Stephanie, Gabriella Gloria (2019) Studi Simulasi: Perbandingan Item Response Theory Dan Classical Test Theory Pada Ability Estimation. Sarjana thesis, Universitas Brawijaya.
Abstract
Penelitian ini bertujuan untuk menguji sejauh mana item response theory (IRT) dan classical test theory (CTT) dapat menghasilkan ability estimation dengan akurat. Peneliti melakukan studi simulasi dengan pendekatan IRT dan CTT untuk menggambarkan studi pada data empiris. Kondisi yang diterapkan pada simulasi adalah panjang tes berjumlah 10 item dan 20 item menggunakan dichotomous scoring model pada dependent data set. Sedangkan pada data empiris, data set yang digunakan memuat pola respons Reading Skill Assessment dan Numeracy Skill Assessment dari 2006 Kenya National Adults Literacy Survey dengan jumlah 14.761 responden yang kemudian disesuaikan dengan ukuran sampel pada studi simulasi, N = 1,000. Ability estimation dilakukan dengan menghitung nilai maximum likelihood estimation IRT (θ) dan penjumlahan total score CTT (TS) untuk masing-masing data set. Analisis Pearson’s product moment dilakukan untuk melihat korelasi hasil ability estimation yang diperoleh menggunakan masing-masing metode. Nilai korelasi yang diperoleh adalah ρθ = .678 dan ρTS = .704 pada data simulasi Tes 10-item; ρθ = .784 dan ρTS = .846 pada data simulasi Tes 20-item; serta ρθ = .307; ρTS = .447 pada data KNALS 2006.
English Abstract
The purpose of this research is to find out to which extent do item response theory (IRT) and classical test theory (CTT) estimate ability accurately. The researcher conducted a simulation study involving the IRT and CTT approaches to illustrate the study on empirical data. The condition implied on the simulation involved test length of 10 and 20 items respectfully with a dichotomous scoring model on dependent data sets. Whereas the empirical data study used data sets that contain response patterns of the Reading Skill Assessment and Numeracy Skill Assessment from the 2006 Kenya National Adults Literacy Survey, involving 14, 761 respondents, which were then adjusted to the sample size in the simulation study, N = 1,000. Ability estimation was calculated with the maximum likelihood estimation of IRT and total score calculation of CTT for each data set. Pearson’s product moment analysis was done to see the correlation between the obtained ability estimation values of each method. The correlation values obtained from this study are ρθ = .678 dan ρTS = .704 on the simulation data of 10-items Test; ρθ = .784 dan ρTS = .846 on the simulation data of 20-items Test; and ρθ = .307; ρTS = .447 on the 2006 KNALS data.
Other obstract
-
| Item Type: | Thesis (Sarjana) |
|---|---|
| Identification Number: | SKR/FISIP/2019/987/051910633 |
| Uncontrolled Keywords: | item response theory, classical test theory, ability estimation, studi simulasi, item response theory, classical test theory, ability estimation, simulation study |
| Subjects: | 100 Philosophy, parapsychology and occultism, psychology > 150 Psychology > 150.2 Miscellany > 150.287 Testing and measurement |
| Divisions: | Fakultas Ilmu Sosial dan Ilmu Politik > Psikologi |
| Depositing User: | Budi Wahyono Wahyono |
| Date Deposited: | 18 Nov 2020 02:09 |
| Last Modified: | 18 Nov 2020 02:09 |
| URI: | http://repository.ub.ac.id/id/eprint/176479 |
Actions (login required)
 |
View Item |