Gaining a deeper understanding of the meaning of the carelessness parameter in the 4PL IRT model and strategies for estimating it

REID (Research and Evaluation in Education), Journal Year: 2023, Volume and Issue: 9(1), P. 86 - 117

Published: June 30, 2023

Three popular models are used to describe the characteristics of test items and estimate ability examinees under dichotomous IRT model, namely one-, two-, three-parameter logistic models. The three-item parameters discriminating power, difficulty, pseudo-guessing. In development carelessness or upper asymptote parameter was proposed, which forms a four-parameter (4PL) model accommodate condition where high-ability examinee gives an incorrect response item when he/she should be able respond correctly. However, 4PL have not been widely accepted due several factors, people’s understanding that strategies for estimating it is still inadequate. Therefore, this study aims shed light on ideas underlying meaning parameter, based extant literature. focus then extended demonstrating practical examples person using empirical data responses 1,000 students from Indonesia Open University (Universitas Terbuka) 21 30 multiple-choice Business English test, paper-and-pencil test. We mainly analyzed ‘mirt’ package in RStudio. present analysis results coherently so users would sufficient they can model.

Language: Английский

---

Revealing the characteristics of Indonesian language test used in the national-standardized school examinations

REID (Research and Evaluation in Education), Journal Year: 2024, Volume and Issue: 9(2), P. 210 - 222

Published: Jan. 2, 2024

This study aimed to determine the characteristics of Indonesian language test used in national standardized school examinations (Ujian Sekolah Berstandar Nasional, USBN). We response data 218 students from a public senior high Special Region Yogyakarta, Indonesia, consisting two packages, A and B, 2018/2019 academic year investigate its items. Quantitative analysis using classical theory (CTT) one-parameter logistic item (1-PL IRT) model was conducted items based on difficulty discriminating power. The results under CTT showed that most both package B have easy category portion difficult is no more than 10%. In addition, while majority packages demonstrated good power, contained number with poor power (47.5%). Under 1-PL IRT, our indicated dominance level moderate packages. addition revealing items, IRT were related topics types conjunction; spelling, grammar, sentence structure; job application letter; observation report text structure. this are expected contribute improving quality learning where category.

Language: Английский

---

Guesses and Slips as Proficiency‐Related Phenomena and Impacts on Parameter Invariance

Educational Measurement Issues and Practice, Journal Year: 2024, Volume and Issue: 43(3), P. 76 - 84

Published: April 8, 2024

Abstract Traditional approaches to the modeling of multiple‐choice item response data (e.g., 3PL, 4PL models) emphasize slips and guesses as random events. In this paper, an model is presented that characterizes both disjunctively interacting guessing conjunctively slipping processes proficiency‐related phenomena. We show how evidence for perspective seen in systematic form invariance violations slip guess parameters under four‐parameter IRT models when compared across populations different mean proficiency levels. Specifically, higher tend lower probabilities than populations. The results undermine use traditional applications require would suggest greater attention alternatives.

Language: Английский

---

Identification of cheating methods in large-scale tests: Islamic education subject matter in the national examination

Educenter Jurnal ilmiah pendidikan, Journal Year: 2024, Volume and Issue: 3(1), P. 10 - 26

Published: Jan. 25, 2024

This study aims to determine item characteristics based on response theory and identify cheating during the administration of National Standardised School Examination (USBN) for Islamic religious education at senior high school level in 2015/2016 year Yogyakarta. The data source consists 2929 answer sheets from USBN participants subject Religious Education (PAI) under Package A KTSP curriculum. Using a quantitative approach with an ex post facto design, analysis uses person fit test acceptance methods identifying cheating. results show that PAI instrument academic has moderate average difficulty, good information function. Test Acceptance method, which identified 153 cases knowledge sharing, 891 ignorance sharing 222 proved be more effective detecting than Person Fit 2 individuals.

Language: Английский

---

Development and calibration of an instrument measuring attitudes toward statistics using classical and modern test theory

Journal of Pedagogical Research, Journal Year: 2024, Volume and Issue: unknown

Published: June 7, 2024

Language: Английский

---