Di.ge.LL (Phase 2)

In this project, an existing digital learning environment (dialog-based comic lab with elements of scientific discovery learning through simulations) is being further developed. In the learning environment, students are to acquire procedural knowledge (instruction phase 2) based on declarative-conceptual knowledge of the density concept (instruction phase 1). The target group is students in grades 5 and 6.
For example, the following research questions will be investigated: Depending on the knowledge acquired in instructional phase 1, which self-explanation prompts (open or assistive) are most effective in promoting the subsequent acquisition of procedural knowledge in instructional phase 2? Are there differential effects of self-explanation prompts on procedural knowledge depending on the acquired declarative-conceptual knowledge? Is the effect of self-explanation prompts on the knowledge application success mediated by the process of cognitive activation? Is the effect of self-explanation prompts on knowledge application success mediated by the use of the variable control strategy?
In order to investigate these and other research questions, three studies will be conducted: 1. pilot study (learning environment and instructional conditions located therein will be tested), 2. impact study (effects of the prompt conditions on knowledge application success will be tested in an experimental design) and 3. mediation study (learning actions triggered by the prompt conditions will be recorded process-based via log files and eye tracking with regard to cognitive activation and strategies and analyzed as mediator variables).

Anthropogenic climate change and its consequences are among the greatest challenges facing us and future generations. Due to the complexity of the climate system and the associated cause-and-effect network for explaining climate change, the topic has rarely been addressed in primary and secondary school (grades 5 and 6). In order to close this research gap, a simulation is used which includes basic interactions between environmental influences, human factors and the consequences of climate change. However, since there is a high degree of variance in students’ learning success after learning with the simulation (some achieve the learning objective, others do not), systemic thinking must be deepened in a subsequent instructional phase. Based on the SOLO taxonomy (Biggs & Collis, 2014), students should first work with the simulation in a self-directed phase in order to acquire predominantly structural knowledge (including knowledge of important elements of the climate system). In a convergent instruction phase, this knowledge should be further developed into relational and even extended abstract target knowledge (understanding the interrelationships between elements of the climate system, making predictions about the global rise in temperature).

Competent handling of data and the ability to make data-driven decisions are key learning objectives of mathematics and statistics education in the 21st century (OECD, 2023). Box plots are particularly important in middle school teaching: they integrate numerous descriptive parameters such as the median, but also measures of variability such as the range and interquartile range, into a single form of representation. At the same time, however, this makes them a complex and challenging subject to learn (Bakker et al., 2004, Edwards et al., 2017). When interpreting the box area in particular, learners regularly make a systematic error when they assume a proportional relationship between the area and the sample proportion represented, as they know from many familiar statistical representations such as bar or pie charts (Lem et al., 2013, Abt et al., 2025). Traditional teaching and textbooks often focus primarily on procedural knowledge—for example, how to create a box plot—and less on the acquisition of conceptual knowledge and the necessary conceptual change (Vosniadou & Skopeliti, 2013) regarding the meaning of the box area. We are therefore investigating how this conceptual knowledge acquisition can be supported in a digital learning environment and how possible systematic errors can be addressed through targeted refutation of the underlying misconceptions. To this end, we are conducting intervention studies in which we examine, for example, the benefits of adaptive feedback during the learning phase or whether dynamic refutations are more effective for learning than static ones. The goal of the research project is to provide teachers with an empirically validated digital learning environment that can be used in statistics classes when teaching box plots.

Price formation in competitive markets is a central concept in economic education (Retzmann et al., 2010). However, learners often find it difficult to link the interaction of individuals in markets with the common economic representation of price formation in price-quantity diagrams, or to understand price formation as a systemic interaction (Franke, 2024). Classroom experiments such as the apple market experiment can help overcome such learning hurdles if they are followed by an instructional phase in which the learners’ concrete experiences are specifically translated into abstract knowledge about price formation (Weyland, 2016; Loibl et al., 2024).

Intervention studies examine a three-phase approach with increasing abstraction (cf. Concreteness Fading: Fyfe et al., 2014; Fyfe and Nathan, 2019). The research interest lies in particular in determining which instructional methods can be used to initiate effective learning processes in a digitally enriched teaching setting. Our hypothesis is that the effectiveness of instruction is moderated by the learners’ prior knowledge.

When comparing the sizes of fractions, students often resort too quickly to a component-by-component comparison strategy, i.e., they only compare numerators or denominators with each other, as they are accustomed to doing with natural numbers (Meert et al. 2010). In order to counteract this natural number bias (Ni & Zhou 2005) in the area of size perception by initiating a concept change (Vosniadou & Verschaffel 2004), appropriate instructional methods are needed that reduce the likelihood of activating natural number concepts in relational situations and instead focus on activating appropriate proportion concepts. Refutations, in which errors are explicitly labeled as incorrect and replaced with correct concepts or procedures, have proven to be an effective instructional support for stimulating such a concept change process (Kendeou et al. 2024). It is not yet sufficiently clear whether, in the acquisition of the fraction concept, such learning from errors through the development of negative knowledge (Oser & Spychiger 2005) actually leads to higher learning outcomes than instruction that does not address errors but systematically builds correct conceptual knowledge (“fraction magnitude,” Siegler et al. 2011). This project aims to gain deeper knowledge about how to effectively address natural number bias in the area of size perception. The focus is on the question of whether building conceptual knowledge or refuting errors proves to be more effective in counteracting natural number bias and whether the effectiveness also depends on the students’ prior knowledge.In an experimental pre-post design, two different digital instruction tools are used: A refutation tool visualizes typical student errors using discretized strips and shows why a component-by-component fraction comparison is insufficient, thereby initiating learning from errors. An integration tool emphasizes the size aspect of fractions through a continuous representation, incorporating existing informal notions of proportion, such as the idea of a half.

In this subproject, digital learning journals are used as a learning method for reviewing a learning unit (e.g., after watching a learning video). Writing learning journals can promote the activation of cognitive and metacognitive learning strategies as well as learning success and interest in the learning content through mechanisms such as cognitive offloading and the genre-free principle (Nückles et al., 2020). The project investigates the extent to which the use of strategies when writing learning journal entries can be further promoted by individual feedback. Initial findings indicate that feedback can have a particular impact on the extent and quality of elaboration strategies (Nückles et al., 2005; Nückles, T., 2019; Roelle et al., 2011). In addition, we are investigating whether and how generative AI language models (e.g., ChatGPT) can be used to generate feedback. As part of this subproject, learning journals are currently being examined in the subject of educational psychology among high school students.

The subproject uses a cognitively activating digital learning environment to promote the basic skill of character comprehension as a sub-skill of literary learning and comprehension processes among elementary school students (cf. Boelmann/König 2021; Spinner 2006). Against the backdrop of increasingly heterogeneous learning groups in primary education, the learning environment builds on the individual prior knowledge of the students and adapts to accompany them in the learning process. A qualitative research design will be used to examine the extent to which different learning environments lead to different learning gains, depending on the individual prior knowledge of elementary school students .

Specifically, the students conduct an initial reception-based character analysis of the main character in the literary work “Das Monster vom blauen Planeten” (The Monster from the Blue Planet) (cf. Funke / Scholz 2015). Learners with elementary analyses recognize the outward appearances and actions of characters, while learners with higher levels of prior knowledge identify causalities between motives for behavior and actions (motivation for action) or higher-level characteristics synthesized from text-related individual pieces of information (explicit/implicit). The elementary school-specific adaptation of the BOLIVE model (cf. Boelmann / König 2021) provides the diagnostic framework for classifying the individual competence levels of the students. Starting from the individual character analyses, the students work in different, digitally presented adaptive learning environments, which are designed to stimulate cognitive processes in order to further develop and refine the character analyses, revise misconceptions, and construct new schemata. Various variants of sample solutions and self-explanation prompts are used to design the learning environments (cf. Renkl et al. 2006; Renkl et al. 2004; Renkl / Reiss 2002; Schnotz 2001) in order to cater to different levels of prior knowledge and enable competence gains depending on specific prior knowledge and the instruction provided (cf. Cronbach 1957).