The Relationship between Response Time and Probability of Answering~Take the decimal multiplication unit as an example

• Main Authors: Chen Mao-Hsiung, 陳茂雄
• Other Authors: Kuo Bor-Chen
• Format: Others
• Language: zh-TW
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/39aub2

碩士 === 國立臺中教育大學 === 教育資訊與測驗統計研究所 === 105 === In compute-based testing and computerized adaptive test, early studies were less likely to involve the response time, but in fact, response time was a significant source of information. With the advent of the information age, and the growing popularity of online testing, now for the examinees in each item, the response time can be accurately recorded and studied, but the empirical data in this field is still insufficient; besides, most of the studies are about speed tests, which is contradictory to the former idea that the prerequisites of item response theory is assumed to be non-speed related. In this study, the data is collected from 589 fifth graders and the range of the test is the multiplication unit of decimal fraction of mathematics. The examinees use the computer as a testing tool, through the network connecting to the test system. The system does not limit the examinee’s answering time but instead records the actual time-span of their responses in each item, meanwhile, asking them to write down how they figure out the answer on paper. In this study, we use the item response theory to estimate the parameters of the items and the ability of the examinees, analyze the relationship between response time and the probability of answering, explore whether the response time can provide useful information in ability estimation, and further inquire the factors that affect the response time. The result of the study reveals that the distribution of the examinee’s response time approximates to lognormal distribution. The response time is also significantly different because of the item types and the complexity of the answering process. With the degree of difficulty on the items being generally simple, the response time of the examinees shows a highly positive correlation between the number of the words of the items and the number of the words correctly calculated. However, the response time still proves a very low correlation between the probability of answering accurately and the correctness of answering reaction. The correlation between response time and the parameters of items in classical test theory is low, and so is the correlation between response time and the parameters of items in item response theory.