Digital Presence and Learning Success: An Empirical Study on the

Impact of Online Engagement on Conceptual Expertise

Benny Platte

1 a

, Anett Platte

2 b

, Christian Roschke

1 c

and Marc Ritter

1 d

University of Applied Sciences Mittweida, Technikumplatz 17, 09648 Mittweida, Germany

Faculty of Psychology and Education, Ludwig-Maximilians-University, Leopoldstraße 13, 80802 Munich, Germany

Keywords:

Online Attendance, Learning Opportunities, Learning Outcomes, Peer-Driven Insights, Higher Education,

Correlation Analysis, Open-Everything Exams, Taxonomic Levels, Virtual Presence.

Abstract:

In the context of the increasing digitalization of higher education, this study examines the relationship between

topic-related attendance and learning success in a fully online module in the ﬁeld of media informatics. It is

well known that the utilization of learning opportunities in online learning environments is generally correlated

with exam success. In technically oriented online learning environments, we further investigate the extent to

which such a correlation differs depending on the taxonomic levels of tasks.

The results show that the correlation of complex synthesis tasks with r = 0.84 is signiﬁcantly higher compared

to repetition tasks (r = 0.21) and also signiﬁcantly higher compared to analysis and calculation tasks (r =

0.11). Both statements hold at a conﬁdence level of 99 %. While repetition and analysis tasks were less

inﬂuenced by attendance time, students who regularly and actively participated achieved signiﬁcantly higher

scores on tasks requiring deeper cognitive processes.

1 PROBLEM STATEMENT

The digitalization of higher education opens up new

opportunities for ﬂexible learning formats but also

presents challenges, particularly in technically ori-

ented modules. These modules require a close inte-

gration of theoretical understanding and practical ap-

plication, which often leads to a gap between knowl-

edge acquisition and actual applicability in digital for-

mats.

The module “Audio – Video – Real-Time Net-

works” exempliﬁes this issue, as it requires both

a solid understanding of fundamental concepts and

their practical implementation. Students often report

difﬁculties in transferring theoretical concepts to real-

world application scenarios and recognizing the un-

derlying structures (Bicak et al., 2023).

Reﬂections on online modules indicate that stu-

dents who actively and regularly participated in online

sessions performed signiﬁcantly better in demanding

tasks. However, speciﬁc studies are lacking on the ex-

https://orcid.org/0000-0001-7754-5170

https://orcid.org/0000-0001-7958-1618

https://orcid.org/0008-0007-2875-5199

https://orcid.org/0009-0004-0204-8275

tent to which online attendance, particularly in mod-

ules with analytical and synthesis-based tasks, inﬂu-

ences student performance quality.

In particular, the relationship between online at-

tendance and the successful completion of complex

synthesis tasks compared to simpler repetition tasks

remains underexplored. A detailed investigation

could provide new insights into the effectiveness of

Figure 1: Main ﬁnding of the results: Correlation coefﬁ-

cients with conﬁdence intervals CI

and CI

for all task

types 1 , 2 , and 3 . Negative values are not deﬁned in

principle. The visualization including negative values is

purely for illustrative purposes and should be interpreted

accordingly. Areas with negative values are drawed using

a dashed pattern and whited out to emphasize this aspect.

Platte, B., Platte, A., Roschke, C. and Ritter, M.

Digital Presence and Learning Success: An Empirical Study on the Impact of Online Engagement on Conceptual Expertise.

DOI: 10.5220/0013472200003932

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 17th International Conference on Computer Supported Education (CSEDU 2025) - Volume 2, pages 893-905

ISBN: 978-989-758-746-7; ISSN: 2184-5026

893

digital teaching methods.

This study aims to address this research gap by ex-

amining how online attendance inﬂuences students’

ability to handle different cognitive demands. The

ﬁndings will contribute to identifying effective teach-

ing and learning strategies for digital formats in tech-

nical disciplines.

2 STATE OF RESEARCH

2.1 Theoretical Framework and

Deﬁnition of Terms

The term presence derives from the Latin “praesen-

tia” (attendance) and traditionally refers to physical

co-presence in a seminar room. However, since the

widespread adoption of digital teaching formats dur-

ing the Covid-19 pandemic (2019), “presence” also

includes virtual attendance. A distinction is made be-

tween “synchronous” and “asynchronous” participa-

tion: Synchrony refers to simultaneous attendance in

a virtual space, regardless of physical location. Asyn-

chronous participation means attending at different

times (Fiedler et al., 2022, p. 2).

Hybrid teaching models combine both ap-

proaches. “Blended learning” integrates online and

in-person teaching. In contrast, “hybrid teaching”

refers to a format where some students are phys-

ically present while others participate online syn-

chronously.

2.2 Impact of Social Presence on

Learning Success

Social presence in online learning environments

refers to the extent to which learners perceive them-

selves as actively engaged. It has been shown to cor-

relate with higher course satisfaction and better per-

formance (Decius et al., 2021). The absence of fa-

cial expressions and gestures in virtual settings can

reduce social presence and negatively impact learn-

ing outcomes (Baumann, 2022, p. 73); (Orvis et al.,

2010). At the same time, studies indicate that tar-

geted, even short-term interactions can enhance social

engagement and foster a sense of community among

learners (Andel et al., 2020).

This implies that online courses should be de-

signed to actively promote social presence—through

interactive formats and adaptive learning environ-

ments. Integrating innovative technologies can help

create a supportive online learning community that

accommodates diverse learning needs.

2.3 Interaction and Learning Outcomes

Students demonstrably beneﬁt from intensive peer in-

teraction and reﬂective group activities, which lead to

higher engagement and improved performance (Kuhn

et al., 2021, p. 40); (Gr

uner, 2018, p. 95). Peer group

discussions, in particular, enhance critical reﬂection

skills (Thielmann & B

ockelmann, 2021, p. 114).

This underscores the need to actively promote so-

cial interaction and connectivity in virtual learning

environments to sustainably enhance learning expe-

riences and outcomes.

2.4 Learning in Online Environments

Digitalization enables ﬂexible access to learning con-

tent and interactive elements that enhance understand-

ing. Studies show that interaction and engagement are

key factors for learning success, whereas the course

format itself has no inﬂuence on actual learning suc-

cess (“keinen Einﬂuss [...] auf den tats

achlichen Lern-

erfolg hat” (Klug & Seethaler, 2021, 1f)). Therefore,

maximizing learning opportunities is crucial. In full-

day online courses, maintaining variety and motiva-

tion is particularly important (Ketter et al., 2022).

Successful online mentoring requires openness to

technology-supported communication and can even

accelerate collaboration processes (Baumann, 2022,

p. 356);(Chrobak, 2004, p. 58).

2.5 Self-Directed Learning in Online

Courses

Even before the pandemic-driven transition in

2019, self-directed learning was extensively studied.

Sharples describes learning as a continuous process

of negotiation and exploration (Sharples, 2005, p. 6).

Students who employ effective self-learning strate-

gies achieve better results in online courses, empha-

sizing the importance of learning strategies and time

management.

A high sense of self-efﬁcacy promotes the use

of self-learning strategies, which can be further sup-

ported by digital platforms (Pelikan et al., 2021,

p. 394), (Baumann, 2022, p. 102). Context manage-

ment and accompanying support are crucial success

factors (L

odermann, 2024, p. 80).

2.6 Factors Inﬂuencing Learning

Success

Examinations are primarily identiﬁed as a factor

that signiﬁcantly inﬂuences student learning success

CSEDU 2025 - 17th International Conference on Computer Supported Education

894

(Heinzel et al., 2020, p. 43). In addition, external con-

ditions that cannot be inﬂuenced by instructors, such

as proximity to home, individual time frames, and

peer networking (see above), also play an important

role. Hussner et al., like Koegler et al., demonstrated

that the perceived workload of online teaching, com-

pared to regular teaching, has increased (Hußner et

al., 2022), and that students’ stress perception is high

(Besa et al., 2022, p. 7), while at the same time, time-

and location-independent learning in online courses

allows for greater individualization compared to face-

to-face courses (Kossack & Bender, 2022, p. 9).

The arrangement and function of online units, as

well as the type of online tools used, can also inﬂu-

ence learning success. Here, the speciﬁc function of

each unit is crucial, particularly whether it serves as

a mere replacement for face-to-face teaching or as

an independent learning unit (Towﬁgh et al., 2022,

p. 18).

2.7 Task Taxonomies

To compare the learning success of different task for-

mats, the impact of various teaching methods must be

considered. Learning success can only be assessed

based on content that has been explicitly taught ac-

cording to Constructive Alignment. It is therefore

necessary to demonstrate that the required competen-

cies for exam tasks could have been acquired. The use

of digital teaching methods depends on factors such as

the number of sessions, didactic design, and curricu-

lar integration (Lerner & Luiz, 2019).

In scientiﬁc and technical disciplines, perfor-

mance assessment through various task types is essen-

tial. Taxonomies such as SOLO, Anderson & Krath-

wohl’s revision of Bloom’s Taxonomy, and Webb’s

Depth of Knowledge (DOK) classify cognitive de-

mands in teaching and learning processes. All models

require extensive knowledge of the underlying litera-

ture (Biggs & Tang, 2011, p. 47).

Type 1 : Repetition Tasks. Aiming to consolidate

basic knowledge and routine skills, these tasks range

from unistructural to multistructural requirements ac-

cording to the SOLO taxonomy. Anderson & Krath-

wohl assign them to the lower levels of the cog-

nitive process dimension, “remember” and “under-

stand” (Wilson, 2016, p. 2). In the knowledge di-

mension, they correspond to “Factual Knowledge”

(Krathwohl, 2002, 215ff).

Type 2 : Analysis Tasks. Requiring the applica-

tion of concepts for problem-solving, these tasks

range from multistructural to relational requirements.

According to SOLO, they promote relational under-

standing by recognizing and utilizing relationships

between concepts. Anderson & Krathwohl classify

them at the higher levels “Apply” and “Analyze”

(Bay, 2018; Krathwohl, 2002).

Type 3 : Synthesis Tasks. Requiring deep under-

standing and the ability to apply knowledge in new

contexts, these tasks reach a relational to extended

abstract level according to SOLO. They correspond

to the highest stages in Anderson & Krathwohl’s tax-

onomy (“Evaluate” and “Create”) and the top lev-

els of Webb’s DOK, which involve critical thinking,

planning, and project development (Krathwohl, 2002,

p. 215).

3 RESEARCH DESIGN

This chapter outlines the general approach to in-

vestigating the relationship between online participa-

tion/engagement and conceptual expertise in the con-

text of a course module on real-time media networks.

The aim is to gain speciﬁc insights into how students’

online presence inﬂuences their ability to handle com-

plex conceptual, analytical, and creative tasks.

3.1 Research Question, Hypotheses

This paper examines the following question, which is

differentiated into three hypotheses listed in Table 1:

“Does the performance of students in solving com-

plex conceptual and synthesis tasks increase dis-

proportionately compared to simpler repetition and

calculation tasks with the online attendance time

in courses conducted entirely online in a technical

module? ”

3.2 Methods

Structured Content Delivery. Learners gain an ad-

equate understanding of the module content through

interactive teaching methods and appropriate content

delivery.

Logical Sequence and Traceability of Content.

As part of the module, students are actively involved,

and targeted questions and discussion prompts are

used to foster critical reﬂection on the topic.

Appropriate Taxonomy. A suitable taxonomy is

chosen to classify learning objectives and tasks for as-

sessing the difﬁculty level.

Digital Presence and Learning Success: An Empirical Study on the Impact of Online Engagement on Conceptual Expertise

895

Topic-Focused Days. Allow for in-depth explo-

ration of speciﬁc topics and improve the structuring

of the learning process. Each block focuses on a spe-

ciﬁc topic.

Increasing Complexity Levels by Demand Areas

within each cohesive session block. Knowledge trans-

fer progresses through application, analysis, and eval-

uation to support the learning process in a targeted

and documentable manner.

Interactive Competency Development through

the use of teaching methods that not only promote

knowledge acquisition but also build on analysis and

evaluation to support the development of synthesis

skills.

Taxonomy. For this study, Anderson and Krath-

wohl’s Revision of Bloom’s Taxonomy was selected as

the most suitable classiﬁcation (Krathwohl, 2002). It

covers a broad cognitive spectrum and enables precise

learning objectives. In Table 3 , the exam tasks are

assigned to the corresponding primary and sublevels

of the competency dimensions.

3.3 Data Collection

A quantitative approach was used for data collec-

tion. The online attendance times of students were

pseudonymized to assign participation to the differ-

ent topic areas. Additionally, the examination results

Table 1: Hypotheses, each formulated as a null hypothesis

) and corresponding alternative hypothesis (H

were divided into the three named cognitive require-

ments (see Task detailiation in Table 3) and recorded

in points. Individual scores from eight tasks were col-

lected for the examination.

4 IMPLEMENTATION OF THE

MODULE

4.1 Module Concept

The courses primarily aimed to provide students with

a deep understanding of the content (Student Learn-

ing First) and were structured in time-blocked units

according to knowledge and taxonomy levels for this

study.

The seven full-day online sessions were structured

so that each topic area was assigned a clearly deﬁned

time block, with increasing complexity throughout.

The goal was a coherent learning experience that ac-

tively engaged students, promoted critical thinking,

and established a measurable link between knowledge

levels and process dimensions. Interactive methods

supported not only knowledge acquisition but also the

development of analytical, evaluative, and synthesis

skills through discussions and practical applications

to tackle complex tasks.

4.2 Courses

Learning settings with “challenging tasks and exer-

cises throughout the learning process” enable needs-

based knowledge acquisition (Erpenbeck et al., 2015,

p. 1). Since prior knowledge is a key factor for learn-

ing success (Hurzlmeier et al., 2024, 108ff), courses

and materials were designed adaptively to “mitigate”

differences in prior knowledge (Hurzlmeier et al.,

2024, p. 111). The structure followed Anderson &

Krathwohl’s taxonomy, and digital materials were

speciﬁcally selected to enhance self-efﬁcacy by acti-

vating self-learning strategies (Pelikan et al., 2021).

The individual sessions are listed in Table 2 . The

module begins with an introductory session and con-

cludes with exam preparation. The content-relevant

sessions are aligned with the exam tasks. Table 2 di-

vides the ﬁrst four days into “morning” and “after-

noon” sections, in which the topics of the exam tasks

were comprehensively covered. Subsequent sessions

only included brief reviews to avoid redundancy.

Each topic block started with an introduction, with

complexity systematically increasing throughout the

session.

CSEDU 2025 - 17th International Conference on Computer Supported Education

896

Table 2: Courses and assigned examination tasks.

4.2.1 Interactive Tools

To address in-depth questions that allow connecting

facts initially not perceived as related, collaborative

maps were used as a central tool. Students worked

both collaboratively and individually on interactive

maps that visually represented individual facts or en-

tire structures as large-scale maps with links (Cross,

2023, p. 88). Additionally, expandable digital posters

were provided, detailing the composition of network

packets across all layers (Ethernet, IP, ...) with typ-

ical variations and exceptions. Figure 2 illustrates

some of the interactive tools used. Furthermore, com-

prehensive sheets with overviews of speciﬁc encod-

ing variants were introduced. Consistent use of these

tools was intended to foster insights into commonali-

ties and fundamental, recurring principles, even when

the underlying procedures initially appeared very dif-

ferent.

Additionally, experiential learning was supported

through “hands-on” activities that enabled individual

progress, particularly in analysis and synthesis skills:

networks could be analyzed “live”, and content could

be decoded. All tools presented and used live re-

lied exclusively on freely available software, such as

“Wireshark” for network analysis (Hein et al., 2021),

“Cisco Packet Tracer”, a cross-platform visual simu-

lation tool for creating and analyzing network topolo-

gies, or “FFmpeg” for encoding/decoding and analyz-

ing audio material (Chappel, 2018; Riselvato, 2020).

The use of exclusively freely accessible tools allowed

for individual (continued) engagement to any extent.

Adams et al. describe such “hands-on” activities as

“just-in-time support” and advocate for these mea-

sures to classify learning strategies into their highest

postulated Level 4 (Adams et al., 2010, p. 6).

Both the exploration of visual maps and the “play-

ing” with analysis and simulation tools enabled a

connected knowledge development through sufﬁcient

engagement of students in the group and individual

follow-up (Blackwell, 2001, p. 10) or “active knowl-

edge construction” (Graham, 2006, p. 31). Sequences

of actions were visually experienced, and a subse-

quent concept could be assembled along these cog-

nitively targeted sequences (Romero-Tejedor, 2021,

p. 68). Both served as a foundation for system de-

velopment in the comprehensive synthesis task of the

exam. The courses were subjectively deemed suf-

ﬁciently suitable to support self-learning, motivate

students, and prepare them for the synthesis tasks (

Type 3 : Synthesis Tasks ).

4.3 Development of Exam Tasks

The exam tasks were based on the taxonomy selected

in Section 3.2 , which follows Anderson and Krath-

wohl’s revision of Bloom’s Taxonomy. This taxon-

omy was divided into 3 blocks: Repetition ( 1 ), Anal-

ysis ( 2 ), and Synthesis ( 3 ), upon which the tasks

were subsequently developed. The tasks were de-

signed to contain typical characteristics of the respec-

tive taxonomic category and are detailed in Table 3.

“Open-Everything” Exam. The assessment was

designed as an “Open-Everything” exam (Jagoe,

2014; Wieman, 2017). “Open-Everything” represents

an extension of “Open-Book” formats: In “Open-

Everything” exams, all resources, including the use of

the Internet, were explicitly allowed. The only prohi-

bition was contacting other individuals, either phys-

ically or online. The “Open-Book” format—or its

extended “Open-Everything” version—reduces stu-

dents’ concerns about failure anxiety and leads to

lower exam tension (Michael et al., 2019, p. 179) and

less stress (Afshin Gharib et al., 2012, p. 476). More-

over, this format aligns more closely with real-world

work environments, where all resources are available

Figure 2: Interactive tools.

Digital Presence and Learning Success: An Empirical Study on the Impact of Online Engagement on Conceptual Expertise

897

Table 3: Classiﬁcation of examination tasks according to

Anderson and Krathwohl’s revision of Bloom’s taxonomy

with the classiﬁcation into “Knowledge Dimension” and

“Cognitive Process Dimension” (Krathwohl, 2002, 214 f.).

for problem-solving (Parker et al., 2021, p. 5). Ac-

cording to Senkova et al., the exam format (“Closed”

or “Open”) plays a role in allowing test questions in

“Open” formats to be designed more deeply, eliciting

elaborative thought processes (Senkova et al., 2018,

p. 22), which is a prerequisite for successfully solv-

ing the Type- 3 question.

Tasks of Type 1 comprised four tasks with funda-

mental cognitive requirements, focusing on the recall

and reproduction of essential content. The level of

difﬁculty was designed so that the relevant content

could be understood through active listening in the

lectures. Figure 3 provides an example task.

The content of the Type 1 tasks was intentionally

Figure 3: Example question for Type 1 : Repetition Tasks

(The illustration is partly labeled in German, as this is a

German exam. The question texts were translated for this

paper).

only brieﬂy addressed during the exam preparation, as

the content could be directly extracted “1:1” from the

provided materials. This ensured that the evaluation

would not be biased by allowing mere participation

in the exam preparation to be entirely sufﬁcient. The

achievable score for these tasks accounted for 40 % of

the total points.

Tasks of Type 2 . This task type required deeper

cognitive processing and consisted of three tasks, col-

lectively accounting for 40 % of the total achievable

score. Combined with the Type 1 tasks, this allowed

for 80 % of the total score to be attained. The key

difference from Type 1 was the necessity to identify

relationships between previously isolated topics and

develop a cohesive understanding. Figure 4 illustrates

this using the example of network packet assembly.

Each task consisted of two parts:

1. Fact Extraction: Students were required to

gather relevant facts from different sources and

correctly name them. This part documented their

ability to systematically retrieve information and

ensured that a minimum score could be achieved

to pass the exam.

2. Application and Calculations: Using the col-

lected information, students had to perform cal-

culations to reinforce their understanding of the

relationships.

Tasks of Type 3 corresponded to the highest levels

of cognitive requirements, as described in Section 2.7

. It presented students with complex challenges re-

quiring deep understanding, the integration of dif-

ferent knowledge domains, and creative problem-

solving skills. The respective task accounted for 20 %

of the total achievable points.

To successfully complete this task, practiced use

of interactive tools was required. The necessary

thought processes were generically developed dur-

ing the corresponding session through structured dis-

cussion. The goal was to enable students to apply

their knowledge to real-world scenarios and, based on

these considerations, design a functional system.

The assigned task is described in Figure 5. Stu-

dents were required to design a redundant setup for a

real-time network for audio transmission. Addition-

ally, they had to create a sketch and provide a detailed

explanation of the chosen architecture. This task not

only assessed their procedural knowledge of network

technologies but also their ability to make creative and

strategic decisions, develop an optimal solution, and

substantiate it with arguments.

CSEDU 2025 - 17th International Conference on Computer Supported Education

898

4.4 Evaluation Plan

The session days were divided into thematic com-

plexes, and the evaluation was conducted based on the

exam tasks, each of which was clearly assigned to a

thematic complex. Initially, basic statistics were cal-

culated, including average results per thematic com-

plex and attendance times in the sessions. Subse-

quently, tasks were grouped by task type to test the

hypotheses from Section 3.1.

Correlations WITHIN Task Types. For each of

the three task types, the correlation of normalized

exam results with attendance classes was calculated.

Since attendance classes were ordinally scaled, Spear-

man’s rank correlation was used (Bortz & Schuster,

2010, p. 388). The ﬁrst vector represented the rank of

the attendance class, while the second vector depicted

the relative score as a percentage of the achievable

points (metric scale level). Since no missing values

occurred in either vector, the application of a “NaN-

Policy” was unnecessary.

Fisher Z-Transformation. To ensure comparabil-

ity of the correlation coefﬁcients across task types

(Behnke & Behnke, 2006, p. 264), Fisher’s Z-

transformation was applied (Tachtsoglou & K

onig,

2017, p. 124). This transformation converts the dis-

tribution of correlation coefﬁcients into an approx-

Figure 4: Example question for Type 2 : Analysis Tasks

(The illustration is partly labeled in German, as this is a

German exam. The question texts were translated for this

paper).

imately normal form, allowing for the calculation

of conﬁdence intervals and coefﬁcient comparisons

(Bortz & Schuster, 2010, p. 156).

The Spearman correlation coefﬁcients r

through

were transformed to Z

, Z

, and Z

as follows:



1 + r

1 −r



Conﬁdence Intervals. The variance of the Z-values

is deﬁned as s

−3

, where n is the number of

observations. Accordingly, the standard deviation is

√

−3

(Rasch et al., 2014, p. 89).

The 95% and 99% conﬁdence intervals for Z

are

given by (Bortz & Schuster, 2010, p. 589):

= Z

±1.96 ·

√

−3

, CI

= Z

±2.576 ·

√

−3

The boundaries of these conﬁdence intervals were

back-transformed to obtain the corresponding inter-

vals for r

r =

−1

+ 1

Comparison of Correlation Coefﬁcients (“BE-

TWEEN”). To determine the difference be-

tween two correlation coefﬁcients, Fisher’s Z-

transformation was also used to standardize the

correlations, stabilize their dispersion, and make the

difference interpretable as an effect size measure

(Rasch et al., 2014, p. 90).

To calculate the statistical conﬁdence of the differ-

ence between two correlation coefﬁcients, the Z-value

for the difference of the coefﬁcients was computed.

For two correlation coefﬁcients r

and r

with cor-

responding transformed values Z

and Z

, and with

the respective Z-space variance σ

n−3

, the critical

Z-value for the difference was calculated, assuming

Figure 5: Example question for Type 3 : Synthesis Tasks

(The illustration is partly labeled in German, as this is a

German exam. The question texts were translated for this

paper).

Digital Presence and Learning Success: An Empirical Study on the Impact of Online Engagement on Conceptual Expertise

899

variances add up



−3



, as:

diff

−Z

where σ =

−3

diff

thus represented the mean distance of the dif-

ference in multiples of the standard deviation. This

diff

could then be used to determine the statistical

power of the difference between the correlation coef-

ﬁcients. A Z-value greater than 1,64 (one-sided test at

the α = 5% level) or 2,33 (one-sided test at α = 1%

level) indicates a signiﬁcant positive difference be-

tween the correlation coefﬁcients.

5 RESULTS

The courses described in Section 4.2 were conducted

entirely online in block format over one semester.

A total of 30 students were enrolled in the module.

Of these, 22 registered for the exam, 3 non-registered

students participated partially in the sessions, and 5

did not attend any session. The cohort under inves-

tigation consisted exclusively of the 22 students who

participated in the exams.

5.1 Methodological Outlier

One participant reported at the beginning of the mod-

ule that they regularly install real-time networks in

their current work and were therefore already very fa-

miliar with all the content. The participant announced

that they would only attend sporadically to see if any

unfamiliar topics were discussed. This individual,

classiﬁed as a methodological outlier due to their ex-

tensive prior knowledge, was excluded from the anal-

ysis. However, in diagrams showing all data, the

data for this methodological outlier are displayed but

marked separately. This outlier was excluded from

the calculation of statistical measures.

5.2 Sessions

The sessions were divided into thematic sections, as

described in Section 4.2 (“Courses”) , which were

clearly assignable to the completed exam tasks. The

assignments are detailed in Table 2 , and Table 3 lists

all exam tasks with their assignment to the task types

described in Section 2.7 based on the taxonomy out-

lined in Section 3.2 .

5.3 Attendance

Attendance at the individual sessions was heteroge-

neous. Since the sessions, as described in Sec-

tion 4.2 (“Courses”) , began with an introduction and

increased in complexity over time, precise attendance

minutes cannot be directly assigned and compared

between sessions. Therefore, attendance per session

was normalized to 0 % to 100 %, and the normal-

ized values were divided into the following 4 ordinal

equidistant classes (“binning”):

• <25 %

• 25 % to 50 %

• 50 % to 75 %

• >75 %

Figure 6 graphically shows the resulting classiﬁ-

cation in the diagram on the right. The same colors

are consistently used for the 3 task types and atten-

dance classes in all diagrams, see Figure 6 .

5.4 Exam Results

The achieved success rates, as shown in the left di-

agram of Figure 6, reveal an average success rate of

over 90 % in half of the tasks, with all Type 1 tasks (

Type 1 : Repetition Tasks ) except for Task 2 showing

success rates above 80 %. Task 2 exhibits very high

variance, with results ranging from 0 % to 100 % and

a standard deviation of 31 %.

Tasks of Type Type 2 : Analysis Tasks , an exam-

ple of which is shown in Figure 4 , exhibit decreasing

success rates: from an average of 99 % in Task 3 to

91 % in Task 5 and 77 % in Task 7.

The task of Type Type 3 : Synthesis Tasks , fully

presented in Figure 5 , shows an average success rate

of 70 % (“Task 8” in Figure 6).

The averaged success rates, as shown in Figure 7

, indicate a range of 12 % (88 % to 100 %) for Task

Type 1 across all attendance classes and a range of

9 % (76 % to 85 %) for Task Type 2 .

The averaged success rates for Task Type

3 show

a range of 57 % (35 % to 92 %), increasing with

higher attendance classes.

5.5 Correlations WITHIN Task Types

The Spearman correlation coefﬁcients for Task Types

1 to 3 are r

= 0.14, r

= 0.27, and r

= 0.85. The

transformed Z-values and calculated conﬁdence in-

tervals indicate that the correlation coefﬁcients differ.

The correlation coefﬁcients increase with the level of

task difﬁculty: r

< r

Figure 1 graphically and numerically presents the

calculated conﬁdence intervals for the correlation co-

efﬁcients. The regression lines shown in Figure 8 vi-

sually match upon inspection.

CSEDU 2025 - 17th International Conference on Computer Supported Education

900

Figure 6: Overview of Attendance.

Left: Distribution of relative success rates by task; The task type colors match those throughout the document; They have

been slightly lightened due to the large areas (see legend).

Right: Frequency of attendance across ordinal attendance classes. Attendance percentages on the metric scale exceed 100 %

in some cases because participants stayed after sessions to clarify questions (see explanation in the text under “Attendance”).

5.6 Comparison of Correlation

Coefﬁcients (BETWEEN)

When comparing the conﬁdence intervals of the co-

efﬁcients for the 3 task types, Figure 1 shows that the

intervals CI

95,1

and CI

95,2

for Task Types 1 and 2 over-

lap signiﬁcantly. The interval CI

95,3

for Task Type 3

does not overlap with the other intervals, neither in

the 95 % conﬁdence interval nor in the 99 % conﬁ-

dence interval.

The distances between the conﬁdence intervals of

the 3 task types (correlations BETWEEN task types)

in Z-space are:

diff

= 0.77 for the comparison of r

and r

diff

= 4.25 for the comparison of r

and r

diff

= 3.64 for the comparison of r

and r

Thus, the difference in correlation coefﬁcients is:

• Z

diff,2−1

= 0

7 (p = 0.2219). The difference

in correlation coefﬁcients is 0,77 standard devia-

tions in Z-space with strongly overlapping conﬁ-

dence intervals.

• Z

diff,3−1

= 4

5 (p = 0.0000). The difference

in correlation coefﬁcients is 4,25 standard devia-

tions in Z-space with no overlap in the 99 % con-

ﬁdence interval.

• Z

diff,3−2

= 3

4 (p = 0.0001). The difference

in correlation coefﬁcients is 4,25 standard devia-

tions in Z-space with no overlap in the 99 % con-

ﬁdence interval.

6 ADDRESSING THE RESEARCH

QUESTION

The research question formulated in “Research Ques-

tion, Hypotheses” is examined below using the hy-

potheses H

, H

, and H

based on the results.

6.1 Analysis of the Hypotheses

Hypothesis H

. The correlation between atten-

dance and exam success for Task Type 3 ( Type 3 :

Synthesis Tasks ) shows a very strong positive rela-

tionship with r = 0.85 and a 99 % conﬁdence interval

of CI

99%

= [0.57, 0.95] (Field et al., 2012, p. 93).

⇒ The hypothesis H

A,0

(“ There is no positive corre-

lation between attendance time and performance

in complex concept tasks

3 . ”) can thus be

rejected at the CI

99%

conﬁdence level.

Hypothesis H

. The positive difference between

the correlation coefﬁcients r

( Type 3 : Synthesis

Tasks ) and r

( Type 1 : Repetition Tasks ) was ob-

served outside a 99,9 % acceptance range from H

A,0

with p = 0.0000, indicating that the correlation co-

efﬁcient r

is signiﬁcantly greater than r

((Cohen,

1988)).

⇒ The hypothesis H

B,0

(“ The correlation between

attendance time and performance in complex con-

cept tasks is not greater than for task type 1. ”)

can be falsiﬁed.

Digital Presence and Learning Success: An Empirical Study on the Impact of Online Engagement on Conceptual Expertise

901

Figure 7: Average success rates by task type and attendance.

Hypothesis H

. The positive difference between

the correlation coefﬁcients r

( Type 3 : Synthesis

Tasks ) and r

( Type 2 : Analysis Tasks ) was ob-

served outside a 99,9 % acceptance range from H

C,0

with p = 0.0001. This indicates that the correlation

coefﬁcient r

is signiﬁcantly greater than r

((Cohen,

1988)).

⇒ The hypothesis H

C,0

(“ The correlation between

attendance time and performance in complex con-

cept tasks is not greater than for task type 2 . ”)

can be falsiﬁed: r

is signiﬁcantly greater than r

within the scope of this study.

6.2 Answering the Research Question

The study results show that student performance in

solving complex conceptual and synthesis tasks (Type

3) in the investigated course module signiﬁcantly cor-

relates with online attendance time.

⇒ H

A,1

(“ The correlation between attendance time

and performance in complex concept tasks 3 is

positive. ”) .

This correlation is signiﬁcantly higher compared to

repetition tasks and also signiﬁcantly higher com-

pared to analysis and calculation tasks.

⇒ H

B,1

(“ The correlation between attendance time

and performance in complex concept tasks is

greater than for task type 1. ”)

⇒ H

C,1

(“ The correlation between attendance time

and performance in complex concept tasks is

greater than for task type 2 . ”)

7 DISCUSSION

The exam results showed that students consistently

achieved high success rates in Type 1 tasks ( Type 1 :

Repetition Tasks ), except for Task 2, which exhibited

very high variance. This could indicate differences

in individual preparation and varying levels of under-

standing of fundamental concepts. Another plausible

explanation is the relatively higher difﬁculty level of

Task 2 ( Table 3 ) compared to the other Type 2 tasks

( Type 2 : Analysis Tasks ). Task 2 is detailed in Fig-

ure 3 (“Example question for Type 1 : Repetition

Tasks (The illustration is partly labeled in German,

as this is a German exam. The question texts were

translated for this paper)”) ; the subject-matter reader

may form their own judgment.

7.1 Correlations Within Task Types

For Type 2 tasks ( Type 2 : Analysis Tasks ), a

stronger relationship between attendance and success

was observed. The correlation coefﬁcient r

= 0.27

is higher than that of Task Type 1 with r

= 0.14,

indicating a general trend. However, the conﬁdence

intervals CI

95,1

and CI

95,2

both include the value 0,

meaning the correlation coefﬁcients r

and r

do not

show a signiﬁcant positive increase.

In this study, a very strong relationship between

attendance and exam success in complex tasks was

observed. This conﬁrms the alternative hypotheses

A,1

(“ The correlation between attendance time and

performance in complex concept tasks 3 is positive.

”) . It should be noted that a strong correlation does

not imply causation.

The correlations between attendance time and

scores across the different task types showed that

higher attendance is associated with better perfor-

mance, particularly in more complex tasks.

7.2 Comparison of Correlation

Coefﬁcients

The relationship between attendance time and exam

success, measured as scores achieved, is signiﬁcantly

higher for complex tasks 3 compared to repetition or

calculation tasks ( 1 and 2 ). This conﬁrms the alter-

native hypotheses H

B,1

and H

C,1

In the respective parts of the courses, less direct

instruction was provided, and more development and

discussion occurred. Numerous solution proposals

CSEDU 2025 - 17th International Conference on Computer Supported Education

902

Figure 8: All individual tasks separated by task type. The

attendance classes are desaturated in the background for ori-

entation. A regression line (excluding the methodological

outlier) was added for visual veriﬁcation.

were presented, which, during plenary discussions,

often led to insights within the student group without

any intervention or comments from the instructor.

Additionally, it was observed that students who

participated in the development-oriented discussion

rounds were able to develop innovative solutions and

creative problem-solving strategies for the complex

task. This suggests that the interactive and collabo-

rative nature of these sessions signiﬁcantly enhanced

students’ critical thinking and problem-solving skills.

These immediate peer-derived insights appear to

embed themselves signiﬁcantly deeper in process and

synthesis memory than mere listening.

Students who voluntarily chose not to partici-

pate in these sessions consistently performed worse.

While learning strategies for repetitive knowledge ap-

pear relatively egalitarian, developmental and synthe-

sis skills seem to be better acquired through group

discussions or individual, in-depth engagement with

interactive tools, such as work posters or simulation

software provided during the courses.

7.3 Outlook and Future Work

Beyond the approach of the present study, the results

highlight the importance of active participation and

engagement in collaborative learning environments to

develop the higher cognitive skills essential for suc-

cess in demanding technical disciplines. These ﬁnd-

ings can serve as a basis for future design and op-

timization of teaching methods to further enhance

learning outcomes in digital educational formats.

Further research could focus on differentiating

this study further into speciﬁc teaching methods to ex-

amine their impact on engagement and learning out-

comes.

Involving students and comparing the results with

their internal perspectives would be another approach

to differentiation, as well as further studies in the pro-

fessional ﬁeld and exploring which content and im-

plementation methods are deemed suitable by stu-

dents in retrospect.

8 SUMMARY

The literature indicates a fundamental relationship

between participation in online courses and learn-

ing success. This study demonstrated that students

who actively and regularly participated in online ses-

sions achieved signiﬁcantly better results in synthesis-

based, cognitively demanding tasks. This supports

the argument that increased participation in online

courses has a signiﬁcantly positive impact on the abil-

ity to solve complex tasks.

In summary, the ﬁndings of this study underline

the importance of online presence for learning suc-

cess in technical modules. Particularly for tasks re-

quiring higher cognitive abilities, active participation

in online courses appears to be crucial for success.

ACKNOWLEDGEMENTS

We would like to thank Kathrin Franke from the

Leipzig University Didactics Center for her very

competent advice on the preparation of the study

(kathrin.franke@hd-sachsen.de).

REFERENCES

Adams, J., Hanesiak, R., Morgan, G., Owston, R., Lup-

shenyuk, D., & Mills, L. (2010-01). Blended Learn-

ing for Soft Skills Development: Testing a Four-Level

Framework for Integrating Work and Learning to

Maximize Personal Practice and Job Performance.

Canadian Council on Learning (CCL).

Afshin Gharib, William Phillips, & Noelle Mathew. (2012-

08-28). Cheat Sheet or Open-Book? A Comparison of

the Effects of Exam Types on Performance, Retention,

and Anxiety. Journal of Psychology Research, 2(8).

https://doi.org/10.17265/2159-5542/2012.08.004

Digital Presence and Learning Success: An Empirical Study on the Impact of Online Engagement on Conceptual Expertise

903

Andel, S. A., De Vreede, T., Spector, P. E., Padmanabhan,

B., Singh, V. K., & De Vreede, G.-J. (2020-12). Do

social features help in video-centric online learning

platforms? A social presence perspective. Computers

in Human Behavior, 113, 106505. https://doi.org/10.

1016/j.chb.2020.106505

Baumann, N. (2022). Zur Wahrnehmung von Online-

Mentoring : Eine qualitative Studie zu den subjektiven

Sichtweisen von Mentorinnen, Mentoren und Mentees.

https://doi.org/10.20378/irb-57500

Bay, B. (2018-11-02). Bloom’s Taxonomy and Depth of

Knowledge (DOK) - Rethinking Learning. Retrieved

2024-03-01, from https://barbarabray.net/2018/11/02/

blooms-taxonomy-and-depth-of-knowledge-dok/

Behnke, J., & Behnke, N. (2006). Grundlagen der statis-

tischen Datenanalyse: eine Einf

uhrung f

ur Politikwis-

senschaftler. VS.

Besa, K., Kochsk

amper, D., Lips, A., Schr

oer, W.,

& Thomas, S. (2022-10-19). Stu diCo III -

Hochschulzukunft gestalten – aus den (digitalen)

Corona-Semestern lernen. https://doi.org/10.18442/

219

Bicak, B. E., Borchert, C., & H

oner, K. (2023-12). Foster-

ing scientiﬁc reasoning with inquiry problem-solving

experiments in an organic chemistry lab course.

CHEMKON, 30(8), 326–333. https://doi.org/10.1002/

ckon.202200024

Biggs, J. B., & Tang, C. S.-k. (2011). Teaching for quality

learning at university: What the student does (4. ed).

McGraw-Hill, Society for Research into Higher Edu-

cation & Open University Press.

OCLC: 838444823.

Blackwell, A. F. (Ed.). (2001). Thinking with Diagrams.

Springer Netherlands. https://doi.org/10.1007/978-

94-017-3524-7

Bortz, J., & Schuster, C. (2010). Statistik f

ur Human- und

Sozialwissenschaftler (7., vollst.

uberarb. und erw.

Auﬂ). Springer.

Chappel, L. (2018-01-31). Wireshark® 101: Einf

uhrung

in die Protokollanalyse - Deutsche Ausgabe

(22018th ed.). mitp.

Chrobak, A. M. (2004-04). Entwicklung von Online-

Trainererfolg in der betrieblichen Weiterbildung

durch die F

orderung von Online-Selbstwirksamkeit, -

Strategie und Koh

asion.

Cohen, J. (1988-07-01). Statistical Power Analysis for the

Behavioral Sciences (Subsequent Edition). Routledge

Member of the Taylor and Francis Group.

Cross, N. (2023). Design thinking: Understanding how de-

signers think and work (Second edition). Bloomsbury

Visual Arts.

Decius, J., Dannowsky, J., & Schaper, N. (2021-12-03).

Die Studierenden im Mittelpunkt: Personale Unter-

schiede beim informellen und formalen Lernen in

Zeiten virtueller Lehre. In I. Neiske, J. Osthushenrich,

N. Schaper, U. Trier, & N. V

oing (Eds.), Hochschul-

bildung: Lehre und Forschung (1st ed., pp. 89–114,

Vol. 3). transcript Verlag. https://doi.org/10.14361/

9783839456903-008

Erpenbeck, J., Sauter, S., & Sauter, W. (2015). E-Learning

und Blended Learning. Selbstgesteuerte Lernprozesse

zum Wissensaufbau und zur Qualiﬁzierung. Springer

Gabler.

Fiedler, M., Kaliske, J., & K

astner, M. (2022-07-05).

Studieren 2.0 - Pr

asenz, Digital oder Hybrid? Lessons

Learned, 2(1). https://doi.org/10.25369/ll.v2i1.34

Field, A., Miles, J., & Field, Z. (2012-04-05). Discovering

Statistics Using R (1st ed.). SAGE Publications Ltd.

Graham, C. R. (2006). Blended learning systems. The hand-

book of blended learning: Global perspectives, local

designs, 1, 3–21.

uner, C. (2018-02). Das Ph

anomen Lurking. In-

dividuelle Lernprozesse

”

aktiver“ und

”

passiver“

Nutzer*innen virtueller Lernumgebungen im Fern-

studium am Beispiel der FernUniversit

at in Hagen.

Hein, M., Ladner, R., & Scholz, U. (2021). Wireshark

- Standard-Tool zur Analyse von Netzwerken: Prax-

iswissen: Netzwerkprotokolle aufzeichnen, darstellen

und analysieren. Independently published.

Heinzel, A., Mottaghy, F. M., Brenner, W., & Marienhagen,

J. (2020-02). Statuserhebung der studentischen Lehre

ur das Fach Nuklearmedizin in Deutschland. Nuk-

learmedizin - NuclearMedicine, 59(01), 38–46. https:

//doi.org/10.1055/a-1011-3481

Hurzlmeier, M., Watzka, B., Hoyer, C., Girwidz, R., &

Ertl, B. (2024-01). Lernergebnisse und individuelle

Prozesse des Physik-Lernens mit auditiven und vi-

suellen Hinweisen. Zeitschrift f

ur P

adagogische Psy-

chologie, 38(1-2), 99–116. https://doi.org/10.1024/

1010-0652/a000331

Hußner, I., Lazarides, R., & Westphal, A. (2022-10-01).

COVID-19-bedingte Online- vs. Pr

asenzlehre: Dif-

ferentielle Entwicklungsverl

aufe von Beanspruchung

und Selbstwirksamkeit in der Lehrkr

aftebildung?

Zeitschrift f

ur Erziehungswissenschaft, 25(5), 1243–

1266. https://doi.org/10.1007/s11618-022-01072-5

Jagoe, C. (2014-01-28). Open Everything Exams. @col-

injagoe.com. Retrieved 2024-08-14, from https : / /

colinjagoe.com/2014/01/28/open-everything-exams/

Ketter, V., Brauch, C., Elser, P., Freier, I., & Roob,

Y. (2022-05-02). Online-LEERprozesse? Von den

Bildungs(un)m

oglichkeiten des digitalen Studierens

aus Lehrenden- und Studierendenperspektive: Ein

Dilemma in drei Akten. Ludwigsburger Beitr

age zur

Medienp

adagogik, 22, 1–16. https://doi.org/10.21240/

lbzm/22/11

Klug, J., & Seethaler, E. (2021-11-30). (Un)Freiwillig

Online-Lehre. Wie Lehramtsstudierende in unter-

schiedlichen Lernsettings Lerngewinn, Motivation

und emotionales Beﬁnden einsch

atzen. Ludwigs-

burger Beitr

age zur Medienp

adagogik, 21, 1–16.

https://doi.org/10.21240/lbzm/21/05

Kossack, F., & Bender, B. (2022). Heterogeneous groups

of students as a challenge in engineering design edu-

cation. DS 119: Proceedings of the 33rd Symposium

Design for X (DFX2022), 10–10. https://doi.org/10.

35199/dfx2022.07

Krathwohl, D. R. (2002-11-01). A Revision of Bloom’s

Taxonomy: An Overview. Theory Into Practice,

41(4), 212–218. https : / / doi . org / 10 . 1207 /

s15430421tip4104

Kuhn, S., Huettl, F., Deutsch, K., Kirchg

assner, E., Huber,

T., & Kneist, W. (2021-02). Chirurgische Ausbildung

CSEDU 2025 - 17th International Conference on Computer Supported Education

904

im digitalen Zeitalter – Virtual Reality, Augmented

Reality und Robotik im Medizinstudium. Zentralblatt

ur Chirurgie - Zeitschrift f

ur Allgemeine, Viszeral-,

Thorax- und Gef

aßchirurgie, 146(01), 37–43. https :

//doi.org/10.1055/a-1265-7259

Lerner, D., & Luiz, T. (2019-03). Nah an der Realit

at. inten-

siv, 27(02), 64–69. https://doi.org/10.1055/a-0821-

3183

odermann, A.-M. (2024-04). Das Mentoring-Programm

als soziales Lernarrangement im wissenschaftlichen

Qualiﬁkationsprozess: Ein Beitrag zur theoretischen

Fundierung von Mentoring-Programmen.

Michael, K., Lyden, E., & Custer, T. (2019-05-01). Open-

Book Examinations (OBEs) in an Ultrasound Physics

Course: A Good Idea or a Bad Experiment? Journal

of Diagnostic Medical Sonography, 35(3), 174–180.

https://doi.org/10.1177/8756479318821075

Orvis, K. A., Brusso, R. C., Wasserman, M. E., & Fisher,

S. L. (2010-12-30). E-nabled for E-Learning? The

Moderating Role of Personality in Determining the

Optimal Degree of Learner Control in an E-Learning

Environment. Human Performance, 24(1), 60–78.

https://doi.org/10.1080/08959285.2010.530633

Parker, A. M., Watson, E., Dyck, N., & Carey, J. P. (2021-

06-15). Traditional versus Open-Book Exams in Re-

mote Course Delivery: A Narrative Review of the Lit-

erature. Proceedings of the Canadian Engineering Ed-

ucation Association (CEEA). https : / / doi . org / 10 .

24908/pceea.vi0.14845

Pelikan, E. R., L

uftenegger, M., Holzer, J., Korlat, S.,

Spiel, C., & Schober, B. (2021-04-01). Learning dur-

ing COVID-19: The role of self-regulated learning,

motivation, and procrastination for perceived compe-

tence. Zeitschrift f

ur Erziehungswissenschaft, 24(2),

393–418. https://doi.org/10.1007/s11618-021-01002-

Rasch, B., Friese, M., Hofmann, W., & Naumann, E.

(2014). Quantitative Methoden 1: Einf

uhrung in die

Statistik f

ur Psychologen und Sozialwissenschaftler.

Springer Berlin Heidelberg. https://doi.org/10.1007/

978-3-662-43524-3

Riselvato, J. (2020-04-16). The FFmpeg Quick Reference

of 100+ Scripts for Video, Audio and Streaming. Inde-

pendently published.

Romero-Tejedor, F. (2021). Visual Storytelling im Design

Thinking: die Kraft von Bildern und Geschichten.

Hanser.

Senkova, O., Otani, H., Skeel, R. L., & Babcock, R. L.

(2018). Testing Effect: A Further Examination of

Open-Book and Closed-Book Test Formats.

Sharples, M. (2005). Learning As Conversation : Trans-

forming Education in the Mobile Age.

Tachtsoglou, S., & K

onig, J. (2017). Statistik f

Erziehungswissenschaftlerinnen und Erziehungswis-

senschaftler. Springer Fachmedien Wiesbaden. https:

//doi.org/10.1007/978-3-658-13437-2

Thielmann, B., & B

ockelmann, I. (2021-03). Die Rolle von

pers

onlichen Ressourcen bei der Stressbew

altigung.

PPmP - Psychotherapie · Psychosomatik · Medizinis-

che Psychologie, 71(03/04), 107–115. https://doi.org/

10.1055/a-1208-5333

Towﬁgh, E. V., Keesen, J., & Ulrich, J. (2022-08-31).

Blended Learning und Flipped Classroom in der

grundst

andigen Lehre. ZDRW Zeitschrift f

ur Didaktik

der Rechtswissenschaft, 9(2), 87–111. https://doi.org/

10.5771/2196-7261-2022-2-87

Wieman, C. (2017-02-01). Preparing physics students for

being marooned on a desert island (and not much

else). The Physics Teacher, 55(2), 68–68. https://doi.

org/10.1119/1.4974110

Wilson, L. O. (2016). Anderson and Krathwohl Bloom’s

Taxonomy Revised.

Digital Presence and Learning Success: An Empirical Study on the Impact of Online Engagement on Conceptual Expertise

905