Workshop 2021

The workshop on learning analytics 2021 is planned as a one-day workshop taking place during the DELFI 2021 conference. This year’s workshop is cooperating with the research cluster D2L2 (FernUniversität Hagen) and is focusing on learning analytics considering student diversity with regard to assessment data and discrimination. In addition, other relevant topics on learning analytics can be discussed.  We are currently accepting submissions to the workshop. Please refer to the call for papers for further details. Deadline extended until June 27, 2021.

We are very happy to announce Professor Ryan Baker as our keynote speaker talking about Algorithmic Bias in Education. Please refer to the program for further details on the keynote and the initial program.

Access the workshop proceedings of the 2021 DELFI workshops.

Workshop topic

As student diversity in higher education is increasing this needs to be also considered in learning analytics approaches. The term diversity in higher education is used in the context of underrepresented groups but refers also to students‘ individual differences regarding for instance prior education, learning strategies and motivation, and preferences.
To offer the increasingly diverse students the support needed, learning analytics are considered as meaningful as they aim at offering feedback and at identifying students at-risk for timely interventions. However, to offer valid and actionable feedback to learners data collection and analysis need to be guided by theory. As assessments determine what students are learning it should not be focus on tasks that are easy to analyze but on valued learning outcomes. To measure complex skills more interdisciplinary efforts are required to integrate assessment models and research with learning analytics.
The validity of the data collection and interpretation is particularly crucial as learning analytics are used for identifying students at-risk which might have negative impact on student motivation or self-efficacy. Therefore, potential discrimination of increasingly diverse students need to be considered in the validation of the analyses. Just as serious as an inappropriate intervention is the lack of intervention, whereby a student does not receive the support needed to increase learning success. For at-risk predictions, the technology applied as well as the variables used are important. With regard to discrimination it is vital which variables are used for the algorithms. Most common variables used are grades, internal assessments and demographic data such as age, gender, family background.

Potential research questions:
Current research might investigate how available assessment models need to be adapted to be integrated into holistic learning analytics and what this means for the learning design of courses. How do current approaches deal with incomplete data in their models and potential self- selection biases of students agreeing to divulge personal data? How are biases of algorithms and discrimination considered in current learning analytics systems in different countries? To what extent are the stakeholders of learning analytics aware of biases and how can this be fostered? Is diversity considered in learning analytics interventions without fostering discrimination but still supporting individual needs? Which variables are necessary to model student diversity without discrimination in learning analytics? How can interdisciplinary communication and research be fostered?

Goals of the workshop

• Networking of the learning analytics community in the German speaking countries to initiate research and joint projects as well as fostering network activities beyond the workshop
• Presentation of current research projects and results
• Enhancing the interdisciplinarity in the working group
• Developing a joint outcome based on the workshop discussions (e.g., publication, digital learning resources, events)