| The interdisciplinary,authenticity and competency-based nature of STEM education pose multiple challenges for teachers to carry out STEM learning design.STEM teachers are faced with the dilemma of lack of learning design competence,which leads to many problems in teaching practice.Improving the STEM learning design competence of pre-service teachers is one of the ways to deal with practical problems.In this context,this study designed a collaborative STEM learning design project,which improves the STEM learning design competence of pre-service teachers through the practice of STEM learning design in real situations.This study aims to analyze the cognitive and non-cognitive performance of pre-service teachers in collaborative STEM learning design based on online discussion discourse data,and then explore how to effectively guide pre-service teachers to carry out STEM learning design.To this end,this study constructed a backward design-based STEM learning design model supported by a virtual internship platform.A total of 33 pre-service teachers(divided into eight groups)participated in the collaborative STEM learning design project,and 9707 utterance data and eight STEM learning design reports were collected.In order to analyze the cognitive performance of pre-service teachers in collaborative STEM learning design,this study compiled a coding framework of STEM learning design competence from the theoretical and practical perspectives,and explored the cognitive performance of groups and individuals by statistical coding frequency and analyzing the network and development trajectory of STEM learning design competence.In terms of groups,according to the score of the STEM learning design report,the eight groups were divided into the high-competence group and the low-competence group.The results showed that the high-competence group had higher coding frequency of the STEM learning design dimension than the low-competence group,paid more attention to the analysis of learning context than the low-competence group,and could maintain the consistency of the STEM learning design better;while the STEM curriculum designed by the low-competence group is often lacking discipline integration,and lack of planning in the collaboration process,there is a clear gap between the high-competence group and the low-competence group in the midstage of the project.At the individual level,according to the number of utterances,the individuals were divided into core participants,medium participants and marginal participants.The results showed that individuals with different levels of participation contribute to STEM learning design,but there were significant differences in collective responsibility consciousness,which was reflected in that marginal participants rarely actively promote collaborative design.In addition,marginal participants have a dual tendency to cope with tasks and interests in discussions,while core participants seldom played a role in facilitating discussion.Moreover,this study found that some groups had a tendency of homogenization of members,and the discussion within the group was lack of doubt.These results can provide empirical evidence for promoting the transformation from low-competence group to high-competence group and improving the initiative of marginal participants.What influences the collaborative STEM learning design includes not only cognitive factors,but also non-cognitive factors.In this regard,based on the constructed non-cognitive online discussion coding framework,this study analyzed the noncognitive performance of pre-service teachers from social,emotional and metacognitive dimensions.The results showed that :(1)Some groups accounted too much low level social interaction,lack of active interdependence within the group,and there were less mutual help behaviors such as spontaneous questioning.(2)Some groups created a positive cooperative atmosphere under the collective contribution of the group,but some groups had the phenomenon that too much negative emotion speech of certain individuals hindered the in-depth discussion;(3)There was insufficient application of metacognitive strategies in each group,especially the lack of monitoring of discussion progress,adjustment of problems,evaluation and reflection,and the efficiency and quality of discussion needed to be improved.The analysis of the performance of STEM learning design in this study reveals the problems existing in the cognitive and non-cognitive aspects.Based on this,this study proposes five strategies to promote cognition development,including concrete design situation strategies,design consistency strategies,learning scaffolding strategies,discipline integration strategies,cognitive conflict strategies,and three guided noncognitive development strategy,including strategies to enhance collaborative identity,strategies to avoid emotional conflicts,and strategies to cultivate metacognition.This study has reference significance for the evaluation of STEM learning design performance and the cultivation of STEM learning design competence. |