At the end of 2019, Corona Virus Disease 2019 (COVID-19) appeared in China. As the virus spread, the World Health Organization (WHO) designated it a pandemic (WHO, 2020). Since then, the ongoing pandemic has been seriously disruptive to teaching/learning activities worldwide (Sharma et al., 2020), and nearly 90 percent of the world’s students have been influenced (UNESCO, 2020). Under this circumstance, school closures have been implemented at all levels from preschool to higher education (Simsek et al., 2021), and most higher education institutions around the world shut down their campuses in late March 2020 (Wong et al., 2020). As a result, higher education institutions have been forced to switch to an online teaching mode. Online learning did not originate during the COVID-19 pandemic, and higher education students had been exposed to this teaching mode for many years before the unprecedented closure. However, online learning on such a large scale is unprecedented in the history of higher education. Compared with normal online classes, this emergency remote teaching/learning could not be well-planned instructors do not have enough time. An important strand of the education literature has focused on online learning issues during the COVID-19 pandemic, especially issues related to students’ online learning.

Research has indicated that online learning has been a must for educational institutions during the pandemic to limit the disruption of education, as online learning is accessible, affordable, and flexible (Faize and Nawaz, 2020). Maqableh and Alia (2021) confirmed that safety and effectiveness are the two most important positive aspects of online learning. By taking an online approach, students and teachers have been less likely to catch the COVID-19 virus. However, the problems of online learning during this special period cannot be ignored. First, from a macro viewpoint, the transition to online learning during the pandemic has aggravated education inequalities. Online learning’s dependency on technological devices and the internet cannot be ignored. Accordingly, instructors and students with bad internet connections have been more likely to be excluded from online learning (Adedoyin and Soykan, 2020). Agormedah et al. (2020) also proposed that students’ ability to have access to e-learning devices and internet connectivity would determine the success of the students’ shift to remote learning. However, students in economic and financial distress are less likely to afford an unlimited and stable internet connection. This result buttressed the study of Adedoyin and Soykan (2020), which found that the online approach was more accessible for advantaged students with access to technology and hindered the participation of disadvantaged learners. Second, from a micro viewpoint, students with access to online learning also have faced challenges such as distraction, workload, technology issues, and inadequate support (Maqableh and Alia, 2021). Similarly, Faize and Nawaz (2020) also reported that students faced challenges such as a lack of resources and internet services and a lack of interaction. Because of these drawbacks to online learning, Mahyoob (2020) found that online learning had failed to allow students to fulfil their expected progress in learning performance, and that most learners were not satisfied with continuing online learning.

There is no denying that online practice is irreplaceable in circumstances such as the COVID-19 pandemic. Still, the problems that impede effective online education should be given more attention, especially those that decrease students’ satisfaction with online learning. In the short term, students’ satisfaction is really important for realizing effective online education practices (Sharma et al., 2020) and meaningful for promoting the sustainable development of online courses. As of May 2022, the epidemic had lasted for more than 2 years. It is uncertain how long the pandemic period will continue. Students have to prepare for long-term online learning as the epidemic becomes normalized. It is indispensable to investigate the factors influencing students’ satisfaction with online learning during the pandemic and to improve the negative aspects of this type of learning. In the long term, this situation will accelerate the transformation of traditional pedagogical approaches and force academic institutions to shift to online teaching and learning (Adedoyin and Soykan, 2020). One day, education activities may shift entirely to online teaching and learning, even if the epidemic is resolved. Students’ adoption of online education and their satisfaction with it is essential for the sustainable development of online learning. Therefore, exploring factors behind students’ dissatisfaction with online learning and dealing with these factors can further improve education equity in different regions.

Few researchers have realized that some studies should be conducted to investigate students’ satisfaction with online learning during the pandemic. There has been limited in-depth exploration of the factors affecting students’ satisfaction with online learning, and no consensus has been reached among these studies’ results about students’ satisfaction. Furthermore, prior studies exploring the configuration of factors influencing students’ satisfaction with online learning during the COVID-19 pandemic in particular have been rare. In terms of research method, the existing studies have used fairly similar approaches; they have mostly been quantitative studies using methods such as cross-sectional design (Faize and Nawaz, 2020; Simsek et al., 2021) or structural equation models (Jiang et al., 2021; Kumar et al., 2021). Although some studies have adopted mixed methods (Landrum et al., 2021), it is challenging to find a single studying using a qualitative comparative analysis method. Motivated by these considerations, this research aimed to explore the factors affecting higher education students’ satisfaction with online learning during the COVID-19 pandemic using the fuzzy-set Qualitative Comparative Analysis (fsQCA) method. The basic idea of the fsQCA method is to analyze the influence of multiple antecedent variable combinations on result variables using Boolean operation. In recent years, the fsQCA method has gained increasing attention because it can explore the influence of collocations of different conditional variables on specific outcome variables. It focuses on the joint action of multiple influencing factors and provides a variety of combinations of condition variables. According to previous research, student satisfaction with online learning during the pandemic is influenced by multiple factors (Baber, 2020; Faize and Nawaz, 2020; Sharma et al., 2020; Jiang et al., 2021; Maqableh and Alia, 2021; Simsek et al., 2021). Student satisfaction may not be fully explained by individual factors acting alone, and the fsQCA method could explore collocations of different factors influencing student satisfaction. Therefore, this research utilized the fsQCA method to analyze student satisfaction using combinations of various factors.

Additionally, many types of research investigating student satisfaction have been conducted in Asian higher education settings. For example, Almusharraf and Khahro (2020) explored the satisfaction of 283 students from the Kingdom of Saudi Arabia with online learning. Besides, Simsek et al. (2021) investigated students’ satisfaction with online education in Turkey. Nevertheless, student satisfaction with online learning during the pandemic needs to get attention in other Asian countries, especially in countries such as China that have taken positive actions to combat the difficulties with online learning during the COVID-19 pandemic. To be specific, the Ministry of Education of the People’s Republic of China (2020a) launched an emergency policy called “Suspending Classes Without Stopping Learning” to ensure that students could continue to study at home through online learning platforms on January 29, 2020. From January 29 to April 3 of that year, 1,454 Chinese universities began their spring semester using online learning platforms. More than 950,000 university teachers offered more than 7,133,000 lectures on online learning platforms (Ministry of Education of the People’s Republic of China, 2020b). The Chinese government also took prompt action to guarantee the availability of fast and stable network services to students to make online education successful (Zhang et al., 2020). This research attempted to measure the factors affecting university students’ satisfaction with online learning during the COVID-19 pandemic in China using the fsQCA approach to provide implications for administrators and policymakers in the education field. This study also provides possible suggestions arising from the findings, especially for the future development of online learning.

The research questions guiding this study were:

To what extent have Chinese higher education students been satisfied with online learning during the COVID-19 pandemic?

Satisfaction is the pleasure or disappointment generated by comparing the perceived relationship Satisfaction is the pleasure or disappointment generated by comparing the perceived relationship between performance and expectations (Kotler and Keller, 2011). Student satisfaction can be defined as students’ perception of the university experience and the perceived value of education in educational institutions (Astin, 1993). As a short-term attitude, student satisfaction originates from a student’s evaluation of their own educational experience (Elliott and Healy, 2001). Considering all factors, Weerasinghe and Fernando (2017) concluded that student satisfaction refers to students’ short-term attitude based on their evaluation of their educational experience, services, and facilities.

Researchers have held different views when regarding the factors affecting higher education students’ satisfaction. Astin (1993) identified that contact time with teachers and administrators, career consultants, campus life, and relationships with teachers and administrators are important. However, Mai (2005) found that the most influential factors are the overall impression of the school, education quality, teachers’ expertise, the quality and availability of technical facilities, and the prospect of career development. Wilkins and Balakrishnan (2013) believed that lecturers’ quality, the quality of facilities, and the effective use of technology are important. Additionally, feedback quality, lecturer–student relationship, interaction between students, and availability of learning equipment and learning materials greatly affect student satisfaction (García-Aracil, 2009; Sojkin et al., 2012). It can be seen that students’ satisfaction is influenced by multidimensional factors and that influential factors in traditional face-to-face classes are relationships with faculty members, curriculum and instruction, campus life, support services, and facilities (Bolliger and Martindale, 2004). Compared with traditional classes, online courses bring challenges to instructors and students. Therefore, exploring student satisfaction within online settings is worthwhile because technologies have changed students’ interaction modes (Kaminski et al., 2009).

Various studies have provided useful insights into online learning students’ satisfaction during the COVID-19 pandemic (Baber, 2020; Faize and Nawaz, 2020; Sharma et al., 2020; Jiang et al., 2021; Maqableh and Alia, 2021; Simsek et al., 2021). Some studies have focused on students from one country and even from a single higher education institution. Sharma et al. (2020) explored online learning students’ satisfaction and its predictors among students at Chitwan Medical College. The results showed that more than half of the students were satisfied with online learning. Learners’ characteristics, technological characteristics, instructors’ characteristics, management, and coordination were factors significantly related to students’ satisfaction with online learning. These researchers also found that gender, Wi-Fi, and other network forms impact learners’ online learning satisfaction.

Regarding instructors’ characteristics, Sultana and Khan (2019) concluded that teachers’ performance significantly influences students’ satisfaction with online learning. Gender as an influencing factor has also been pointed out by Simsek et al. (2021), and they revealed that gender, discipline, education level, and grade level could cause changes in online learning students’ satisfaction levels during the pandemic. However, gender is a demographic feature that cannot be influenced or changed from the outside to improve the online learning process further. Furthermore, the other variables in Simsek et al.’s (2021) research—discipline, education level, and grade level—are also demographic features. Factors that are demographic features will not be considered in this research.

There have also been studies that have compared students from different backgrounds when exploring student satisfaction levels and the factors affecting them. For example, Jiang et al. (2021) explored online learning satisfaction in higher education by comparing students from two Chinese universities. The findings from this study showed students’ computer self-efficacy directly impacted their satisfaction with online learning. In addition, Baber (2020) conducted a cross-country study examining the determinants of undergraduate student satisfaction in South Korea and India. The study found that interaction, study motivation, course structure, instructor knowledge, and facilitation positively influence student satisfaction. Baber (2020) also found that there is no significant difference in student satisfaction between South Korea and India. Sharma et al.’s (2020) results supported instructor knowledge as a factor influencing student satisfaction. Faize and Nawaz (2020) also pointed out the importance of interaction to students’ satisfaction. They measured students’ satisfaction levels using Kranzow’s (2013) guidelines as a theoretical framework. Kranzow (2013) addressed students’ satisfaction through two channels. The first one was faculty–student interaction, which ensures the course content and the technology required to attend online courses are accessible to students. If students have any technical problems, the instructor will assist them in solving them. The second was the interaction between students and their peers. However, Maqableh and Alia (2021) revealed that reduced focus and psychological and administrative issues are also influential factors. This result is different from the findings of other previous literature reviewed in this section. In addition, there are many contradictions in their findings. For example, Maqableh and Alia (2021) found that more than a third of their participants were dissatisfied with their online learning experience. They identified that the respondents were not satisfied that they had more assignments. However, Simsek et al. (2021) found that students’ satisfaction level with online learning during the pandemic was moderate.

It can be deduced from the above-listed literature that previous studies have had inconsistent findings regarding the factors influencing student satisfaction with online learning during the COVID-19 pandemic, although there have been some overlapping factors. Besides, no study has considered combinations of various causal factors rather than single factors influencing student satisfaction.

This research explored how the non-demographic variables mentioned in section 2.2 (instructors, assignments, interactions, faculty services, and students’ internet self-efficacy) that have been proved to separately impact students’ satisfaction with online learning during the COVID-19 pandemic work as configurations. The conceptual framework for this research was based on Astin’s (1984) theory of student involvement.

Astin (1984) put forward his student involvement theory while investigating how the environment affects students’ learning and development. “Student involvement” refers to students’ physical and psychological effort in learning. A highly involved student will devote more time and energy to their studies, actively participate in various school activities, and interact with their classmates, instructors, and other staff. Accordingly, their satisfaction with school life will also improve. Astin’s theory has three key points: students’ input, environment, and output.

Students’ input refers to their potential and characteristics, family background, and academic and interpersonal relationships before the learning process. Environment refers to students’ actual experience in educational activities, including management strategies and practices in schools, courses, teaching, peer relations, and campus facilities. Output refers to educational activities’ direct and indirect impact, including students’ personality, knowledge, skills, attitudes, and behaviors. Astin’s theory emphasizes the role of the school environment, and a supportive environment is essential. In addition, it also pays attention to students’ non-classroom input. Although high-quality non-classroom experience is not an integral part of the curriculum, it can effectively complement education at school. In this research, students’ input factors were assignments and students’ internet self-efficacy. In addition, interaction and services provided by faculty were environmental factors. Students’ satisfaction with online learning during the COVID-19 pandemic was the result. Under this conceptual framework, the following sections will review instructors, assignments, interactions with instructors and other students, faculty services, and students’ internet self-efficacy.

This section reviewed a representative sample of the available literature about student satisfaction and, in particular, student satisfaction with online learning during the COVID-19 pandemic. It can be concluded from the literature that various factors can influence student satisfaction, not only in traditional teaching modes but also in an online setting. Various studies on student satisfaction with online learning have investigated the following factors: instructor interaction, communication, students’ ability to control their actions in the learning environment, efficient assessment, and technology (Cole et al., 2014; Dziuban et al., 2015). However, the research thus far has explored student satisfaction during the COVID-19 pandemic on the surface without in-depth analysis. Student satisfaction with online learning during the COVID-19 pandemic may not have been fully explained. Therefore, this research explored the combinations of various causal factors (instructors, assignments, interaction, faculty services, and internet self-efficacy).

In order to obtain empirical data for this study, self-administered questionnaires were distributed to Chinese higher education students. We believe that this survey research approach was suitable for the data we wanted to collect and could answer the research questions mainly due the following considerations. First, a questionnaire was easy to implement, and it allowed us to collect data online during the pandemic. Also, the results of the questionnaire were convenient for statistical processing and analysis. Figure 1 shows an image of the research design.

When it came to data analysis, we adopted the fsQCA method, a set-theoretical approach introduced by Ragin (2000) and recently applied increasingly in many fields. The method is intended to analyze the influence of multiple antecedent variable combinations on result variables using Boolean operations. Additionally, it focuses on the joint action of multiple influencing factors and provides a variety of combinations of condition variables. Unlike other correlation-based quantitative methods, the fsQCA method seeks to establish a logical link between combinations of causal conditions (conjunctural causation) and an outcome, thereby generating rules that summarize the adequacy of all possible combinations of subsets based on causal conditions (or their supplements) and the outcome (Mendel and Korjani, 2012). The fsQCA approach attempts to avoid some limitations of multiple regression. Woodside (2013) proposed that a negative or positive relationship can exist between the two variables. Still, regression analysis does not account for or explain all cases in a dataset, as in any given dataset, not all cases support a negative or positive relationship between the independent and dependent variables (Skarmeas et al., 2014).

In order to overcome the defects of existing research on students’ satisfaction, this study adopted the fsQCA method to analyze this construct. Kraus et al. (2018) argued that fsQCA is a method that employs qualitative inquiry with quantitative exploration to explicate complex phenomena by using configuration analysis. It can be concluded from the literature review that various factors can influence student satisfaction with online learning during the pandemic (instructors, assignments, interaction, faculty services, and internet self-efficacy). The fsQCA approach in this research could identify all necessary and sufficient conditions that lead to student satisfaction with online learning.

This research utilized a survey administered to university students from 10 colleges and universities in Mainland China during the second semester of the 2021–2022 academic year. In May 2022, the anonymous questionnaire was distributed via an online survey system (Wenjuanxing). The snowball method of sampling was used for recruitment: We gave each participant a discount coupon when they shared the questionnaire link with their classmates—the coupon was an incentive. Snowball sampling is suitable for sampling members of a special population segment. In our research, we needed to collect data from Chinese university students. It was really hard for us to get enough data through sending questionnaires to students because of the school closures during the pandemic. In such a case, snowball sampling was the best sampling method for us to choose.

In total, 424 questionnaires were collected, of which 67 were invalid (including 49 samples in which the same option was chosen continuously, five in which the responses were logically contradictory, 10 samples obtained from postgraduates, and three that were self-test samples); finally, there were 357 valid samples. Table 1 shows the demographic characteristics of the valid samples. As one can see, 48.7% (174) of participants were male, and 51.3% (183) were female. As the participants were higher education students, the age of almost 71.4% (255) of them was between 19 and 24 years old, 18.8% (67) of them were under 18 years old, and about 9.8% (35) of them were between 25 and 30 years old. At the same time, all participants were undergraduate students. Regarding the residential area where these students lived, the number (148) of those who lived in an urban area was nearly the same as that of those from a smaller town (142). Additionally, 18.8% (67) of these respondents were from rural areas.

The questionnaire consisted of four domains: demographic information, academic background, perceptions of online learning from the learners’ perspectives, and satisfaction with online learning. The first 11 questions included in the first part of the survey were intended for understanding the respondents’ demographic information. Then, there were multiple-choice questions for learning about participants’ academic background (seven items). In the third part, six questions revealed whether the students were able to participate in face-to-face classes that semester and their difficulties with online learning. Finally, six questions were used to identify their satisfaction with online learning. Each item in the last domain was rated on a 5-point Likert scale (1: strongly disagree, 2: disagree, 3: neutral, 4: agree, 5: strongly agree). Before issuing the questionnaire, we invited three higher education students from different disciplines (who were not included in the final analysis) to pretest the questionnaire.

In this study, the variables were measured using a 5-point Likert scale, and each variable was measured using the average score of the corresponding items on the scale. SPSS 25.0 was used to analyze the reliability and validity of the scale items. Table 4 shows the reliability and validity of the scale. First, the Cronbach’s coefficients of all six scales were greater than 0.8, indicating that the scale questionnaire had good reliability. Secondly, the KMO values of these scales were more than 0.8, and the cumulative variance contribution rates were more than 60%, indicating that the scale questionnaire had good content validity. In general, the factor loading of each item was above 0.5, the combined reliability (CR) of each item was greater than 0.6, and the average extraction variance (AVE) was greater than 0.5, indicating that the questionnaire had good convergent validity. In conclusion, the scale had good reliability and validity and could be used for further research.

According to Table 5, 31.4% of these students preferred online learning to the face-to-face learning mode, and 23.2% strongly agreed that they preferred online learning. As for the continuation of online learning, 33.3% of the respondents agreed with it, and 21.9% strongly agreed. It can be inferred from the data that more than half of these respondents held a positive attitude toward the continuation of online learning in the future. This calls on universities to pay attention to student satisfaction with online learning and improve online learning from all aspects gradually.

Table 5 shows the extent to which Chinese higher education students were satisfied with online learning during the pandemic. Each item was rated using a 5-point Likert scale (1: strongly disagree, 2: disagree, 3: neutral, 4: agree, 5: strongly agree). As can be seen from the table, the average point of students’ satisfaction with online learning was 3.66, which is between neutral and agree. It can be inferred from the data that many participants were not satisfied with online learning during the pandemic. However, nearly 36.4% of the respondents reported that they were satisfied with online learning, and 26.9% strongly agreed with that statement. In order to know the reasons why the remaining Chinese higher education students were not satisfied with online learning, our questionnaire included a multiple-choice question that asked respondents to identify the problems they faced during the online process. In Figure 2, one can see that deferred examination and graduation were what students cared about most (43.7%). This is a reminder that universities need to arrange examinations and graduation reasonably.

Additionally, universities should give students clear instructions and explain planning about examinations and graduation to avoid students’ concerns. Lower self-control ability and interaction with students and teachers were the second and third problems that students were concerned about (38.7 and 37.2%, respectively). Surprisingly, 35.3% of the respondents believed that the value for money of online courses was not high because the tuition for them was the same as for face-to-face courses. Students were also troubled by the informal learning environment and low learning efficiency (31.4 and 24.1%). These two problems can be solved effectively through students’ efforts, such as by finding a comfortable place to learn and planning their tasks to avoid procrastination and distraction. However, quality of online classes was the problem that students cared about least (7.8%). These results show that not only should universities provide technical support to students, but also teachers should improve the classroom atmosphere to help students concentrate.

Unlike the traditional regression method, fsQCA does not pay attention to the influence of a single variable on the results but focuses on the necessary conditions or sufficient condition combinations that lead to the outcome variables. As mentioned in 3.3, the operation of fsQCA includes three steps. After the above calibration, we used fsQCA to perform the necessity test of single-condition variables, and the results are shown in the Table 6. The analysis results of necessity are reflected in consistency and coverage. Consistency measures the degree to which each condition variable is a subset of the outcome variable. Coverage measures the coverage of each condition variable to the outcome variable degree. Both values are between 0 and 1. Generally speaking, when the consistency is greater than 0.9, this condition variable can be considered a necessary condition for the results.

It can be seen that the consistency of Instructors (IN), Assignment (AS), Interaction (IT), Faculty services (FS), and internet self-efficacy (IS) were all less than 0.9, which was not enough for any of them to constitute a necessary condition that affected Chinese university students’ satisfaction with online learning during the pandemic. This indicates that these variables would weakly explain the outcome variables independently. Therefore, it was necessary to analyze further the configurations of these condition variables and their impact on the outcome variables.

This research used fsQCA3.0 software to analyze the configurations with 357 samples who responded about higher education students’ satisfaction with online learning during the COVID-19 pandemic in China. The consistency threshold, PRI consistency, and case threshold were set to 0.8, 0.876, and 9, respectively. Finally, we obtained three solutions: a complex solution, a parsimonious solution, and an intermediate solution. In addition to the differences in complexity, the three solutions were also different in terms of their enlightenment and universality. To be specific, the complex solution was the most comprehensive, but its universality was poor, resulting in a lack of applicability and generalizability of the conclusion. The parsimonious solution was the simplest one. Still, it lost more information in the process of omission, so the enlightenment of the conclusion was weak and might have been contrary to the actual situation. The explanatory power of the intermediate solution was between the former two. Generally, it represented the combination of the theoretical basis of the research with the analysis results. Therefore, it had strong enlightenment and universality. In general, the intermediate solution determines the final configuration result. Additionally, the intermediate solution and the parsimonious solution jointly determine the core conditions and peripheral conditions in the configuration. Fiss defined the conditions that appear in both a parsimonious solution and an intermediate solution as the core conditions and the conditions that appear in the intermediate solution but are excluded from the parsimonious solution as the secondary conditions.

In the complex solution, we got five path solutions. The solution coverage and solution consistency were 0.685704 and 0.844349, respectively. As for the parsimonious solution, there were three path solutions, and the solution coverage and solution consistency were 0.741548 and 0.802272, respectively. In addition, there were five paths in the intermediate solution. The complex solution, parsimonious solution, and intermediate solution of this research are listed in Table 7.

Outcomes of the fuzzy-set analysis for students’ satisfaction with online learning during the pandemic are presented in Table 7. There were five configurations with which Chinese higher education students could achieve satisfaction with online learning during the COVID-19 pandemic. The overall consistency of these five configurations was higher than 0.8 (0.844349), which indicated that these five configurations fully explained satisfaction with online learning during the pandemic. The coverage of the solution was 0.685704, which means that these five configurations accounted for nearly 69% of the satisfaction with online learning during the pandemic among Chinese higher education students. Table 8 presents each configuration that constituted high satisfaction with online learning during the pandemic in fsQCA. Each configuration will be explained and discussed in detail in the following parts.