ability estimation while using zero imputation and non-imputation with One Parameter Logistic model. The method that were used in this study is simulation study and illustration with real data. The condition used in the simulation study are two imputation condition (zero imputation, non-imputation), two test length (20, 40) and three missing data percentage (10%, 20%, 50%) with dichotomous response choice (0, 1). The data then tested with Item Response Theory (IRT) 1PL model and then the ability were estimated using Expected A Posteriori (EAP). The result of this study indicate that missing data treatment technique zero imputation conduce a small change toward ability estimation, therefore the ability estimation will slightly be better and accurate if the missing data were treated with zero imputation in certain condition, first when the missing data percentage are high and second when the test item’s are low. Furthermore, the outcome of simulation study and illustration with real data indicate a similarity, where on both study ability estimation will slightly be better if missing data were treated with zero imputation.
Other obstract
-
| Item Type: | Thesis (Sarjana) |
|---|---|
| Identification Number: | SKR/FISIP/2019/493/051905324 |
| Uncontrolled Keywords: | studi simulasi, missing data, zero imputation, ability estimation, simulation study, missing data, zero imputation, ability estimation |
| Subjects: | 500 Natural sciences and mathematics > 511 General principles of mathematics > 511.8 Mathematical models (Mathematical simulation) |
| Divisions: | Fakultas Ilmu Sosial dan Ilmu Politik > Ilmu Politik |
| Depositing User: | Nur Cholis |
| Date Deposited: | 24 Jul 2020 09:48 |
| Last Modified: | 08 Mar 2024 03:48 |
| URI: | http://repository.ub.ac.id/id/eprint/171330 |
![[thumbnail of Wisnu Wardhana - 1.pdf]](http://repository.ub.ac.id/style/images/fileicons/text.png) |
Text
Wisnu Wardhana - 1.pdf Download (1MB) |
Actions (login required)
 |
View Item |