A process consisting of 03 stages for developing CCOCVHSS is based on the overview of scale development studies by Morgado et al. (2017). The stages are: (1) Item generation; (2) Evaluation of items; (3) Psychometric analysis.

First, the study establishes a set of items using a deductive method, which is a scale development approach based on theoretical definitions of the concept, as well as theoretical and empirical studies related to the central concept. This method is used to create an “items pool”. This approach is widely used in research to develop a set of items suitable for the study [with 84.7% of the 89 studies evaluated by Morgado et al. (2017) using this method].

Secondly, the study assessed the items identified in the first step using an expert judge method, based on the recommendations of DeVellis (2017). This tool is commonly used to evaluate the face validity of items, which provides a basis for eliminating unsuitable items from the scale (Kapuscinski and Masters, 2010; Ladhari, 2010).

Finally, with the items identified in the second stage, the study proceeds to collect data to determine specific aspects of CCOCVHSS. This process is carried out sequentially through two steps as follows EFA and CFA.

With 32 items developed, we designed a questionnaire to collect data from high school students and teachers. The questionnaire consists of three parts. The first part provides information about the study and includes a consent statement from participants regarding their involvement in this research. The second part is designed to gather demographic information such as (1) Province/City; (2) Living place; (3) Gender; and Subjects, without collecting personal information like name, age, address, or the name of the school attended. The final part, which is the main content of the questionnaire, consists of 33 items presented with the question: “Please indicate your level of agreement with the following codes of conduct in online classrooms of high school students.” These items are rated using a 5-point Likert scale with the following specific values: 1—Strongly disagree; 2—Disagree; 3—Neither agree nor disagree; 4—Agree; 5—Strongly agree.

This section provides a description of the data distribution according to demographic information to give the audience a general overview of the research context. Table 1 below presents the demographic characteristics of the study sample. The survey participants include both teachers and students, with the proportion of students nearly double that of teachers (with a difference of 31.6%). These participants are individuals who have engaged in teaching and learning in synchronous online classrooms at various locations, both in the past and present, using devices such as desktop computers, laptops, smartphones and tablets. Teachers manage their classrooms through features provided by online teaching applications or video conferencing tools to perform tasks such as taking attendance, requesting students to turn on their cameras, controlling microphones and monitoring student participation.

The research sample was selected from high schools in three provinces/cities of Vietnam, including Hanoi (36.3%), Nghe An (38.4%), and Ho Chi Minh City (25.4%). Among them, Hanoi and Ho Chi Minh City are the two largest cities in Vietnam, economic and social development centers that respectively represent the northern and southern regions of the country. Nghe An is one of the provinces with a high population, particularly known for its long-standing tradition of education in Vietnam. Additionally, the research sample was also classified into two regions with differing levels of economic-social development and telecommunication infrastructure: rural (55.9%) and urban (44.1%). Meanwhile, the distribution across different living areas shows no significant disparity.

With the items developed as described above, the study proceeded to collect data. Regarding the survey sample, the study targeted high school teachers and students in Vietnam using a convenience sampling method. This method involves selecting individuals who best meet the research requirements and continuing the selection process until the necessary sample size is achieved (Cohen et al., 2012). Vietnam is one of the countries with the highest internet and mobile network (3G, 4G) coverage in the world (VNA, 2024). Additionally, the ownership of internet-connected devices among both teachers and students in Vietnam is relatively high, thanks to government support policies in recent years aimed at accelerating digital transformation across various sectors, including education. Therefore, the data collection was conducted through an online survey to facilitate data gathering and optimize research costs. After obtaining permission from school administrators, the research team distributed the online questionnaire to two target groups: teachers and students. For teachers, the research team directly sent the online questionnaire link to the teachers of the school. For students, the research team distributed the survey through their homeroom teachers in an online format designed using Google Forms. All questions in the questionnaire were mandatory, ensuring that no missing data existed once the form was completed. The survey link was sent to participants via internet platforms such as email, Facebook Messenger, and Zalo.

Data collection took place from February 18 to March 22, 2024, with 1,014 responses, including 244 teachers and 670 students. The data were then downloaded in ^*.csv format and cleaned based on the recommendations of Hair et al. (2014) to remove records according to specific criteria such as: (1) Providing the same answer for all questions; (2) Responses following a descending or ascending pattern; (3) Leaving all answers blank. The final result was a dataset with 923 records that met the requirements for Exploratory Factor Analysis. This file was then imported into SPSS software for data analysis.

The items in this study are derived from a review of research literature on the code of conduct of teachers and students in online teaching and learning. As a result of this process, 34 specific items were identified. The selection of experts for evaluating this system of items was based on three criteria: (1) Experts must have deep expertise in one of the following fields: Social Psychology or Educational Psychology; Measurement Scale Development or Scientific Measurement; Linguistics (Vietnamese); Educational studies; or Educational Management. In this study, we also included two additional groups for feedback on the items: High school teachers and high school students, as representatives of target population opinions (Bastos et al., 2010; Uzunboylu and Ozdamli, 2011); (2) Experts must have at least 5 years of experience working or researching in their field, although this criterion was not required for students; (3) Experts must volunteer to participate and agree to certain research ethics and confidentiality conditions. We contacted 15 experts via email, Zalo, and Messenger. The results of the expert selection for the item evaluation group are as follows: (1) 13 experts responded, of whom 11 agreed and 2 declined to participate; (2) Two experts did not respond to the invitation. Therefore, a final expert group consisting of 11 members was selected.

Subsequently, the expert group will be sent a file (.doc) designed with the items on the left and the rating scale on the right. For this study, we apply the sumscore decision rule in combination with the complete rule (Hardesty and Bearden, 2004) to determine whether an item is selected or removed from the toolset. Regarding the sumscore decision rule, an item will receive evaluations from 11 experts using a 3-point Likert scale with the following specific levels: (i) Completely representative: 3 points; (ii) Somewhat representative: 2 points; (iii) Not representative: 1 point. As for the complete rule, an item needs to receive at least 50% of experts rating it as completely representative of the construct (Saxe and Weitz, 1982). Therefore, for an item to be selected, it must satisfy the two decision rules mentioned above, achieving a minimum of 16.5 points. Items that meet this point threshold will be retained for evaluation in the third stage. The evaluation results from the experts proposed 32 items, of which two were excluded: (1) C23: 'Students mute the microphone when there is no discussion related to the lesson', which scored only 16 points (below the 16.5 threshold); and (2) C30: “Students always mute the microphone when another student is speaking or discussing”, which scored above 16.5 (25 points) but was rated as “completely representative” by only 36.6% of experts.

In the first step, the EFA method was used to identify the variables and the corresponding items for those variables. Before conducting EFA, the research team checked the mean values of the observed variable C25 “Students do not engage in unrelated activities during class” and the distractor variable C29 “Students always respond to private messages from other students during online classes.” The results indicated that these two items needed to be removed because their mean values were too similar (C25 = 4.17; C290 = 3.18). The requirement is that the distractor variable and the main observed variable should have opposite values, or in other words, the mean values of these two items should be at opposite extremes. Thus, the remaining 30 items were used for EFA. With Varimax rotation, the analysis results were as follows: The Kaiser–Meyer–Olkin measure (KMO = 0.972) and Bartlett's test of sphericity (Chi-Square = 28,213.787, df = 435, p = 0.000) confirmed the adequacy of the sample for factor analysis. The initial direct oblimin rotation yielded four factors with eigenvalues exceeding 1, which shows a dramatic slope change on the screen plot at the fourth factor, with extraction sums of squared loadings at 71.831% [as recommended by Hair et al. (2010), this value should be >50%]. This indicates that the four-factor solution was highly interpretable. The analysis also showed that all items had cross-loadings >0.5, and no item loaded on multiple factors (see Table 2). The EFA results indicate that four constructs were extracted, including: (1) Factor 1: consisting of 09 items representing students' attitudes toward teachers in online classes, the lessons, and the principles of these classes; (2) Factor 2: consisting of 08 items reflecting communication principles between students and teachers, as well as among students themselves in online classes, such as language and classroom behaviors; (3) Factor 3: consisting of 07 items describing students' behaviors when participating in online classes on aspects such as proper camera usage and the principles for starting and ending online classes; and (4) Factor 4: consisting of 06 items reflecting principles related to students' clothing, posture, and body language in online classes.

In the third step, the four factors or scales identified through EFA were tested for reliability based on Cronbach's Alpha values. The results in Table 3 show that the Cronbach's Alpha values for all factors are >0.6. Additionally, all items have a Corrected Item-Total Correlation >0.3. Therefore, the factors meet the criteria for conducting further analyses.

With 30 items explored through EFA, CFA was conducted with four constructs (n=923) to assess the factorial validity of the final version of the CCOCVHSS. To ensure the discriminant validity of the latent variables, 14 items were removed due to loading on multiple latent variables across CFA iterations (Hair et al., 2010). These items were: (i) Factor 1: C7, C8, C15, C11; (ii) Factor 2: C22, C23, C24, C28; (iii) Factor 3: C16, C17, C18, C33; (iv) Factor 4: C5, C6. With the final 16 items, the results indicated that the four-factor model of the CCOCVHSS fit the data well with values meeting the threshold criteria, specifically: Chi square/df = 4.353 ≤ 5 (Hu and Bentler, 1999); GFI = 0.927 ≥ 0.9 (Forza and Filippini, 1998); TLI = 0.963 ≥ 0.9 (Hu and Bentler, 1999); NFI = 0.961 ≥ 0.95 (Awang, 2012); CFI =0.969 ≥ 0.95 (Hair et al., 2010); RMSEA =0.06 ≤ 0.06 (Hair et al., 2010); and SRMR = 0.028 < 0.08 (Hair et al., 2010) (see Figure 1). Inter-factor correlations were ranged from 0.39 to 0.78, all of which were below 0.85, indicating that the discriminant validity was acceptable (Kline, 2023).

Table 4 shows that all Composite Reliability (CR) values, which indicate levels of internal consistency, are >0.7, thus ensuring the reliability of the four factors (Hair et al., 2020). To evaluate convergent and discriminant validity, this study uses the results from Fornell and Larcker (Hair et al., 2010). The results presented in this table also indicate that the Average Variance Extracted (AVE) values for all factors are >0.5, Maximum Shared Variance (MSV) is less than AVE, and the Square Root of AVE is greater than the Inter-Construct Correlation. Therefore, the factors in the tool meet the criteria for both convergent and discriminant validity.

Thus, the CCOCVHSS includes four scales with 16 items to measure four aspects of the Code of Conduct. Based on the meaning of the items in each scale, we have named these scales as specified: (1) Students' Attitudes Toward Teachers and Classmates: 5 items; (2) Communication Conduct of Students Toward Teachers and Classmates: 4 items; (3) Conduct in Using Camera by Students: 3 items; and (4) Dress Code, Sitting Posture and Learning Location Regulations: 4 items.

Students' attitudes play a critical role in the learning process across all learning modes. Many studies have shown that students' attitudes in online learning environments, as well as in online classes, have a direct impact on their attention and academic performance. Additionally, according to the theory of planned behavior, a person's attitude predicts and motivates the occurrence of their behavior in a specific activity (Ajzen and Fishbein, 2005). For students participating in online classes, where the online learning format requires a high level of self-study skills and self-discipline, students' attitudes need to be aligned with the online learning environment. Iliyas et al. (2023) emphasize that students' conduct in online classes goes beyond merely following rules; it also involves mutual respect, active listening and the ability to respond positively. Clark and Mayer (2023) argue that maintaining respectful and friendly language is essential to creating a positive learning environment. Students feel more confident and secure when communicating with teachers if teachers show respect and listen to their opinions. Likewise, when students demonstrate respect toward their teachers and peers, it helps create a more friendly and collaborative learning space.

Polite communication and interaction in online classes play an important role in fostering a friendly and effective learning environment. Hodges et al. (2020) point out that simple actions like greeting others when joining a class or expressing gratitude for help received can strengthen connections among class members. Barbour et al. (2020) also highlight that online communication tools cannot fully replace gestures and face-to-face interactions. This calls for more creative use of online tools to foster positive interactions and build stronger connections between students. Barbour et al. (2020) emphasize that students should show respect for teachers and classmates by avoiding teasing or disruptive behavior, apologizing for mistakes and thanking others for their support. Bozkurt and Sharma (2020) recommend having clear online classroom conduct rules and measures for handling violations to ensure fairness and mutual respect among students. In online learning environments, students need to build relationships and cooperation with teachers and peers, as Crawford et al. (2020) note that students may feel isolated in online settings. According Garrison et al. (1999), when students receive help from teachers and peers, they can show appreciation and respect by demonstrating gratitude and apologizing when they make mistakes. This not only helps establish positive relationships but also encourages mutual cooperation and cultivates respectful behavior habits among students toward teachers and peers. This is the foundation for gradually building a culture of respectful conduct, not only in online classes but also across online education settings in general.

In online learning, students' use of cameras as regulated by the school or teacher is also an important behavior that helps enhance the effectiveness of students' learning. It ensures a certain level of supervision from teachers and stakeholders regarding students' study. Online classes on platforms like Zoom, Microsoft Teams and Google Meet display videos in different ways, which affects students' decisions to turn their cameras on or off. Some reasons for turning off their cameras during online classes include feeling uncomfortable with constantly appearing on the screen (Castelli and Sarvary, 2021), or concerns about their personal space (Lowenthal, 2022). The type of device students use significantly influences their decision to turn the camera on or off, as those using mobile phones are more likely to disable their cameras due to battery limitations and inconvenient camera angles (Gherheş et al., 2021). Moreover, the quality of the internet connection is an important factor influencing the decision to use the camera. Several studies figure out that some students turn off their cameras for the following reasons: (i) Saving bandwidth to avoid lag and ensure better audio quality in the classroom, especially for students in rural areas (Fabriz et al., 2021); (ii) Related to technical issues like weak internet connection, poor-quality devices, and unsuitable devices (Williams and Pica-Smith, 2022). The survey results of this study show that Vietnamese students tend to turn off their cameras in online classes, similar to the reasons found in previous studies in different countries around the world. In addition to technical factors, psychology and the learning environment also significantly influence students' decisions to turn their cameras on or off. Many students have to study in spaces that do not ensure privacy, such as sharing a room with family members or being disturbed by surrounding noise. Others feel hesitant to turn on their cameras due to fear of being judged on their appearance or not wanting teachers and classmates to see their living environment. Furthermore, turning on the camera can also create pressure, making it difficult for students to learn comfortably. Although some teachers and schools require students to turn on their cameras during class, if this rule is not flexible, it can create stress and have the opposite effect. In many cases, students only turn on the camera when the teacher asks but then turn it off again due to personal or technical reasons. These factors indicate that the decision to turn the camera on or off is not solely based on the student's personal preference, but is also influenced by the device, connection quality, learning environment and school regulations. Therefore, there needs to be rules that encourage, but do not mandate, turning on the camera, allowing students the right to choose based on their circumstances.

Clothing, sitting posture and learning environment in online classrooms, which are visible on the shared screen of video conferencing tools, also play a significant role in establishing digital etiquette, fostering a sense of responsibility toward one's learning and contributing to a respectful learning environment with teachers and classmates. This is one of the fundamental principles of interpersonal communication, helping to form a system of culturally normative behaviors. With established regulations on camera use, students in online classes should also ensure they wear attire suitable for a formal setting. According to Anderson (2008), when students dress neatly, they become aware that they are participating in an important activity, which enhances their self-discipline and seriousness in learning, potentially impacting their academic performance (Volpe et al., 2023) As for sitting posture, Garrison (2016) found that when students sit with correct posture, teachers can more easily observe students' expressions, and students can focus better on the lesson. Regarding the study environment, ensuring a suitable, quiet space optimizes knowledge acquisition and engagement in learning activities (Bao, 2020). In an online learning environment, students often study from home, where they may be affected by various external factors such as noise from family, neighbors or household devices. This directly impacts the student's focus on the lesson, as well as that of other students and the teacher. Therefore, clear regulations on attire and sitting posture within an appropriate learning environment help establish specific norms for online classrooms, fostering a class culture that promotes effective learning and teaching. Additionally, these guidelines contribute to shaping students' behaviors and habits in formal online settings, gradually cultivating professional manners and digital etiquette, fostering a cultured individual even beyond the online environment.