Introduction

One of the expected outcomes of education is permanent learning. To this end, different teaching methods are used in classrooms, but other learning environments are also tested. In particular, among the teaching methods, student-centered learning and individual learning come to the fore. Specifically, following the development of internet technologies, learning environments have shifted from face-to-face to virtual. Thus, time and space boundaries have been eliminated for learning environments.

One of classroom management goals is improving learning and making students’ learning experiences enjoyable (Turan, 2015a, b). However, effective classroom teaching in classrooms is possible only by choosing the right methods and techniques (Taspinar, 2012) and student-centered environments. Students actively participate in tasks and undertake their learning (Hannafin et al., 1997; Means, 1994; Shea et al., 2012). The integration of various technologies into the education process provides many physical and instructional advantages (Arabaci & Polat, 2013). To ensure effective learning, it is necessary to apply student-centered contemporary teaching methods and classroom management approaches (Carswell & Venkatesh, 2002; Polat, 2016), one of which is distance education. Unlike the face-to-face education environment, distance education offers global learning opportunities with temporal and spatial flexibility. Therefore, distance education environments support the individualized teaching process by considering the stages appropriate to individuals’ learning speeds (Altıparmak et al., 2011). Under the concept of distance education, different learning types are known by other names, including web-based teaching/learning.

French (1999) defined web-based teaching as providing information to the students via computer networks (cited by Oral & Kenanoglu, 2012). Web-based distance education involves creating a meaningful learning environment in which learning is encouraged and supported by a hypermedia-based curriculum using the resources available on the web (Bay & Tuzun, 2002; Chrisman & Harvey, 1998). Because of its independence from time and place, web-based distance education is widely acknowledged (Aslanturk, 2002; Ohler, 1991; Zaborova et al., 2017). While web-based distance education enables fast and interactive learning, its tools and materials provide a rich source of information and encourage users to structure and exchange information. Moreover, it facilitates more counseling, discussion, and information exchange; it is considered student-centered, democratic, and individual-based (Keser et al., 2002). The essential advantages of web-based education are creating a virtual campus and enabling asynchronous education. Students and teachers are not in the same classroom at the same time. It is an ideal model that allows students to access the content at any time and to make use of resources as quickly and economically as possible (Carswell & Venkatesh, 2002).

Despite the benefits of web-based distance education, there are various problems or concerns, including: (1) failure in program planning; (2) trainers' insensitivity to the change in their roles; (3) low quality in teaching; (4) uncertainties in the use of resources and materials; (5) lack of technical expertise; (6) uncertainty in access to technical support (MacDonald et al., 2001). Another problem in distance education is students' lack of motivation because of deficiencies in social interaction (Karabatak & Turhan, 2017; Yolcu, 2015). As an extension of this problem, the rate of withdrawal of students participating in a distance education environment is higher than for the traditional education (Rovai, 2001). In other words, inadequate interaction, lack of motivation, failure to meet expectations, and low satisfaction rates cause distance education students to withdraw from the program as the physical remote environments in distance education ensure that students cannot feel that they belong to a community (Emmanuel-Frenel, 2017; Rovai, 2002; Tinto, 1993). Therefore, creating a sense of belonging in distance education enables individuals to keep their interest in education alive and will have a positive influence on success and satisfaction (Ilgaz & Askar, 2009).

Sense of belonging refers to the subjective state of feeling that the student is individually approved, respected, involved, and supported by other school individuals (Goodenow, 1992; Goodenow & Grady, 1993). Although the sense of belonging to school is one of the essential affective characteristics that students are expected to have for their school (Sari, 2013), this feature is often missing in distance education (Alexander, 2001). If the sense of belonging in distance education is not adequately addressed, students are unlikely to acquire learning resources if they do not feel part of a community (Lapointe & Reiseter, 2008). But a limited amount of research has focused on the sense of belonging in distance education (Emmanuel-Frenel, 2017; Peacock & Cowan, 2019; Thomas et al., 2014).

Emmanuel-Frenel (2017) investigated how the perceived value placed on student support services correlates with students’ sense of belonging to the academic community. A qualitative study by Thomas et al. (2014) involved how the sense of belonging was understood and experienced by students and the strategies used by academics to foster belonging in online learning. Peacock and Cowan’s (2019) study recognized that sense of belonging could promote and consolidate online learning and suggest how tutors can nurture online learners’ sense of belonging. Lapointe and Reisetter (2008) sought to understand students’ perceptions of the online learning community and current interactive pedagogy effectiveness. Moreover, in the study, the issue of whether these students had different expectations of and value for online and traditional learning communities was investigated. Koole (2010) examined the relationship between self and belonging in online social settings and the factors that would affect this relationship.

As in distance education, there are various problems in face-to-face education: the lack of time flexibility, the lack of equal opportunities in education, failure to adapt technological tools, and inadequate classroom environments because of the increasing number of students (Yolcu, 2015). To minimize the drawbacks in both learning environments, researchers have developed a new classroom layout by combining the classroom environment with the web, called ‘flipped classrooms’. The flipped classroom model is designed to improve teaching and learning quality despite crowded classes, a shift in teaching from face-to-face to online environment, decreased resources over time, and difficulty updating resources (Larcara, 2014).

Flipped learning refers to a blended approach that can be contrasted with the traditional approach to education (Gilboy, 2015; Strayer, 2012). Flipped learning is student-centered, and the focus of learning passes from teachers to students, thus allowing students to have a more flexible learning environment for building knowledge (Chao et al., 2002). This system enables learners to reach out to the content they will learn in the classroom environment with the help of asynchronous plans (lecture videos and articles, various electronic data sources, visuals, pictures, and presentations) outside the school environment and, thus, ensures effective learning. The flipped classroom is a learning environment that replaces homework with classroom teaching and enables learners to focus on the problems that they face in their learning processes (Gencer et al., 2014; Verleger & Bishop, 2013). In other words, the flipped classroom system encourages individuals to learn the theoretical content at home and to practice it at school (Strayer, 2012; Zownorega, 2013). Therefore, it promotes student–teacher interaction more, provides opportunities for real-time feedback, increases student participation, and allows learning according to students’ pace (Goodwin & Miller, 2013). Among the flipped classroom method’s most significant advantages are that it provides learning environments where students can access information regardless of time and place and support individual learning (Bergman & Sams, 2012; Davies et al., 2013; Talbert, 2012).

The flipped classroom environment has been considered frequently both in the education process and academic research because of its advantages. In the studies related to the flipped classroom model, generally, students’ academic achievement and satisfaction, motivations, and engagement levels (Stone, 2012) were investigated. For example, by Alamri’s (2019) mixed-method research involved the effectiveness of a flipped classroom for students’ academic performance and satisfaction. In this study, when researchers compared traditionally and flipped classrooms, the academic performance of students in the flipped classroom group was at a high level. Also, almost all students were found to have a high level of satisfaction and generally enjoy learning in the flipped classroom environment. Mixed-method research by Turan (2015a, b) identified students’ perspectives about the flipped classroom and revealed the impact of this environment on student achievement, cognitive load, and motivation. The flipped classroom method increased students’ accomplishments and motivation and decreased their mental load levels. Furthermore students demonstrated positive views about the technique. Another mixed-method study by El-Banna et al. (2017) examined differences in academic achievement and teaching satisfaction between a flipped and traditional classroom environment in a nursing school. The results of the study revealed that students in the flipped classroom scored higher in examinations. It was also stated that instructors should provide students with sufficient knowledge and logic to use the flipped classroom.

A quasi-experimental study by Missildine et al. (2013) aimed to determine the effects of a flipped classroom and innovative learning activities on academic success and satisfaction. The traditional course only, course and course capture back-up, and the flipped classroom environment of course capture with innovative classroom activities were compared. Examination scores of students in the flipped classroom group were higher than in the other groups. Students were less satisfied with flipped classrooms. Another quasi‐experimental study by Kazanidis et al. (2019) investigated the effectiveness of the flipped classroom approach in teaching instructional media design subjects by comparing students’ academic performance and training satisfaction in traditional course‐based instruction with those in a flipped classroom. There were significant differences in academic performance and educational satisfaction between the two groups, with students in the flipped classroom performing better. An experimental study by Karabatak and Polat (2020) compared the traditional classroom model, distance education model, and flipped classroom model designed according to ARCS (attention, relevance, confidence, and satisfaction) motivation strategies of motivation and academic achievement. The training process did not lead to a significant change in the motivation of the students for the traditional classroom model and distance education classroom model but did for the flipped classroom model. Goedhart et al. (2019) examined the effects of a flipped classroom trial conducted during a Master’s course. Half of the course was taught in a traditional course-style while flipped classrooms replaced the remaining half. Although activities facilitated deeper learning, not all students agreed that the flipped classroom environment contributed to positive learning outcomes. The researchers suggested that this result be investigated further.

The quasi-experimental mixed-methods research by Davies et al. (2013) explored how technology can teach technical skills and determine the benefits of flipping the classroom in terms of student achievement and satisfaction with the class. The environments with the flipped approach and simulation-based approach were both more efficient than the traditional classroom environment. The aim of Limniou et al.’s (2018) study was for students to express their views of traditional and flipped classroom teaching approaches delivered by two teachers. Although all students had similar preferences for following either the traditional or the flipped classroom, a significant difference was observed in students’ views related to teachers’ contributions to teaching approaches, students’ higher-order thinking skills development, and students’ choice of learning material. Yilmaz’s (2017) study aimed to explore the impact of the students' e-learning readiness on student satisfaction and motivation for the flipped classroom model of instruction. Path analyses with structural equation modeling verified that students' e-learning readiness was related to their satisfaction and motivation while undertaking academic tasks in the flipped classroom model of instruction.

McLaughlin et al. (2013) investigated whether ‘flipping’ a traditional course delivered synchronously to two campuses would improve student academic performance, engagement, and perception. It was found that the flipped classroom can enhance the quality of the students’ experiences in the course through thoughtful course design, enriched dialogue, and promotion of learner autonomy. The effect of flipped and non-flipped classrooms on students’ active learning and engagement in a course was investigated in an experimental study by Stone (2012), who reported that flipping the class in a small classroom flipped the class result in large learning gains and positive attitudes towards the learning process. Many students agreed that flipped classrooms positively influenced their learning and performance in the classroom environment.

In the studies reported in the literature, students’ sense of belonging in different learning environments (e.g., face-to-face, distance learning) has been examined. However, no study involved the effects of the flipped classroom environment on students' sense of belonging. Our study fills a gap in the literature. We investigated the flipped classroom model’s impact on learners’ sense of belonging and their academic success and academic satisfaction. These three variables were examined for three different classroom model environments in this study: the flipped classroom model, distance education model, and traditional classroom model.

Method

In this section, the research model, participants, Implementation Process, data collection tools, and data analysis are discussed.

Research model

In this quantitative study, an experimental design was used to investigate the effect of the flipped classroom model on participants' academic achievement, academic satisfaction, and general belongingness. The experimental design is called an intervention or group comparisons study that keeps various groups under observation to establish any possible change within groups (Creswell, 2012).

In the experimental model, a true experimental research design involves a pretest, a posttest, control groups, and random assignment of participants from a previously-determined subject pool (Creswell, 2012). In this study, one experiment and two control groups were formed. The research groups and procedures are described in Table 1. As seen in the table, the learning period took eight weeks for all groups, and the same questionnaires were used for both pretest and posttest.

Table 1 Research design and procedures
Full size table

Table 1 shows the random design with the pretest–posttest control group. For eight weeks, the traditional classroom model was implemented with the Control-1 group in a face-to-face environment Distance education was implemented with the Control-2 group in a web-based environment. The flipped classroom model was implemented with the Experimental group in both face-to-face and web-based environments together. The same questionnaires were used for both pretest and posttest for all groups.

Participants

We used purposeful sampling in which the participants are determined by the researcher (Cohen et al., 2005; Silverman, 2006). For this reason, the study was carried out with the selected undergraduate students who have the skills to use information and communication technologies and have easy access to the internet. Participation in the study was voluntary.

This study was conducted in a faculty of education at a state university in middle-eastern Turkey. In this education faculty, 72.50% of these students are female, and 27.50% are male. Therefore, 75 (79.79%) of the 94 students who participated in the application were girls, and 19 (20.21%) were boys. Regarding daily internet use, 33 students who used the internet for 1–3 h, 47 students used it for 4–6 h, and 14 students used it for more than seven hours. Almost all students accessed the resources from home/dormitory, and four of them accessed them from school. Regarding accessing the resources, 72 of the students stated that they had no problems, 12 of them indicated that they had issues, and 10 indicated that they had a few problems. The main problems mentioned by the participants were technical issues, internet access or connection problems, login issues, and time restriction issues.

One critical element in experimental studies is the homogeneous structure of the groups (Creswell, 2012). A single instructor conducted courses. Before the implementation process, an academic achievement test was administered to all participants to ensure homogeneity and the equivalence of the groups. Therefore, students were divided into three groups by the researchers. But most of the students in the education faculty were female, but there were 6 or 7 male students in each group. Students were informed about the operation of the courses in three groups (face-to-face education, distance education, and flipped classroom). Later, students who wanted to change their groups were allowed to do so without disrupting the equivalence of the groups. The distribution of the participants in the groups is given in Table 2.

Table 2 Gender distribution of participants
Full size table

As shown in Table 2, while experimental and control groups were formed, the genders of participants were also taken into account and their academic achievement levels. As shown in Table 2, there was no significant difference between the pretest scores of the Control-1 (\(\stackrel{-}{X}\hspace{0.17em}\) = 30.57), Control-2 (\(\stackrel{-}{X}\) = 29.56) and Experimental (\(\stackrel{-}{X}\hspace{0.17em}\) = 30.15) groups [F(2, 91) = 0.08, p > 0.05]. For this reason, the groups were considered similar before the implementation processes.

Implementation process

Before the implementation process, necessary permissions were obtained from the university where the study was carried out. Then, the participants were informed about the implementation, and the pretests were administered. Then, groups were formed based on participants’ academic achievement test scores and genders, as shown in Table 2. The syllabus and all planned activities were followed in the same sequence in three groups. The course was prepared in a modular structure. The implementation of the Information Technology course lasted for eight weeks, and each module was completed over a week. The modules are given in Table 3.

Table 3 Modules, environment and materials
Full size table

As seen in Table 3, in the implementation process, the traditional classroom model was used for the Control-1 group, distance education model for the Control-2 group, and flipped classroom model for the Experimental group. In all three classroom models, the instructor delivered all the resources (documents, videos, and study files prepared by the instructor) to the participants for their use every week. Within the classroom models framework, the participants in the Control-1 group attended the courses in face-to-face sessions, and practices were conducted in the computer laboratory. With the Control-2 group, web-based synchronous (virtual classroom) and asynchronous (documents, videos, and study files prepared by the instructor) sessions were conducted, and participants were allowed to use the computer laboratory at any time.

Participants in the experimental group benefitted from the web-based asynchronous environment and resources (documents, videos, and study files prepared by the instructor) used by the Control-2 group. Also, they attended face-to-face sessions. The procedures in the flipped classroom out as follows:

  • The instructor gave the experimental group an assignment with all the web-based environment resources before coming to the class.

  • Participants watched videos and presentations, took notes, and completed their homework (wherever they were), and prepared questions about what they could not do in their homework.

  • Participants communicated with the instructor in a web-based environment about what they needed to ask before the face-to-face sessions.

  • In the face-to-face sessions, in line with the nature of the questions, the instructor grouped participants before discussing the issues they had in the group and helped each other. The instructor assisted when necessary in these sessions.

  • For questions that could not be answered, the instructor directed the participants to resources to find answers.

  • The instructor asked students to do another example to test whether they had learned the module.

At the end of the implementation process, all posttests were administrated to the participants in three groups.

Instruments

For data collection, three questionnaires were used: Academic achievement-test, Academic Satisfaction Scale, and General Belongingness Scale. Detailed explanations of the scales are given below.

Academic achievement test

The Academic achievement test developed by Bingöl and Halisdemir (2017) consists of 26 questions. Item difficulty index values (Pj) ranged from 0.21 to 0.49. The average difficulty (p) of the test was 0.59. The test’s Kuder Richardson-20 reliability was found to be 0.51.

Academic satisfaction scale

A five-point Likert-type Academic Satisfaction Scale developed by Schmitt et al. (2008) was used to determine the satisfaction level of university students. For this scale, individuals are asked to evaluate themselves for each item within the range from Strongly Disagree (1) to Strongly Agree (5). The internal consistency reliability of the scale was reported as 0.81 (Schmitt et al., 2008). The scale was adapted to Turkish by Balkis (2013). The internal consistency coefficient of the scale was reported as 0.86 for the Turkish version. Some of the 5-item scale items are "All in all, I am satisfied with the education I can get in this school" and "I’m happy with the amount I learn in my classes". In our study, this reliability coefficient was 0.85.

General belongingness scale

The General Belongingness Scale, developed by Malone et al. (2012), consists of 12 items with two dimensions (Acceptance/Inclusion and Rejection/Exclusion). The scale uses a seven-point Likert-type response ranging from Strongly Disagree (1) to Strongly Agree (7). The scale was adapted to Turkish by Duru and Balkis (2015). The internal consistency coefficient of the whole scale and its dimensions called “Acceptance”, “Inclusion and Rejection”, and “Exclusion” were calculated as 0.92, 0.89, and 0.09, respectively. Analyzes for confirmatory factor analysis of the scale supported the model (χ2/df = 2.076. GFI = 0.92, RMSEA = 0.07, SRMR = 0.04, CFI = 0.96, TLI = 0.95, RFI = 0.91, NFI = 0.93). Some of the items are “When I am with other people, I feel included” and “I feel accepted by others” in the Acceptance/Inclusion dimension and “I feel like an outsider” and “I feel as if people do not care about me” in the Rejection/Exclusion dimension. In this study, the internal consistency coefficient of the scale was calculated as 0.93.

Data analysis

The Kolmogorov–Smirnov (K-S) test was used to examine whether the data were normally distributed (Dewan & Somanathan, 2007). Values between − 1.5 and + 1.5 are acceptable for normal distribution (Can, 2013; Tabachnick & Fidell, 2007). In this study, the K-S test, skewness, and kurtosis coefficients were calculated to ascertain whether group scores showed normal distribution or not for this purpose.

In this study, statistical analyses were performed with parametric tests (dependent groups t-test) or non-parametric tests (Wilcoxon signed-rank test). ANCOVA (Analysis of Covariance) was used to compare control and experimental groups. For the ANCOVA to give accurate results, the data of the groups are required to show normal distribution, their variances should be equal, a linear relationship should exist between the common variable and the dependent variable, and the slope of the regression lines should be equal (Buyukozturk, 2011; Can, 2013; Kalayci, 2010; Tabachnick & Fidell, 2007). However, ANOVA (Analysis of Variance) is used when ANCOVA does not satisfy assumptions. In cases where significant differences were observed, post hoc tests were employed to find the source of difference. LSD was used when the equal variance assumption was met and Tamhane’s T2 test was used if there was a violation of the equal variance assumption.

Effect sizes were calculated in addition to statistical significance (Cohen et al., 2005). Statistical tests reveal whether there is a significant difference between two or more means, but they do not give information about the magnitude of differences (Can, 2013). The effect size d was calculated for the dependent groups with significant differences between scores, and η2 was calculated for the independent groups with significant differences between scores. The value of d is considered to indicate a very high effect size if it is greater than 1, a high effect if it is between 1 and 0.8, a medium effect if it is between 0.7 and 0.5, and a low effect if it is between 0.4 and 0.2 (Cohen, 1992; Thalheimer & Cook, 2002). The value of η2 is considered a high effect if it is greater than 0.14, a medium effect if it is between 0.13 and 0.06, and a low effect if it is between 0.05 and 0.01 (Can, 2013).

Findings

In this section, results are reported based on students’ academic achievement, academic satisfaction, and general belongingness levels.

Findings related to academic achievement

Because data related to the academic achievement of the groups were normally distributed, dependent groups t-tests were performed to compare groups’ pretest and posttest scores. The results are given in Table 4.

Table 4 Difference between pretest and posttest in academic achievement for three instructional groups
Full size table

As shown in Table 4, significant differences (p < 0.001) were found between the pretest (\(\stackrel{-}{X}\hspace{0.17em}\) = 30.57) and posttest (\(\stackrel{-}{X}\hspace{0.17em}\) = 45.13) scores of the Control-1 group, between the pretest (\(\stackrel{-}{X}\hspace{0.17em}\) = 29.56) and posttest (\(\stackrel{-}{X}\hspace{0.17em}\) = 48.44) scores of the Control-2 group, and between the pretest (\(\stackrel{-}{X}\hspace{0.17em}\) = 30.34) and post-test (\(\stackrel{-}{X}\hspace{0.17em}\) = 53.50) scores of the Experimental group. Also, high effect size values (dcontrol-1 =  − 1.09; dcontrol-2 =  − 2.09; dexperimental =  − 2.13) were observed in all groups after the implementation process. The most important finding is that the experimental group had the highest increase in terms of academic achievement.

ANCOVA was used to compare the academic achievement posttest scores of the Control-1, Control-2, and Experimental groups. Before conducting ANCOVA, four assumptions were tested. According to the first assumption, the data showed a normal distribution. For the second assumption, the variances of the achievement posttest scores were determined to be equal (F = 6.955, p = 0.54). For the third assumption, the slopes of the regression lines were close to each other and there was a linear relationship between the dependent variable (pretest) and the common variable (posttest), thus justifying conducting ANCOVA. For the last assumption, the group x pretest interaction for academic achievement posttest scores was found to be statistically nonsignificant (F = 1.826; p = 0.167). These findings show that the regression lines for testing the posttest scores apply equally to the pretest scores. This means that the fourth assumption was also met.

After all assumptions had been tested, ANCOVA was applied. Adjusted academic achievement pretest scores for each group were calculated along with descriptive statistics. The new value for the Control-1 group was 45.00, for the Control-2 group was 48.63, and for the Experimental group was 53.44. ANCOVA was conducted to determine whether this change between academic achievement pretest and posttest scores was statistically significant. ANCOVA results are shown in Table 5.

Table 5 ANCOVA results comparing posttest achievement of three groups while controlling pretest achievement
Full size table

Academic achievement pretest scores were included in the analysis as a covariate or control variable, and the significance of the difference between the three groups’ achievement posttest scores was tested (Table 5). There was a significant difference between the adjusted academic achievement posttest scores according to the pretest scores [F(2–90) = 8.056, p = 0.001]. To determine the magnitude of the difference, the effect size (η2 = 0.13) was calculated and was found to be moderate. According to the LSD test conducted to identify the differences between the adjusted posttest scores of the three groups, the academic achievement of the experimental group using the flipped classroom model was higher than for Control-1 using the face-to-face classroom model and for Control-2 using the distance education.

A dependent groups t-test was performed to compare the pretest and posttest satisfaction scores of the three groups. The results are provided in Table 6.

Table 6 Changes in academic satisfaction between pretest and posttest for three instructional groups
Full size table

As seen in Table 6, although no significant difference was observed between pretest (\(\stackrel{-}{X}\hspace{0.17em}\) = 3.85) and posttest (\(\stackrel{-}{X}\hspace{0.17em}\) = 3.74) academic satisfaction for Control-1 group and between pretest (\(\stackrel{-}{X}\hspace{0.17em}\) = 3.48) and posttest (\(\stackrel{-}{X}\hspace{0.17em}\) = 3.68) academic satisfaction for Control-2 group (p > 0.05), a significant difference was found between pretest (\(\stackrel{-}{X}\hspace{0.17em}\) = 3.70) and posttest (\(\stackrel{-}{X}\hspace{0.17em}\) = 4.29) academic satisfaction of the Experimental group (p < 0.001). Moreover, the effect size for the experimental group is high (d =  − 0.89).

ANCOVA was used to compare the academic satisfaction posttest scores of Control-1, Control-2, and Experimental groups. According to the first assumption of ANCOVA, the data showed normal distribution. For the second assumption, the variances of posttest scores were equal (F = 5.929, p = 0.17). For the third assumption, the slope of the regression lines was close to each other. There was a linear relationship between the dependent variable (pretest) and the common variable (posttest) and it, suggesting suitability for ANCOVA. For the last assumption, the group x pretest interaction on academic achievement posttest scores was statistically nonsignificant [F = 0.457; p = 0.635]. These findings show that the regression lines calculated for testing the post-test scores were equal based on the pre-test scores. This means that the fourth assumption was also met.

The adjusted academic satisfaction pre-test scores for each group were calculated with descriptive statistics. The new value for Control-1 group was 3.66, for Control-2 group was 3.77, and for the Experimental group was 4.28. ANCOVA was conducted to determine whether the change in academic satisfaction between pretest and posttest was statistically significant (Table 7).

Table 7 ANCOVA results comparing posttest satisfaction of three instructional groups while controlling for pretest satisfaction
Full size table

Academic satisfaction pretest scores were included as a control variable while the significance of between-group differences in academic satisfaction posttest scores was tested (Table 7). There was a significant difference between the three instructional groups in their posttest satisfaction scores when pretest was controled [F(2–90) = 6.213, p = 0.003]. Regarding the magnitude of the difference, the effect size (η2 = 0.12) size was moderate. According to the LSD test conducted to reveal the differences between the adjusted posttest scores of the groups, the academic satisfaction of the experimental group using the flipped classroom model was higher than for Control-1 using the face-to-face classroom model and for Control-2 using distance education.

Findings related to general belongingness

To compare the three instgructional groups on acceptance/inclusion, pretest and posttest scores for the General Belongingness Scale were compared using a dependent groups t-test (Table 8).

Table 8 Difference between pretest and posttest acceptance/inclusion for three instructional groups
Full size table

As seen in Table 8, there was no significant difference between pretest (\(\stackrel{-}{X}\hspace{0.17em}\) = 4.13) and posttest (\(\stackrel{-}{X}\hspace{0.17em}\) = 4.03) scores for acceptance/inclusion of the Control-1 group, or between pretest (\(\stackrel{-}{X}\hspace{0.17em}\) = 3.96) and posttest (\(\stackrel{-}{X}\hspace{0.17em}\) = 4.10) scores of the Control-2 group. But a significant change was found between the pre-test (\(\stackrel{-}{X}\hspace{0.17em}\) = 4.33) and posttest (\(\stackrel{-}{X}\hspace{0.17em}\) = 4.59) scores of the experimental group (p < 0.001). The effect size of 0.61 for the experimental group was medium. Thus, the acceptance/inclusion dimension of the general belongingness scale had the highest pretest–posttest increase for the experimental group with the flipped classroom model.

It was checked whether ANCOVA test assumptions were met before attempting to compare acceptance/inclusion posttest scores of the three groups. The data did not show a normal distribution, and the variances were not equal (F = 6.106, p = 0.006). Therefore, ANOVA was used to compare acceptance/inclusion posttest scores of the three groups (Table 9).

Table 9 ANOVA results for comparison of three instructional groups on acceptance/inclusion posttest scores
Full size table

Table 9 reveals a significant difference between the three groups in terms of their acceptance/inclusion posttest scores [F(2,91) = 5.481; p < 0.05]. To identify the source of the difference, Tamhane’s T2 test used to identify that the difference was between the experimental group (\(\stackrel{-}{X}\hspace{0.17em}\) = 4.59) and Control-1 (\(\stackrel{-}{X}\hspace{0.17em}\) = 4.03) and Control-2 (\(\stackrel{-}{X}\hspace{0.17em}\) = 4.11) in favor of the experimental group. The effect size was medium level (η2 = 0.12). In summary, the experimental group with the flipped classroom model had higher scores than the two control groups in terms of the acceptance/inclusion dimension of the General Belongingness Scale.

Pretest and posttest scores for the rejection/exclusion dimension showed a normal distribution for Control-1 data, but not for Control-2 and Experimental groups. Therefore, for Control-1, pretest and posttest rejection/exclusion scores were compared using a dependent group t-test. The results provided in Table 10 confirm no significant difference between pretest (\(\stackrel{-}{X}\hspace{0.17em}\) = 4.05) and posttest (\(\stackrel{-}{X}\hspace{0.17em}\) = 3.96) rejection/exclusion scores in Control-1 group.

Table 10 Difference between pretest and posttest rejection/exclusion scores for Control-1 group
Full size table

To compare pretest and posttest rejection/exclusion scores for Control-2 students, the Wilcoxon signed-rank test was performed. The results Table 11 confirm no significant difference between the pretest and posttest scores for Control-2 group [Z =  − 2.868; p = 0.062].

Table 11 Difference between pretest and posttest scores for rejection/exclusion for Control-2 group
Full size table

The Wilcoxon signed-rank test was also performed for the experimental group to compare rejection/exclusion pretest and posttest scores. The results in Table 12 show no significant pretest–posttest difference in rejection/inclusion scores for the Experimental group [Z =  − 1.217; p = 0.224].

Table 12 Comparisons of pretest and posttest rejection/exclusion scores for Experimental group
Full size table

ANOVA was performed to compare the three groups on their rejection/exclusion posttest scores. The results in Table 13 show a significant difference among the groups in rejection/exclusion posttest scores [F(2,91) = 5.551; p < 0.05]. According to the Tamhane’s T2 test, a significant difference existed between the experimental group (\(\stackrel{-}{X}\hspace{0.17em}\) = 4.65) and Control-1 group (\(\stackrel{-}{X}\hspace{0.17em}\) = 3.96), in favor of the experimental group and with a medium effect size (η2 = 0.10). The experimental group scores using the flipped classroom model had higher rejection/exclusion posttest scores.

Table 13 ANOVA results for comparison of three instructional groups on posttest rejection/exclusion scores
Full size table

Conclusion and discussion

In this study, the effects of the flipped classroom model on students’ academic achievement, academic satisfaction, and general belongingness were investigated. For over eight weeks, an Information Technology course was taught to experimental and control groups using different learning techniques and classroom models.

When the three instructional groups were compared in terms of academic achievement, significant differences were observed between the pretest–posttest changes of the groups and among the groups’ posttest scores. One of the critical findings is that the experimental group with the flipped classroom model had the largest increase in academic achievement, with a very high effect size. Results of the study by Karabatak and Polat (2020) indicated that the academic achievement levels of students in the flipped classroom model using ARCS motivation strategies were significantly higher than those of the students in the distance education classroom model and the traditional classroom model. For the flipped classroom model, the instructor gives the course resources to the students before face-to-face sessions. Regardless of time and environment, students study these resources and mostly take advantage of the videos. Then, by using these resources, students try to do the homework given by the instructor. Also, they have an opportunity to ask the instructors and their group members about the points that they do not understand while doing the homework in a face-to-face environment. These steps in flipped classrooms can make learning more effective. Naturally, this situation can be interpreted as an essential explanation of why students in the flipped classroom model have higher academic achievement than other classroom models.

This finding is parallel to the results of some other studies of the effect of flipped classrooms on academic achievement (Alamri, 2019; Chen Hsieh et al., 2017; Davies et al., 2013; Karabatak & Polat, 2020; El-Banna, 2017; Evseeva & Solozhenko, 2015; Karadeniz, 2018; Karagol & Esen, 2018; Kazanidis et al., 2019; Kim et al., 2014a, b; McLaughlin et al., 2013; Orhan, 2019; Peterson, 2016; Sergis et al., 2018; Stone, 2012; Zainuddin & Halili, 2016; Zhonggen & Guifang, 2016). In addition to academic achievement, a few studies also confirmed an increase in motivation, which is a critical element in academic achievement, in the flipped classrooms (Chao et al., 2015; Chen Hsieh et al., 2017; Evseeva & Solozhenko, 2015; Peterson, 2016; Zainuddin & Halili, 2016). Moreover, in other studies, increases in students’ academic skills (Asiksoy & Ozdamlı, 2016; Deslauriers et al., 2011; González-Gómez et al., 2016; Hung, 2014; Kong, 2014; Kostaris et al., 2017; Love et al., 2013; O’Flaherty & Phillips, 2015; Roach, 2014; Schultz et al., 2014; Turel & Sanal, 2018) and participation (Chen et al., 2014; Deslauriers et al., 2011; Hung, 2014; Willey & Gardner, 2013) were observed.

On the other hand, other empirical studies reported no significant effect of the flipped classroom model on students’ academic achievement (Frydenberg, 2013; Winter, 2013) or small but statistically-significant improvements (Mason et al., 2013; Murphree, 2014; Talley & Scherer, 2013; Tune et al., 2013; Wilson, 2013). In another study by Goedhart et al. (2019), the combination of personalized pre-classroom learning and peer-learning classroom activities facilitated deeper learning in the flipped classroom. However, although students' experience was positive in our study, not all students agreed that flipped classroom environments contributed to positive learning outcomes. Therefore, future research is needed to clarify this situation by considering various variables that could affect students’ academic achievement in flipped classrooms.

When students’ academic satisfaction scores were compared for the three groups, nonsignificant results were found in the distance education and traditional education environments. However, a statistically significant increase occurred in students’ academic satisfaction in the flipped classroom. Students in the flipped classroom had a more considerable increase in their satisfaction level than students in distance education and traditional education classrooms. In the flipped classroom model, students attend face-to-face sessions and have an opportunity to learn subjects or topics that they do not understand from their friends or instructors. This situation can lead to an increase not only in academic success but also in student satisfaction.

Similar results were found in other studies focusing on the flipped classroom model (Alamri, 2019; Alsowat, 2016; Cigdem et al., 2016; Davies et al., 2013; El-Banna, 2017; Kazanidis et al., 2019; Moraros et al., 2015; Yilmaz, 2017; Zhonggen & Guifang, 2016). Cigdem et al. (2016) found that self-regulation, interaction among learners, and course content are the critical predictors of academic satisfaction in the flipped classrooms. Also, they posited that instructor–student interaction did not influence students’ academic satisfaction. Despite the significant results of some prior studies, other studies reported a non-insignificant link between flipped classrooms and academic satisfaction (Missildine et al., 2013; Whillier & Lystad, 2015). Moreover, Strayer (2012) concluded that students were less satisfied in the flipped classrooms than in traditional classrooms. Thus, to examine academic satisfaction, future research should include various variables, including students’ readiness to be part of a flipped classroom, their self-regulation and learning abilities, their interactions among students, and the course content that affects learning in a flipped classroom.

Finally, in this study, the three groups were compared in terms of the general belongingness level of students. Despite nonsignificant pretest–posttest changes for the control groups, a significant difference was observed for flipped classroom students. In terms of posttest scores, students in the flipped classroom had higher belongingness than the control groups. In the flipped classroom model, the instructor divided students into groups according to the topics they did not understand during the face-to-face session and allowed them to learn from each other. It can be said that peer learning occurs while students study with their group friends in the same classroom environment. Thus, this situation is likely to enable students to feel belonging to a group or a community, and therefore students' sense of belonging develops.

In the literature, the belongingness level of participants was not included as a variable in studies related to the flipped classroom model. Our study is among the first to combine general belongingness as a research variable focusing on the flipped classroom model. However, it has been included in studies related to blended learning (Keskin & Seferoglu, 2017; Rovai & Jordan, 2004; Tayebinik & Puteh, 2012) and online learning (Alexander, 2001; Emmanuel-Frenel, 2017; Koole, 2010; Lapointe & Reiseter, 2008; Peacock & Cowan, 2019; Thomas et al., 2014) approaches. In these studies, it was found that belongingness was higher in blended learning environments. Moreover, Keskin and Seferoglu (2017) found that preservice teachers in blended learning environments had a good sense of belonging to their profession and their family and friends. Thus, future research should include the sense of general belonging when evaluating the flipped classroom model. Thomas et al. (2014) found that a sense of belonging could lead to greater learner satisfaction, and Emmanuel-Frenel (2017) found that increased use and satisfaction regarding student support services in distance education were associated with students' sense of belonging to the learning community.

Limitations of the study

There are some limitations to this study. Because participants consisted of undergraduate students, this study cannot be generalized to other groups regarding the effects of the flipped classroom model on students’ academic achievement, academic satisfaction, and general belongingness level. Thus, future research should be conducted with different participants from various levels of education. Another limitation is related to the implementation process. For this study, the implementation lasted for eight weeks because of the difficulty in preparing course materials, but more time is needed to prepare course materials. Future research should take this time limitation factor into account. In line with this issue, the Ministry of National Education of Turkey, with support from universities, should provide resources for the preparation of digital materials, videos, and presentations for flipped classrooms. The last limitation is that the criteria for selecting the participants and the course might have affected participants’ academic achievement, academic satisfaction, and general belongingness because, in this study, the flipped classroom was formed for the Information Technologies course and participation in the study was voluntary and involved participants who could use technology effectively. Therefore, replication of the study is suggested with various courses and participants to identify the effects of the flipped classroom model.

Conclusion

In conclusion, technology integration into education influences students’ attention towards course content, learning pace, motivation, self-confidence, and attitudes towards courses, as well as increasing self-evaluation and academic achievement (Arabaci & Polat, 2013). This study confirmed that a flipped classroom model that integrates technology into the educational environment improved students’ academic achievement, academic satisfaction, and general belongingness. Currently, international organizations such as the European Union (EU) and the Organization for Economic Development and Cooperation (OECD) and countries such as the United States, Japan, Singapore, and Australia have allocated large budgets for projects based on the use of information technologies in education (Gokoglu, 2014). To increase the quality of education in Turkey, similar projects and investments have been employed. Specifically, the Digital Transformation in Higher Education Project that started in February 2019 requires the transformation of learning processes into the digital environment and the flipped classroom model. Similar projects should be introduced to the Turkish educational system to improve the quality of education.