Veranstaltungsprogramm

Personalized, individualized, differentiated, and adaptive learning, instruction, and teaching (LIT) are big educational ideas that are an alternative to the prevailing one-size-fits-all educational model of many school systems (Ohanian, 1999). These approaches have grown popular in recent years due to technological innovations and affordances that open up new ways of designing, implementing, and scaling them (Kerr, 2016; Plass & Pawar, 2020). While their meanings seem straightforward, their definitions are actually not clear cut. Consequently, it is also nebulous how these educational approaches relate to and are distinct from each other. Definitions in educational literature range from adaptive as an umbrella term for personalized and individualized instruction (Tetzlaff et al., 2021), to these terms being inseparably connected (e.g., Peng et al., 2019), to defining one term through the other(s) (e.g., Li et al., 2021; Taylor et al. 2021), to those that make no explicit reference to the overlap of these terms and use multiple terms interchangeably (e.g., Gómez at al., 2014; Park et al., 2019; Serra & Gilabert, 2021). As we put high hopes into what these educational approaches can achieve, especially with respect to supporting each student to reach their full potential in increasingly diverse and heterogeneous student populations, we are conducting a multiple phases systematic thematic review to pinpoint the common core and distinguishing features of the four big ideas.

We report on the first phase of the review in which we focus solely on adaptive LIT. We searched the databases ERIC and PsycInfo to identify all peer-reviewed publications about adaptive LIT in primary and secondary education from 2018-2022. We used the software CADIMA (Kohl et al., 2018) to conduct and document the systematic review process. From 605 initial search results, 555 were unique records. Through trained screening of titles and abstracts, we identified 136 relevant publications. We developed and iterated a codebook for full text screening and data extraction to answer our research question if and how adaptive LIT is defined and whether it is explicitly or implicitly separated or linked to the other big ideas. All 136 full papers were coded by two researchers who also annotated text passages that fell under the codes “conceptualization” or “operationalization” using the software Dedoose (2021). Annotated data was extracted, discussed and combined to a final data set of conceptualizations and operationalizations. This data was thematically coded.

Our goal for this work is to find the consensus and discern distinctions among the conceptualizations and operationalizations of adaptive LIT. This analysis approach allows us to present the frequencies as well as rich insights into the core of adaptive LIT as well as its dimensions. We will present the results of our systematic thematic review on adaptive LIT and talk about our next steps to complete the review in its full scope spanning not only adaptive but also personalized, individualized, and differentiated LIT. We argue that this work can encourage researchers to be explicit about the meaning of terms we use to decrease siloing in educational research and increase collaborative knowledge construction. This will allow us to connect research evidence that is generated under various terms and help identify what we already know about adaptive LIT, its common core, and distinct features. Breaking down the big idea into actionable pieces will also support its implementation into practice. When adaptive LIT is discussed as a set of strategies or specific activity structures that are tangible, practitioners will realize the ways in which they already teach adaptively or how they can adopt adaptive practice into their teaching routines. This might ultimately drive change towards a more tailored educational practice.

Integrating educational technology in learning environments has become paramount for subject-specific teaching, as it allows to contribute to students’ cognitive (learning) and meta-cognitive (monitoring accuracy) competencies. A crucial potential of educational technology is the design of adaptive and personalized learning environments (Aleven et al., 2016), as it allows to productively handle the increasing heterogeneity among students in the classroom. However, little is still known about how educational technology should be implemented in classrooms to realize adaptive learning environments. Thus, evidence-based practice examples are needed that demonstrate an effective use of educational technology for adaptive teaching.

Against this background, we adopted a co-design approach (Roschelle et al., 2006) in which teachers and researchers equitably contributed to the design of adaptive teaching environments, combining research evidence with the demands and requirements of educational practice. Within the co-design, we developed eight adaptive teaching units (duration: 3-4 weeks) across central topics and subjects of the German curriculum (e.g., German, Mathematics, Ethics). We followed a quasi-experimental control-group pre-post-(four-week)delayed design with a total of 16 classes and N = 395 secondary school students. To examine the generalizability of adaptive teaching effects, we adopted a ManyClasses approach (Fyfe et al., 2021) to investigate whether the technology-based adaptive teaching units resulted in higher learning outcomes and better monitoring accuracy compared to the control classes which were engaged in business-as-usual teaching units (pre-registered via as.predicted).

We used multiple imputations to handle missing data and used cluster-robust estimation of fixed effect models to account for the correlated error terms within a cluster (i.e., students within classes) but independent error terms across clusters (see Cameron & Miller, 2015). Since data collection has just been finished, we report preliminary results in this submission and will present more extensive results at the GEBF conference. Preliminary results revealed a small effect of the technology-based teaching units compared to the control group regarding students’ learning outcome (b = 0.22, p = .005). No differences regarding students’ monitoring accuracy were found (b = - 0.18, p = .220). However, the adaptive teaching units did not result in lasting learning, since there were no differences among teaching units – neither regarding students’ learning outcome (b = 0.06, p = .781) nor regarding their monitoring accuracy (b = 0.09, p = .505). To unpack these findings, we will explore potential boundary conditions which we will be presented at the conference.

The findings provide first evidence that technology-based adaptive teaching lessons may be a fruitful approach to address students’ heterogeneity and to improve their learning. In this context, we regard the co-design as an effective approach to investigate research questions in “the wild” as it combines research evidence with the needs and experiences of educational practice. Our teaching units have increased students’ learning but did not result in lasting learning. Even though students participated in the teaching units for three to four weeks, there were no differences regarding students’ learning outcomes in the technology-based adaptive teaching units or the control groups in the four-weeks delayed posttest. More research is needed on how adaptive teaching can result in lasting learning. All in all, our research will contribute to a better understanding of adaptive teaching processes in authentic contexts and illustrates that co-designs can be a successful approach to productively investigate research questions in classrooms.

Learning in digital environments requires self-regulation. While these environments allow learners to study more flexibly, learners typically receive less instructional support than they would in a classroom. For self-study to be successful, a choice of effective learning activities is crucial (Dunlosky et al., 2013). One such learning activity is the distribution of study over multiple sessions (distributed practice; Cepeda et al., 2006; Dunlosky et al., 2013). However, studies have shown that younger students in particular tend to make little use of this strategy (Dirkx et al., 2019). Therefore, interventions are needed that have long-lasting effects and can be integrated into students’ daily lives.

Mobile technology allows interventions to be integrated into daily life and intervention prompts to be repeated over time for lasting impact. Such interventions could take the form of reminders instructing students to study. However, since failure to start and maintain goal striving is a volitional problem even among adults (Gollwitzer & Sheeran, 2006), it seems reasonable that younger students would struggle in that regard as well. Simple reminders might not be enough to bridge the gap between intention and action. We therefore employed a volitional strategy – implementation intentions (Gollwitzer & Sheeran, 2006) – to help younger students use a vocabulary learning app more regularly.

Implementation intentions are if-then plans that link a situational cue to a goal-directed action (Gollwitzer, 1999). In forming implementation intentions for distributed practice, students need to determine when and where to study. Not all students will share an ideal situation that should cue studying. This means that the plans should be personalized – students should create their own plan based on what works best for them.

In an intensive longitudinal study, we compared the effects of simple reminders and implementation intentions on children’s study behavior. The sample consisted of N = 130 fifth graders (Mage=10.75 years). The students were asked to use both a study app created by us and a vocabulary learning app for 37 days. This allowed us to base our analyses on objective study data. The study app included daily questionnaires as well as the intervention. All students first watched a video explaining the benefits of distributed practice. A full intervention group additionally was asked to create their own implementation intention building on situations regularly occurring in their lives. Over the course of the following 36 days, they were regularly reminded of both the benefits of distributed practice and their implementation intention. The reminder group only watched the video on distributed practice and received the reminder on the benefits of distributed practice but did not form an implementation intention. The plan group created a personalized implementation intention but received no reminders at all.

Children in the full intervention group studied more often than those in the other two groups (𝛘2(2) = 7.78, p = .020). In the groups that received reminders, children were more likely to study on days when they received a reminder (b = 0.87, 𝛘2(1) = 18.50, p < .001). When looking into the development of students’ study behavior over time, we found that the likelihood that students studied decreased faster in the group that did not form an implementation intention than in the other two groups (b = -0.05, 𝛘2(1) = 7.38, p = .007; b = -0.04, 𝛘2(1) = 4.26, p = .039). Forming a personalized implementation intention thus buffered the decrease in likelihood to study.

Overall, all groups seemed to profit from our intervention. However, the personalized implementation intentions based on children’s individual lives had a longer lasting positive effect than the simple distributed practice reminder.