Wednesday, 24 May 2017

Menganalisa Hasil Ujian Pilihan Ganda dengan IRT (Item Response Theory)

Status : Draft


Output IRT menghasilkan dua cluster informasi, theta siswa dan item parameter.

Θ, theta, adalah tingkat kepandaian siswa.

IRT modelling dapat menggunakan satu sampai dengan 3 parameter, parameter A, B, dan G.

  1. Parameter B : Item difficulty, difficulty parameter. Parameter ini menunjukkan, seberapa mudah atau sulit sebuah item. Paramater B digunakan dalam one-parameter (1P) IRT model.
  2. Parameter A = Discrimination parameter. Nilai parameter ini, menginformasikan seberapa efektif sebuah item dapat membedakan siswa yang pandai dan kurang pandai. Two-parameter (2P) IRT model menggunakan parameter A dan B.
  3. Parameter C   = Dikenal juga dengan nama G parameter atau guessing parameter. Nilai ini memberi informasi kepada kita, seberapa mungkin siswa menjawab suatu item dengan cara menebak. Model yang menggunakan parameter A, B dan G, disebut three-parameter (3P) IRT model. 
Untuk menentukan item difficulty dan item discrimination kita bisa menggunakan verbal terms untuk mempermudah membedakannya;

Item difficulty dapat dibagi menjadi beberapa level : 
No Verbal label Typical value
1 very easy -2.625
2 easy -1.5
3 medium 0
4 hard 1.5
5 very hard 2.625

Sedangkan item discrimination dapat dibagi menjadi beberapa level sebagai berikut :
No Verbal label Typical value
1 none 0
2 low 0.4
3 moderate 1
4 high 2.1
5 perfect 999

Untuk item discrimination, verbal label-nya, dapat kita bagi sebagai berikut :
No Verbal label Range of values Typical value
1 None 0 0.00
2 Very Low 0.01 - 0.34 0.18
3 Low 0.35 - 0.64 0.50
4 Moderate 0.65 s.d 1.34 1.00
5 High 1.35 s.d 1.69 1.50
6 Very High > 1.70 2.00
7 Perfect +∞ +∞

ICC, Item Charasteristic Curve 


A typical item characteristic curve, sumber : The Basics of Item Response Theory Using R, Hal 13


IIF, Item Information Function 
TIF, Test Information Function
Odd ratio, 
Logit,
Item-Person Map (IPM)
Misfit, outliers
Misfit indices,
Model fit,
Infit Standardized Residuals,
Outfit Standardized Residuals,
Regression Analysis
Systematic Bias
Chi-square,
Item Fit,
Infit Mean Squared,
Outfit Mean Squared,
Degress of  freedom (df),
Parallel coordinate,
Person fit,



IRF, Item Response Function,
TRF, Test Response Function,

Referensi

  1. A Simple Guide to the Item Response Theory (IRT) and Rasch Modeling Chong, Ho Yu, Ph.Ds, https://www.creative-wisdom.com/computer/sas/IRT.pdf
  2. Introduction to Item Response Theory, https://www.slideshare.net/NathanThompson54/introduction-to-item-response-theory-70872990
  3. The Basics of Item Response Theory Using R, by Frank B. Baker, Seock-Ho Kim, https://amzn.to/2UM849e
  4. psych: Procedures for Psychological, Psychometric, and Personality Research, https://cran.r-project.org/web/packages/psych/index.html
  5. ltm: Latent Trait Models under IRT, https://cran.r-project.org/web/packages/ltm/index.html
  6. IRT workshop Spring 2014, https://github.com/cddesja/IRTS2014
  7. Introduction to IRT Using R (2PL), https://wnarifin.github.io/simpler/irt_2PL.html
  8. How should we handle missing responses?, https://www.researchgate.net/post/How_should_we_handle_missing_responses
  9. 7 Ways To Handle Missing Data, https://measuringu.com/handle-missing-data/
  10. Evaluating Performance of Missing Data Imputation Methods in IRT Analyses, http://ijate.net/index.php/ijate/article/view/549
  11. link: IRT Separate Calibration Linking Methods, https://cran.r-project.org/web/packages/plink/index.html
  12. A method for designing IRT-based item banks, https://research.utwente.nl/en/publications/a-method-for-designing-irt-based-item-banks
  13. WrightMap Tutorial - Part 1, http://wrightmap.org/post/80523814110/wrightmap-tutorial-part-1
  14. Data Analysis Using Item Response Theory Methodology: An Introduction to Selected Programs and Applications., https://digitalcommons.library.umaine.edu/cgi/viewcontent.cgi?article=1019&context=psy_facpub
  15. PSYCHOMETRICS, Item Response Theory (Part 1), https://rpubs.com/castro/156912
  16. PSYCHOMETRICS, Item Response Theory (Part 2), https://rstudio-pubs-static.s3.amazonaws.com/156125_d88cf281bd2546c294c519afb2577768.html#
  17. PerFit: Person Fit, https://rdrr.io/cran/PerFit/
  18. Person fit assessment using the PerFit package in R, https://www.tqmp.org/RegularArticles/vol12-3/p232/p232.pdf
  19. HOW DOES DIRECTOR FINANCIAL LITERACY INFLUENCE FINANCIAL MONITORING?, https://eprints.qut.edu.au/213554/1/Jacqueline_Bettington_Thesis.pdf

Friday, 13 January 2017