The relationship between personal-collaborative motivation profiles and students’ performance in collaborative problem solving

Abstract

Collaborative problem solving (CPS), as a twenty-first century skill, is critically important for both living and life-long learning. Motivation in CPS equates to students’ needs to recognize their efforts in collaboration. Given the complexity of CPS contexts, the intertwined relationship among different types of motivations was investigated using a person-centred approach to identify different configural profiles of collaborators based on PSI ($\psi$) theory. A total of 9398 Chinese students in 264 schools were included in this study. Latent profile analysis was used to identify four profiles of students with different CPS motivations: “Achievement-oriented cooperator” (n = 773), “Instrument-oriented Cooperator” (n = 1539), “Achievement-oriented nonteam player” (n = 1202), and “Instrument-oriented nonteam player” (n = 5884). Their psychological attributes and CPS behaviours were further analysed, with the following results: (1) achievement-oriented cooperators performed the best, while instrument-oriented nonteam players performed the worst; and (2) students who are achievement-oriented and value collaboration with others are more likely to engage in CPS tasks. These findings provide insights into how different motivational configurations influence CPS behaviours, offering practical implications for enhancing collaboration in educational settings.

Introduction

Collaborative problem solving (CPS), which includes both social skills (collaboration skills) and cognitive skills (problem-solving skills) (Griffin et al., 2012; OECD, 2017a), is widely known as a 21st-century skill (Graesser et al., 2017). Students who have a high level of CPS tend to be more active and involved in complex work that requires cooperation, making them more competitive when entering into and participating in the modern labour market (Graesser et al., 2018). Thus, CPS has also been incorporated into international assessments (such as PISA 2015) to evaluate students’ collaborative behaviours.

According to the theoretical framework of CPS in PISA 2015 (OECD, 2017c), team-member characteristics, including knowledge, motivation, and dispositions, are one of the most important factors based on the available theory and empirical research on CPS (Graesser et al., 2018). Motivation, defined as the process that initiates, guides, and sustains goal-oriented behaviours (Schunk, et al., 2008), has been thought to play an important role in influencing CPS performance since it explains why students exert persistent efforts in collaboration (Taylor & Baek, 2017; Wang & Ning, 2019; Xu & Li, 2019). Furthermore, the PISA 2015 framework for measuring collaborative problem-solving (CPS) skills, defined motivation in this context as the willingness of students to engage in the collaborative process, referred to as CPS motivation (OECD, 2017b). The measurement of CPS motivation in PISA 2015 included both collaborative motivation and personal motivation. Collaborative motivation captures students' values and enjoyment in participating with others in problem-solving activities, while personal motivation evaluates their attitudes and persistence in overcoming challenges to solve problems as part of a team (Chen et al., 2020; OECD, 2017b).

According to Schunk et al. (2008), motivation is described as the process whereby goal-directed activity is instigated and sustained. While recognizing motivation’s dynamic and ongoing nature, PSI (ψ) theory highlighted that different sources of motivations influence our behaviour through distinct systems in the hypothalamus, depending on the intensity of the needs (Dörner & Gerdes, 2012). Within the complexity of CPS contexts, different sources of motivations work as a unified whole rather than as isolated factors when making behavioural decisions (Bechtel & Abrahamsen, 2002), thereby forming diverse combinations depending on the task at hand (Dörner & Güss, 2013).

However, little is known about how different sources of motivation interact with each other during the process of CPS. The dearth of information prevents a deeper understanding of the underlying reasons for the behaviour and hinders effective intervention that is meant to develop students’ CPS. From the perspective of PSI theory, which emphasizes that different motivations use separate systems in the hypothalamus depending on the intensity of needs (Dörner & Gerdes, 2012), this study aims to classify students by investigating different sources of CPS motivations. In such a way, by going beyond previous studies, the present study provides new insight into the intertwined relationship among different sources of motivations, with the goal to deepen and expand our current understanding of the representative characteristics of students’ motivation during CPS. The findings can also provide more explainable information for teachers to better understand the students’ core demands and needs regarding CPS.

Latent profile analysis (LPA) is a person-centred approach used to identify distinct configural profiles of collaborators based on combinations of diverse attributes. Unlike traditional methods that may treat motivation as a single continuum, LPA allows for the identification of distinct motivational profiles or patterns. While several researchers have acknowledged the importance of using a person-centered approach to identify diverse profiles of collaborators, much of the focus has been on students’ performance. For instance, Herborn (2017) analyzed students’ behavioral patterns, personal attributes, and motivation in CPS, classifying them into three profiles: passive low-performing non-collaborators, active high-performing collaborators, and compensating collaborators. However, classification analyses specifically focusing on CPS motivations remain limited and warrant further investigation. By utilizing LPA, we can identify different motivational profiles exhibited by individuals. This subgroup identification offers a nuanced understanding of the varied patterns of motivation within a population, which is crucial for comprehending the complexity of motivation traits.

Motivations are closely tied with other internal perception processes and directly inform external behaviours in different collaborative contexts (Zhai, 2021). For example, Graesser et al. (2018) pointed out that when working as a group, individuals might withhold or reduce their efforts according to their level of motivation, which influences their engagement and commitment to collaborative tasks. However, previous research has generically considered motivation as a separate rather than an interwoven construct for predicting student outcomes. Thus, the second goal of the present study is to further reveal whether different combinations of student motivation cause differences in students' attributes and CPS performance. Differentiating the variance in psychological dynamics and skills among different profiles of CPS motivation can provide a more explainable means of understanding students’ performance.

Therefore, our study focuses on both personal motivation (instrumental motivation and achievement motivation) and collaborative motivation (enjoy cooperation and value cooperation) in CPS based on PSI theory to further identify the different profiles of collaborators. Furthermore, to clarify the distinctions in CPS performance. In our study, CPS performance not only refers to student final scores in CPS tasks but also considers students’ behaviours that happened during the CPS. We will compare different groups in terms of their psychological attributes and behaviours in CPS by using log data. Finally, different motivation profiles will be modelled along with other student-level and school-level factors to predict students’ CPS achievement.

Literature review

CPS motivation in PISA

In the assessment framework of CPS in PISA 2015, CPS is divided into cooperative and personal problem solving processes (Griffin et al., 2012; OECD, 2017a). As in the CPS process, CPS motivation in PISA 2015 can also be divided into two dimensions: personal motivation and collaborative motivation (Peterson & Schreiber, 2006).

Personal motivation in CPS refers to motivation aroused during the personal process and comprises perceived problem-solving capacity, willingness to complete the task, and perceived CPS values (Linnenbrink-Garcia et al., 2016). Achievement motivation, as one type of the personal motivation, indicates the willingness to finish the events properly (Wigfield & Eccles, 2002). Instrumental motivation, as another type of the personal motivation, indicates the value or importance of finishing certain things (Liu et al., 2020; Wigfield & Eccles, 2002). The former has a tendency to finish difficult CPS tasks, while the latter has a tendency to treat the CPS task as a goal or as a means. Both are closely related to CPS awareness (Chen et al., 2020). These two types of personal motivation were measured in PISA 2015 (OCED, 2017b).

Collaborative motivation in CPS refers to the motivation aroused in the cooperative process. It includes not only the willingness to work in a group (Taylor & Baek, 2017) but also perceptions about the value of cooperation (Eccles & Wigfield, 2002). Willingness is an important aspect of motivation. It can be aroused by certain actions, such as assigning a group role to a student to arouse his or her interest or desire to complete the task. Once students’ willingness is activated, they are not only more likely to stay on task but also to more deeply engage in collaborative working (Taylor & Baek, 2017). According to the expectancy-value theory, task values have several components, such as the interest of a task, the meaning of a task, and the usefulness of a task (Jiang et al., 2018). When individuals recognize task values, they are more likely to instrumentally engage in collaborative tasks or to make changes through their own efforts. In PISA 2015, two dimensions of collaboration were measured: students’ interest and enjoyment in collaboration, and their values regarding the meaning and usefulness of CPS (OECD, 2017c).

Given the multidimensional structure of CPS motivations, some researchers place significant emphasis on examining their interrelationships. Some studies have discovered that personal motivation played a mediating role between cooperative motivation and collaborative participation, which further influences students’ collaborative performance and knowledge sharing (Wang & Ning, 2019).

CPS motivations in PSI

Understanding the interrelationships of CPS motivations in PSI

The PSI theory, named after Greek letter “$\psi$”, is one of three motivation theories in CPS (Güss et al., 2017). It integrates cognition, motivation, and emotion into a cohesive framework (Dörner & Güss, 2013). PSI theory is particularly well-suited for this study because it captures the dynamic interplay of different motivations in CPS. Unlike other theories that treat motivations as isolated factors, PSI theory conceptualizes them as interconnected systems, providing a comprehensive explanation for the interaction between personal and cooperative motivations within CPS environments.

According to Dörner and Gerdes (2012), different motivations can be understood as dynamic systems that fluctuate in response to the level of need satisfaction. For instance, a low level of achievement motivation indicates a higher demand for fulfilling that need, which in turn influences other motivational factors. Within CPS, these various motivations continuously interact and adjust in response to the collaborative context, thereby influencing students' behavior and outcomes (Bechtel & Abrahamsen, 2002).

Personal attributes, action and CPS motivations in PSI

According to PSI theory, CPS involves four core processes: motivation, perception, cognition, and action (Dörner & Güss, 2013). Previous studies have found that motivation has a close relationship with students’ attributes. Zhai (2021) found that although student perceptions and motivations fluctuate due to specific projects' constraints and challenges, their growing awareness of perceived collaborative outcomes and achievement can remotivate them. Moreover, perceptions such as belonging (Goodenow & Grady, 1993), anxiety (Celik & Yildirim, 2019) and emotional support from parents (Heilat & Seifert, 2019) are closely related to motivations. In addition, motivation and action are highly related to each other in the fields of psychology and cognitive science (Gendolla, 2017). Psychology and cognitive science emphasize that what motivates people to act directly are their beliefs regarding appropriate or right ways to behave (Markus, 2016). This means that the “benefit neurocognitive processes” push people to prefer decision-making (Ji et al., 2021).

Log-file is a good means of process monitoring and is used to detect student actions, such as disengagement and persistence (Gobert et al., 2015; Ventura & Shute, 2013). Timing is a common process variable that reflects the degree of students’ effort (Wise & Kong, 2005). For example, students who spend more time on tasks are considered to be highly motivated. Another common process variable that relates to test-taking motivation is the number of actions (Ivanova et al., 2020). A student that makes a moderate number of actions is considered to actively engage in tasks (Goldhammer et al., 2017). Thus, analysing these process recordings among different motivation profiles can further elucidate the relationships in a more refined way than previous research that only focused on superficial behaviour.

Factors that affect the CPS

In addition to motivation, prior values, prior knowledge or experience, and cognitive load, which have all been shown to affect CPS (Jung & Lim, 2020; OECD, 2017b; Sun, 2020), researchers have also found that some student-level and school-level contextual variables may confound the relationship between CPS motivation and performance.

Student-level contextual variables in PISA that affect CPS competency include gender, grade, and the economic, social, and cultural status of students (ESCS). Notably, the gender effect on CPS varied among different countries. Gender influences were found in Taiwan (China) but could not be statistically identified in Finland (Ahonen & Harding, 2018; Li & Liu, 2017). In mainland China, not only gender effects but also gender pairings affect CPS (Lin et al., 2020; Tang et al., 2021). Regarding grade, previous research points out that upper-grade students have a higher level of CPS competency (Ahonen & Harding, 2018; Tang et al., 2021). ESCS, as the most affected factor in many fields, has positive correlations with CPS in China (Sun, 2020; Tang et al., 2021; Wang, 2018).

CPS is also affected by school-level-related factors, indicating that school quality was found to influence students’ CPS. For example, the location of a school is an important factor in that students from city schools have higher levels of CPS (Sun, 2020). CPS has also been found to be highly influenced by ESCS at the school level (Lee & Namwook, 2020; Wang, 2018). Moreover, the proportion of science teachers who are fully certified can benefit students engaged in CPS (Tang et al., 2021). Notably, ICT ability and literacy are critical factors in CPS (Lee & Namwook, 2020), because the current CPS test mainly relies on a computer. In the current study, we attempt to answer three major research questions:

• 1. What kind of different motivation profiles could be identified among students by using a person-centred approach?

• 2. What are the differences in perception and behaviour of different motivation profiles in CPS?

• 3. How do different groups (or different motivation profiles) predict student CPS performance when both student-level and school-level confounding factors are considered? (Fig. 1)

Fig. 1

Framework of factors that affect CPS. The suffix_W indicates a variable at the student level, the suffix_{_}B indicates a variable at the school level. Method interprets a detailed description of all variables

Method

Sample

This study utilized data from four regions in mainland China (Beijing, Shanghai, Jiangsu, and Guangdong; B-S-J-G) collected in PISA 2015, including the information from the CPS assessments and responses from both student and school questionnaires. Beijing, the capital of China, is located in the northern part of the country; Shanghai, the largest city in China, is situated along the eastern coast; Jiangsu and Guangdong are medium-sized provinces in southeastern China. In these four regions, all school samples were selected according to the procedures and standards of PISA, thus ensuring the sample could represent each region. A total of 9841 students (M _age = 15.74, SD = 0.005) participated. Considering that the missing rate of all variables was below 5%, with the maximum missing rate (SOIAICT) of 2.53%, this study deleted missing observations by variable. The final sample comprised 9398 students (47.67% girls) from 264 schools. The valid sample rate was 95.50%. Moreover, the student-level sample weight (W_FSTUWT) and school-level sample weight (W_FSCHWT) were used to decrease the unequal probability of sample selection and nonresponse bias (Hu et al., 2018).

Statistical approach and corresponding variables

Latent profile analysis (LPA)

Model introduction for LPA

To address RQ1, this study conducted an explanatory LPA on predictors of CPS motivation. LPA, a person-centred approach, classifies students into subgroups based on their individual response patterns (Muthén & Muthén, 2000). Unlike traditional methods that treat motivation as a single continuum, LPA considers motivation holistically, grouping subjects according to their response patterns, with each group exhibiting distinct motivational profiles (Fisher & Robie, 2019). This person-centred approach overcomes the limitations of variable-centred methods by prioritizing individual characteristics before examining shared traits. In this study, Mplus 8.3 was used to construct the LPA.

Variables in LPA

CPS motivation. As we explained, CPS motivation comprised personal motivation and collaborative motivation. Personal motivation in CPS consists of instrumental motivation (INSTSCIE) and achievement motivation (MOTIVAT). Instrumental motivation, which indicates the attitude towards learning science for some instrumental purpose, included four items (ST113, e.g., Many things I learn in my < school science > subject(s) will help me to get a job). Achievement motivation, which indicates the willingness to achieve higher scores, included five items (ST119, e.g., I want to be the best, whatever I do.). The Cronbach's α for instrumental motivation was 0.901, and for achievement motivation was 0.780.

Collaborative motivation in CPS contains two dimensions, enjoy cooperation (COOPER) and value cooperation (VALUE), including four items for each dimension. Enjoy cooperation indicates how much students enjoy working in teams (ST082, e.g., I enjoy seeing my classmates be successful.). Value cooperation indicates the motivation of a student to cooperate with their peers. (ST082, e.g., I find that teamwork raises my own efficiency.) The Cronbach's α for enjoy cooperation was 0.677, and for value cooperation was 0.821.

A total of four variables representing different sources of motivation were used in the LPA to identify the motivation profiles. These variables were measured on a four-point Likert scale, ranging from “strongly agree” to “strongly disagree,” with some items requiring reverse-coding. Additionally, higher weighted likelihood estimates (WLEs) corresponded to higher levels of motivation (OCED, 2017b, p305, 307, 312, 315, 317, 318).

Personal perception of students. Belonging (BELONG), indicating students' sense of belonging to their school, includes six items (ST034, e.g., Other students seem to like me). Cronbach’s $\alpha$ was 0.792.

Test anxiety (ANXTEST), indicating how nervous or scared students are on the test, includes five items (ST118, e.g., I get very tense when I study for a test). Cronbach’s $\alpha$ was 0.824.

Parental support (EMOSUPS), indicating perceived emotional support from students’ parents, includes four items (ST123, e.g., My parents support me when I am facing difficulties at school). Cronbach’s $\alpha$ was 0.788.

Three variables listed above on a four-point Likert scale, ranging from strongly agree to strongly disagree (some required reverse-coding). In addition, the higher weighted likelihood estimates (WLEs) corresponded to higher levels (OCED, 2017b, p. 305, 307, 312, 315, 317, 318).

Personal timing and action in CPS items. The index of personal behaviours in CPS items includes timings and activities. “Timings”, the personal behaviour index of time in CPS, was measured by the time that students took to finish the CPS items. “Activities”, the personal behaviour index of action in CPS, was measured by the number of actions that occurred during the items in each item’s log-data.

MANOVA

To address RQ2, this study uncovered differences among several personal perceptions and CPS behaviours in each profile based on RQ1’s results. For personal perception, MANOVA and post hoc were conducted on three perception variables. For personal CPS behaviours, 6 CPS items (e.g., “The Garden”) from the CPS test were selected. SPSS 24.0 was used to conduct the MANOVA.

Hierarchy linear model (HLM)

Model introduction for HLM

To address RQ3, HLM was employed to explain the impact of each profile, accounting for school-level factors in Mplus 8.3. The ICC of the CPS was 0.391, indicating a high variance between schools. Thus, a multilevel analysis was needed. The model function composed of student-level and school-level variables is given below. The effect size ${{f}_{i}}^{2}$ was calculated in $\frac{{\tau }_{null}-{\tau }_{i}}{{\tau }_{null}}$ at the student level and in $\frac{{{\sigma }^{2}}_{null}-{{\sigma }^{2}}_{i}}{{{\sigma }^{2}}_{null}}$ at the school level.

Null model:

$${y}_{cpsij}={\beta }_{0j}+{\varepsilon }_{ij}$$

$${\beta }_{0j}={\gamma }_{00}+{\mu }_{0j}$$

Random intercept models:

$${y}_{cps}={\beta }_{0j}+{{\beta }_{10}{X}_{gender}+{\beta }_{20}{X}_{ESCS\_w}+{\beta }_{30}{X}_{SOIAICT\_w}+{\beta }_{K0}{X}_{profileK}+\varepsilon }_{ij}$$

$${\beta }_{0j}={\gamma }_{00}+{\gamma }_{01}{\omega }_{ESCS\_b}+{\gamma }_{02}{\omega }_{PROATCE}+{\gamma }_{03}{\omega }_{SOIAICT\_b}+{\gamma }_{04}{\omega }_{ICTCLUB}+{\mu }_{00}$$

$${\beta }_{10}={\gamma }_{10} ;{\beta }_{20}={\gamma }_{20};{\beta }_{30}={\gamma }_{30};{\beta }_{K0}={\gamma }_{K0}$$

$$K\in \text{1,2},\dots ,k(k+1 \text{indicates the number of profiles in LPA})$$

Variables of HLM

Student performance in CPS was measured by using the dependent variable of student performance in the PISA computer-based CPS assessment, represented by 10 plausible values (PVs) for each student in PISA 2015 and assessed as “the capacity of an individual to effectively engage in a process whereby two or more agents attempt to solve a problem” (OCED, 2017b, p. 43). The CPS scores were scaled to have a mean of 500 and a standard deviation of 100 (OCED, 2017b, p. 234). All plausible values of CPS were treated as the dependent variable, with estimation carried out through imputation using Rubin's rules code in Mplus 8.3. Detailed code can be found in Appendix C.

Student-level variables. At the student level, ESCS, gender, and SOIAICT were chosen as covariates. ESCS, an index of economic, social, and cultural status, was calculated by parental education, highest parental occupation, and home possessions via principal component analysis (OCED, 2017b, p. 339). Gender was a dichotomous variable, where zero represented female. Students’ ICT as a topic in social interaction (SOIAICT), which indicates the degree to which ICT is a part of students' daily social life, included five items (IC016, e.g., I learn a lot about digital media by discussing with my friends and relatives.) Cronbach’s $\alpha$ was 0.840.

School-level variables. At the school level, PROSTCE, ICTCLUB, aggerated ESCS, and SOIAICT were included. PROSTCE represented the proportion of science teachers who were fully certified (SC019) and was computed by dividing the number of certified science teachers by the total number of teachers (OCED, 2017b, p. 322). Additionally, the ICT-relevant club in schools provided a good place for students’ social learning, which may benefit their ability development due to a school collaboration atmosphere that may affect CPS. ICTCLUB indicated a club that focuses on computer information and communication technology (SC053Q08TA, yes = 1, no = 0). Last, due to the positive prediction of students' CPS competency, SOIAICT and ESCS were added to the analysis as school-level factors (Tang et al., 2021).

Result

Model selection and nomination

Descriptive statistics showed in Appendix A. Model selection considered classification from 2 to 6 classes, and all models had a good classification result (see Appendix B). Therefore, selecting a robust latent class should balance the interpretability of the models according to both relevant theory and the statistical results. Although the six-class model had the most minimized information criterion and the highest classification consistency, the fourth, fifth, and sixth groups in the six-class model were highly similar except in instrumental motivation, and the third, fourth, and sixth groups had insufficient proportions to be representative. In summary, the four-class model with the highest interpretability and proper proportions of each group was selected as the final model.

Figure 2 presents the achievement-oriented collaborator (n = 773), which is the first motivation profile of students and contains the highest scores in both personal and collaborative motivation. It represents a stronger preference for achievement motivation as well as cooperation enjoyment and value. The instrument-oriented collaborator (n = 1539), which is the second profile, represents a strong interest in cooperation, especially in cooperation value rather than enjoyment, and it includes more instrumental aims towards personal motivation. The achievement-oriented nonteam player (n = 1202), which is the third profile, was uninterested in cooperation value and had a negative attitude towards cooperation enjoyment. However, this profile represents a very high level of achievement motivation. The instrument-oriented nonteam player (n = 5884), which is the last profile, contains the largest number of students but the lowest (including even negative) levels of personal motivation and collaborative motivation. This kind of student is only motivated by instrumental motivation and shows little interest in achievement motivation, cooperation value, or cooperation enjoyment.

Fig. 2

Characteristics of each profile

The personal perception difference of each motivation profile

We further conducted an ANOVA for deeper analysis of the different profiles in terms of their psychological attributes to draw a clearer portrait of each CPS motivation profile.

As shown in Table 1, the differences among the four profiles’ attributes were significant: sense of belonging, test anxiety, and parental emotional support. The post hoc comparisons revealed that (1) belonging varied among all students; (2) test anxiety showed no significant difference between achievement-oriented collaborators and instrument-oriented collaborators, nor between achievement-oriented nonteam players and achievement-oriented collaborators; and (3) achievement-oriented nonteam players and instrument-oriented collaborators showed no significant differences in the perception of parental emotional support.

Table 1 ANOVA of each profile student’s perception (M ± SD)

The personal CPS behaviour difference of each motivation profile

To examine differences in personal behaviour in CPS, six CPS items were included, by which students were divided into four groups (abnormal responses by students on each item were excluded using the $3\sigma$ Guidelines). As depicted in Tables 2 and 3, the results showed behaviour (including timings and activities) differences among the four groups.

Table 2 Different groups’ timing on six CPS items

Table 3 Different groups’ activities of six CPS items

As for timings (Table 2), achievement-oriented collaborators tended to take a longer time to finish the collaboration problem (bolded in Table 2), whereas instrument-oriented nonteam players tended to spend less time on this problem (bolded in Table 2). This tendency occurred in some CPS items; however, it was only significant in “Making a Film” and “The Garden.” In addition, the post hoc test indicated that the characteristics of other groups of students did not show such a strong tendency in timing.

For activities (Table 3), achievement-oriented nonteam players tended to have the highest number of explorations (number of activities) in most items (bolded in Table 3). This tendency was significant in the “Making a Film”, “Field Trip”, and “Preparing a Presentation” items. However, the tendency toward the least exploration was not that clear. Instrument-oriented nonteam players spent less exploration time in “Making a Film” and “The Garden” than the other groups did. However, achievement-oriented collaborators tended to explore the least (bolded in Table 3) on average.

Results of multilevel linear models

As the results displayed in Table 4, CPS’s ICC was much higher than the suggested principle of 0.059, which means that the multilevel analysis was highly necessary. Model 2 focused on the effects of student- and school-level factors that have been shown to be the most influential variables or that were likely to affect CPS achievement. All student-level variables were found to appropriately affect CPS achievement. As expected, school-level ESCS positively influenced CPS achievement, while $PROATCE$ was a significant positive predictor, which is consistent with previous research (Tang et al., 2021). However, for SOIAICT, the factors associated with ICT at the school level and ICTCLUB were not significant. These predictors explained 2.23% of the student-level variance in CPS achievement and 70.61% of the school-level variance in CPS achievement.

Table 4 The analysis result of the multilevel model

Specifically, as shown in Model 3, when student-level and school-level variables were controlled, the results indicate two conclusions regarding the CPS performance of the four groups of students. Setting “instrument-orient nonteam player” as the baseline in dummy encoding, the regression coefficients represent the average difference between a certain group and the “instrument-orient nonteam player”. Since three dummy groups’ coefficients are greater than zero, this means “instrument-orient nonteam player” had the lowest average score, and all other groups scored higher than it. “Achievement-oriented collaborator” scored highest.

In summary, a student driven by achievement motivation appears to have a higher ability in CPS than a student who is driven by instrumental motivation. The other conclusion appears to be that students can perform well in CPS even if they have no interest in cooperating but have a high level of achievement motivation. On the other hand, once these students obtain a certain level of cooperative or personal motivation, they perform better than students who lacked both personal and cooperative motivation.

Discussion

Investigating the complex relationship between the different motivational profiles in CPS improves our understanding of how motivations affect CPS (Herborn et al., 2017; Tze et al., 2021;  Wang & Ning, 2019). To further advance knowledge in this area, this study focused on the interrelationship of CPS motivations and uncovered the profiles of students who completed a CPS item in PISA by using instrumental motivation, achievement motivation, enjoyment of cooperation, and value of cooperation as indicators in LPA. Based on PSI theory, the current study extends previous studies in several ways: (1) typical CPS motivation profiles provide a more specific and in-depth understanding of students’ core needs for CPS; (2) differences in psychological attributes, personal behaviour, and CPS performance among profiles provides a more explainable and refined way for teachers and parents to acknowledge students’ performance.

The profiles of students’ personal-collaborative motivations

Given the interrelationship of personal and collaborative motivation in CPS, only CPS motivation-related indicators were utilized for LPA classification, and the impacts of other factors (Herborn, 2017; Tze et al., 2022) were not blended. Detailed interpretations of the characteristics, behaviour, and performance of each profile are as follows:

Achievement-oriented collaborators (8.2% of the sample), achieving the highest CPS scores among the four classes, are recognized for their exceptional academic performance and strong motivation in both achievement and cooperation. Achievement-oriented collaborators are typically “socially oriented” individuals who place more emphasis on the judgement of others within a particular social setting. These students exhibit a combination of needs for achievement and affiliation, which may explain their pronounced sense of belonging within the school environment. This finding aligns with a longitudinal study (Gillen-O'Neel & Fuligni, 2013) that showed that academic motivation had a strong correlation with school belonging. Students with this profile tend to attach much importance to others’ comments (Chen, 2015), which may lead them to collaborate with teammates and pay close attention to the comments of others regarding their achievements (Cheung et al., 2011; Hung, 2017). However, this might make them excessively focus on elements that they cannot control, such as others' praise and criticism, which easily induce worry (Weiner, 1972). This concern about external judgment might explain why these students experience relatively high levels of test anxiety compared to the other groups, as they fear being judged negatively if they fail to achieve high grades. Consequently, they tended to spend the most time on CPS tasks for a better score.

Achievement-oriented nonteam players (12.8% of the sample) achieved the second highest CPS scores and are characterized as independent and highly capable individuals who tend to avoid teamwork. Despite their talent, they exhibit very low levels of enjoyment and perceived value in cooperation, which leads them to place little importance on CPS tasks. While achievement-oriented nonteam players were likely to be “individually oriented”, individuals who internalized performance to a high degree retained higher levels of autonomy and had minimal affiliation with teammates (Xu et al., 2012). As a result, their sense of belonging was the second lowest out of the four groups, and their perception of parental emotional support was low. This inclination toward seclusion and isolation makes them feel very anxious, as evidenced by their highest test anxiety scores among all student types. Their hesitation to interact might be due to their low social standing (Levy et al., 2004). Although achievement-oriented nonteam players might not be comfortable collaborating with others during CPS, their strong motivation to achieve good results drive them to engage more deeply and to do their best to explore additional options because of a strong drive to properly complete CPS to obtain good results. This was also revealed by their tendency to explore the most (greatest number of activities) during CPS activities among the four groups.

Instrument-oriented collaborators (16.4% of the sample), having the third highest CPS scores, are individuals who prioritize their own interests while demonstrating a strong inclination towards cooperation. Previous research has shown that people who have a higher requirement for affiliation engage more in teamwork (Hilkenmeier, 2018), just as achievement-oriented collaborators do. Despite a strong appreciation for working in a group, instrument-oriented collaborators have a negative sense of belonging and a relevantly high level of instrumental motivation. This result suggests that instrument-oriented collaborators, who value affiliation with teams, have motivations that are slightly different from those of achievement-oriented collaborators. While they have a high willingness to cooperate, their focus remains primarily on their own interests, using cooperation as a means to an end. Although they may explore more actions or spend more time on CPS tasks, their willingness to cooperate likely stems from a reliance on peers to help them achieve their instrumental goals rather than from a genuine commitment to the goal of cooperation itself.

Instrument-oriented nonteam players (62.6% of the sample), having the lowest CPS scores, are characterized by their minimal effort and disengagement. With a very low level of both personal and collaborative motivation, they tend to be content with more attainable achievements and only complete the bare minimum required, which is reflected in negative achievement motivation and low instrumental motivation. In collaborative settings, instrument-oriented nonteam players quickly lose interest and enthusiasm, which is reflected in their negative cooperative motivation. However, they constituted the largest group, highlighting an environment where competition is intensifying and cooperation is less valued (Sommet et al., 2022). Consequently, students in this profile often feel exhausted by the pressures to perform, resulting in a pervasive sense of disengagement. Their sense of belonging, test anxiety, perceived emotional support, and the frequency of their engagement in CPS tasks were the lowest among the four groups. This lack of interest in communication and cooperation aligns with previous findings (McCormick et al., 2013).

This study confirmed four distinct profiles with different specificities and features in CPS, indicating the different interrelationships among CPS motivations. To conclude, it was determined that achievement-motivated students performed better than instrument-motivated students, which was consistent with previous findings (Liu et al., 2020). Furthermore, research indicated that students who prioritize collaboration with others are more likely to achieve higher academic performance, aligning with earlier studies (Hilkenmeier, 2018; Wang & Ning, 2019). In Chinese culture, both “individual-oriented” and “social-oriented” achievement-driven students are commonly recognized as “Excellent students” (Chen, 2015). It is noteworthy, however, that a significant number of students fall into the category of “instrument-oriented non-team players” with lower levels of motivation. This trend may be attributed to external pressures, such as the standardized tests. The prevalence of academic stress in Chinese educational institutions often compels students to concentrate solely on their performance in entrance examinations (Zhao et al., 2015). While these four student categories were identified within the Chinese sample, we posit that these findings could be generalized to other cultural contexts sharing similar characteristics.

The factor related to CPS according to HLM

At the student level, gender was found to be the most important factor, which indicates that males perform worse than females. Females tend to solve problems through conversation and collaboration and showed more regulation in collaborative contexts (Li & Liu, 2017; Lin et al., 2020), which made females outperform males. ESCS was the second most important factor. Students with higher ESCS had more opportunities to develop their CPS skills in public places with a strong cooperative atmosphere (Tang et al., 2021). The last was SOIAICT, which showed that the higher the frequency of ICT in students' daily social lives is, the lower the student’s CPS score. This result was consistent with previous studies that pointed out that students' reliance on ICT reduced their opportunities to learn how to interact, collaborate, communicate, and negotiate with teammates (Xu & Li, 2019).

At the school level, only schools’ ECSC and the proportion of science teachers who were fully certified (PROSTCE) consistently impacted CPS performance. PROSTCE was the most important variable, which means that the more certified science teachers a school has, the higher the CPS score of the school's students. Science teachers could design activities and develop students’ CPS strategies, thus increasing students’ CPS skills (Li & Liu, 2017). The second most important variable was schools’ ESCS, where the higher the schools’ ESCS is, the higher the CPS score. According to previous research, schools with higher ESCS could provide more ICT resources and CPS courses (Tang et al., 2021).

Limitations and future directions

This study has several limitations. First, our research, which focused only on a few types of motivation that might be insufficient, did not provide a complete picture of the relationship between motivation and its explicit features. Second, CPS student perception is not only limited to belonging, anxiety, and emotional support. Meanwhile, this study only picked topic-level process data (e.g., item response time, action sequences) and did not focus on operational-level process data (e.g., operation records and operation sequences), which might provide additional insight into individual behaviours (Liu et al., 2022). Third, as a result of the secondary data analysis of PISA, which is a large-scale survey, there is a limitation in terms of the tools and participants that were utilized in the analysis. Motivations, for instance, were not built expressly to analyse the variation of different motivation levels in different tasks or moments. According to PSI theory, motivation levels might fluctuate continuously according to certain situations due to interactions with peers or with the environment (Cai et al., 2012). Future studies should also investigate motivational variation across tasks or specific situations by analysing CPS process data or multimodal data (Liu et al., 2022).