Symposium
Using person-centred approaches to examine intraindividual and interindividual differences in students’ motivational beliefs.
Chair(s): Wendy Symes (University of Potsdam, Germany), Rebecca Lazarides (University of Potsdam)
Discussant(s): Katrin Arens (DIPF: Leibniz-Institut für Bildungsforschung und Bildungsinformation)
Previous research has shown that students’ motivational beliefs significantly impact their learning and achievement (Eccles & Wigfield, 2020). Understanding more about how these beliefs arise, and how they change over time is therefore an important goal for educational psychology. However, much of the previous research in the area has made use of variable-centred approaches. Although these approaches provide an understanding of how motivational beliefs typically develop or influence outcomes, they have some important limitations. Firstly, even though such studies typically report significant variance in levels of motivational beliefs or their development over time, they do not examine the different trajectories students can follow. Secondly, statistical constraints typically lead to these studies exploring the impact of motivational beliefs individually, limiting the extent to which we can know how different motivational beliefs might influence outcomes in tandem. Finally, although these studies can identify factors that predict developmental change, they cannot show how they relate to the probability of following specific developmental trajectories. To address these limitations, researchers are increasingly turning to person-centred approaches of analysis. The purpose of our symposium is to showcase these alternative approaches, and to answer the question: ‘How can person-centred approaches increase our theoretical understanding of intraindividual and interindividual differences in students’ motivational beliefs?’
Our proposed symposium brings together four studies that use different person-centred approaches to analyse motivational and/or emotion data collected from students in different national contexts, at different stages of schooling, and from different theoretical perspectives. The four studies include samples from England (paper 1), USA (paper 2), Germany (paper 3) and Finland (paper 4). With regards to the different approaches, paper 1 and paper 2 use two methods for exploring interindividual differences cross-sectionally, namely latent profile analysis (LPA) and latent class analysis (LCA) respectively. Paper 3 and paper 4 utilise two methods for exploring interindividual differences in intraindividual development longitudinally. More specifically, paper 3 uses latent transition analysis (LTA) with two waves of data, whilst paper 4 uses growth mixture modelling (GMM) with three waves of data. In terms of stage of schooling, the fours studies cover primary school (papers 1 and 2), middle school (paper 2), and early secondary school (papers 3 and 4). Finally, our studies show how these approaches can be adopted to explore motivational and emotional development from a variety of theoretical perspectives, including control-value theory (Pekrun, 2006; paper 1), the dual-factor model of mental health (Suldo et al, 2014; paper 2), situated expectancy value theory (Eccles & Wigfield, 2020; study 3) and dimensional comparison theory (Möller & Marsh, 2013; papers 3 and 4).
Alongside all using person-centred approaches, the four studies also include similar covariates and/or outcomes. For example, two studies explore how gender relates to the likelihood of belonging to a particular profile (papers 1 and 3), whilst one study examines how gender is related to developmental change (paper 4). In terms of outcomes, papers 1, 2 and 3 all explore how class or profile membership relates to mathematics competence. Papers 2 and 3 additionally explore how class or profile membership relates to outcomes in language (e.g. German) as well. Thus, the combination of different approaches but similar research questions allow for a more nuanced understanding of the factors that might influence the development of multiple motivational beliefs simultaneously, or how different combinations of motivational beliefs might relate to outcomes, thereby increasing our theoretical understanding of motivational development. Together, the four studies demonstrate the theoretical and methodological added value of adopting person-centred approaches to explore both intraindividual and interindividual differences in motivational beliefs.
Presentations of the Symposium
Profiles of Control, Value and Achievement Emotions in Primary School Mathematics Lessons
Stephanie Lichtenfeld1, Wendy Symes2, David Putwain3
1University of Hamburg, 2University of Potsdam, 3Liverpool John Moores University
Theoretical Background
According to control-value theory (CVT, Pekrun, 2006), student emotions about their schoolwork (‘achievement emotions’) arise because of simultaneous and mutually reinforcing control and value appraisals. Although the theory proposes that distinct achievement emotions are related to certain combinations of control and value, it is less clear how these combinations of motivational beliefs relate to the experience of multiple domain-specific emotions. Furthermore, pleasant emotions (e.g., enjoyment) are proposed to positively, and unpleasant emotions (e.g., boredom) negatively, influence achievement via their impact on motivation and learning strategies. Yet it remains unclear how the experience of multiple emotions relates to competencies in a relevant domain. This is because previous research has typically adopted a variable-centred approach that examines average relations between control and value appraisals, discrete emotions, and achievement. Recent research in the field however suggests students may be heterogenous in their profiles of achievement emotions (e.g. Jarell et al., 2016; Tze et al., 2022), or achievement emotions and control and value appraisals (Parker et al., 2021), and that these profiles may relate differentially to achievement (Parker et al., 2021). The present study adopted a person-centred approach to build on this research and to answer the following two research questions:
Research Questions
1. Are heterogenous profiles of control and value beliefs and achievement emotions found in a large sample of primary school students?
2. Do these profiles relate differently to mathematics test scores?
Method
Our sample comprised 883 primary school students (50% girls, mean age = 9.34 years, SD = .48) from 23 primary schools in England. Students sat a mathematics test in January (T1) and June (T3) of the school year, and provided self-reported control and value beliefs, and emotions (T2), approximately one week before T3 data was collected. Control was measured using 4 items from the Academic Self-Description Questionnaire II (Marsh, 1990). Subjective task value was measured using 12 items each from the Michigan Study of Adolescent Life Transitions scales (Eccles et al., 2005). Enjoyment, anxiety and boredom were measured using 4 items each from the Achievement Emotions Questionnaire – Mathematics Elementary Version (Lichtenfeld et al., 2012). Participants responded to all items on a 5-point Likert scale from 1 = strongly disagree to 5= strongly agree. Mathematics tests were worth 20 marks each. We conducted a latent profile analysis in Mplus v. 7 to examine the presence of heterogenous profiles of control and value beliefs and emotions in our sample.
Results
The 3-profile model was selected as the optimal model based on statistical fit indices, theoretical considerations and class size. Profile one (the ‘high enjoyment’ profile) comprised 52% of the sample and was characterised by high intrinsic value and enjoyment, and low boredom. Profile two (the ‘high boredom’ profile) comprised 12% of the sample and was characterised by low intrinsic value and enjoyment, and high boredom. Profile three (the ‘moderate’ profile) comprised 36% of the sample and was characterised by moderate levels of control, value, enjoyment, anxiety and boredom. We explored profile differences in mathematics test scores at T3, controlling for gender and mathematics test scores at T1. A Wald test indicated a main effect of profile on mathematics test scores, 𝜒2(2) = 25.596, p = <.001. Pairwise difference tests revealed that students in the ‘moderate’ profile had significantly lower mathematics test scores than students in the ‘high enjoyment’ profile (z = -4.344, p <.001). Higher levels of anxiety seemed to compensate for high boredom and low intrinsic value in the ‘high boredom’ group. Studying the impact of emotions and motivational beliefs in combination may be a fruitful area of future research.
Profiles of Student Emotion as Predictive of Academic Achievement
Nathaniel von der Embse, Caroline Mierzwa
University of South Florida
Theoretical Background
Nearly 20% of school-aged children exhibit signs of significant mental health needs with this concern rising over the last several years since the pandemic (De France et al., 2022; Merikangas et al., 2016); however, few of these students are identified and receive timely intervention services. Mental health challenges that are left unaddressed can lead to long-term, negative outcomes for youth, including social problems (e.g., poor relations with peers; Hall‐Lande et al., 2007), increased behavioral concerns (e.g., aggressive behaviours; Taras et al., 2003), as well as substance abuse and school dropout (Perfect et al., 2016). Schools have become an ideal setting for both identifying and intervening for students with mental health concerns (Nemeroff et al., 2008). Screening tools offer benefits over more reactive methods such as teacher referrals, as they are proactive (i.e., regularly scheduled), predictive of salient long-term outcomes (e.g., academic performance), and encompassing complete mental health (i.e., assets and needs). The Dual-Factor Model of Mental Health (Suldo et al., 2014) posits that complete mental health consists of psychological strengths and the absence of psychological problems. As such, (screening) assessment tools should identify both strengths and weaknesses to accurately reflect which students need support.
Research on these screeners has yielded promising psychometric evidence (Kilgus et al., 2016; von der Embse et al., 2017). However, schools are often forced to prioritize limited resources when identifying student complete mental health with screening tools. Research is needed to identify the grouping of students in different types (emotional, social) or intensities of need (low, medium, high) to allow for triaging of intervention supports. Evidence that these groupings are linked to socially valid outcomes such as academic performance, would be important to support these critical decisions.
Research Questions
The purpose of the present investigation is to (1) use latent class analysis (LCA) to identify classes of student emotional behavior and (2) if these classes are linked to student math and reading performance. Identified classes of need could lead to better utilization of limited school resources by aligning intervention to comorbid risk domains (e.g., academic and social risk; Iaccarino et al., 2018).
Methods
The current study attempts to resolve these gaps in the literature regarding student self‐report scales with a large sample of elementary and middle school‐aged students in the USA. A de-identified dataset was used and included students' responses on the Social, Academic, and Emotional Behavior Risk Screener (SAEBRS), which is a screening tool based on a dual-factor model of mental health (Kilgus & von der Embse, 2014). An LCA was used to identify latent risk classes of children with similar patterns of behaviour risk and to explore their impact on academic outcomes.
Results
Three distinct classes emerged as the best-fitting model for the SAEBRS scores, including Class 1 with low risk (46.8%), Class 2 with low-moderate risk (33.7%), and Class 3 with high risk (20.8%). Differences indicated significant variations in reading and math outcomes between the classes, with effect sizes ranging from small (-.24) to large (.81). Class 1 indicated higher reading and math scores than Class 2 and Class 3. Class 2 had slightly better reading scores than Class 3 but lower math scores. Overall, there is a statistically significant difference between math and reading outcomes and student emotions between the classes (p=0.000). These findings indicate that different combinations of psychological strengths and problems may relate to academic outcomes in different ways. Screening for students using comprehensive mental health screening tools may therefore be an effective option for offering targeted support.
Stability and change in profiles of mathematics and German self-concept and intrinsic value: Relations to perceived teaching behaviours and competence.
Wendy Symes, Rebecca Lazarides
University of Potsdam
Theoretical background:
Variable-centred studies have shown that students’ motivational beliefs, more specifically their expectancies of success and subjective task values, decline with age (e.g. Jacobs et al., 2002). Findings from research using person-centred approaches have shown, however, that there may be interindividual differences in time-related motivational change (e.g. Archambault et al., 2010). Situated expectancy-value theory (SEVT; Eccles & Wigfield, 2020) proposes that interindividual differences in level and change of motivational beliefs result from socialisation processes. These processes include the beliefs and behaviours of socialisers, such as teachers, which may be influenced by salient student characteristics, such as gender. Furthermore, whilst some students may have high or low levels of motivation across academic domains (Gaspard et al., 2019) others may have higher or lower motivation in some domains than others. According to dimensional comparison theory (DCT; Möller & Marsh, 2013), these domain-related differences arise due to the experiences in one domain influencing the development of expectancies and values in a contrasting domain. Although perceptions of teaching behaviour in one domain have been shown to be influenced by achievement in a contrasting domain (Arens & Möller, 2016), it is less clear how such comparisons relate to the interindividually different development of expectancies and values.
Research Questions
Accordingly, we used latent transition analysis to answer the following research questions:
1. Which profiles of mathematics and German motivation do students belong to in grades 5 and 6, and how stable is profile membership?
2. Are there grade 6 profile differences in mathematics and German competence, even when controlling for prior levels of competence?
3. How does gender and perceived teaching behaviour relate to grade 5 profile membership and transitions from grade 5 to grade 6.
Methods
Our sample comprised 721 secondary school students from Germany (54.6% female) who participated in the longitudinal study ‘Educational process, competence development and selection decisions in preschool and school age’ (BiKS 8-14, Artelt, Blossfeld, Faust, Roßbach & Weinert, 2013). Our study focused on data collected in wave 4 (grade 5) and wave 5 (grade 6). Mathematics and German self-concept and intrinsic value were measured using three items each from adapted versions of established scales (Baumert et al., 1997; Marsh, 1992). Students responded to all items on a 5-point Likert scale. Perceived teaching behaviour was measured using an established scale (Rakoczy, Buff, & Lipowsky, 2005). Students responded to all items on a 4-point Likert scale. German competence was assessed with 43 itemsin grade 5 and 31 items in grade 6. Mathematics competence was assessed using 44 items in grade 5 and 40 items in grade 6. We used latent transition analysis and the three-step approach to explore outcomes and covariates of profile membership.
Results
We selected the 5-class model as the best fitting model at both waves. In grade 5, girls were significantly more likely than boys to belong to a Low Motivation or High German profile than a Low Value profile. Perceived teaching behaviours related to the likelihood of profile membership in grade 5, but not transitions in grade 6. Students in the High German profile in grade 6 scored significantly lower in mathematics than German than students in the High Mathematics or Low Value profile in grade 6, even after controlling for prior competence. Students in the High Mathematics profile in grade 6 scored significantly higher in mathematics than German than students in the High Motivation profile. The findings indicate that perceived teaching behaviour in one domain is related to motivation and outcomes in the same and a contrasting domain, and that these motivational beliefs are relatively stable over time.
Growth Trajectories of Self-Concept and Interest in Mathematics and Language – Individual Differences and Cross-Domain Relations
Anna Widlund1, Markku Niemivirta2, Heta Tuominen2, Johan Korhonen1
1Åbo Akademi University, 2University of Eastern Finland
Theoretical Background
During secondary education, adolescents’ motivation seems to decline on average (Gaspard et al., 2020). Although this may reflect a mismatch between students’ needs and the secondary schools’ resources, general declines may also stem from increasing intraindividual differentiation in competence- and value-beliefs across domains (Möller et al., 2020; Wan et al., 2021). As adolescents become increasingly aware of their abilities and their interests become more specialized, they likely maintain high self-concept and interest in only a few domains, while disengaging from others. Yet, inter- and intraindividual differences in students' co-developmental processes of self-concept and interest across domains have rarely been studied. In addition, considering some consistent gender differences in both mathematics and L1, it seems relevant to investigate whether such dimensional comparison processes differ between genders.
Research Questions
This study aimed to examine:
1. what kinds of developmental trajectories of self-concept and interest in mathematics and L1 can be identified among adolescents across lower-secondary education, and
2. whether trajectories and cross-domain relations differ between genders.
Methods
612 Finnish students were followed across Grades 7–9 (13 to 15-year-olds). Participation was voluntary, and informed consent forms were collected from the student’s parents. In Finland, lower secondary education spans from Grade 7 to Grade 9, after which the students can make an important decision about whether they opt for vocational or general upper secondary education. Students completed electronic questionnaires on their math and L1 self-concept and interest (Marsh, 1992) across four time points. The analyses started with confirming longitudinal measurement invariance for each construct. Next, a series of growth models were tested, comparing both linear and quadratic growth curves to confirm the best-fitting growth trajectory for each construct. Based on the models chosen in the first step, growth mixture models were applied to identify distinct motivational trajectories of math and L1 self-concept and interest across Grades 7–9. Lastly, multi-group growth models were used to compare growth trajectories and cross-domain relations between genders.
Results
The results revealed that students’ math and L1 motivation development was rather homogenous across grades 7–9, as a one-profile solution fitted the data best. Although variances were found in the overall levels of self-concept and interest in math and L1 (i.e., intercepts), no significant variances were found in the growth rates (i.e., slopes). The growth trajectories indicated that most students experienced a decline in both their math and L1 motivation, but interest and self-concept in math were levelled out by a positive quadratic trend towards the ninth grade. Nevertheless, these findings suggest that many experience challenges while simultaneously entering both early adolescence and lower secondary education, highlighting the importance of supporting students' motivation and well-being during this critical time period when several co-occurring changes take place and while they are making important decisions about their future. When a gendered multi-group growth model was applied, we found that there was a clear differentiation in girls' motivational beliefs across domains: they showed significantly higher L1 motivation as compared to math, whereas boys had similar levels of motivational beliefs across domains. Girls also engaged in more dimensional comparison processes, as we identified several negative cross-domain relations among girls that were not identified among boys. This, coupled with prevailing gender stereotypes, may be harmful in some cases. It could, for example, unnecessarily hinder some from engaging in and aspiring for math-related educational and career alternatives, despite having high performance in math. These findings should be considered in schools while supporting students’ motivation and goal setting, and, while planning targeted interventions to, for example, support girls' relatively low motivation in math.
|
