Measurement Model Test And Improvement Of Cooperative Problem Solving For Junior High School Students

In the 21st century,the non-collaborative work of a single person is no longer enough to solve complex problems,and people prefer to solve problems together through collaboration and communication with other team members.Therefore,cooperative problem solving is widely recognized as a very important 21st century skill and one of the key student competencies that needs special attention in the current educational reform.The "Evaluation of Middle School Students' Cooperative Problem Solving Ability" research group has explored and developed its own measurement model of cooperative problem solving since 2017 to evaluate Chinese students' cooperative problem solving ability.Based on the evaluation model of cooperative problem solving ability(CPS at Group Discussion,referred to as CPSaGD)developed by the research group of "Evaluation of Cooperative Problem Solving Ability of Middle School Students",this paper adds data from different schools and different task scenarios.Firstly,the data were randomly selected for two rounds of manual trial coding.During the process,the measurement indicators of the original framework were continuously organized,which improved the readability of the indicators and the operability of the measurement tool CPSaGD.The generalization theory is used to conduct a random double-sided crossover design,which proves the consistency of manual coding.At the same time,the number of raters who can maximize the consistency is analyzed under the premise of considering the cost.In semi-automatic CPSaGD design of convolution neural network evaluation method,on the basis of considering the context of the text data word order problems,increase both short-term and long-term memory neural network and two-way short-and long-term memory neural networks,by comparing the three results of neural network training,found a more suitable model in this paper,the automatic coding of data-two-way short-and long-term memory neural networks,The automatic coding model was used to replace the original convolutional neural network model,and the semi-automatic scoring method was optimized to improve the measurement accuracy of CPSaGD.Cooperative problem solving ability is a potential trait.In order to test the measurement effect of CPSaGD,a many-facet Rasch model was constructed to analyze the degree to which students' ability was differentiated by different indicators,different time periods and different groups from three aspects of index,time and group,thus further improving the measurement analysis of cooperative problem solving ability.