Exploring Colocated Synchronous Use of Tablets Based on Split Screen Feature

This study analyzes task performance, perceived satisfaction, and behavior of students completing a similar set of mathematical tasks on paper, tablet in 1:1 (one-per-one), and tablet in 1:m (one-per-many) settings with implemented split screen feature. The primary goal of the study was to explore whether the proposed split screen feature could support the learning activities of students sharing a mobile device. A customized tablet mobile application was designed and developed to support up to four colocated users, generating independent instances of a math activity in a visually independent subinterface. This feature was tested among 80 students attending the first four grades of elementary education and the results showed a statistically significant difference in task performance between the paper and tablet conditions, but no significant differences between the 1:1 and the 1:m conditions. Moreover, no significant differences were found between the tested 1:m settings (1:2, 1:3, and 1:4). The students’ perceived satisfaction was significantly higher while working with tablets compared to the paper condition, but the results showed no significant difference between the 1:1 and the 1:m conditions. Furthermore, the observed higher frequency of verbal interaction between group peers in the 1:3 and 1:4 distributions compared to the 1:2 setting could indicate that working in pairs provides the most similar learning environment to the 1:1 distribution. The overall results suggest that the proposed split screen approach could effectively support the organization of individual educational activities and collaborative learning on a shared mobile device implementing the split screen feature.


I. INTRODUCTION
The use of mobile technologies in education, nowadays mostly smartphones and tablet computers, has been proven to be effective and applicable in different contexts [1], [2], [3], supporting learning processes by primarily enabling mobility, but also empowering it with interactivity, adaptivity and personalization of educational content.Taking into consideration multiple factors such as the learning environment (e.g., in-classroom, out-of-classroom, formal, informal), the ratio of available devices and students engaged in a learning activity, classroom seating arrangements, or educational The associate editor coordinating the review of this manuscript and approving it for publication was Nkaepe Olaniyi .
approach applied (e.g., teacher-centered instruction, collaborative learning), learning with mobile devices can be organized in two distributions: 1:1 (one-per-one) or 1:m (oneper-many).1:1 distribution refers to one device per user ratio, whereas 1:m distribution means that one device is used simultaneously by multiple users.Although, for the last two decades, the tendency was to support learning with one device per student ratio [4], [5], the rationale for using either distribution should be perceived as highly contextual and resource dependent.For example, due to the COVID-19 outbreak, the worldwide shift from in-classroom to online learning, and health recommendations to keep physical distance from others, there has been a consequential favoritism of 1:1 device usage in education [6].Moreover, throughout the past two decades numerous ''one device per child'' initiatives attempted to provide every student with a personal digital learning device [4], [5], [7].Still, the reality of the device-to-student ratio somewhat differs for the majority of schools worldwide, and learning in groups in 1:m settings is often organized as such due to limited technological resources [8], [9].For example, the ''ICT in Education'' research [10], conducted for the European Commission, surveyed teachers and students from the European Union, Norway, Iceland, and Turkey to explore the usage of technology in education.The survey pinpointed there is still an insufficient number of digital resources (i.e., personal computers, tablets) available in schools.Due to a number of factors such as financial and organizational limitations, this and similar studies and reports suggest that 1:1 initiatives remain a desired outcome rather than a global standard [7], [11], [12].Moreover, some studies, for example, the research on the potential and prerequisites of effective tablet integration in rural Kenya [11], suggest that device sharing alone could facilitate peer-to-peer learning and collaboration, indicating a 1:1 ratio is not a necessary condition for learning with tablets, thus questioning the justification of the ''one device per student'' initiative.
Nevertheless, numerous studies [13], [14], [15] have explored and identified the positive effects of 1:1 distribution on students' achievements, motivation, and engagement, given that each student has their own device, hence supporting independent content research and seamless learning.However, a very small number of studies have been undertaken to determine which device distribution, 1:1 or 1:m, is more effective and in what context.Moreover, only a few studies have focused specifically on the exploration of the 1:m distribution of mobile devices in the educational context.This leaves an open question of whether learning with mobile devices in 1:m settings is underestimated.
The majority of the research on the usage and comparison of 1:1 and the 1:m distributions can be found in the area of computer supported collaborative learning (CSCL) which strives to contribute to the application of technology in education as ''a branch of learning sciences concerned with studying how people learn together with the help of computers'' and emphasizes the importance of group learning analysis at both the individual and group levels [16].Specifically, the domain of mobile computer supported collaborative learning (mCSCL) studies mobile devices as tools for in-classroom and out-of-classroom teaching and learning.A predominance of the 1:1 distribution can also be observed in mCSCL in the existing literature [17].Moreover, only a few studies have explored the effects and possible benefits of shared mobile device use.Among few, Lin et al. [18] investigated the effects of collaborative concept mapping in 1:1 and 1:m settings using tablet computers, and the results indicated that in both 1:1 and 1:m settings students improved their learning results and retention.Nevertheless, while the 1:1 group demonstrated more consistency in participation, the results indicated that working in a 1:m distribution can result in the creation of superior artifacts.More recently, Wang et al. [19] compared primary school students' learning performance and group satisfaction, and the analysis revealed that students in the 1:m condition significantly outperformed their peers in the 1:1 condition, in both conceptual understanding and problem solving.However, owing to the limited number of studies, there is an evident research gap in the exploration of 1:m potential in mobile collaborative learning.
As previously stated, among the predominant reasons for applying 1:m distribution, a financial factor most often stands out, usually due to the impossibility of obtaining a separate work unit for each student, and consequently, work on devices is organized in 1:m distribution.The underlying problem in the circumstances of integrating technology with limited resources is that teachers often ask students to share a single device and work in pairs or larger groups, yet learning in these settings is usually organized using mobile applications designed for a single user.Surprisingly little research has been done to explore how students work and collaborate around single user mobile applications and to establish whether the shared use of a mobile device constitutes a collaborative situation that engenders interactions triggering learning and how this happens.A case study conducted by Fleck et al. [20] aimed to identify ways in which successful and less successful collaborations occur for pairs around tablets with applications for single users.This study demonstrated the significance of approaching turn-taking as part of a co-constructed environment in which both children have the responsibility to contribute to the process, either verbally or physically.They concluded that fostering collaborative orientation is even more important when providing children with applications that lack a shared goal.
On the other hand, collaborative scaffolds are of great importance in CSCL.Thus, many studies have focused on the design and development of methods to support and encourage the collaborative process.Scaffolding, as a learning metaphor, is a process in which the learner receives help or aid to effectively complete a goal or task that they would be unable to complete on their own [21].However, research literature presenting or analyzing existing scaffolded support for colocated mobile collaborative learning around a single device in a 1:m setting is almost nonexistent.Considering that very little is known about the nature of the learning process when students share a mobile device, further research on this topic would deepen the understanding of possible benefits and constraints of colocated learning in different 1:m settings.
Nevertheless, examples of systems supporting colocated learning with one device can be found outside the mobile learning research field, particularly in the single display groupware (SDG) domain [22] which focuses on the development of systems that allow people who are physically located next to each other to simultaneously interact with one VOLUME 11, 2023 123419 Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.common output interface (e.g., computer screen).In this setting, each user has his or her input unit (e.g., mouse, keyboard) enabling interaction with the computer.The output interface can be divided in a way that each user either interacts with a programmatically separated and independent subinterface as presented in the research by Kumar [23] and Nussbaum et al. [24], or all users work on a shared common interface as proposed by Heimerl et al. [25].The nature of the activities in which users are engaged can simply be based on the principle of splitting the screen into multiple independent segments enabling individual work, or a user interface can be designed to encourage collaboration.Throughout the last three decades, the SDG model has been most widely used in activities on personal computers, where the boundary between the input and output components is clear.The question remains whether these principles can be applied efficiently and to what extent on touchscreen devices, where input and output units are conjoined.Application examples can be found in studies focusing on tabletops.However, although previous research has demonstrated the benefits of using interactive tabletops to support colocated collaborative learning [26], [27], only a few studies on tabletop computers have explored the concept of sharing and dividing the display into multiple segments to support the independent interaction of multiple users.For instance, Hornecker et al. [28] analyzed different aspects of awareness in an empirical study that compared two types of input, multitouch and multiple mice, while Tse et al. [29] explored multimodal split view interaction on a tabletop whose surface was split into two adjacent projected views.Lissermann et al. [30] presented Permulin, an integrated set of interaction and visualization techniques for multiview tabletops to support colocated collaboration across a wide variety of collaborative coupling styles.
Interestingly, very little research has focused on the possibility of screen splitting on mobile devices, such as tablets and smartphones, into user-independent segments.Considering the gaps in existing research on 1:m mobile learning, this type of segmentation could enable a different approach to individual and group work of colocated users sharing a device.To some extent similar, Rick [31] proposed Proportion, a tablet mobile application developed to support colocated collaborative learning for pairs where students work together to solve a series of ratio/proportion problems.Each user works on a separate part of the screen, and together they must size the columns in proportion to their respective numerical labels.Still, due to the limited research on the topic, it remains an open question whether and in what way individual and collaborative activities of colocated users sharing a mobile device can be supported by dividing the display of a single mobile touchscreen device into several independent work units.
Moreover, there is uncertainty about what impact splitting the screen into smaller segments has on student's performance and perceived satisfaction.Some studies have suggested that there is no correlation between screen size and student performance.For example, Amornchewin and Sitdhisanguan [32] showed that children preferred a larger screen, but screen size did not affect the efficiency, speed, and accuracy of task execution.In addition, Park et al. [33] investigated the effects of screen size (3.5'' vs. 7'' vs. 10.1'') on the efficiency of mobile learning over time and the results indicated that screen size did not make a significant difference.On the other hand, research findings by Alasmari [34] revealed that a small screen display size produces the lowest cognitive load among university students as compared with larger displays.

A. RESEARCH QUESTIONS
Based on previous research and the aforementioned considerations, we assume that there is an undiscovered potential in the 1:m mobile device usage by implementing a single display (i.e., touchscreen) multi-input (i.e., touch) SDG principle.An interface with a split screen feature, where each user in a group is provided with an independent segment on a touchscreen display, could be useful in supporting both individual educational activities, for example in schools with limited technological resources, and collaborative learning of colocated students sharing a single mobile device.The authors presume that this principle can be applied in educational activities without a negative correlation between a small working screen area on a shared device and students' performance and satisfaction.The authors surmise that by applying the SDG principle to tablets (i.e., designing an appropriate interface to support the interaction of colocated users on a shared device), their use in specific contexts can be as effective as the 1:1 setting, which was proven effective.
Therefore, this study is based on the following hypothesis: H 0 : There is no significant difference in students' task performance between 1:1 distribution and 1:m distribution when implementing the split screen feature on a touchscreen mobile device.
Hence, given the complexity of the hypothesis set, this study aims to research possible differences in task performance, perceived satisfaction, and behavior of students completing a similar set of tasks in 1:1 and different 1:m settings with the split screen feature, and provides answers to the following research questions (RQ): RQ1: Is there a significant difference in students' task performance between 1:1 and 1:m settings when implementing the split screen feature?RQ2: What is the students' perceived satisfaction with the 1:m setting implementing the split screen feature?
RQ3: What behavioral patterns do students display in 1:m settings implementing the split screen feature?

A. THE MOBILE APPLICATION WITH THE SPLIT SCREEN FEATURE
To answer the research questions a customized mobile application was designed and developed for iPadOS.The application generates mathematical assignments within a 123420 VOLUME 11, 2023 Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.given time for selected operations (addition, subtraction, multiplication, and division) and a defined range of integers.A unique feature of the application is the possibility of running the math activity in split screen mode, simultaneously supporting up to four users.Besides running the application in the 1:1 mode designed for a single user, this feature enables device-student use in ratios of 1:2, 1:3, or 1:4.For each user, the application generates an independent instance of mathematical activity, with a unique set of tasks, which is visually segregated from the other segments by unique background color and adjusted in orientation as shown in Fig. 1.The orientation of segments is programmatically predefined and cannot be changed by users.At the end of the activity, every student is prompted with the results on his or her screen segment.
The display segmentation was designed with two aspects in mind: every individual segment must be large enough to give the impression of at least a small smartphone (phone screen sizes smaller than 4.5'' are usually considered small mobile phones), and the group sharing a device should be organized in pairs or smaller groups with 3 to 5 students, as studies have demonstrated that increasing the number of participants in a group may reduce the individual motivation and effort to work on a task [35].Thus, it was decided that the maximum number of segments that should be implemented on tablet computers with 7.9'' and 9.7'' displays, which were at the disposal for the research, is four.It should be noted that argumentation for using these parameters is prominent for further exploration since very little research has been conducted to evaluate the impact of screen size on students' performance.

B. RESEARCH CONTEXT AND PARTICIPANTS
The participants in this study were 80 students attending the first four grades of elementary education (7 to 10 years of age) from one public school in Zagreb, Croatia.There were 35 female and 45 male students participating in the study, with varying academic abilities.A total of five different classes participated in the study: one first grade (n 1 = 20), one second grade (n 2 = 21), one third grade (n 3 = 16), and two fourth grade classes, 4A (n 4a = 11) and 4B (n 4b = 12).In each of the five classes, there was a dedicated teacher during the intervention whom the students knew and who teaches them a group of subjects during their first four grades VOLUME 11, 2023 123421 Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.
of elementary education.The students have already had some previous experience in using tablets in the classroom, but this was the first time they used them in the 1:m setting with the split screen.The research was conducted over a period of two weeks in May 2023, with only one intervention in every class.A total of five class-independent interventions were made during the research period.Participation in the research was voluntary, and participants and their legal representatives gave written informed consent, as did all teachers.A small number of students, present on the day of the intervention, did not give consent to participate in the study (n = 14).Additionally, one student from class 4A did not participate in the study due to the Down syndrome (DS) condition.Statistics of the research context and participants' demographics are provided in Table 1.The research methodology was approved by the school's principal and institutional research ethics committee.

C. RESEARCH PROCEDURE
The research procedure was designed to minimize interference with the regular school curriculum.Therefore, the assignments for each grade were adjusted according to the curriculum and served as a repetition of recently learned topics in mathematics.The first-graders solved tasks by subtracting numbers 10 to 20 with numbers ranging from 1 to 9, second-grade assignments included multiplication of numbers in the range 2 to 9, third-grade students divided two-digit numbers with numbers 2 to 9, and fourth-graders divided three-digit numbers with numbers 2 to 9.
For research purposes, a total of 11 tablet computers were at the disposal, of which 9 iPad minis with 7.9'' display and 2 iPads with 9.7'' display.Given the average number of students in one class and the number of devices available, the research procedure was designed as a repeated-measures or within-subjects study where all students participated in three different conditions: 1) paper, 2) tablet 1:1, and 3) tablet 1:m, where m = CA, and A = {2, 3, 4}.Examples of seating arrangements of the students under different conditions are shown in Fig. 2.
Although the research objective was to explore possible differences between 1:1 and different 1:m conditions, the paper condition was included in the procedure due to the limited number of tablets and to ensure students' preoccupation with assignments during the intervention.The goal was to reduce occurrences of intervention-unrelated interactions between students, hence reducing the possibility of biased results.Every student participated only once in each condition during three sequential 5-minute time slots, throughout one school hour (45 minutes).The order in which students were assigned into conditions was randomized to reduce systematic variation (i.e., practice effects, boredom effect).Fig. 3 shows the example of intra-class student randomization across three different conditions.The similarity of assignments across conditions was achieved by generating tasks for the paper condition with the application developed for tablets.The example of the paper assignment is presented in Appendix A. Additionally, the frequency of different 1:m conditions was counterbalanced at the study level.Three different 1:m modes were simultaneously included in the research procedure due to the limited number and differences in the size of displays of the devices available.It should be noted that iPads were used only in groups of 3 to 4 students (1:m distribution), and iPad minis were used by pairs (1:m distribution) and by single users (1:1 distribution) to ensure a similar size of segments across all conditions.The detailed distribution of devices across iterations and conditions, and for each class separately, is presented in Table 2.For instance, during the first iteration, the students from the first grade were divided into two 1:m groups with 4 students in one, and 3 students in another group sharing an iPad, while at the same time, 6 students were using iPad mini in the 1:1 mode, and 7 students were doing the paper assignment.
At the beginning of class, students were greeted by researchers and detailed about the procedure, conditions, and assignments.Students were instructed that the assignments served as a repetition and would not be graded.The explanation of the research goal was omitted.When needed, nonparticipating students were seated in different parts of the classroom to isolate them from participating students.These students were given the paper assignment with the same set of tasks given to the participating students.The paper had enough tasks to keep these students occupied for the duration of one school hour, so they did not interfere with participating students.It should be noted that participating students were informed that the paper assignment contained a large number of tasks and presumably could not be solved in a given time of 5 minutes.Therefore, the students were instructed to solve them from the top of the paper to the bottom without skipping tasks they did not know or found hard to solve.They were informed that if they skipped a task, it would be marked as false.With this procedure design, the goal was to achieve a similar situation as when generating tasks with the tablet application since each condition had a fixed time limit of 5 minutes and the objective was to analyze the differences in students' task performance across conditions.The paper assignments were checked long after the intervention.Therefore, students did not get the results after the activity, whereas on tablets they were prompted with the results immediately upon completion of the assignments.

D. DATA COLLECTION AND INSTRUMENTS
The data for the paper condition were collected in written form (i.e., assignment paper given to a student) and contained the student's name, grade, and mathematical tasks with the provided solutions, whereas data collected from tablets were logged into a database.The dataset for each student consisted of the student's name, device ID, tablet configuration setting (1:1 or 1:m, where m = CA, and A = {2, 3, 4}), timestamp of the beginning and end of the activity, duration of the activity in minutes, number of solved tasks, number of correct tasks, and number of incorrect tasks.
Additionally, the students in the 1:m distribution were audio and video recorded to complement quantitative data, gain deeper insights into verbal and physical interactions between students sharing a device, and extract the possible challenges they experienced.An example of a research setting is depicted in Fig. 4.Moreover, during the interventions, some teachers gave their verbal remarks and observations to researchers regarding working in a 1:m setting and additional information about students (e.g., grades, affirmations towards math).This information was collected to enrich data analysis.It should be noted that the teachers were not obliged to give comments, therefore some of them were more involved than others.
At the end of the intervention, a simple Likert-type threepoint scale questionnaire (Appendix B) with smiley faces (sad, neutral, and happy), in combination with open-ended questions, was distributed to gain a better understanding of VOLUME 11, 2023 123423 Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.It should be noted that in Croatia at the end of the first grade, all students know how to read and write.Thus, all students were able to read the questionnaire content and write comments.Although it was mandatory for students to circle one of the three smiley faces in order to rate the conditions, the additional explanation of their selection was not.In addition to the questionnaire, students were able to give a short voluntary video statement at the end of class.

A. DATA PREPARATION
The data from the paper assignments and database logs were combined for further in-depth examination.Several significant outliers were observed in the results of the paper condition, and further analysis indicated that for a few students (n = 9), the number of tasks completed in the paper condition was more than 30% higher than that in the other two conditions.We concluded that the data were most likely skewed due to oversight of the 5-minute time slot expiry during the paper condition, which was measured more loosely than under tablet conditions with a software built-in timer.The papers in question were double-checked and the results were altered to match the student's highest number of tasks completed in one of the other two conditions.
Additionally, it was observed that the database entries for two students (n = 2) were corrupted since they did not participate constructively in tablet conditions but were, for some reason, continuously clicking on the task submission button.Since the number of answers provided by these students in tablet conditions was very high (more than 300% above the class average) and the percentage of correct answers for the paper and tablet conditions differed greatly (more than 60%), these data were excluded from the analysis of students' task performance and perceived satisfaction.
Furthermore, it was decided that the results of the two fourth grades (4A and 4B) that participated in the study will be analyzed as a single group since these students solved the same type of assignments.In addition, after testing for normality, the independent samples t-test showed no significant difference in means between these two grades regarding the percentage of correct answers provided on paper [t( 20

B. STUDENTS' TASK PERFORMANCE
Due to the observed outliers, a total of 78 participants were included in the analysis of the task performance.Since the research procedure included repeated measures of outcomes (i.e., scores) on three different levels (paper, tablet 1:1, and tablet 1:m) of the factor variable involving four grades of students who were assigned mathematical assignments with between-grade differences, to answer the first research question, the student's task performance was analyzed as a percentage of correct answers, which was perceived as a proportion of correct answers versus total answers given by a student, hence providing a unified performance scale for all four grades.The descriptive statistics for all four grades showing the percentage of correct answers across conditions is shown in Table 3.
To determine whether there are any statistically significant differences between the means of the three conditions, the one-way repeated-measures ANOVA was run on the study sample data (n = 78).Mauchly's test of sphericity indicated that the assumption of sphericity had not been violated, χ 2 (2) = 2.840, p = 0.242.The test of within-subjects effects showed the percentage of correct answers was statistically significantly different across conditions during the intervention [F(2, 154) = 10.997,p < 0.001, partial η 2 = 0.125].A further post hoc analysis with a Bonferroni adjustment (TABLE 4) revealed significant differences between the paper and the 1:1 condition (p = 0.008), and the paper and the 1:m condition (p < 0.001), but no significant difference in the task performance between the 1:1 and the 1:m conditions (p = 0.242).The results also revealed students solved the tasks with around 10% more accuracy during the paper condition in comparison to working alone on a tablet (µd = 6.624) and while sharing a tablet with other students in the 1:m setting (µd = 10.455).
Since different students participated in one of the three 1:m groups, a one-way ANOVA was conducted to look for possible differences between the percentage of correct answers achieved in different 1:m distributions: pairs or 1:2 (n = 23), 1:3 (n = 23), and 1:4 (n = 32).The analysis showed there was homogeneity of variances, as assessed by Levene's test for equality of variances (p = 0.267).Although the mean was the highest for the groups working in the 1:3 setup (M = 80.5617, SD = 22.16564) compared to the 1:2 distribution (M = 72.2013,SD = 26.26963),followed by the 123424 VOLUME 11, 2023 Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.Overall, these results support the hypothesis that there is no significant difference in students' task performance when working alone on a tablet or when doing a similar task while sharing a tablet with the implemented split screen feature.

C. STUDENTS' PERCEIVED SATISFACTION
The overall results of students' perceived satisfaction gathered through the questionnaires (n = 78) are shown in Table 5.These results indicated that students were the most satisfied while working alone on a tablet (M = 2.87, SD = 0.373), followed by doing assignments in the 1:m distribution (M = 2.81, SD = 0.457), and were at least satisfied while solving the tasks on paper (M = 2.33, SD = 0.715).The results also suggest students were generally satisfied with the activities, rating all conditions mostly with a happy smiley face.Some of the students from the first, second, and fourth grades rated the paper condition with a sad smiley face, and only two students from the fourth grade rated the 1:1 and the 1:4 conditions with a sad smiley face.
In addition, the results of the Friedman test revealed that the students' perceived satisfaction ranks were statistically significantly different across conditions [χ 2 (2) = 43.986,p < 0.001], and pairwise comparisons were performed with Bonferroni correction for multiple comparisons (Table 6).The results revealed significant differences between the paper and the 1:1 condition (p < 0.001), and the paper and the 1:m condition (p = 0.001), but no significant difference between the 1:1 and the 1:m conditions was observed (p = 1.000).The results suggest that students preferred doing the assignments on tablets over doing the same type of assignment on paper.
In addition, the independent samples Kruskal-Wallis test showed no significant difference between the 1:m modes (p = 0.401), although the descriptive statistic indicated that, while working in a 1:m setting, the most satisfied students were those working in the 1:3 device-student ratio (M = 2.91, SD = 0.288), followed by students working in pairs (M = 2.83, SD = 0.388) and working in the 1:4 settings (M = 2.72, SD = 0.581).
To detect possible reasons affecting students' perceived satisfaction, a correlation analysis (Table 7) between questionnaire answers and task performance based on the percentage of correct answers was performed.The results showed that there was a positive correlation between the percentage of correct answers achieved in the paper condition and the 1:1 condition [r s = 0.541, p < 0.001], the paper and the 1:m condition [r s = 0.461, p < 0.001], as well as between the 1:1 and 1:m conditions [r s = 0.667, p < 0.001], indicating that students who did well in one condition also did well in other conditions.Although a negative correlation was found between the percentage of correct answers submitted in the paper condition and perceived satisfaction with working with paper (r s = −0.029), the result was not statistically significant (p = 0.801).
On the other hand, a significant positive correlation was observed between task performance and perceived satisfaction when working alone on a tablet [r s = 0.291, p = 0.010], as well as between task performance and perceived satisfaction when sharing a tablet [r s = 0.279, p = 0.013].These results could indicate that satisfaction with working on a tablet is mildly related to the scores.For example, one of the only two students who rated the 1:m condition with a sad smiley face, more precisely the 1:4 distribution, had an error rate of 82.61% in the 1:m condition.The same student rated the 1:1 condition with a happy smiley face in which the student had an error rate of 7.69%.This student also made no errors in the paper condition but rated the condition with a neutral smiley face and made a comment in the survey that doing the paper assignment was boring.However, this student also wrote ''I prefer to work alone.''as the reason to rate the 1:m condition with a sad smiley face.Moreover, the other student who rated the 1:4 mode with a sad smiley face also rated all other conditions with a sad smiley face and wrote an additional comment ''I did not like it.''for each condition.What is important, this student performed below average in all three conditions with more than a 76% error rate.
Apart from the scale data, text data related to students' perceptions of satisfaction were also collected through the survey.The majority of first-graders (80%) did not write any comments, but the two students who gave sad and neutral smiley faces for the paper condition and happy smiley faces for tablet conditions pointed out that they do not like to write.One of the two students who wrote comments and rated all conditions with a happy smiley face stated ''I love math.'',''I love to learn.'', and ''I love to learn with others.'',while the other student wrote a general comment ''I love to work on a tablet.''.
The second-grade students pointed out that ''There were too many tasks'' as the reason they rated the paper condition with a sad or neutral smiley face.In addition, a few of them wrote that solving the tasks on paper was boring.According to the written statements, the main reason the students liked to work alone on a tablet was related to the fact they were not interrupted by others.On the contrary, some of them also pointed out that working in a group was fun and easier because they were collaborating.One student, who rated the 1:1 condition with a happy smiley face and the 1:2 mode with a neutral smiley face, pointed out that she would preferred if she had shared a tablet with a specific friend.In addition, one student who rated both the 1:1 and the 1:3 conditions with a happy smile face wrote a comment ''I just do not care about others and win by solving the most tasks and having the most correct answers.''.As the reason to rate the paper condition with a neutral smiley face, one student wrote that the paper assignment contained too many tasks, and the other student commented it was because she did not perform well.On the contrary, the third student wrote that the paper assignment was too easy and rated it with a neutral smiley face while rating the other conditions with a happy smiley face.One student who rated the 1:3 condition with a neutral smiley face wrote ''I was second because I had one point less''.This student also wrote the comment ''I solved everything correctly.''for the 1:1 condition which the student rated with a happy smiley face.Correspondingly, one student who rated the 1:1 condition with a neutral smiley face wrote ''I did not know how to solve some tasks!'', but rated the 1:4 condition with a happy smiley face and wrote, ''It was great!''.Others pointed out that they liked working on a tablet because it was fun.
Fourth graders commented they did not like the paper condition mostly because ''it was boring'' and ''it was no fun''.On the contrary, they pointed out that sharing a tablet ''was the best'', ''was fun'', and that ''it was great because everyone had different tasks on their part of the screen, but we were working together in a group''.Two students who preferred the 1:1 condition over the 1:m condition commented they preferred to work alone in general.The students who preferred the 1:m condition over the 1:1 condition commented it was harder to work alone or that they performed better in the 1:m condition, and that it was more fun.

D. STUDENTS' BEHAVIORAL PATTERNS AND VERBAL INTERACTION IN 1:M SETTINGS
In addition to quantitative data obtained from data logs, video recordings were analyzed to acquire a better understanding of the group dynamics and challenges students face while working in 1:m settings with the split screen feature.All participating students (n = 80) were included in the video analysis, and a total of 28 groups were inspected, including twelve 1:2 groups, eight 1:3 groups, and eight 1:4 devicestudent ratios, as shown in more detail in Table 2. Group dynamics were observed across physical and verbal dimensions, and for each group, we examined the position of group members during the activity (sitting, standing, or leaning towards the device), whether there were any device usurpations (e.g., tilting), and the verbal interaction in terms of quarreling, seeking help, or helping others.
As for the physical dimension, a surprisingly small percentage of students had done device usurpations (7 out of 80 students, or 8.75% of students) with an average of 1.43 device tilting or pulling per student.Since this number is relatively small compared to the duration of the whole activity, it should be emphasized that a tablet was not moved the majority of the time during the activity, regardless of the differences in students' positions around the device.In 15 of 28 groups (53.57%), an unequal distance between the device and peers was observed, with the device usually positioned closer to one student.This was observed almost equally across every setting.Students who were more distant from the device were found kneeling on the chair and leaning towards the device or standing.This behavior was observed in 6 out of 15 groups with unequal device-student distance, and the majority of these groups were from the first grade (4 groups).Generally, non-sitting positions were mostly observed in the first and second grades.Only one student from the fourth grade was doing the assignment while mostly kneeling on a chair.Moreover, it was observed that students from the first grade were more physically active and leaned towards the device during the activity compared to higher grades.The examples of students kneeling or standing are shown in Fig. 5.
In addition, the analysis indicated differences in the verbal nature of group interaction regarding the device-student ratio, pointing out that within-group communication was more frequent in the 1:3 settings (36.36%) and the 1:4 settings (63.64%).On the contrary, interaction was very low or nonexistent while students were working in pairs, except for one detected between-group verbal interaction of pairs while sharing a desk as depicted in Fig. 6.This suggests that working in pairs might be more similar to 1:1 setting than to the other 1:m settings.
The majority of verbal within-group interaction included seeking or providing help, and discussions about possible solutions.Interestingly, only one moment of quarreling  between two students was observed, in which one student was appointing a remark ''Do not copy!'' to another student.More interestingly, since the assignments generated for each student were different there was no possibility of ''copying''.On a positive note, two different situations in which one VOLUME 11, 2023 123427 Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.
student was trying to motivate the other student to do the assignment were observed.After noticing that her peer was not interacting with the tablet, the student tried to motivate the other by saying ''Are you even doing the assignment?''.In another situation, the student was telling her peer ''Please try to solve at least one task.''.

E. TEACHERS' OBSERVATIONS AND ANALYSIS OF SPECIFIC CASES
As previously pointed out, the teachers were not obliged to give observations, therefore some of them were more involved during the intervention than others.Thus, their comments were used purely to get a deeper understanding of the obtained results, the observed students' behavior, and the peer interaction in the 1:m settings.
One of the comments the second-grade teacher gave was that she perceived one student with average abilities in math as more motivated than usual while working in the 1:4 group.During the survey, this student rated working in the 1:m condition with a happy smiley face and wrote that the reason he liked sharing a tablet was ''Because I was learning and at the same time I was spending time with friends.''.Interestingly, this student solved approximately 40% fewer tasks during the 1:m work, and submitted around 40% more incorrect answers, compared to the paper and the 1:1 condition.However, given that this student first participated in the 1:m condition, followed by the paper and the 1:1 condition, it could have been that the group mode motivated this student to solve more tasks afterward.On the contrary, the teacher also observed one student with average abilities in math being confused while working in the 1:3 group.Although this student rated all conditions with a happy smiley face in the questionnaire, he wrote a nonnegligible general comment ''I solved the assignments better when I was working alone on a tablet.''.However, this student's performance results were approximately similar in all conditions.In addition, the performance results of one student from the second grade had to be excluded from the analysis due to the bias in data caused by the student submitting a large number of answers in tablet conditions with a high error rate.Interestingly, in the questionnaire, this student wrote comments that the tasks were easy and that the mobile application was easy to use.This student also rated both tablet conditions with a happy smiley face and the paper condition with a neutral smiley face.
The teacher from the third grade singled out four students who usually perform above the class average in math.One of these students gave a neutral smiley face to the 1:3 condition in the questionnaire and wrote ''I was second because I had one point less.''as the reason for rating the 1:m condition lower than the other two, which he rated with a happy smiley face.This student also wrote ''I solved everything correctly.''for the 1:1 condition, and ''I was quick and it was fun.''for the paper condition.
The teacher from grade 4A gave a remark that there is one student with very low performance in math.This student also performed low in all experimental conditions, with a 100% error rate.In the questionnaire, this student rated the paper condition with a sad smiley face and commented ''I did not like it.''.This student rated the 1:1 condition with a neutral smiley face and commented ''This was not so bad.'', and the 1:2 condition with a neutral smiley face and commented ''I did not like it much.''.It should be taken into consideration that this student also wrote a general comment ''I was not feeling well during the class.''.On the contrary, according to the teacher's insight, there was one gifted student in the class (i.e., according to tests, this student's abilities and knowledge were the same as those of an average seventhgrade student).Interestingly, the results of this student had to be excluded from the performance and satisfaction data analysis due to the bias in data caused by submitting a large number of incorrect answers, with a 99% error rate in both tablet conditions.On the contrary, this student achieved a great result in the paper condition and rated all conditions in the questionnaire with a happy smiley face and commented ''It was great.'',and ''It was great working together.''.
In grade 4B teacher appointed three higher-ability students, one of whom rated the 1:4 condition with a sad smiley face and wrote a comment ''I like to work alone.''.This could be grounded in the fact that this student's error rate in the 1:m condition was around 80% higher compared to the other two conditions.In addition, this student rated the paper condition as neutral explaining it was boring, and the 1:1 condition with a happy smiley face stating it was easy and fun.
These observations and the analysis of specific cases additionally pinpoint the complex nature of the educational process, suggesting possible correlations between learning outcomes or satisfaction and student's abilities in the given subject, personality traits, and other factors.

IV. DISCUSSION
This study analyzed differences in students' task performance and perceived satisfaction by comparing three conditions: paper versus 1:1 tablet setting versus three different 1:m tablet distributions.However, the motivation for this research was to contribute to the general understanding of the differences between 1:1 and 1:m settings and to explore whether the proposed split screen approach could effectively support the organization of both individual educational activities and collaborative learning on a single mobile device.The applications implementing the proposed feature could benefit schools with limited technological resources but also bring a novelty in supporting and encouraging collaboration between students sharing a mobile device.Given the limited number of studies exploring the effects of learning in 1:m settings, the research hypothesis presumed that a mobile device with the split screen feature can be used with no significant difference in task performance and perceived satisfaction compared to doing a similar assignment alone in 1:1 setting, which was 123428 VOLUME 11, 2023 Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.proven effective.To detect possible differences between these two distributions, elementary school students were given a set of mathematical assignments in three experimental conditions.Afterward, a survey was conducted to analyze their perceived satisfaction.
Furthermore, to gain a deeper insight into the nature of the interaction between students sharing a device, the behavior of students in three different 1:m settings (1:2, 1:3, and 1:4) was analyzed through video recordings.In addition, teachers' observations were collected during the interventions.
The analysis results are summarized and presented in the following sections answering defined research questions.

A. IS THERE A SIGNIFICANT DIFFERENCE IN STUDENTS' TASK PERFORMANCE BETWEEN 1:1 AND 1:M SETTINGS WHEN IMPLEMENTING THE SPLIT SCREEN FEATURE?
To answer the first research question of whether there is a significant difference in students' task performance between 1:1 and 1:m settings when implementing the split screen feature, the performance data collected from the tablets was analyzed.The results showed no statistically significant differences between these two conditions.However, since the research combined three different 1:m distributions (1:2, 1:3, and 1:4) into one 1:m condition, the effect of each condition could be mixed.Therefore, additional research should be conducted separately for each of the 1:m distributions, preferably with a larger data sample, to effectively distinguish possible differences.Nevertheless, no statistically significant differences were observed in pairwise comparisons of the 1:m distributions (1:2, 1:3, and 1:4) in the research data sample suggesting students' task performance was equal across tested 1:m settings.
Conversely, a significant difference was observed between the paper and the 1:1 condition, and the paper and the 1:m condition, indicating students solved the tasks on paper with slightly more accuracy.Despite limited existing research on the comparability between tablet and paper tests, some studies in which authors explored reading comprehension suggest that higher familiarity with tablets can have a positive effect on educational outcomes [36], [37].It should be noted that the students included in this study usually do not use tablets in school.Therefore, this could have influenced the indicated better performance in the paper condition compared to the tablet conditions.Furthermore, some studies suggest that the differences between scores also depend on the characteristics of the students (e.g., experience with the device, socio-economic status) as well as the test characteristics, such as the subject matter being tested, user-friendliness of the device, and how the test items are presented [38].This correlates with the fact that students included in this study were using the developed mobile application with the split screen feature for the first time.In addition, when interpreting the results, variations in the study methods of time measurement should be considered.The duration of tablet conditions was controlled by the software's built-in timer, whereas the beginning and the ending of the paper assignment were controlled by the teachers and researchers, which could have caused inconsistencies in measurement (e.g., giving students more or less time to solve the paper assignment).

B. WHAT IS THE STUDENTS' PERCEIVED SATISFACTION WITH THE 1:M SETTING IMPLEMENTING THE SPLIT SCREEN FEATURE?
To address a second research question of what is the students' perceived satisfaction with the 1:m setting implementing the split screen feature, a survey was conducted upon competition of the math assignments.According to the analyzed questionnaire results, students were almost equally satisfied working with tablets in the 1:1 and the 1:m conditions and rated the conditions mostly with a happy smiley face.Moreover, there were no statistically significant differences observed when comparing the tested 1:m modes which students also rated mostly with a happy smiley face.The quantitative analysis results suggest students' perceived satisfaction with working on tablets is mildly related to task performance.This is also supported by the questionnaire comments written by the students.Furthermore, the majority of the students who preferred the 1:1 condition over the 1:m condition explained their response by noting that they were not interrupted when working alone on a tablet compared to sharing a device.On the other hand, the most prominent reason they liked working in a group was because it was fun and they liked collaborating, although each student had an individual assignment.This could suggest that the split screen approach is less appropriate for individual assignments where a student is oriented towards accomplishing a good personal result through tests and more suitable for exploratory learning and encouragement of collaboration between peers.However, in the analysis of the survey results and teachers' comments, it was also observed that several other parameters could have had an influence on students' satisfaction, such as how they perceive working in a group (e.g., competition, collaboration), who they work with (e.g., friends, peers with lower or higher abilities), what are their achievements in the subject being presented (e.g., low, average), how are they feeling during the activity (e.g., worried about something, sad, happy), and other nonnegligible factors.These findings can also be observed in other studies researching the influence on students' performance and perceived satisfaction with group work [39], [40].
Although the results showed that students solved mathematical tasks on paper with more accuracy compared to the tablet conditions, the survey analysis indicated students were significantly more satisfied while working on tablets compared to working on paper.Moreover, the students' comments suggest that a large number of tasks presented on the two-sided paper could have influenced their satisfaction, possibly making them feel like underachieving.In addition, the comments suggest students perceived the paper assignment as boring compared to doing the same assignments on a tablet, regardless of whether they were doing it in 1:1 or 1:m distribution.It should be taken into consideration that tablets could have caused a wow effect because these students usually work with paper in school [41].Hence, to deepen the understanding of the observed differences in perceived satisfaction, future investigations should replicate the research design of this study with students who are exposed to tablets more frequently and possibly without the paper condition included.Furthermore, the paper assignments were checked long after the intervention.Therefore, students did not get the results after the activity, whereas on tablets they were prompted with the results immediately upon completion of the assignments.This could have caused a lack of enjoyment since the general observation was that students wanted to know how well they performed and in addition, according to survey commentary, some students experienced the 1:m mode as competing.Moreover, the students were instructed that the assignments served as a repetition and would not be graded by teachers, which could have influenced their motivation to participate.This might explain the circumstances of the two students who were excluded from the analysis since they were not participating constructively throughout all of the assignments but were instead just clicking on the ''Submit'' button on the tablet application.

C. WHAT BEHAVIORAL PATTERNS DO STUDENTS DISPLAY IN 1:M SETTINGS IMPLEMENTING THE SPLIT SCREEN FEATURE?
To answer the third research question video recordings of groups working in the 1:m settings (1:2, 1:3, and 1:4) were analyzed.These results together with the collected teachers' comments complemented the performance and satisfaction survey results and provided a deeper insight into the nature of peer interaction.
According to video analysis, the occurrences of negative behavioral patterns (e.g., device usurpation, quarreling), during the work in the 1:m condition, were minimal.Conversely, the observed nature and high frequency of verbal interactions (i.e., students asking for help or helping others) indicate the proposed split screen approach could have the potential to support and encourage collaborative learning.However, the observed frequency of verbal interaction between group peers in the 1:3 and 1:4 distributions was higher compared to the 1:2 setting suggesting that students are more likely to engage in discussion in smaller groups (i.e., triads, groups of 4 to 5 students) than in pairs.This could mean that working in 1:1 distribution is more similar to pairs than to the other 1:m settings.To support this assumption, more research is needed addressing the differences between various 1:m distributions implementing the split screen feature since there is a very small number of studies exploring the topic.
Interestingly, it was observed that a tablet was not moved the majority of the time during the activity, regardless of the differences in students' positions around the device.This suggests that, even though students worked on a part of a screen that was on a shared device, they did not show a need to bring the device closer to do the assignment and were able to use it as if they were using it alone.However, the video analysis showed that students from the first grade were more physically active and leaned towards the device during the 1:m activity compared to students from higher grades (i.e., second, third, and fourth).The possible effect of the observed higher physical activity on task performance should be researched in more detail, preferably on a larger sample of first grade students.

V. CONCLUSION
This study contributes to the understanding of differences between 1:1 and 1:m distributions in the context of math learning in early grades of elementary education.A comparison of the students' task performance across conditions confirmed the research hypothesis that there is no significant difference in students' task performance when working alone on a tablet or doing a similar task while sharing a device with the implemented split screen feature.This indicates that splitting a tablet screen into smaller independent segments can be as effective as the 1:1 mode, in certain contexts.However, to generalize findings, future studies should examine in more detail the effect of different device-student ratios on students' performance and perceived satisfaction, as well as the application of the split screen approach in different educational contexts, such as other courses or subjects, and with diverse groups of students, such as other grades.
Nevertheless, the research results indicate that the split screen feature could lead to a different approach in work organization around a single mobile device and we hope that further research on the topic will provide additional guidelines for designing educational activities for both independent and collaborative work of colocated users sharing a mobile device.

APPENDIX A EXAMPLE OF THE PAPER ASSIGNMENT
123430 VOLUME 11, 2023 Authorized licensed use limited to the terms of the applicable license agreement with IEEE.Restrictions apply.

FIGURE 1 .
FIGURE 1.The iPadOS mobile math application with the split screen feature in the single mode (top left), the 1:2 mode (top right), the 1:3 mode (bottom left), and the 1:4 mode (bottom right).

FIGURE 2 .
FIGURE 2. Seating arrangements and work modes from left to right: the paper and the 1:1 condition, the 1:2 mode, the 1:3 mode, and the 1:4 mode.

FIGURE 3 .
FIGURE 3. Representation of the research procedure with the participants from the same class, randomized across three different conditions (paper, 1:1, and 1:m) over three time points.

FIGURE 4 .
FIGURE 4. Research settings in the class 4A.
students' perceived satisfaction and the possible difficulties they encountered across conditions.The questionnaire was constructed out of three Likert-type questions unified under a single question ''Rate the satisfaction of working in different activities during the class (circle the smiley face):''.Each of the three sub-questions for different conditions provided an extra text area with the open-ended question ''Please explain your answer. . .'' enabling additional commenting.

TABLE 4 .
Results of pairwise comparison for all participants based on the percentage of correct answers.1:4 distribution (M = 67.2788,SD 27.37823), there were no statistically significant differences found between students' scores across the 1:m distributions [F(2, 75) = 1.804, p = 0.172].

FIGURE 5 .
FIGURE 5. Examples of students in the 1:3 and the 1:4 groups standing or kneeling during the activity.

6 .
Examples of verbal interaction between peers engaged in inter-group discussion between pairs (top), discussion in the group of three (bottom left), and discussion in the group of four (bottom right).

TABLE 1 .
An overview of participants' demographics and research context.

TABLE 2 .
Detailed distribution of devices across conditions and iterations.

TABLE 3 .
Descriptive statistics showing the percentage of correct answers across conditions for all participants.

TABLE 5 .
Overall descriptive statistics (means and standard deviations), number and percentage of respondents (%) rating their perceived satisfaction across conditions.

TABLE 6 .
Pairwise comparison of students' perceived satisfaction.

TABLE 7 .
Correlations between the percentage of correct answers and questionnaire results.