Analisis model simultan model logistik satu parameter dengan waktu respon berdasarkan data simulasi

Abstract

AbstrakPenelitian ini bertujuan untuk menganalisis model simultan model logistik satu parameter (ML1P) dengan waktu respon. Analisis terhadap model menggunakan data simulasi, yang skenario pembangkitan data simulasi dilakukan berdasarkan banyaknya peserta tes (500, 1000) dan banyaknya soal tes (11, 20, 40 ANALYSIS OF SIMULTANEOUS MODEL OF ONE PARAMETER LOGISTIC MODEL AND RESPONSE TIME BASED ON SIMULATION DATA AbstractThe aim of this research is to analyse simultaneous model One Parameter Logistic Model (1-PLM) and respon time. The analysis of model used the simulation data, where the data generation scenario was done based on the number of test takers (500, 1000) and the number of test items (11, 20, 40). Parameter estimation method used the Bayesian method, Markov Chain Monte Carlo. The analysis of model was done with the accounting of the distance of true value and estimated parameter. The Analysis methods use Root Mean Square Error (RMSE), Standart Error (SE) and bias.The result of research reveals the performance of parameter estimation result for the test item (the test item difficulty, test item slowness, and the effort to complete the item test) is not influenced by the number of the test items. However, the performance of parameter estimation result for the test takers (the speed and ability of the test takers) is influenced by the number of the test items. The more test items there are, the closer is the parameter estimation result to the true parameter.

Two Parameter Logistics Model with Lognormal Response Time for Computer-based Testing

“…According to Ingrisone II [16], the developed simulataneous model is suitable for power test. However, in the next study, Hidayah and Kumaidi [17] shows that, in the same amount of response time and the different level of test takers' ability, the probability of answering the test questions correctly is almost the same (speed test). Model of Ingrisone [24] is as follows.…”

“…Research on the development of simultaneous model between response accuracy and response time has not been done much. Some experts who develop this are Roskam [20], Verhelst, Verstralen and Jansen [21], Wang and Hanson [22], Wang [23], Ingrisone [24], Ingrisone II [16] and Hidayah and Kumaidi [17].…”

“…The limitation of Ingrisone II [16] is improved by Hidayah and Kumaidi [17] by implementing the parameter of response time into one-parameter logistic model using joint distribution concept for power test. Hidayah and Kumaidi [17]…”

“…Based on the model above, the test taker's ability is symbolized as , the level of difficulty is symbolized as b, and response time is symbolyzed as . If the value of t is infinite (test does not have time contraint), model of Hidayah and Kumaidi [17] is the same with one-parameter logistic model. The limitation of defining in model of Hidayah and Kumaidi [17] can be used by other researchers as the basis of developing simultaneous model between response accuracy and response time to be a more realistic model.…”

“…Therefore, the development of model considering the response time is continuously done to improve the current existing model to get the most reliable model. Not every response time model developed for power test is proven effective for power test, for instance, Ingrisone II [16] that is intended to be used in power test, according to Hidayah and Kumaidi [17] is proven more suitable for speed test because each of the questions is easy to answer correctly in limited time. This study aim to develop twoparameter logistics model with random variable of response time for power test, whose chosen parameters are limited to the difference of the questions (aj), the difficulty level of the questions (bj), the ability of the test takers (Ɵi), the delay due to the questions (di) and response time (tij).…”