Science career expectation and science-related motivation: a latent profile analysis using PISA 2015 data

ABSTRACT

In response to growing concerns about students’ low aspiration for science-related careers, more understanding on how student science motivation is related to science career aspiration is needed. This study utilised a person-centred approach to investigating the association between students’ science-related motivation profiles and their science career expectation using the 2015 Programme for International Student Assessment (PISA) data set. Analysing responses from 9841 15-year-old students in China through Latent Profile Analysis (LPA), the study found that: (1) five distinct students’ motivation profiles were identified based on science-related motivation factors; (2) significant variations were observed in science career expectations across students’ motivation profiles, and different profiles predicted science career expectations differently; and (3) the perceived instructional practices and science learning engagement, social and economic status, and students’ gender significantly predicted students’ motivation profiles differently. These research findings contribute to new insight on the non-linear relationships between science-related motivation and science career expectation, and how student motivation profiles are impacted by demographic and context factors such as instructional practices, learning engagement, students’ social and economic status, and gender.

KEYWORDS:

• Science career expectation
• science-related motivation
• latent profile analysis

Introduction

With the prevailing global consensus that in order to maintain a prominent position in the fields of science and to drive the progress in technological innovation, it is crucial to actively encourage a greater number of young individuals from diverse backgrounds to pursue a career path in science, technology, engineering, and mathematics (STEM) (OECD, 2017a; Rotermund & Burke, 2021). Specifically, research has unveiled a concerning lack of desire among students around 15 years old to pursue science-related careers (Archer et al., 2020; Aschbacher et al., 2010; Susan & Amy, 2021; Vedder-Weiss & Fortus, 2011). For instance, in a comparative study conducted by Du and Wong (2019), it was found that Chinese students had a similar science achievement to that of teenagers from the United Kingdom, but lower science career expectation. A similar pattern also has been identified between China and the United States; i.e. despite a high level of scientific achievement, Chinese 15-year-olds have substantially less interest in pursuing a career in science than their American counterparts (OECD, 2017b).

Through examination of underlying factors that influence student science career expectation, it has been found that students’ aspirations to pursue a career in science are not necessarily determined by their science achievement (Normandeau, 2017). Instead, a range of motivation factors have been found to be significant (Badri et al., 2016; Guo, 2022), such as science-related attitudes (e.g. Andersen & Chen, 2016; Chi & Wang, 2023; Tytler & Osborne, 2012), self-efficacy (e.g. Avargil et al., 2020; Sahin et al., 2018), enjoyment and interest towards science (e.g. J. Wang et al., 2020; H.-H. Wang et al., 2021), and epistemological beliefs (e.g. She et al., 2019; Trautwein & Lüdtke, 2007). Further, recent research has revealed that these factors do not act in isolation, but rather in interwoven ways, and their mechanism to influence science career expectation is complex (Archer et al., 2020). The interdependency among these factors has also been investigated (Mason et al., 2013).

In addition to factors associated with student motivation mentioned above, contextual factors within the educational environment, such as teachers’ teaching practices and students’ learning engagements, also play crucial roles. Research has found that teachers’ teaching practices or instruction approaches significantly impact students’ learning outcomes (e.g. Teig & Nilsen, 2022). Furthermore, these teaching practices have been linked to students’ interest and attitudes towards science (Hampden-Thompson & Bennett, 2013; Wang, 2012). However, how the relationships function among teachers’ teaching practices, students’ science-related motivations, and further students’ science career expectation remains unclear. Existing research suggests that specific teaching practices, such as teacher-directed and inquiry-based practices, have the potential to foster positive science dispositions among students (Areepattamannil et al., 2020; Lei et al., 2018). Yet, further exploration is needed to fully understand how teaching strategies relate to students’ motivation and career expectations in science. Additionally, student learning engagement relates to both academic and non-academic outcomes (Wang, 2012; Yang et al., 2022). Gender of students and socioeconomic status of students also affect learning performance and career choices (Napp & Breda, 2022; Perry et al., 2022; Wegemer & Eccles, 2019), mediated by beliefs and attitudes (Jeffries et al., 2020; Peng & Yue, 2022).

Overall, there is evidence suggesting a relationship between students’ science-related motivation and their career expectation. However, it would be beneficial to have a more holistic understanding of the nature of this relationship by taking multiple motivation variables into consideration. Methodologically, previous studies have been almost based solely on variables that assume linear relations. Research has found that when variables (e.g. interest) are used as units of analysis, findings can be misleading (Kaplan et al., 2012; Molenaar et al., 2009).

To address these research limitations, this study aims to utilise a person-centred approach to examining how different profiles of students’ motivation factors are associated with students’ science career expectation. The purpose of this study is to identity latent profiles of students’ motivation and find out how these latent profiles relate to students’ science career expectation. Specific research questions are:

1. What are the latent profiles of student science-related motivation?

2. How do student science career expectation differ in terms of student motivation profiles?

3. How do instructional practices and student engagement predict different student motivation profiles?

4. How do gender and socioeconomic status predict different student motivation profiles?

The findings of this study will help understand the nature (e.g. non-linear) of relationships between science-related motivation and science career expectation. Furthermore, the utilisation of a person-centred approach in identifying distinct subgroups of students enhances the recognition of the heterogeneity among students, which offers methodological insight for future studies on student science career expectation. Finally, research findings will also provide insights into possible ways to enhance student science career expectation through cultivating positive science-related motivation factors and improving contextual variables.

Theoretical framework

Multiple theories have been put forth on career development, including expectancy-value theory (EVT) (Eccles et al., 1983; Eccles & Wigfield, 2002; Wigfield & Eccles, 2000), social cognitive career theory (Lent et al., 1994), and the career development systems theory framework (STF) (McMahon & Patton, 2018; Patton & McMahon, 1999, 2006, 2014, 2015, 2017). This current study is informed by an integrated theoretical framework that combines the STF and EVT to systematically explore the synergistic effects of multiple motivational factors, demographic and contextual variables on science career expectation (see Figure 1).

Figure 1. The synthesised science-related motivation and career expectation model.

According to STF, an individual's career aspiration is influenced by multiple interconnected systems including intrapersonal, social and environmental-societal systems (McMahon & Patton, 2018). Each system comprises various factors that influence individual's career choice and development. For example, the individual's gender, age, self-concept, abilities, beliefs, interests, values, and aptitudes impact career aspiration (McMahon & Patton, 2018). Instead of considering these factors in isolation, the STF emphasises their interrelationships from a holistic perspective of systems thinking (McMahon, 2014). Drawing on the STF, we propose that multiple factors of the intrapersonal system function not only as independent influences but also interact synergistically within the system to shape students’ science career expectation.

EVT is another theoretical framework that elucidates the influence of motivational factors and self-related beliefs on the decision-making process, perseverance, and overall performance of individuals (Eccles et al., 1983; Eccles & Wigfield, 2002; Wigfield & Eccles, 2000); it has been widely used in the research of science career expectation (e.g. Wang et al., 2020). Motivation is a fundamental driving force behind students’ engagement, learning, and career choices. It is defined as ‘the energy students bring to their educational tasks, the beliefs, values, and goals that determine which tasks they pursue, and their persistence in achieving them’ (Wentzel & Miele, 2009). Grounded in the EVT, individuals’ career expectation is determined by a multidimensional construct of motivation, encompassing both individuals’ beliefs about their capabilities (expectancy) and their perceptions of the task's importance and value (value) (Wentzel & Miele, 2009). Four aspects of values were identified: utility value, cost, attainment value, and intrinsic value (Eccles et al., 1983; Eccles & Wigfield, 2002).

Within EVT, science self-efficacy, representing students’ belief in their ability to perform science tasks, can be identified as an element of expectancy. Moreover, tasks with high intrinsic value, where students find inherent interest and enjoyment, are more likely to elicit motivation. The enjoyment of science (i.e. students finding learning science and working on science problems enjoyable) and interest in broad science topics can be considered intrinsic motivation (intrinsic value in EVT). Additionally, instrumental motivation to learn science (i.e. students perceiving learning science as useful for their future studies and careers) can be regarded as extrinsic motivation (utility value in EVT). Besides, students’ epistemic beliefs are defined as their tacit beliefs about the nature of knowledge and the process of knowing (Hofer & Pintrich, 1997). Prior research in science education has demonstrated the influence of students’ epistemic beliefs about science on their motivation to learn science (Ho & Liang, 2015). Thus, science epistemological beliefs were integrated into the overall construct of motivation in our study.

In addition, it is widely recognised that demographic variables, such as gender, SES, and school environment, as well as school teaching and learning have an indirect impact on the career expectations of teenagers within both theories. This influence is thought to occur through their assessments of personal ability and values (Gao & Eccles, 2020; Vedder-Weiss & Fortus, 2011).

Literature review

Science knowledge acquisition does not inevitably lead to a strong aspiration for pursuing a science-related career. The science-related motivation significantly impacts individuals’ career decision-making (Bathgate et al., 2014). Many empirical studies have been conducted to explore the relationship between a series of science-related motivation factors and science career expectation.

Science-related motivation

There is no universally agreed upon definition of motivation (Jones & Carter, 2007; Liu, 2020). Motivation can be considered as ‘an internal state that arouses, directs, and sustains students’ behaviour’ (Koballa & Glynn, 2007, p. 85). There are four approaches to defining motivation: the behavioural, humanistic, cognitive, and social (Koballa & Glynn, 2007). The behavioural perspective on motivation highlights the role of external rewards and reinforcement, while the humanistic viewpoint prioritises human's intrinsic needs like self-actualisation and autonomy. Cognitive approach focuses on the importance of goals, expectations, and beliefs in motivating behaviour. The sociocultural approach, meanwhile, examines how social identities and interactions within a community influence motivation. In this study, we take a synthesised approach of behavioural, humanistic and cognitive perspectives to conceptualising motivation related to science to include five components: enjoyment of science, interest in science, instrumental motivation of learning science, science self-efficacy, and epistemological beliefs about science.

Laukenmann et al. (2003) stated that enjoyment of science reflects students’ attachment to learning science and experiencing it as a meaningful activity. Interest in broad science topics refers to an individual's engagement and positive feelings towards a comprehensive range of scientific subjects and domains (Krapp & Prenzel, 2011). According to Krapp and Prenzel (2011), both enjoyment and interest are intrinsically motivated and self-determined. This aligns with the humanistic perspective of motivation, which emphasises the importance of intrinsic needs and self-autonomy.

Instrumental motivation in science learning is defined as students’ perceptions of the utility of science for their future lives focusing on the outcomes and consequences of specific actions rather than the act of learning itself. This form of motivation emphasised the practical benefits of science education for achieving desired future goals (Schiepe-Tiska et al., 2016). It is often driven by external rewards or purposes beyond the intrinsic value of learning, thus aligning with the behavioural motivation perspective that underscores the importance of external rewards and reinforcement.

Science self-efficacy, as defined by Bandura (1977), relates to an individual's belief in their capability to successfully execute specific science-related tasks. This concept aligns with the cognitive perspective of motivation, focusing on the role of personal beliefs and perceptions in driving behaviour. Epistemological beliefs about science are beliefs about the nature of knowledge and knowing (Hofer & Pintrich, 2002). Epistemological beliefs about science can influence individuals’ motivational processes in specific domains or tasks (Khine et al., 2020) According to the self-regulated learning theory (Zimmerman, 1989), epistemological beliefs play a critical role in how students approach, engage with, and perceive the value of learning tasks within specific domains. If students believe that science knowledge is complex, evolving, and can be understood through effort, they are more likely to engage deeply with science tasks, showing higher motivation and self-regulation in learning. Moreover, sophisticated epistemological beliefs are associated with higher self-efficacy (Brandmo & Bråten, 2018). Both science self-efficacy and epistemological beliefs about science as two belief systems shape individuals’ perceptions of the external world and influence their developed beliefs, leading them to exhibit certain behaviours (Aksan, 2009). In this sense, epistemological beliefs about science can also be considered as a component of science-related motivation from the cognitive perspective.

The above-synthesised approach to conceptualising motivation, employing behavioural, humanistic, and cognitive perspectives aims to provide a comprehensive description of students’ profiles of science-related motivation. This effort seeks to elucidate the reason students pursue specific goals in science learning, emotions and beliefs they experience throughout this process, and to explore how a series of background factors affect science-related motivation and how such motivation influences students’ science career expectations.

Science-related motivation and Science career expectation. Science career expectation is defined as ‘those career expectations whose realization requires further engagement with the study of science beyond compulsory education, typically in formal tertiary education settings’ (OECD, 2016b, p. 282). Studies have shown that enjoyment and interest in science can lead to a greater likelihood of choosing science-related occupations (Badri et al., 2016; Chi & Wang, 2023). These two variables have been found to contribute to students’ career aspirations (Badri et al., 2016; Guo, 2022). Positive science engagement experiences enhance their understanding of potential science careers (Mujtaba et al., 2018). Both of them, as intrinsic motivation, are the strongest predictors of science career expectations (Ustun, 2023). The predictive value of intrinsic motivation for science career expectations is more consistent across different groups and stronger than instrumental motivation (Ahmed & Mudrey, 2019).

Instrumental motivation of learning science, as an external driven factor to engage in tasks or activities for practical or pragmatic reasons, plays a role in shaping students’ science career expectations. Students learn science because of the usefulness of learning science to their future studies and careers (Guo, 2022; Liu et al., 2023; Ustun, 2023; Wigfield & Eccles, 2000). It has been found that instrumental motivation to learn science is one of the most important predictors of course selection and career choices (Eccles, 1994; Eccles & Wigfield, 1995). Previous research supports that students with higher instrumental motivation to learn science are more likely to continue learning science-related subjects even when they are not compulsory and tend to have higher achievement in science (Rozek et al., 2015; Xiao & Sun, 2021). However, instrumental motivation appears to have a weaker association with science career expectations compared to intrinsic motivation (Ustun, 2023). Furthermore, instrumental motivations may undermine or crowd out the effects of internal motivations which are more strongly associated with long-term engagement and success in a field, such as pursuing a science career (Wrzesniewski et al., 2014).

Science self-efficacy, as students’ belief in their ability to perform science tasks, is a significant factor influencing students’ science learning performance and their career orientation. Previous studies have identified a positive relationship between self-efficacy and career expectations among students (Blotnicky et al., 2018; Chi & Wang, 2023; Liou, 2017; Ustun, 2023). It has also been found that self-efficacy significantly predicts career orientation through science interest (Schar et al., 2017). However, the strength of self-efficacy in predicting science career expectations varies across countries, as indicated by Ustun (2023).

Students’ epistemological beliefs about science, which refer to their views on the nature, development, and justification of scientific knowledge, can influence their science career expectation. Sophisticated epistemological beliefs are associated with higher motivation, achievement, and aspirations in science (Chai et al., 2021; Guo, 2022). Furthermore, epistemological beliefs can mediate the relationship between students’ science learning experiences and their career expectations (Chi & Wang, 2023).

Contextual factors and science-related motivation

Factors that have the greatest impact on students’ aspirations to pursue careers in science are the amount of time they dedicate to studying science and the pedagogical approaches employed by their teachers. This holds true even when socioeconomic status and school performance in science are taken into consideration (OECD, 2016b).

Teachers’ instructional practices and science-related motivation

Classroom instructional practices in science education, including teacher-directed and inquiry-based instructional methods, significantly influence students’ interest and motivation to engage in science-related STEM fields (Areepattamannil et al., 2020). Inquiry-based instruction, a constructivist teaching approach, involves ‘engaging students in experimentation and hands-on activities, as well as challenging students and encouraging them to develop a conceptual understanding of scientific ideas’ (OECD, 2016b; Prince & Felder, 2006). This instructional practice allows students to experience, act, and work in science as participants, reflecting the authentic practice of scientists as they generate scientific knowledge (National Research Council, 1996; Patall et al., 2018). Inquiry-based instruction has been recognised for its positive effects on students’ science-related motivation and science career expectations (Chi & Wang, 2023). Gibson and Chase (2002) found that an inquiry-based programme led to a more positive attitude towards science and a higher interest in science careers. There has been an extensive literature on the positive impact of inquiry-based instruction on students’ enjoyment of science, interest in broad science topics, instrumental motivation in science learning, science self-efficacy, and epistemological beliefs about science (Areepattamannil et al., 2020; Cairns & Areepattamannil, 2017; Chi & Wang, 2023; Kang & Keinonen, 2017; Mostafa et al., 2018).

Teacher-directed instruction is defined as teaching practices in which the teacher plays a central role in initiating and structuring the learning process, primarily responsible for guiding students through carefully designed educational content and methodologies (Long et al., 2022). Despite the popularity of inquiry-based instructional practice, teacher-directed instruction remains common across numerous education systems, including those in Arabic-speaking countries, some English-speaking countries, and B-S-J-G (China) (OECD, 2016b, p. 44). Teacher-directed instructional approaches are reported to have a consistently positive impact on students’ learning outcomes (Areepattamannil et al., 2020; Cairns & Areepattamannil, 2022; Stockard et al., 2018). For instance, the PISA 2015 report revealed a clear, positive, and significant association between student-reported exposure to teacher-directed science instruction and students’ enjoyment of science across all countries and economies. Similarly, a positive and significant correlation between science self-efficacy and epistemic beliefs and teacher-directed science instruction was also observed in all countries and economies (OECD, 2016b, p. 47). Teacher-supported teaching, which fosters autonomy, competence, and relatedness, can enhance students’ intrinsic motivation and improve their academic performance (Ginzburg & Barak, 2023). Moreover, logistic regression analyses investigating the association between exposure to teacher-directed science instruction and the expectation of a science career at age 30 demonstrated a positive and statistically significant relationship in 30 out of 67 countries and economies (OECD, 2016b, p. 50).

Students’ learning engagement and science-related motivation

Students learning time in and out of school as well as students’ participation in science activities are objective indicators of students’ learning engagement. Expanding learning time in school has been proven to promote students’ achievement (Kolbe & O’Reilly, 2017) as well as non-academic outcomes (Redd et al., 2012). A meta-analysis showed that there would be a positive effect on students’ motivation by increasing learning time out of school (Kidron & Lindsay, 2014). Deci and Ryan (2013) asserted that the amount of time students engage in science learning activities can have an impact on their enjoyment of science. Engaging and well-structured science activities that allow sufficient time for exploration and understanding can enhance students’ enjoyment. Likewise, the time dedicated to exploring a variety of science topics in and out of the curriculum can impact students’ interest in broad science topics. Exposure to diverse scientific concepts and phenomena over time can foster a wider interest in science (Krapp & Prenzel, 2011). Instrumental motivation, or the motivation to learn science due to its perceived utility (Eccles & Wigfield, 2002), can be influenced by the amount of time students perceive as useful and relevant to their future goals. Extended learning time that connects science education with real-world applications and future career opportunities can enhance instrumental motivation. Conceptually, increased learning time, especially when that time is spent on activities that provide opportunities for mastery experiences, positive feedback, and overcoming challenges in science learning can positively affect students’ science self-efficacy. According to Hofer and Pintrich (1997), students’ epistemological beliefs about science can develop and become more sophisticated with increased exposure to science learning experiences over time. From this, engaging with scientific inquiry and problem-solving activities can help students appreciate the complexity and dynamic nature of scientific knowledge. Participation in informal science learning activities also promotes diverse youth’s career aspirations and developments (Caspi et al., 2023; Zhao et al., 2023). The general consensus in science education research is that the quality and quantity of science learning experiences play a crucial role in shaping students’ attitudes, motivations, and beliefs about science and further impact science learning outcomes, such as science career expectations.

Students’ gender, socioeconomic status, and science-related motivation

Gender differences in science-related motivation have been observed in various studies. Specifically, female students often report lower levels of interest and enjoyment in science (Else-Quest & Peterca, 2015), lower science self-efficacy (Fahle et al., 2019), but more sophisticated beliefs in certain dimensions (Ozkan & Tekkaya, 2011), compared to male students.

Socio-economic status (SES) can be defined as a family’s or a person’s position on a hierarchy that depends on social status, access to wealth, and possessions (Mueller & Parcel, 1981). Differences among students from different SES can also be observed in their enjoyment and interest in science, and levels of instrumental motivation to learn science (Alhadabi, 2021; Grabau, 2016). Students from high-SES families were more likely to have higher science self-efficacy (Tan et al., 2023) and believe that knowledge is uncertain and not handed down by authority (Ozkal et al., 2011) compared to students from low SES family. Students from high-SES families were more likely to be ready to major in science-related fields and have science-related career aspirations (Cooper & Berry, 2020; Tilleczek & Lewko, 2001).

Methods

This study employed a person-centred approach to considering multiple science-related motivation factors simultaneously (Bergman et al., 2003). This approach uses individuals as unit of analysis, acknowledging the multiple factors that operate simultaneously in individuals’ lives (Bergman & Trost, 2006). It allows for a holistic examination of various factors related to science-related motivation, avoiding oversimplified thinking on individual factors and effectively bridging theory and practice. Through this approach, we aimed to identifying latent profiles of science-related motivation within the student population, moving beyond categorisations of individual factors as ‘positive’ or ‘negative’ in isolation (Byrd, 2017). This analysis offers a deeper understanding of the complex interactions among different factors, shedding light on students’ science-related motivation patterns and its impact on their career expectations in science.

Subjects

We utilised the science assessment data derived from PISA 2015 for four provinces/municipalities in China, including Beijing, Shanghai, Jiangsu, and Guangdong (B-S-J-G-China). The Chinese sample was chosen as China is a unique case with observed high achievement but low career expectation when compared to other countries in the PISA 2015(OECD, 2016a). The sample included a total of 9,841 (4682 girls, 47.6%) 15-year-old students from 268 schools. The weighted sample represented a student population of 1,331,794 individuals.

Measures

PISA constructs weighted likelihood estimates (WLEs) as an index for items that collectively measure a single latent trait or construct, such as practices in enjoyment of science by utilising the item response theory framework and the generalised partial credit model. These WLE scores are assigned to individuals, normalised to have an OECD mean of 0 and an OECD standard deviation of 1 (OECD, 2017c). The PISA 2015 technical report provides a comprehensive explanation of the scaling methods and the validation of the constructs (OECD, 2017c).

The present study employs five science-related motivation variables, seven demographic and contextual variables, and one outcome variable. All the science-related motivation variables, contextual variables and the outcome variable are WLE scores taken directly from the PISA 2015 dataset. Details of the variables are summarised in Table 1.

Table 1. Description of variables and their descriptive statistics.

Variables		Description	Scale reliability	N	Min	Max	Means	Sd.	Missing rate (%)
Science motivation-related variables	JOYSCIE (WLE)	Continuous variable^a. Derived from ST094Q01∼05^b	0.940	1314351	−2.115	2.164	0.370	0.889	1.31
	INTBRSCI (WLE)	Continuous variable^a. Derived from ST095Q04,07,08,13 and15^b	0.787	1302168	−2.532	2.603	0.454	0.809	2.23
	INSTSCIE (WLE)	Continuous variable^a. Derived from ST113Q01∼04 ^b	0.901	1314463	−1.930	1.736	0.531	0.751	1.30
	SCIEEFF (WLE)	Continuous variable^a. Derived from ST129Q01∼08^b	0.891	1315488	−3.757	3.278	−0.010	1.146	1.23
	EPIST (WLE)	Continuous variable^a. Derived from ST131Q01∼05^b	0.857	1314276	−2.790	2.155	−0.081	0.848	1.32
Science career expectation	SCEXP	Dummy variable. Coded as 0 = non-science-related career expectation; 1 = science-related career expectation^c	/	912861	0	1	/	/	31.47
Demographic variables	GENDER	Categorical, coded as 1 = Female; 2 = Male	/	1331794	0	1	/	/	0.00
	ESCS (WLE)	Continuous variable^a. Derived from the indicators parental education (PARED), highest parental occupation (HISEI), and home possessions (HOMEPOS)^b	0.740	1323060	−4.973	3.614	−1.074	1.103	0.66
Science instructional practices and science learning engagement variables	TDTEACH (WLE)	Continuous variable^a. Derived from ST103Q01,03,08 and11^b	0.858	1237952	−2.448	2.078	0.008	1.022	7.05
	IBTEACH (WLE)	Continuous variable^a. Derived from ST098Q01∼09, except Q04^b,d	0.898	1240606	−3.341	3.183	−0.282	1.100	6.85
	OUTHOURS (SUM)	Continuous variable. Sum of ST071Q01NA	/	1196951	0.000	30.000	5.745	5.980	10.12
	SHOURS	Continuous variable. Multiplication of ST059Q03TA and ST061Q01NA	/	1294324	0.000	33.330	5.563	3.408	2.82
	SCIEACT (WLE)	Continuous variable^a. Derived from ST146Q01∼09^b	0.922	1315743	−1.761	3.362	0.518	0.964	1.21

^aThis index is standardised to have a mean of 0 and a standard deviation of 1 across OECD countries.

^b These items are all reverse coded in order that the higher of the WLE scores corresponded to the higher level of indices.

^c The index of science-related career expectations is derived from a question asking students about ‘what kind of job (they) expect to have when (they) are about 30 years old’.

^d These items are all reverse coded in order that the higher values of the index corresponded to the higher frequency of inquiry-based science teaching.

Science-related motivation factors

In this study, five WLE scores of science-related motivation variables were used: enjoyment of science (JOYSCIE), interest in broad science topics (INTBRSCI), instrument motivation (INSTSCIE), science self-efficacy (SCIEEFF), and Epistemological beliefs about science (EPIST). The derived variable JOYSCIE is constructed from five items about how students generally have fun when they are learning science (OECD, 2016a, p. 123). INTBRSCI is constructed from five questions assessing students’ interest in the topics like the biosphere, motion and forces, energy and so on. The PISA 2015 constructs INSTSCIE from the responses to four questions to measure the extent to which students feel that science is relevant to them. SCIEEFF is constructed using eight items asking students on how they could perform different science tasks. EPIST is derived from students’ views on scientific approaches using six questions. Higher values of these indices indicate that students have more positive science-related motivation.

Science career expectation

In the PISA 2015 student questionnaire, students were asked to report their expected occupation at age 30 using the question (ST114) ‘what kind of job [they] expect to have when [they] are about 30 years old’ (OECD, 2016a). Students’ responses were coded into four-digit ISCO-08 codes and then classified into science-related and non-science-related occupations. Four groups of science-related jobs, i.e. Science and engineering professionals, Health professionals, Information and communications technology (ICT) professionals, and Science technicians, and associate professionals, were coded into Science-related career expectations, and the rest into non-Science-related career expectations to derive a binary variable Science Career Expectation (SCEXP).

Demographic variables

Gender. Gender is a dichotomous variable from student questionnaire of PISA 2015.

ESCS. The index of economic, social and cultural status (ESCS) is a composite score (WLE) constructed based on parental education, highest parental occupation, and home possessions from student questionnaire of PISA 2015 (OECD, 2017c). Higher values of the indexes correspond to a higher levels of students economic, social and cultural status.

Science instructional practices and science learning engagement variables

Regarding science instructional practices, two indicators: teacher-directed instruction (TDTEACH), and inquiry-based science teaching and learning practices (IBTEACH) were selected. PISA 2015 constructs TDTEACH from four questions and IBTEACH from eight items related to students’ perceptions of how frequently teacher-directed and inquiry-based activities took place at school. Higher values of these indices correspond to higher frequencies of science teaching activities.

In terms of science learning engagement, three indicators were selected, out-of-school science learning time per week, in-school science learning time per week and students ́ science activities. In students’ questionnaire of PISA 2015, the learning time (minutes per week) was computed by multiplying the average minutes in a class period (ST061Q01NA) by number of class periods per week attended in science (ST059Q03TA). The variable SMINS was used to represent this combined information. The out-of-school study time in science was also surveyed through asking students ‘how much time they spent studying science in addition to their required school schedule (ST071Q01NA)’. The variable OUTHOURS was used to represent this information. In this study, we converted the unit of SMINS from minutes to hours for comparability between the two indicators within the group. At the same time, we changed the variable name accordingly to SHOURS to avoid confusion. Higher values of time correspond to higher levels of science learning engagement.

Finally, the index of science activities (SCIEACT) regarding how often they engaged in science-related activities was also selected. It is a derived variable from nine items. Higher values of this index correspond to higher levels of science learning engagement.

Data analysis

For the first research question (#1), Latent Profile Analysis (LPA) was conducted. Five variables, JOYSCIE, INTBRSCI, INSTSCIE, SCIEEFF, EPIST, were used to identify distinct student profiles of student science-related motivation. Models with two through five latent classes (k = 2 to 5) were tested to determine the number of emerging profiles from the data using Mplus 8.3 (Muthén & Muthén, 1998–2019). The fit indices of profiles include maximum Log-Likelihood (LL), Akaike's Information Criterion (AIC), Bayesian Information Criterion (BIC), Sample Adjusted Bayesian Information Criterion (SABIC), p-Value for Lo-Mendell-Rubin likelihood ratio test (LMR (p)), p-Value for the parametric bootstrapped likelihood ratio test (BLRT (p)) and Entropy.

For the second question (#2) to compare differences in science career expectations among students across different students’ motivation profiles. SCEXP was treated as the dependent variable and latent profiles as independent variables. For this purpose, the AUXILIARY (DCAT) function in Mplus 8.3 was utilised to test whether there was a statistical difference in science-related career expectation (SCEXP) by different student profiles (Collier & Leite, 2017; McLarnon & O’Neill, 2018). Furthermore, the logistic regression in SPSS 22.0 was employed to investigate the effect of distinct students’ motivation profiles on their science-related or non-science-related careers expectation.

For research questions (#3) and (#4) to investigate how the emerged student profiles could be affected differently by instructional practices, student engagement, gender and socioeconomic status, these variables were treated as predictors of the latent profiles. The 3-steps approach (Asparouhov & Muthén, 2014) was employed to identify the significant predictors of the previously established distinct profiles. The AUXILIARY (R3STEP) function in Mplus 8.3 was used.

Results

The descriptive statistics, e.g. means and standard deviations (Sd.), of all variables are presented in Table 1; correlation coefficients among five science-related motivation variables are showed in Table 2. Consistent with recommended practices in LPA research (Geiser, 2013; Johnson, 2021; Lubke & Muthén, 2005; Pastor et al., 2007), solutions with varying numbers of latent classes were tested. The stepwise approach was used to determine the number of final latent profiles that best characterise the data and sample, starting with LPA with two profiles and successively adding profiles (Nylund et al., 2007). The fit index criteria (see Table 3) were examined. Further, the theoretical coherence, past empirical evidence, the characteristics of each profile, and the interpretability of the solution were considered for the decision on the final profile solution.

Table 2. Correlations between the motivation indices used in latent profile analyses.

	(1)	(2)	(3)	(4)	(5)
JOYSCIE (1)	1	0.452**	0.297**	0.332**	0.369**
INTBRSCI (2)	—	1	0.308**	0.310**	0.272**
INSTSCIE (3)	—	—	1	0.219**	0.2229**
SCIEEFF (4)	—	—	—	1	0.248**
EPIST (5)	—	—	—	—	1

Note: * p < .05, **p < .01, ***p < .001.

Table 3. Fit indexes of LPA models.

No. of groups	Log-likelihood	FP	AIC	BIC	SABIC	Entropy	LMR (p)	BLRT (p)	Smallest profile
2	−61387.921	16	122807.842	122922.843	122871.997	0.862	0.000	0.000	14.44%
3	−59965.783	22	119975.566	120133.69	120063.780	0.910	0.000	0.000	13.57%
4	−58017.755	28	116091.511	116292.763	116203.783	0.960	0.000	0.000	3.12%
5	−57750.762	34	115569.523	115813.901	115705.854	0.962	0.049	0.000	0.72%

Note: FP, No. of free parameter; Smallest profile: smallest class frequency.

The values of AIC, BIC, and SABIC kept descending as additional profiles were included, with significant LMR and BLRT values for all models. The additional solutions of increasing numbers of profiles continued to be examined and finally we found the five-class profile solution to be the most appropriate, because the six-class solution resulted in the first nonsignificant LMR value (p = .170). Furthermore, the five-class profile solution had a relatively high entropy value (0.962), reaching the optimum size of 0.80, indicating that student cases could be appropriately allocated to different latent profiles with acceptable certainty. Finally, the five-class profile solution showed a number of qualitatively different profiles of theoretical interest that further added significant new information about students.

To consider the reliability of the profiles, we randomly split the entire sample into two halves. Compared to the full student sample, identical patterns were uncovered with split samples, providing evidence for the reliability of the selected profiles.

RQ1. Latent profiles of student science-related motivation

Figure 2 presents the five profiles, mean scores for each science-related motivational variable, along with the proportion represented by each profile. The description of the five-class profile as follows:

1. The Average profile (61.1%), consisting of the largest proportion of students, was associated with average scores on all five indicators (Mean = −0.062 ∼0.532).

2. The Unconfident profile (22.0%) had lower scores than the Average profile across five indicators. It was also associated with the lowest scores in science self-efficacy indicator (Mean = −0.537 ∼ 0.221).

3. The Enthusiast profile (13.0%) demonstrated the highest scores in both instrumental motivation and epistemological beliefs indicators, and second highest scores across the other three indicators (Mean = 0.898 ∼2.041).

4. The Uncommitted profile (3.1%) exhibited the lowest scores on all five indicators, indicating a lack of enjoyment, interest, and belief in science, showing hesitation in committing to science learning (Mean = −2.033 to 0.155).

5. The Scientist profile (0.7%) showed the highest levels across three indicators of interest, enjoyment, and self-efficacy in science (Mean = −2.197 ∼ 2.068).

Figure 2. Latent profiles of student motivation.

RQ2. Students’ science-related motivation profiles and their science career expectation

Table 4 presents comparison results among the motivation profiles in terms of science career expectation. As seen in Table 4, compared with Average profile, students in Enthusiast profile were more likely to be engaged in science-related career (OR = 165.6), whereas students in Uncommitted profile were less likely to be engaged in a science-related career. Group differences in science career expectation among the profiles were statistically significant(χ²=  64.067 p = 0.000). A pairwise comparison among five profiles was performed. Results indicated that ‘Enthusiast profile’ with the highest probability (Prob = 0.368) of expecting a science-related career was significantly different from all other profiles, ‘Scientist’ profile was significantly different from ‘Unconfident’ profile, and ‘Average’ profile was significantly different from the ‘Unconfident’ profile.

Table 4. Students’ profile differences in terms of science career expectation (SCEXP).

Profile	SCIEEXP	Prob.	S.E.	OR	Pairwise comparison (Chi-Square)
					Vs. 2	Vs. 3	Vs. 4	Vs. 5
Uncommitted (1)	Non-science	0.728	0.030	1.000	0.616	3.767	15.605***	1.902
	Science	0.218	0.030	0.793
Unconfident (2)	Non-science	0.807	0.011	1.000	—	15.205***	45.472***	26.917***
	Science	0.193	0.011	0.681
Scientist (3)	Non-science	0.710	0.022	1.000	—	—	4.732*	1.648
	Science	0.290	0.022	1.164
Enthusiast (4)	Non-science	0.632	0.023	1.000	—	—	—	19.308***
	Science	0.368	0.023	1.656
Average (5)	Non-science	0.740	0.006	1.000	—	—	—	—
	Science	0.260	0.006	1.000

Note: Reference profile is Averaged profile. S.E.: Standard Error. OR: Odds Ratio. *: p < 0.05; **: p < 0.01; ***: p < 0.001.

Furthermore, logistic regression analysis (Table 5) revealed that the Enthusiast profile of students was a significant positive predictor of science career expectation (B = 0.451 p = 0.000,), while all other profiles of students were a significantly negative predictor of science career expectation, suggesting that the relationship between student profiles and science career expectations was non-uniform.

Table 5. Logistic regression of students’ profiles on science career expectation (SCEXP).

Profile	B	SE	P-value	Exp(B)
Uncommitted (1)	−0.291	0.016	0.000	0.747
Unconfident (2)	−0.435	0.007	0.000	0.647
Scientist (3)	−0.831	0.043	0.000	0.436
Enthusiast (4)	0.451	0.007	0.000	1.570

Note: Reference profile is Averaged profile. *p < 0.05; **: p < 0.01; ***: p < 0.001.

RQ3. Perceived instructional practices and science learning engagement on students’ motivation profiles

Table 6 presents the prediction results of perceived instructional practices and science learning engagement on students’ motivation profiles. Differences in profiles were significantly predicted by teacher-directed instruction (TDTEACH) and inquiry-based science teaching practices (IBTEACH). Compared to Average profile, inquiry-based science teaching practices could make students to have a higher likelihood (38.9%) of belonging to the Enthusiast profile, but teacher-directed instruction could make students to have a 70.9% higher likelihood of being in the Scientist profile and a 56.0% higher likelihood of being in the Enthusiast profile.

Table 6. Odds-ratio being in a profile as a function of IBTEACH, TDTEACH, SHOURS, OUTHOURS, and SCIEACT.

Profile	IBTEACH			TDTEACH			SHOURS			OUTHOURS			SCIEACT
	Estimate	S.E.	OR	Estimate	S.E.	OR	Estimate	S.E.	OR	Estimate	S.E.	OR	Estimate	S.E.	OR
1	−0.655***	0.079	0.520***	−0.522***	0.087	0.594***	0.014	0.018	1.014	0.008	0.012	1.008	−0.699***	0.082	0.497***
2	−0.459***	0.026	0.632***	−0.414***	0.028	0.661***	−0.039***	0.009	0.961***	−0.032***	0.006	0.968***	−0.726***	0.030	0.484***
4	0.314	0.050	1.369	0.536**	0.193	1.709*	0.036	0.027	1.037	0.075***	0.017	1.078***	1.338***	0.253	3.811**
5	0.329***	0.231	1.389***	0.445***	0.037	1.560***	0.040 ***	0.008	1.041***	0.038***	0.005	1.038***	0.793***	0.050	2.209***

Note: Reference profile is profile 3: Averaged profile. Profile: 1: Uncommitted profile; 2: Unconfident profile; 4. Scientist profile; 5: Enthusiast profile. Estimate: polynomial logistic regression coefficients; S.E.: Standard Error. OR: Odds Ratio. *: p < 0.05; **: p < 0.01; ***: p < 0.001.

Regarding science learning engagement, compared to ‘Average profile’, the higher science learning time engagement in both in-school and out-of-school, the greater likelihood of belonging to the ‘Enthusiast profile’ (the higher likelihood is 4.1% for in-school and 3.8% for out-of-school). Increased engagement in out-of-school learning time could increase the likelihood of students being categorised as ‘Scientist profile’ by 7.8%. Students showing higher frequency to science activities were more likely to be in ‘Scientist profile’ with a 281.1% higher likelihood and in ‘Enthusiast profile’ with a 120.9% higher likelihood, compared to ‘Average profile’.

RQ4. Students’ profile differentiation by gender and ESCS

Table 7 presents effects of gender and ESCS on student motivation profiles. Differences in profiles were observed based on gender and ESES. Boys were more likely to be in the ‘Enthusiast’, ‘Scientist’, and ‘Uncommitted’ profiles compared to the ‘Average’ group. Girls had a higher likelihood of being in the ‘Unconfident profile’. Specifically, boys had a 130.5% higher likelihood of being in the ‘Scientist profile’ and a 72.5% higher likelihood of being in the ‘Enthusiast profile’ than the ‘Average profile’. Girls in the ‘Unconfident profile’ had a 75.8% higher likelihood than the ‘Average profile’.

Table 7. Odds-ratio being in a profile as a function of gender and ESCS.

Profile	Gender			ESCS
	Estimate	S.E.	OR	Estimate	S.E.	OR
1	0.726***	0.129	2.067***	0.021	0.054	1.021
2	−0.277***	0.053	0.758***	−0.245***	0.026	0.783***
4	0.835**	0.303	2.305*	0.395**	0.131	1.484*
5	0.545***	0.066	1.725***	0.433***	0.029	1.542***

Note: Reference profile is profile 3: Averaged profile. Profile: 1: Uncommitted profile; 2: Unconfident profile; 4. Scientist profile; 5: Enthusiast profile. Estimate: polynomial logistic regression coefficients; S.E.: Standard Error. OR: Odds Ratio. *: p < 0.05; **: p < 0.01; ***: p < 0.001.

Students with lower ESCS were more likely to belong to the ‘Uncommitted profile’, while those with higher ESCS were more likely to belong to the ‘Scientist’ and ‘Enthusiast’ profiles. Students with higher ESCS had 48.4% higher odds of belonging to the ‘Scientist profile’ and 54.2% higher odds of belonging to the ‘Enthusiast profile’, compared to the ‘Average profile’.

Discussions

The profiles of science-related motivation

Regarding the first research question, our study identified five distinct profiles, with some profiles similar to the profiles found in previous research (Hofverberg et al., 2022; Radišić et al., 2021; She et al., 2019). Specifically, previous studies identified higher-level profiles corresponding to ‘Scientist’ and ‘Enthusiast’ profiles in our study, marked by higher values on nearly all indicators compared to the lower-level profiles. Notably, the Scientist profile is characterised by the lowest epistemological beliefs about science yet the highest self-efficacy in science. Although it constitutes the smallest proportion (0.7% of the total student population), these two profiles are similar to the ‘practical inquirer’ profile identified by Radišić et al. (2021) in an Italian sample and the ‘Mixed’ profile found by Hofverberg et al. (2022) in a Swedish sample. This suggests that different education systems demonstrate both common motivation profiles as well as different profiles.

The relationship between science-related motivation and science career expectation

Our study found that students’ science career expectations are significantly different by their different motivational profiles. Specifically, students with the Enthusiast profile demonstrated the highest science career expectations. Compared to the Average profile, Enthusiast profile students are 57% more likely to expect a science career, the highest probability within all profiles. The finding is consistent with previous studies (Grabau & Ma, 2017; Jeffries et al., 2020; Radišić et al., 2021). Conversely, students with Unconfident profile show a clear preference for non-science careers, which is evident from their higher probabilities of non-science career expectations and significant differences in science career expectations between Unconfident profile and other profiles.

Researchers adopting the EVT perspective have suggested that students’ aspirations for science careers are typically influenced by their value, interests, and motivation for learning or engaging in science (Chi & Wang, 2023; Jones & Hite, 2020). This study further verifies that science-related motivations – including enjoyment of science, interest in science, instrumental motivation for learning science, science self-efficacy, and epistemological beliefs about science – have a positive effect on students’ science-related career expectations. However, our findings also reveal that such an effect was non-linear, contributing to our understanding of EVT. Compared to the Average profile, only Enthusiast profile had a positive effect on student science career expectation, while other profiles had a negative effect on student science career expectation. Particularly, students with a Scientist profile are 56.4% less likely to expect a science career than the average profile students. This finding is somewhat counter-intuitive, as one might expect students with a Scientist profile to have a positive association with science careers. This anomaly suggests that the relationship between science-related motivation and science career expectation may not be a simple linear one.

Specifically, the non-linear relationship between student motivation profiles and science career expectation is further demonstrated by the different effects of Enthusiast profile and Scientist profile. Students with Enthusiast profile scored relatively lower on science-related motivation factors compared to Scientist profile students who scored relatively higher on science-related motivation factor scores in this study, yet students of Scientist profile had lower probability of and negatively predicated science career expectation. Our results provide new insights into the complex relationships between science-related motivation and science career expectation. They also suggest that employing a person-centred approach, such as latent profile analysis, has the capability of revealing complex relationships than variable-centred approaches with the assumption of linearity between independent variables and the dependent variable.

The relationship between contextual factors and science-related motivation

Our study found that teachers’ instruction, including teacher-directed instruction and inquiry-based science teaching practices, can significantly predict the profile category, which is in line with the related research that demonstrated the positive impact of teacher-directed instruction and inquiry-based science teaching on science dispositions among students (Areepattamannil et al., 2020; Lei et al., 2018). This finding may suggest that a combination of teacher-directed and inquiry-based instruction could nurture students’ positive science-related motivation. As for the students learning engagement, the finding shows that there is a significant correlation between the amount of time dedicated to studying science and the likelihood for developing a positive motivation profile. Moreover, the more time spent outside school, the more likely they are to demonstrate the ‘Scientist’ and ‘Enthusiast’ profiles. Thus, it is reasonable to suggest that the frequency of various science-related activities plays a positive role in shaping ‘Scientist’ and ‘Enthusiast’ profiles among students.

Students from higher ESCS backgrounds are more likely to be categorised in the ‘Scientist’ and ‘Enthusiast’ profiles, which are associated with a stronger intention to pursue science-related careers. The observed effect sizes underscore the magnitude of these associations between ESCS and motivation profiles. For instance, each unit increase in ESCS is associated with a 1.481-fold increase in the odds of a student belonging to the Scientist’ profile and a 1.542-fold increase in the odds of a student belonging to the Enthusiast profile. These findings align with previous research indicating that family socioeconomic status significantly predicts students’ science literacy and has a notable influence on their interest and attitudes towards science (Betancur et al., 2018; Diemer & Rasheed Ali, 2009). This study adds to the evidence of the positive impact of family socioeconomic status on students’ non-cognitive abilities. The results also highlight the importance of ESCS in shaping students’ motivation profiles, primarily through enhanced access to resources and support systems. Addressing socioeconomic disparities in access to resources and support systems through policy and educational practice is essential for promoting equity and improving educational outcomes for students.

Our research has revealed that boys have higher odds of belonging to the optimal patterns of science-related motivation, such as the Scientist profile and Enthusiast profiles. Girls exhibit a higher tendency to belong to the Unconfident profile in comparison to male students. The finding implies that boys exhibit relatively more motivation towards science when the profiles are considered as a whole.

Previous studies in the extant literature have identified gender gaps in students’ science-related interest, attitudes, and motivation (Ozkal et al., 2011; Wegemer & Eccles, 2019). Our findings reported above align with these previous research findings. Previous studies found that female students often express interest in science and hold positive views about scientists, yet they struggle to connect the image of scientists with themselves (Makarova et al., 2019). Our findings may be attributed to the underrepresentation of women in scientific fields in the modern society and stereotypes associated with scientists and science-related professions (Kim et al., 2018; Makarova et al., 2019). Furthermore, the findings of the present study indicate that educational systems and institutions should inspire and motivate female students in the field of science so that they develop optimal motivational profiles.

Conclusions

The objective of this study is to explore the association between different profiles of students’ science-related motivation and their science career expectation. The research makes four significant contributions. Firstly, the study identified non-linear relationships between science-related motivation and science career expectation, differing from the conventional view of a simple linear relationship. Further, the findings revealed significant differences in science-related motivation profiles among students based on their ESCS backgrounds and gender. These results underscore the importance of equity related to aims and goals in science education by extending beyond students’ academic performance to encompass their motivation. Thirdly, this study employed a person-centred methodology, i.e. laten profile analysis, to identify distinct student subgroups based on their motivation factors. By recognising the heterogeneity among students rather than treating them as a homogenous group, this study offers valuable methodological implications for future studies on student science career expectation. It opens avenues for exploring a range of characteristics such as identity within diverse student samples. Finally, this study provides nuanced understanding on the complex (i.e. non-linear) relationships between motivation and career aspirations as implied by the expectancy-value theory (EVT) and the systems theory framework (STF). Overall, this study contributes a holistic and systematic understanding of the predictors of students’ career expectation related to motivation and how motivation may be affected by student demographics and contextual factors.

Authors’ contributions

YF was responsible for designing the study, conducting literature searches, analysing data, as well as taking responsibility for manuscript writing. XF provided continuous feedback during manuscript preparation, supervising the project's progression, and making substantial revisions to this manuscript. The final manuscript was read and approved by both writers.

Acknowledgements

The authors would like to thank the editors and anonymous reviewers for their constructive feedback. In addition, we would like to thank our participants for their willingness to participate in the study.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The study represents a secondary data analysis of the public use PISA 2015 file provided by the OECD. The PISA 2015 Science data for China have been made publicly available by the OECD and can be accessed at: https://www.oecd.org/pisa/data/2015database/

Additional information

Funding

This study was not supported by any internal or external funding sources.