The IMMERSE Fellowship program is a one-year fellowship experience aimed at supporting education researchers integrate mixture modeling to address critical research questions. Generally funded by the Institute of Education Studies (IES; R305B220021), selected fellows will receive indepth training on how to use mixture models as well as one year of continued support to ensure fellows feel supported while engaging in their own research using mixture models.
In early May, the fellows will start their online training. The objective is to introduce the program and lay a foundation for collaborative, reproducible data science foundation which will serve them throughout the year long training. During these pre-training workshops, we will introduce the structure of the training program, introduce the staff and trainees, and an introduction to data science (Day 1). Following that, we will continue data science topics (Days 2-4) and an introduction to Mplus and Mplus Automation in R (Day 5) led by an advanced graduate student. Pre-training workshops will consist of 20 hours of synchronous and asynchronous online instruction (4 hours of instruction, 5 days). Fellows will be given their own copy of Mplus for use during the one year training. The topics are as follows:
- Day 1: Introductions, training goals, introduction to data science
- Day 2: Collaborative, reproducible workflows with R, RStudio, git and GitHub
- Day 3: Data science principles: data organization, storage, sharing, and code
- Day 4: Data wrangling, exploration and visualization
- Day 5: Introduction to Mplus and MplusAutomation in R
Modern data science requires researchers to work together, using workflows and tools that facilitate computational reproducibility, project management, and collaboration. However, researchers with domain expertise often lack formal training in data science, which can lead to non-reproducible analyses lacking clear organization and documentation, and without a good system for sharing and reviewing code with others. In pre-training workshops, fellows will build essential skills and good practices for working with data (e.g. how to store and organize datasets), organizing projects and building reproducible workflows, exploring and analyzing data (e.g., in scripted code and/or R Markdown), and collaborating and managing projects with git and GitHub.
The in-person training will be three full days of workshops for the 12 training fellows in June (2 cohorts in total) held at UCSB. Workshops will be held in a large classroom in the Gevirtz School of Education (GSE), which has an adjoining computer lab (see letter of support from the GSE Dean). The fellows will stay at The Club, a hotel on the campus of UCSB and walking distance to the GSE, which also includes a free breakfast and an afternoon happy hour. Training fellows will be encouraged to arrive in Santa Barbara no later than the Sunday afternoon, traveling into the airport in Santa Barbara (SBA), and stay through Wednesday afternoon.
Workshop instruction will be a mixture of lecture and activities from Drs. Karen Nylund-Gibson and Katherine Masyn. Instruction will be a combination of lecture-based delivery of content, structured cooperative learning activities, and hands-on lab sessions led by graduate students. Throughout these sessions, a substantial focus will be on the practice of mixture modeling, i.e., the systematic, effective, purposeful, and principled application of mixture modeling to real data. Mastering these skills is arguably the greatest learning challenge. To support the learning goals, the lab sessions (described below) will provide and demonstrate step-by-step guidelines and troubleshoot common issues when they arise.
The workshop timeline for the in-person portion includes a combination of lecture, lab, and consultation. Each day we have a working lunch either with consultation with workshop personnel or where fellows are grouped together based on overlapping interests (either in modeling or research area). This intentional use of lunch as working meetings is to support networking and collaboration amongst the fellows, as well as and between fellows and grant personnel (including the graduate students). Fostering collaborative relationships is a critical component of this “wrap around,” training model, where the training fellows feel comfortable asking for support and project staff know the individual needs and wants of each fellow.
After the initial training at UCSB, participants will be engaged in continued training throughout the year. This will be in continued methods training, mentoring meetings with project staff, brown bag talks, consulting on statistical analysis from graduate students, and one-on-one mentoring from experts in their identified field.
Brown-bag Talks/Mentoring Sessions. These talks will be held virtually, once a month September through May. They will include talks from identified consultants who will give a talk on a project in their respective area that used mixture modeling and how they think the models could be used for future work. Any fellows who are in their related area will also have an opportunity to network and consult with these individuals (see letters of support), providing feedback on their ideas and projects. For example, Dr. Danielle Harlow, is a science educator who has used admixture modeling in research programs, will give a brown-bag talk on her research using mixtures models and will be available to consult with fellows interested in science education.
Methods Training. While the pre- and in-person workshops will have provided the foundational knowledge and resources to run models, continued methods training will be provided throughout the training year, taught by the Drs. Nylund-Gibson, Maysn, and Ing. Topics may include advanced covariate/distal outcome, graphical presentation of results, using GitHub for effective collaborations, and the presentation of new modeling developments. These will be both synchronous and asynchronous resources, depending on the month and will be available to fellows when they are ready to access them.
Ongoing Virtual Consulting. Recognizing the importance of ongoing support throughout the analysis process, we will provide one year of support to facilitate the use of the modeling in their own research. This includes meeting one-on-one with the leadership team (including graduate students) to discuss research questions, support with data management and coding questions, or troubleshoot data problems dealing with their own data. Each month, there will be “office hours” that the fellows can sign up for. These online meetings will be with project personnel (Nylund-Gibson, Masyn, or Ing) or the graduate students. Depending on the needs of the fellow in a given month, they can decide what office hours to sign up for, and once signed up the nature of what is discussed at the office hours. For example, an office hour with a graduate student may help with data coding and merging data files and an office hour with Dr. Nylund-Gibson could include conceptualizing the larger model or to talk through modeling results. While each fellow will have a primary mentor from project staff, they are able to meet with any available project staff.