With the movement in education towards collaborative learning, it is becoming more important that learners be able to work together in groups and teams. Intelligent tutoring systems (ITSs) have been used successfully to teach individuals, but so far only a few ITSs have been used for the purpose of training teams. This is due to the difficulty of creating such systems. An ITS for teams must be able to assess complex interactions between team members (team skills) as well as the way they interact with the system itself (task skills). Assessing team skills can be difficult because they contain social components such as communication and coordination that are not readily quantifiable. This article addresses these difficulties by developing a framework to guide the authoring process for team tutors. The framework is demonstrated using a case study about a particular team tutor that was developed using a military surveillance scenario for teams of two. The Generalized Intelligent Framework for Tutoring (GIFT) software provided the team tutoring infrastructure for this task. A new software architecture required to support the team tutor is described. This theoretical framework and the lessons learned from its implementation offer conceptual scaffolding for future authors of ITSs.

This paper describes the development and evaluation of an Intelligent Team Tutoring System (ITTS) for pairs of learners working collaboratively to monitor an area. In the Surveillance Team Tutor (STT), learners performed a surveillance task in a virtual environment, communicating to track hostile moving soldiers. This collaborative problem solving task required significant communication to achieve the common goal of perfect surveillance. In a pilot evaluation, 16 twoperson teams performed the task within one of three feedback conditions (Individual, Team, or None) across four trials each. The STT used a unique approach to filtering feedback so that teams in both individual and team conditions received a similar amount of feedback. In one performance measure, Team condition participants made fewer errors in one task than those in other conditions, though at a potential cost of mental workload. Feedback condition also significantly affected participants' subjective rating of both their own performance and their teammate's. This ITTS is one of the first automated team tutoring systems that provided real-time feedback during task execution. Recommendations are offered for the design of the optimal team task for future ITTSs that offer tutoring for small teams performing collaborative problem solving.

Intelligent Tutoring Systems have been useful for individual instruction and training, but have not been widely created for teams, despite the widespread use of team training and learning in groups. This paper reviews two projects that developed team tutors: the Team Multiple Errands Task (TMET) and the Recon Task developed using the Generalized Intelligent Framework for Tutoring (GIFT). Specifically, this paper 1) analyzes why team tasks have significantly more complexity than an individual task, 2) describes the two team-based platforms for team research, and 3) explores the complexities of team tutor authoring. Results include a recommended process for authoring a team intelligent tutoring system based on our lessons learned that highlights the differences between tutors for individuals and team tutors.

ObjectivesSPECTRUM is a unique multi-disciplinary, cross-sector collaborative research partnership between academics, government, and community organizations in Manitoba working together to address complex social problems. In our first research project, partners identified the need for quantitative evidence around outcomes of children involved with Child Protection Services (CPS), using linked administrative data. ApproachFrom the SPECTRUM partnership, a research team with CPS expertise was established, including government policy-makers, community organizations representing First Nations families, and academics from multiple disciplines. The research is guided by an Advisory Circle of First Nations Knowledge Keepers. Linking health, education, CPS, and justice data we developed a matched cohort identifying children involved with CPS (2007-2018) for whom there was discretion in the decision to 1) place them in out-of-home care (n=19,718), or 2) keep them in their family home while providing services (n=28,154). Instrumental Variable analysis, with CPS agency as the instrument, will be used to compare outcomes. ResultsFollowing the trajectories of these two groups of CPS-involved children over time, we will compare their mental and physical health, educational achievement, and justice system involvement while accounting for individual-level (e.g., age, sex, chronic health conditions) and family-level (e.g., family income, maternal mental health, number of siblings) factors that may contribute to these outcomes. Preliminary findings will be workshopped with the SPECTRUM partnership to facilitate discussions on framing the evidence for policy makers. The SPECTRUM policy team will then prepare policy recommendations for government to consider. The Advisory Circle and a youth advisory squad will facilitate contextualizing and mobilizing findings. Actions taken by government in response to material provided will be monitored and will inform the development of subsequent research projects conducted by SPECTRUM. ConclusionGovernment and community stakeholder involvement throughout bolsters the likelihood of evidence translating into program and policy changes. This is not a situation where academics are telling government how to do their jobs – this is government, community organizations, and academics working together with the shared goal of better outcomes for children.

No abstract