| Parallel test papers are widely used in large-scale educational examinations,which can solve the problem of test fairness caused by different testing times or locations.Digitalization is promoting a new round of development and transformation in the field of education.In this context,automated test assembly are constantly mentioned,and relevant research is increasingly prominent.Compared to manual test assembly,automated test assembly reduces the human burden and improves the quality of test paper generation.The core of automated test assembly is algorithm.This study focuses on a heuristic algorithm that has been proven to be effective in test paper formation: the maximum priority index method.Currently,the maximum priority index has good results in generating test papers in the context of a single content item bank,but the method is not effective in generating test papers in the context of a multi content item bank.Most studies only focus on dichotomously scored items,with very limited scope for expansion.In addition,this method is only applicable to generating test papers with the parameters of each item in the sample paper,and cannot generate test papers without the parameters of each item in the sample paper.Therefore,this study intends to develop Maximum Priority Index-Control methods(MPI-C and MPI-C *)in an attempt to address the above issues.This article is divided into four studies: Study 1: Conduct mathematical analysis to verify the effectiveness of the two new methods,Maximum Priority Index Control(MPI-C)and Maximum Priority Indicator Control*(MPI-C*),in generating test papers in the context of multiple content item banks and without sample test,respectively;In study 2,a simulation study was conducted to explore the effectiveness of Maximum Priority Index Control(MPI-C)in generating test papers under the influence of multiple factors,including dichotomously or polytomously scored items,item bank size,and constraint form;Study 3: Through a simulation study,explore the effectiveness of the Maximum Priority Index Control*(MPI-C*)method in generating test papers under the influence of multiple factors,including scoring forms and constraint forms;Study 4: Using empirical data to further verify the effectiveness of the MPI-C* method in generating test papers in reality.The main conclusions of the study are as follows:(1)From the perspective of mathematical theoretical analysis,the two new methods(MPI-C and MPI-C*)constructed are reasonable.(2)Under the condition of multi content item bank test paper formation,MPI-C method has a better overall performance than MPI method.(3)When there are no parameters for each item in the sample paper,MPI* and MPI-C* can effectively form the parallel papers,and the overall performance of MPI-C* is better than MPI*.(4)The MPI-C and MPI-C* methods can be applied to test assembly of polytomously scored items. |
