Students’ Well-Being in Digital Learning Environments: A Multilevel Analysis of Sixth-Graders in Comprehensive Schools

Abstract

Despite increasing digitalization of schools, the impact of digital learning environments on students’ well-being at school remains insufficiently understood. This study examines how individual- and class-level characteristics of digital learning environments affect sixth-graders’ well-being (N = 1033; 45 classes in Germany). Data were collected using standardized questionnaires measuring students’ well-being at school, digital competencies, digital media use in school, perception of gaining digital skills at school, and teachers’ use of digital tools in the classroom. Multilevel analyses were conducted to assess the influence of individual and class-level factors. Results indicate that students’ perceived gains in digital skills in school are positively linked to their well-being at school, while digital teaching practices focused on individualization have a small negative effect. Other aspects of digital learning environments, such as the frequency of digital media use and additional dimensions of teaching quality, did not show significant effects. Home resources (e.g., piano, cars) were positively linked to students’ well-being at school. These findings underscore the importance of fostering the development of skills, rather than simply increasing media usage to support students’ well-being at school. The study also suggests that individualized digital differentiation in classrooms should be implemented thoughtfully to avoid potential negative social consequences.

1. Introduction

Although digitalization in schools continues to advance, the impact of digital learning environments on students’ well-being at school remains poorly understood. The surge in digitalization—accelerated by the COVID-19 pandemic—has resulted in schools being better equipped with digital media, particularly tablet computers for students (Lewalter-Manhart et al., 2023). Digitalization in general, but especially in schools, brings about changes not only in communication and social structures but also in how learning content is visualized and delivered, thereby affecting the entire classroom (Jusufi et al., 2024; KMK, 2021; Pettersson, 2021; Timotheou et al., 2023). However, despite these advancements, the use of digital media in German schools remains comparatively low on an international scale (Eickelmann et al., 2019). Moreover, digital skills are increasingly recognized as essential for participation in 21st-century society and are implemented in both international and national educational standards (European Commission, 2019; International ICT Literacy Panel, 2002; OECD, 2019). Consequently, teachers are expected to empower their students to effectively use digital media and digital information (KMK, 2016) as school in the 21st century can be seen as a digital learning environment. Nevertheless, both students’ self-assessed competence in using digital media (Jusufi et al., 2024) and their objectively measured digital competence remain rather low in Germany and have declined over time, although they still remain significantly above the international average (Eickelmann et al., 2019).

The ongoing digitalization of education not only transforms how teaching and learning are organized but also shapes students’ everyday experiences at school. As digital technologies become more deeply integrated into educational settings, it is crucial to understand their broader implications beyond academic achievement, particularly their impact on students’ well-being at school, which is vital for in motivation, engagement, and personal development.

While previous research has primarily focused on technological infrastructure and students’ digital competencies, the wider effects of digital learning environments on students’ emotional and social experiences at school have received less attention. This article seeks to address this gap by examining how specific elements of digital learning environments—such as students’ perceived acquisition of digital skills, the frequency and quality of digital media use in classroom settings, and digital teaching practices including individualization and structuring—are related to various dimensions of students’ well-being at school. The focus lies on understanding both cognitive–evaluative and affective–social aspects of students’ well-being at school within the broader context of students’ school experience.

2. Theoretical Background

2.1. Students’ Well-Being at School

The concept of students’ well-being at school, as defined by Hascher (2004a), is closely linked to the theory of subjective well-being and rooted in students’ school experiences (Hascher, 2023). Both constructs are multidimensional rather than unidimensional (Hascher & Hagenauer, 2011; Lucas & Diener, 2015), relating to the evaluation of one’s quality of life (Ryff, 1995).

Students’ well-being at school includes both cognitive and affective evaluations of school experiences (Eder, 1995; Lucas & Diener, 2015). It is important to distinguish between positive and negative evaluations, as the overall balance of these determines individual well-being (Diener, 1994). Joy and happiness are considered key components (Hascher et al., 2018). According to Hascher (2004b), well-being at school includes positive attitudes toward school, enjoyment of attending school, academic self-concept, absence of worry or physical discomfort in the school context, and a sense of social inclusion within the class and school structures. Additionally, both individually self-related conditions and contextual factors within the school environment contribute to well-being (Obermeier, 2021). On the individual level, factors such as personality and academic buoyancy are relevant (Hoferichter et al., 2021), while at the school level, social relationships, teaching quality, and organizational conditions (Holzer et al., 2024) play significant roles. Additionally, out-of-school factors, such as socioeconomic background, can also influence students’ well-being at school (Hadjar & de Moll, 2022).

Accordingly, we define students’ well-being at school as a multidimensional construct that includes cognitive–evaluative and affective–social components of the school experience. Based on Hascher’s (2004b) definition, this construct encompasses students’ perceived ability to manage school demands (e.g., concentration, learning success), emotional comfort and safety within the school environment, positive attitudes toward school attendance, and a sense of social belonging in the classroom.

2.2. (Digital) School Environment

A supportive, structured, and motivating school environment fosters positive emotions and cognitive evaluations of school (Hascher & Mori, 2024). An increasingly important aspect of the school environment is the integration of digital media in teaching (KMK, 2016; Schmid et al., 2023). Digital media not only influences learning processes but also shapes social interactions at school (Timotheou et al., 2023), which in turn affect students’ well-being (Donoso et al., 2021; Mourlam et al., 2020). In this context, both the quality of digital instruction and students’ digital competencies are crucial.

2.3. Digital Competencies

Digital competencies are regarded as key transversal functional skills essential for learning in both school and vocational settings, as well as for social participation (SWK, 2025). Young people, in particular, need digital competencies to navigate increasingly media-rich environments in a critical and autonomous manner. The ability to use digital tools for information retrieval, communication, problem-solving, and participation is considered as a key component of future-oriented basic education (Krumsvik, 2011; SWK, 2025).

There are many different ways to define and use the terms digital literacy, information and communication technology (ICT) literacy, digital competencies, and digital skills, with some being used interchangeably (Godhe, 2019; Tinmaz et al., 2022). The international framework for ICT literacy (International ICT Literacy Panel, 2002) defines ICT literacy across five components, each increasing in complexity of knowledge and expertise. The European Commission (2019) defined digital competence in 2006 as a key competence for lifelong learning, involving “the confident, critical, and responsible use of, and engagement with, digital technologies for learning, at work, and for participation in society” (p. 10). The Digital Competence Framework for Citizens (DigComp; Carretero et al., 2017; Vuorikari et al., 2022) provides a reference framework describing digital competence across five areas and up to eight proficiency levels. Mattar et al. (2022) review various instruments for assessing digital competence based on DigComp-related frameworks, including self-report measures, knowledge-based tests, problem-based questions, and performance assessments. In Germany, the DigComp framework serves a key theoretical reference for the national strategy “Education in the Digital World” (KMK, 2016, 2021).

2.4. Digital Competencies and Digital Media in Education

In Germany, each of the 16 federal states has authority over its own education system, and most use the DigComp framework as a guide for their federal frameworks for media education in schools (e.g., Niedersächsisches Kultusministerium, 2020). However, these frameworks typically describe the dimensions of digital competence without offering robust means for assessment.

Beyond equipping schools with digital media—resources that are sufficiently available in Germany and comparable to international averages and OECD values (Jusufi et al., 2024)—learning specifically about media and how to use it (e.g., for research) and developing digital competence/ICT literacy more broadly is essential (Lewalter-Manhart et al., 2023). The use of digital media is associated with added value for teaching and learning. Scheiter (2021) distinguishes technology-enhanced learning, which focuses on individual learner characteristics, and technology-enhanced teaching, which emphasizes the teacher’s role and the classroom context. She recommends that digital media should always be integrated with didactic planning for effective use, which requires teachers to possess both didactic and technical (including digital) skills.

3. Current State of Research

3.1. Digital Media and Digital Competence at School and Among Young People

Many schools lack a structured and curricular-based integration of digital competence (OECD, 2015). As a result, digital education often relies on the individual initiative of teachers, leading to fragmented and inconsistent digital implementation across schools (Schmid et al., 2023; Starkey, 2020). Despite the widespread availability of technology, digital tools are underutilized in classrooms and are primarily used for basic, receptive, or mechanical tasks such as information searches (Halonen et al., 2017). When shifting the focus from the school or class level to the individual level, significant disparities emerge regarding the frequency of digital media use and levels of digital competence. The International Computer and Information Literacy Study (ICILS) found that German students generally achieve only basic proficiency in essential digital skills, such as evaluating information and problem-solving (Eickelmann et al., 2019). Furthermore, there was a notable decline in the ICT skills of eighth-graders between 2018 and 2023 (Eickelmann et al., 2024). Adolescents often rate their digital skills highly; however, objective assessments consistently reveal significant gaps (van Deursen & van Dijk, 2014). Given this discrepancy, the use of standardized, objective assessment tools is especially important for accurately measuring digital competence.

3.2. Students’ Well-Being in Digital Learning Environments

With digital technologies now integrated into nearly all phases of education, it is increasingly important to understand how digital learning environments affect students’ psychological, social, and school-related well-being. Beyond the general influence of digital environments, research suggests that it is crucial to recognize variations in students’ well-being at school. For example, Hascher and Hagenauer (2011) and Obermeier and Gläser-Zikuda (2022) report a decline in well-being during the transition to secondary school. Gender-specific differences are also evident: adolescent girls more frequently report emotional distress and school-related concerns, such as worries and physical complaints, while boys are more likely to demonstrate disengagement (Lewalter et al., 2023; Obermeier et al., 2021). A multilevel analysis by Konu et al. (2002) found that most of the variance in students’ well-being at school is explained at the individual level, highlighting the importance of personal resources and experiences. However, institutional and organizational factors also contribute, with social support, perceived autonomy, and a sense of belonging emerging as significant predictors. The design of digital learning environments can either strengthen or weaken these factors depending on school culture and the formats of digital interaction.

In the digital context, it is important to differentiate the types of media use to accurately assess their impact on students’ well-being at school, as digital transformation can have positive and negative effects (OECD, 2019). Hietajärvi et al. (2019) found that not all digital activities influence well-being equally: knowledge-oriented activities are associated with greater school engagement, while social networking and action gaming are linked to burnout symptoms and school detachment. The PISA special analysis by Kastorff et al. (2025) highlights the role of digital information literacy in shaping school experiences. Students with high self-efficacy in digital information processing report significantly greater intrinsic learning motivation, perceive school content as more meaningful, and experience less subjective overwhelm. These findings underscore the importance of cross-cutting media-related competencies for students’ well-being at school. From a systemic perspective, there is a clear connection between digitalization and subjective well-being. Donoso et al. (2021), in a study of Chilean students, found that a high level of school digital development as measured by the School Digital Development Index (SDDI) positively correlates with subjective well-being. This index includes three dimensions: technical infrastructure, resource management, and pedagogical integration. Effective and learning-supportive use of technology, rather than mere availability, is crucial for well-being. Boniel-Nissim et al. (2024) found that problematic social media use among school-aged children is associated with lower school-related well-being. Several studies focus specifically on psychological well-being: Dienlin and Johannes (2020) systematically reviewed the varied effects of digital media use on psychological well-being, finding that passive and consumptive usage (e.g., scrolling through social media or viewing algorithmically curated content) tends to negatively correlate with well-being indicators. By contrast, active, creative, and socially interactive usage (e.g., blogging, digital collaboration) is more positively associated with well-being. Vissenberg et al. (2022), in a systematic review, identified digital competence as a factor that promotes well-being, although all studies included in the review measured digital competence through self-report measures. Bahramian et al. (2018) found a small positive association between digital competences and psychological well-being, emphasizing the importance of educational interventions aimed at enhancing young people’s digital skills to strengthen their psychological well-being. Wilmers et al. (2022) present a meta-review of policy-oriented findings on teaching and learning with digital media in German-speaking countries. Their analysis reveals substantial sectoral heterogeneity and a lack of systematically developed, evidence-based models linking digitalization and psychosocial well-being. Nonetheless, pilot projects show promise, particularly in areas such as personalized learning support, competency-based diagnostics, and socio-emotional scaffolding through digital tools. However, students experienced digital media in school in very different ways: while some perceive it supportive, others find it unhelpful or distracting (Labusch et al., 2024).

4. Aims of the Study and Research Question

Existing research on the interplay between the increasing digitalization of schools and students’ well-being at school has primarily focused on either infrastructural factors or individual competencies in isolation, while often overlooking the broader learning environment shaped by digitalization. Yet, digital tools influence not only how students learn, but also how they interact, communicate, and experience daily school life—all of which are central to students’ well-being at school.

Furthermore, digital learning environments are embedded within the social and instructional context of classrooms, indicating that the effects of digitalization may manifest not only at the individual level but also at the collective class level. To date, only a limited number of studies (e.g., Konu et al., 2002) have adopted a multilevel perspective to examine how both individual and shared experiences with digital media are linked to school-related well-being.

In light of these considerations, the present study addresses the following overarching research question: How are individual and class-level characteristics of digital learning environments associated with students’ well-being at school?

To specify this guiding question, we differentiate between predictors at the individual and the classroom level, and formulate the following sub-questions:

• Individual level

• To what extent do individual background characteristics (e.g., gender, home resources, and intelligence) predict students’ well-being at school?
• How are students’ perceived digital skills, their digital competencies, and their frequency of digital media use in school related to their well-being at school?
• How do students’ perceptions of digitally supported teaching practices (e.g., individualization, structuring, cognitive activation, and constructive support) relate to their well-being at school?

• Classroom level

• Are class-level compositional characteristics (e.g., average language use, books at home and home resources) associated with students’ well-being at school?
• Are class-level aggregates of digital learning conditions (e.g., average digital skills, school digital media use and instructional practices) linked to differences in students’ well-being at school?

5. Materials and Methods

5.1. Study Design and Sample

In the German ‘Deisel’ research project, 16 comprehensive schools (‘Gesamtschulen’) were monitored over two school years (2023/24 and 2024/25), with data collected and reported throughout the study. Specifically, 1.033 fifth- and sixth-grade students (50.9% female) aged 10 to 14 years (M = 11.51, SD = 0.61) across 45 classes in Lower Saxony and North Rhine-Westphalia were analyzed at a total of four measurement points, each approximately six months apart (t0–t3) (Nonte et al., in press). Solitary students whose parents provided written consent in advance were eligible to participate. Of the 1207 eligible students, 1055 returned signed parental consent forms, yielding a response rate of about 87.4%. The average class size in this sample is 21.2, compared to the general average of 24 in German comprehensive schools.

The research project was funded by the Federal Ministry of Research, Technology and Space (in German ‘Bundesministerium für Forschung, Technologie und Raumfahrt’ BMFTR; funding label O1UP2217). Employing a multi-method longitudinal cohort design, the study aims to establish a reliable basis for understanding the conditions for successful implementation of, the stability of the effects of, and the sustainability of promoting digital competencies. Data collection included standardized surveys of students and teachers, as well as qualitative–empirical key informant interviews with various stakeholders. The data used in this paper derive from three measurement points: t0 (November 2023), t1 (April 2024), and t2 (November 2024). Students completed standardized questionnaires, while class teachers filled out a student-tracking list (STL), assigning each student an identification number (with first and last names retained at the school to ensure anonymization) and recording age, sex, and any special educational needs. This process ensured an anonymized longitudinal tracking and accurate student assignment.

5.2. Students’ Well-Being at School

The scale measuring students’ well-being at school comprises eight items drawn from established instruments (KINDL-R and KIDSCREEN) and adapted items from the StEG study (German title: Studie zur Entwicklung von Ganztagsschulen, Furthmüller, 2015), as well as self-developed items. This combination enables the assessment of multiple dimensions of well-being, including learning-related aspects (see

Table 1. Detailed description of students’ well-being at school—learning-related components.

Item (Source)	1 (Never)	2 (Rarely)	3 (Sometimes)	4 (Often)	5 (Always)	Mean (SD), n
Last week…
… I managed my schoolwork well overall (adapted from Ravens-Sieberer & Bullinger, 2000).	n = 16	n = 38	n = 201	n = 437	n = 151	3.79
	1.9%	4.5%	23.8%	51.8%	17.9%	(0.85), n = 843
… I was able to concentrate well (adapted from Ravens-Sieberer et al., 2014).	n = 23	n = 108	n = 266	n = 343	n = 96	3.46
	2.8%	12.9%	31.8%	41.0%	11.5%	(0.95), n = 836
… I generally did well with learning (adapted from Ravens-Sieberer et al., 2014).	n = 36	n = 67	n = 200	n = 318	n = 206	3.71
	4.4%	6.3%	24.2%	38.5%	24.9%	(1.17), n = 827

for details). Example items include the following: “Last week, I managed my schoolwork well overall” (KINDL-R, Ravens-Sieberer & Bullinger, 2000) and “Last week, I was able to concentrate well” (KIDSCREEN_10, Ravens-Sieberer et al., 2014). These items were rated on a scale from 1 (“never”) to 5 (“always”). Social–affective components of school experience (see

Table 2. Detailed description of students’ well-being at school—social–affective components.

Item (Source)	1 (Not True at All)	2 (Rather Not True)	3 (Rather True)	4 (Completely True)	Mean (SD) n
How are you personally doing at school right now?
I like being at this school (adapted from Furthmüller, 2015).	n = 110	n = 151	n = 317	n = 292	2.91
	12.6%	17.4%	36.4%	33.6%	(1.00) n = 870
I wouldn’t want to miss out on school lessons anymore (adapted from Furthmüller, 2015).	n = 147	n = 281	n = 290	n = 119	2.46
	17.6%	33.6%	34.6%	14.2%	(0.94) n = 837
I feel comfortable at school (adapted from Furthmüller, 2015).	n = 117	n = 192	n = 297	n = 251	2.80
	13.7%	22.4%	34.7%	29.3%	(1.01) n = 857
I feel safe at school (adapted from Gerecht et al., 2012).	n = 127	n = 215	n = 278	n = 232	2.72
	14.9%	25.2%	32.6%	27.2%	(1.02) n = 852
I enjoy being with my classmates at school (self-development).	n = 60	n = 119	n = 216	n = 462	3.26
	7.0%	13.9%	25.2%	53.9%	(0.94) n = 857

for details) were assessed with the prompt, “How are you personally doing at school right now?” and included items such as “I like being at this school”, “I wouldn’t want to miss out on school lessons anymore”, “I feel comfortable at school” (adapted from Furthmüller, 2015), “I feel safe at school” (adapted from Gerecht et al., 2012), and “I enjoy being with my classmates at school” (self-developed). These were rated on a scale from 1 (“not true at all”) to 4 (“completely true”).

Because the two scales use different response formats, both were z-standardized at the individual level before being combined. The mean of the standardized values was then calculated to create a unified scale. The resulting dependent variable represents a standardized measure of students’ well-being at school, with both subscales weighted equally. The internal consistency of the combined scale is considered good (Cronbach’s Alpha = 0.84).

5.3. Digital Competencies

As part of the ‘Deisel’ project, a Digital Competence Test was developed, drawing on a longitudinal study from the Netherlands (Lazonder et al., 2020) and aligned with the Lower Saxony Media Education Framework (Niedersächsisches Kultusministerium, 2020). The goal was to map digital competencies as broadly as possible across six competence areas and three competence levels and create a reliable and valid test instrument that does not rely on self-reports, as adolescents often overestimate their abilities. The Digital Competence Test was administered longitudinally; this article refers to the second test administration (D2). The test included information-based response tasks, including both closed formats (multiple choice, drag and drop) and constructed response formats, but did not include interactive tasks. Designed as a power test, students were given 35 min (M = 18.53, SE = 4.98) to complete 29 tasks (dichotomously coded; summary score M = 14.1, SE = 3.9, n = 874). The assessment was conducted in the classroom under the supervision of a test administrator. Analysis using the dichotomous Rasch model indicated a normal distribution across all competence levels. After refining the test and excluding certain items, the Weighted Likelihood Estimate (WLE) reliability was 0.66. The WLE values from the Digital Competence Test are used for further analyses in this article.

5.4. Culture Fair Intelligence Test

At the first measurement point (t0), all participating students completed two subscales of the Culture Fair Intelligence Test 20-R (CFT 20-R, Weiß, 2019), a standardized intelligence test. The subscale ‘series completion’ measures the ability to recognize patterns and logical sequences, and the subscale ‘matrices’ evaluates abstract reasoning and the ability to identify logical relationships within a matrix of figures. Both subscales assess non-verbal cognitive abilities. For the subsequent analyses, a WLE value was calculated from both subscales (CFT, EAP/PV reliability = 0.74).

5.5. Digital Learning Environment

At the second measurement point (t1), the student questionnaire focused on digital media use in school from multiple perspectives, focusing on the digital learning environment. Students were asked to what extent they had learned to perform 10 different tasks at school (1 = not at all, 2 = to a small extent, 3 = to a large extent) such as “search for information using digital media” or “work out whether to trust information from the Internet”. These items were combined to form the “Student’s Perception of Gaining Digital Skills at School’ (Perceived Digital Skills) scale (UK2), which demonstrated good internal consistency (Cronbach’s Alpha of 0.82).

Another scale, “Frequency of Digital Media Use in School” (School Digital Media Use), asked how often they used nine different digital tools during class (HM2, 1 = never, 2 = in some lessons, 3 = in most lessons, 4 = in every or almost every lesson), with examples such as “edit worksheets digitally or do exercises” or “write tests digitally”. This scale also showed good reliability (Cronbach’s Alpha of 0.82).

To assess digital teaching features (DTFs) across four subdimensions, items from Quast et al. (2021)—originally designed for teachers to evaluate the quality of teaching with digital media—were adapted for student responses. Quast et al. (2021) derived four latent factors, based on theoretically and empirically established teaching quality dimensions for classroom management, cognitive activation, and student support (Praetorius et al., 2018): individualization, structuring, cognitive activation, and constructive support.

Students rated how often their teachers used digital media in these ways (1 = not at all, 2 = in very few lessons, 3 = in some lessons, 4 = in most lessons, 5 = in every lesson), with each subdimension measured by three items. The internal consistency for each subscale was as follows: individualization (UM2_a; e.g., “My teachers give students who need more practice additional tasks with digital devices”), Cronbach’s Alpha: 0.80; structuring (UM2_s; e.g., “My teachers use digital media to show us what we should memorize.”), Cronbach’s Alpha: 0.77; cognitive activation (UM2_ka; e.g., “My teachers ask us to present different solutions for tasks using digital media.”), Cronbach’s Alpha: 0.82; and constructive support (UM2_ku; e.g., “My teachers discuss difficult tasks with us using digital media.”), Cronbach’s Alpha: 0.80. The overall DTF scale had excellent reliability (Cronbach’s Alpha: 0.91).

Sociodemographic characteristics

Age and sex were recorded by teachers using a student-tracking list. Sex was coded as binary (SEXD; 0 = male, 1 = female), as the third option (“diverse”) was selected only once. At the second measurement point, students also reported their home resources (WG2; e.g., “At home there … is a piano, … is a person who helps with the housework at least once a week (cleaner), … are two or more cars…” Wendt et al. (2017)) and the number of books at home (B2; 1 = 0–10 books, 2 = 11–25 books, 3 = 26–100 books, 4 = 101–200 books, 5 = more than 200 books; Bos et al., 2008). Additionally, students indicated how often they spoke German at home, with responses coded dichotomously (LANGU_2; 1 = always speak German, 2 = do not always speak German).

All measurement instruments used in this article are documented in

Table 3. Descriptive data of students’ sociodemographic characteristics, well-being at school, competencies, and digital learning environment.

Construct	Level	Item Description	Source	Measurement Point (t0–t2)
Age		M = 11.51, SD = 0.61 (min 10, max 14), n = 955	Student-tracking-list	t0
Sex		Male: n = 468 (49.0%) Female: n = 486 (50.9%) Divers: n = 1 (0.1%)	Student-tracking-list	t0
Home Resources (WG)	Individual Level	t1: M = 6.03, SD = 2.21, (min 0–max 12), n = 884	Adapted from Mullis et al. (2021)	t0, t1
	Class Level	t1: M = 5.99, SD = 0.86, (min 3.27–max 7.43), n = 45
Books at Home (B)	Individual Level	t1: M = 2.89, SD = 1.13 n = 877	Bos et al. (2008)	t0, t1
	Class Level	t1: M = 2.83, SD = 0.49 n = 45
Language Use (German) (LANGU)		Always German: n = 474 (54.5%) Not always German: n = 395 (45.5%)	Beese et al. (2022)	t1
Students’ Well-Being at School (WBL, WBS)	Individual Level	Learning-Related Components (WBL2, t2) M = 3.50, SD = 0.70, n = 791	Adapted from Ravens-Sieberer et al. (2014); Ravens-Sieberer and Bullinger (2000)	t0, t2
		Social-Affective Components (WBS2, t2) M = 3.10, SD = 0.68, n = 815	Furthmüller (2015); Gerecht et al. (2012); self-development	t0, t2
	Class Level	Learning-Related Components (WBL2, t2) M = 3.42, SD = 0.24, n = 45	Adapted from Ravens-Sieberer et al. (2014); Ravens-Sieberer and Bullinger (2000)	t0, t2
		Social-Affective Components (WBS2, t2) M = 2.80, SD = 0.27, n = 45	Furthmüller (2015); Gerecht et al. (2012); self-development	t0, t2
Digital Competencies (D2)		Dichotomy coded and summarized (t2) M = 14.1, SD = 3.9, n = 874 EAP/PV Reliability = 0.66	Self-development based on Lazonder et al. (2020); Niedersächsisches Kultusministerium (2020); Vuorikari et al. (2022)	t0, t2
Intelligence (CFT 20-R) (CFT_WLE)		EAP/PV Reliability = 0.74	Weiß (2019)	t0
Student’s Perception of Gaining Digital Skills at School (Perceived Digital Skills) (UK2)	Individual Level	M = 2.16, SD = 0.39, n = 856	Adapted from Vennemann et al. (2021)	t1
	Class Level	M = 2.17, SD = 0.10, n = 45
Frequency of Digital Media Use in School (School Digital Media Use) (HM2)	Individual Level	M = 2.19, SD = 0.63, n = 856	Adapted from Mullis et al. (2021)	t1
	Class Level	M = 2.21, SD = 0.30, n = 45
Digital Teaching Features	Individual Level	Individualization: M = 1.91, SD = 0.92, n = 839 Structuring: M = 2.69, SD = 0.99, n = 835 Cognitive Activation: M = 2.00, SD = 0.94, n = 828 Constructive Support: M = 2.51, SD = 1.02, n = 823	Adapted from Quast et al. (2021)	t1
	Class Level	Individualization: M = 1.93, SD = 0.42, n = 45 Structuring: M = 2.69, SD = 0.32, n = 45 Cognitive Activation: M = 2.02, SD = 0.38, n = 45 Constructive Support: M = 2.51, SD = 0.34, n = 45

.

5.6. Data Analyses

Descriptive analyses were conducted using IBM SPSS© Statistics (IBM Corp, 2021), while all subsequent analyses were performed with Mplus 8.8 (Muthén & Muthén, 1998–2017). Missing values for all variables except sex were imputed using multiple imputation framework (m = 10). Multiple imputation is a well-established method for handling missing data in multilevel analyses (Grund et al., 2025). In our study, the survey was conducted in classroom settings with research staff present, resulting in a high overall response rate of 87.4% and generally low item-level missingness (e.g., sex: 2%). However, some classrooms experienced interruptions due to time constraints, leading to substantial item-level missingness in certain variables, most notably the frequency of digital media use in school (39%). Despite this, case-level missingness remained low overall, and numerous auxiliary variables with minimal missing data were available to inform the imputation model. This supports the use of multiple imputation even for variables with higher missingness rates (Graham, 2009). Given the controlled data collection process, we assumed the missingness mechanism can primarily be classified as Missing at Random (MAR), justifying the application of multiple imputation techniques (Rubin, 2018). Nonetheless, it is important to recognize that imputed values are model-based estimates and introduce some degree of uncertainty. Therefore, while the applied imputation approach is considered valid and methodologically sound, results involving variables with higher missingness should be interpreted with appropriate caution.

The predictors were implemented within a multilevel modeling framework in a stepwise manner. At the class level, the dataset included 45 clusters, with an average of 21.2 students per class. This sample size is considered adequate for two-level regression models, which require at least 30 clusters to produce reliable estimates of main effects (Maas & Hox, 2005). The dependent variable and all predictors except sex were standardized (M = 0, SD = 1) to ensure comparability of effects, facilitate interpretation in terms of standard deviation units, and avoid scale-related distortions due to differing metrics (Hox et al., 2017). Accordingly, unstandardized regression coefficients were reported, interpretable as changes in the outcome variable expressed in standard deviation units.

First, an empty model (Model 0) was tested to determine the intraclass correlation coefficient (ICC), indicating the proportion of variance attributable to between-class differences. Next, a random intercept multilevel model (Model 1) was specified, regressing individual-level predictors on the outcome variable while accounting for class-level variation by allowing the intercept to vary randomly across classes. Finally, an intercepts-and-slopes-as-outcomes model (Model 2) was specified. Aggregated class-level variables were included at the between level to explain variance in the intercepts and account for between-class differences in the outcome (Lüdtke et al., 2008).

Model fit was evaluated using the deviance statistic (–2*log-likelihood), AIC, and sample-size-adjusted BIC, with the model exhibiting the lowest values selected for interpretation (Burnham & Anderson, 2004).

6. Results

The following section presents the results of the multilevel analyses which were conducted in a stepwise manner (see

Table 4. Multilevel analysis of students’ well-being at school (n = 955, significant effects in bold, unstandardized betas).

	Model 0		Model 1			Model 2
			b		SE	b		SE
Within (Student)
Sex (Male)			−0.01		0.04	−0.01		0.04
Home Resources			0.09	**	0.03	0.08	**	0.03
Digital Competencies			0.02		0.03	0.02		0.03
Intelligence (CFT)			−0.02		0.03	−0.02		0.02
Perceived Digital Skills			0.13	***	0.03	0.12	***	0.03
School Digital Media Use			−0.03		0.03	−0.03		0.03
DTF Individualization 1			−0.09	**	0.03	−0.10	**	0.03
DTF Structuring 1			0.01		0.05	0.01		0.05
DTF Cognitive Activation 1			0.01		0.04	0.02		0.04
DTF Constructive Support 1			0.00		0.04	0.01		0.04
Between (Class) 2
Language Use						−0.16		0.24
Home Resources						0.05		0.04
Books at Home						−0.02		0.08
Perceived Digital Skills						0.70		0.37
School Digital Media Use						0.04		0.17
DTF Individualization 1						0.03		0.12
DTF Structuring 1						0.26		0.19
DTF Cognitive Activation 1						−0.26		0.19
DTF Constructive Support 1						−0.07		0.20
ICC	0.07				0.07		0.07
R2 (Within)					7%		7%
R2 (Between)							33%
Deviance					1893.46		1883.70
AIC					1919.46		1927.70
aBIC					1941.36		1964.76

¹ DTF = Digital Teaching Feature. ² All predictors as class average; ** p ≤ 0.01, *** p ≤ 0.001.

). As previously described, the dependent variable is a standardized composite score representing students’ well-being at school. Model 0, specified as an empty model, was used to estimate the intraclass correlation coefficient (ICC), quantifying the proportion of variance in students’ well-being at school attributable to differences between classes. The ICC was 0.07, indicating that 7% of the total variance in students’ well-being at school lies between classes.

In Model 1, all individual-level predictors were included. These comprised control variables (sex, home resources, intelligence) as well as aspects of the digital learning environment: students’ digital competencies, frequency of digital media use in school, students’ perception of gaining digital skills at school, and instructional characteristics related to digital media (individualization, structuring, cognitive activation, and constructive support). Among these, only home resources (b = 0.08, SE = 0.03) and students’ perception of gaining digital skills at school (b = 0.12, SE = 0.03) showed significant positive effects, each demonstrating small but statistically significant associations with higher levels of student well-being at school. Conversely, the instructional feature “individualization”—referring to the differentiation of tasks according to students’ strengths and weaknesses—was negatively associated with student well-being at school, also with a small effect size (b = −0.09, SE = 0.03). Although this finding may appear counterintuitive, it will be explored further in the discussion. Overall, the individual-level predictors accounted for 7% of the variance in students’ well-being at school.

Model 2 extended the analysis to include all class-level predictors, operationalized as class averages. This allowed for the examination of compositional effects, such as the overall socioeconomic background within classes (measured by students’ at-home language use, availability of home learning resources, and the number of books at home). Additionally, class-level aspects of the digital learning environment were considered, including average digital competencies, frequency of digital media use in school, and the extent to which digital skills were taught, as well as instructional practices related to digital media (individualization, structuring, cognitive activation, and constructive support). Although no significant class-level predictors of students’ well-being at school were identified, the model explained 33% of the variance at the class level. However, model fit indices (deviance = 1883.70, AIC = 1927.70, aBIC = 1964.76) indicated a poorer fit compared to Model 1. Therefore, Model 1 was deemed the more appropriate representation of the data structure.

7. Discussion

The primary aim of this study was to examine the relationship between students’ well-being at school and digital learning environments at both the individual and class levels. In this context, the digital learning environment was defined as a combination of students’ perception of gaining digital skills at school (Perceived Digital Skills), the frequency of digital media use in school (School Digital Media Use) and digital teaching features (DTFs) with four subdimensions. The findings indicate that the relationship between digital learning environments and students’ well-being at school is shaped predominantly by students’ individual experiences rather than by shared class-level characteristics.

Results show that the extent of gaining digital skills at school has a small but significant positive effect on students’ well-being at school. In other words, the more students perceived that they were gaining digital skills at school, the higher their reported school-related well-being. This finding aligns with Donoso et al. (2021), who highlight the critical role of pedagogical integration. Conversely, it is noteworthy that the individualization of lessons by teachers (digital teaching features—individualization), such as assigning extra tasks to students who are either advanced or need additional practice, was associated with a small but significant negative effect on well-being. This may suggest that students perceive such individual digital tasks as stigmatizing or exposing, especially in classroom settings where differentiation is visible to peers. While this interpretation is speculative, it underscores the importance of considering the social-emotional dynamics of digital differentiation in future research. According to Self-Determination Theory, controlling teaching styles or negative feedback (which may be perceived through additional tasks) can evoke feelings of exposure and shame, thereby negatively impacting students’ well-being at school (Niemiec & Ryan, 2009; Ryan & Deci, 2000).

All other aspects of the digital learning environment—such as the frequency of digital media use in school and the other subdimensions of digital teaching features (structuring, cognitive activation, constructive support)—showed no significant effects on well-being at either the individual or class level. These findings support those of Lewalter-Manhart et al. (2023), who argue that while schools may be well equipped with digital media, the development of digital skills is even more important. The results also align with Boniel-Nissim et al. (2024), who advocate for the teaching of digital skills in school. The only other significant predictor of students’ well-being at school was “home resources”: a greater number of resources at home was associated with higher well-being at school. This is consistent with the findings of Konu et al. (2002), whose multilevel analyses also showed that students’ well-being at school is primarily influenced by individual-level factors, particularly home resources. In this context, the availability of digital devices and other learning-related equipment at home may play an important role in shaping students’ well-being at school, as they can affect both perceived competence and the ability to participate in digital learning activities. No significant effect of sex was found at either the individual or class level. This contrasts with previous studies (Konu et al., 2002; Lewalter-Manhart et al., 2023; Obermeier et al., 2021), but it may be attributable to the younger age of the students surveyed (sixth grade), with significant differences possibly emerging in later adolescence.

Although the multilevel analysis did not show significant effects of students’ digital competences on their well-being at school (see Table 4), small positive correlations were found between digital competences and students’ well-being at school (Appendix A 

Table A1. Correlation table.

	1	2	3	4	5	6	7	8
1. Student Well-Being at School
2. Digital Competencies	0.049 *
3. Intelligence	0.027	0.402 ***
4. Extensive	0.113 ***	−0.003	−0.037
5. Frequency	−0.025	−0.182 ***	−0.167 ***	0.288 ***
6. DTF Individualization	−0.065 **	−0.147 ***	−0.168 ***	0.185 ***	0.392 ***
7. DTF Structuring	0.014	0.012	0.015	0.280 ***	0.290 ***	0.487 ***
8. DTF Cognitive Activation	−0.039	−0.166 ***	−0.194 ***	0.211 ***	0.457 ***	0.669 ***	0.498 ***
9. DTF Constructive Support	−0.010	−0.062	−0.079	0.256 ***	0.315 ***	0.493 ***	0.664 ***	0.597 ***

Notes. All variables are standardized, which means that all means equal 0 and all standard deviations equal 1. * indicates p < 0.05, ** indicates p < 0.01, *** indicates p < 0.001.

), consistent with Vissenberg et al. (2022). Another noteworthy finding is the negative correlation between digital competence and aspects of the digital learning environment: there are small but significant associations between digital competence and both the frequency of digital media use in school and certain digital teaching features (individualization and cognitive activation). This variation likely reflects differences in individual schools’ digital teaching concepts and infrastructure (Tönsing et al., 2025). For instance, ANOVA revealed a significant effect regarding the frequency of digital media use in class F (15, 840) = 11.64, p < 0.001, η² = 0.17, indicating a large effect size. The most substantial significant differences were observed between schools in disadvantaged and privileged areas, with higher media usage reported in disadvantaged schools. Some schools opt not to use digital devices and media in fifth grade, while others introduce daily media use or implement dedicated tablet classes. Thus, both student-level characteristics and perceptions, as well as class-level factors, were considered in the multilevel analysis. However, due to the small number of participating schools (n = 16), school-level effects could not be analyzed. At the class level, no significant predictors of students’ well-being at school were identified. The intraclass correlation (ICC) across all models was small, with only 7% of the variance in students’ well-being at school explained by the predictors included. This suggests that the variance in students’ well-being at school is likely not primarily attributable to the digital learning environment as operationalized here. As Konu et al. (2002) emphasize, social support and a sense of belonging remain key factors.

Limitations and Future Research

Several limitations of the current study should be acknowledged. First, many items—with exception for the standardized tests of digital competencies and intelligence (CFT)—are based on student self-reports. While self-perceptions of sixth-graders may introduce bias, student’s perceptions of gaining digital skills at school or the frequency of digital media use in school may also be subject to misinterpretation. Second, the correlational design of the study limits the ability to draw causal inferences. Longitudinal research would be necessary to track the development of digital competencies from the start of schooling (e.g., Grade 1). Third, neither school-level nor teacher-level data were included in the analysis. Future studies should examine the impact of internal school media policies, digital infrastructure, and teachers’ attitudes toward digital media—areas already being explored in studies such as TIMSS (Trends in International Mathematics and Science Study) (Schwippert et al., 2024).

Future research should expand the current analytical framework to include individual and contextual factors beyond the school setting. While this study focused on digital learning environments within school, the concept of students’ well-being at school, as defined by Hascher (2004b), is influenced not only by school-based experiences but also by out-of-school conditions (Hadjar & de Moll, 2022). In particular, aspects of the home learning environment, including media usage behaviors, parental support, and the availability of digital resources, may shape students’ sense of academic competence, emotional security, and social inclusion, all of which are core dimensions of school-related well-being (Hascher et al., 2018).

Our findings underscore that perceived digital skill acquisition at school is positively associated with students’ well-being at school, but they also show that individual background factors (such as home resources) remain significant predictors. This highlights the importance of adopting a broader perspective that considers interactions between school-based and home-based influences.

Further studies might therefore integrate home digital practices, parental mediation styles, or students’ digital self-efficacy as additional predictors. In this context, Kastorff et al. (2025) emphasize that high digital self-efficacy can enhance students’ motivation to learn—an effect that may also extend to their well-being at school. Such extensions would help to better understand the interplay between school and home environments in shaping students’ well-being and contribute to more targeted approaches in educational practice and digital learning design.

In addition to these considerations, the unexpected negative association between perceived digital individualization and students’ well-being at school observed in our study warrants further investigation. While individualized instruction is typically intended to foster learning by addressing diverse student needs, it may, under certain classroom conditions, be perceived as socially exposing or controlling—particularly when such differentiation is visible to peers. From the perspective of Self-Determination Theory (Ryan & Deci, 2000), this perception may interfere with the satisfaction of students’ basic psychological needs, such as autonomy and relatedness, thereby negatively affecting their well-being at school.

Moreover, this finding contrasts with research emphasizing the potential of pedagogically integrated digital tools to enhance student engagement and emotional outcomes (Donoso et al., 2021; Wilmers et al., 2022). At the same time, studies such as that by Labusch et al. (2024) underline that students’ perceptions of digital media in school are highly diverse—some students perceive digitally supported practices as empowering, while others find them alienating or even stressful. Future research should therefore explore how digital individualization is implemented and perceived in classroom settings, and which pedagogical strategies foster inclusive, supportive experiences rather than feelings of exposure or difference.

8. Conclusions and Implications

This study explored the relationship between students’ well-being at school and digital learning environments at both the individual and class level. The findings indicate that students’ perceived gaining of digital skills is positively associated with their well-being at school, while digitally supported individualized instruction has a small but significant negative effect, potentially due to increased social exposure in classroom settings. Other aspects of digital media use and digital teaching features showed no significant impact on well-being. By contrast, home resources emerged as a significant predictor, consistent with prior research emphasizing the importance of individual-level influences. No significant effects were observed at the class level, and gender was not associated with differences in well-being. Overall, these results underscore the importance of students’ subjective experiences and development of digital skills, rather than the mere frequency of media use, in promoting well-being in school contexts. This analysis represents an important first step in understanding how digital learning environments relate to students’ well-being at school. Despite the growing integration of digital tools in education, meaningful integration and explicit instruction in digital tool use appear to be more relevant for supporting students’ well-being at school. Consequently, greater emphasis should be placed on embedding digital competence education into curricula at both the individual and school levels, beginning in the early grades. Further research is essential and should expand beyond technological infrastructure to include pedagogical approaches and classroom practices. Educational policies prioritize comprehensive teacher training that addresses not only the technical aspects of digital tools, but also the socio-emotional dynamics of digital differentiation in classroom settings. Ensuring equitable access to digital infrastructure and learning resources at home, including devices, internet connectivity, and parental support, should be a central component of educational equity strategies. This will help ensure that digital learning opportunities are not only available but also accessible to all students. By aligning educational policy with the factors that truly influence student well-being, schools can become environments where digitalization supports both academic achievement and the emotional and social development of students.