Adaptív tesztek minimális hosszának, hibájának, értékelési szintjének és a megoldók számának összefüggései - általános megoldási aránnyal
Loading...
Date
2021
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
THE RELATIONSHIPS OF THE MINIMUM LENGTH, ERROR, EVALUATION LEVELS AND NUMBER OF RESPONDENTS OF ADAPTIVE TESTS WITH GENERAL SOLUTION PROBABILITY Judit Karász T., Szabolcs Takács Correlations between the minimum length, error, evaluation level and number of solvers of adaptive tests - with overall solution rate Our article basically aims to create a general formalism that can be used to address one of the problems of digitization of paper-and-pencil-based tests. The developed formalism was constructed in a general way, but the results are also described in a practical example. In our article, we present the limitations and benefits of a future adaptive version of a survey based essentially on a paper-and-pencil test, such as the OECD PISA [8] and the National Assesment of Basic Competencies [9]. In the case of international measurements, the development work that has prepared this direction has already begun, and the recent online educational experience has paved the way for the transition to digital and then adaptive measurement in the case of the National Assesment of Basic Competencies. The necessary item-level logistics (task bank, modeling environment) are basically available. However, the literature generally covers these technical details - as well as how many people are needed for reliability. In our article, we were not interested in this question, because the itembank will be of sufficient size, or typically in a national measurement, the sufficient filling will be given by default. In contrast, the most important question for us is to what extent the test can be shortened to a test that is the same - or better, more reliable than the current one. So we are not interested in the number of respondents, but in the length of the test, we performed simulation calculations based on the commonly published indicators of the current [8], [9] measurements.