Longitudinal associations of student–teacher relationships and behavioural and emotional strengths on academic achievement

Abstract Positive student–teacher relationships are related to students’ academic achievement and behavioural and emotional adjustment. How a student’s behavioural and emotional strengths are associated with these relationships and how the relationships influence students’ academic performance remains unknown. We examined this framework using a cross-lagged panel model with a group of Finnish students and their parents from Grade 5 to Grade 7. The results revealed that the parents rated behavioural and emotional strengths are stable over a 1-year (r = .78) and 2-year (r = .71) period and that students’ perceptions of student–teacher relationships demonstrated greater change over time (r’s = .54, .35). Behavioural and emotional strengths demonstrated a positive relationship with student–teacher relationships as well as academic achievement (β = .39, p < .01). Strengths were also indirectly associated with academic achievement via student–teacher relationships. Study limitations, implications and future research are discussed.


Introduction
The rationale for the behavioural and emotional strength-based assessment perspective can be reasoned from two perspectives: resiliency and positive youth development (PyD) (see, e.g. Epstein, 2004;Hotulainen, Lappalainen, & Sointu, 2014). Resiliency itself is generally defined as a process reducing vulnerability to risk experiences, overcoming adversity and stress and obtaining relatively positive outcomes despite risk experiences, and it has two components: (a) successful adaption to life situations for achieving positive outcomes (i.e. protective factors) and (b) the presence of problematic, challenging or even threatening life situations (i.e. risk factors) (Luthar, Crossman, & Small, 2015;Rutter, 1987Rutter, , 2006Rutter, , 2012Werner & Smith, 2001). For example, when students have strong protective factors, they are more likely to put effort into school achievement (Sutherland, Conroy, Abrams, & Vo, 2010). Another approach to explain the importance students' strength identification is the PyD framework, which considers the importance of aligning individual strengths and assets around individuals to enable the positive development, learning and growth of youth (Geldhof, Bowers, & Lerner, 2013;Lerner, Bowers, Geldhof, Gestsdóttir, & DeSouza, 2012).
There is some evidence that students' behavioural and emotional strengths are associated with students' overall social skills and competencies, adjustment and appropriate behaviour in school and academic performance (Harniss & Epstein, 2005;Harniss, Epstein, Ryser, & Pearson, 1999;Trout, Ryan, La Vigne, & Epstein, 2003). in these studies, teachers have rated the emotional and behavioural strengths as well as the skills, behaviour, academic performance and learning of elementary age students, and significant correlations have been found between these ratings; however, the relationship between youth strengths and actual academic achievement has not been investigated. in addition, according to Sutherland et al. (2010), highlighting students' behavioural and emotional strengths increases supportive teacher-student relationships and promotes a positive school environment for them. Furthermore, it has been suggested that the relationship between strengths, specifically prosocial behaviour and academics may be mediated by the quality of teacher-student relationships (Spinard & Eisenberg, 2009). Therefore, it is important to investigate not only the associations between relationships and strengths on academic achievement, but also possible indirect effects as well as and stability over time to obtain more solid evidence of the phenomenon.
in summary, previous research suggests that student-teacher relationships are one of the most important factors in the school achievement and outcomes of students with behaviour and emotional problems (e.g. Baker et al., 2008;Capern & Hammond, 2014;Murray & Murray, 2004). Recently, the value of strength-based assessments has been increasingly recognised in assessing students' socio-emotional behaviour (Kauffman & Landrum, 2013). According to our knowledge, there is no previous research in which student-teacher relationships, behavioural and emotional strengths and academic achievement were examined together. Therefore, the present study aims (a) to investigate the relationships between behavioural and emotional strengths and student-teacher relationships over time and (b) to test the efficacy of these variables to predict academic achievement. We hypothesised that strengths and student-teacher relationships have a bidirectional positive relationship over time (i.e. students with greater strengths will develop better relationships in future grades and students with better relationships will developed greater strengths in future grades) and that both predictors will have a significant positive relationship with academic achievement (see also Figure 1).

Participants
The sample was a part of a larger study of school inclusion in Eastern Finland involving 46 schools during 2010-2012. in this longitudinal study, the participating students were in grades 5-7 (11-14 years old) in the area's basic education schools. in this study, student status such as special educational designation or high-risk status was not separately identified. The assessments were administered to parents when their children were in 5h (N = 295), 6th (N = 363) and 7th (N = 328) grades. Similarly, the student questionnaires were administered to them when they were in 5th (N = 588), 6th (N = 631) and 7th (N = 595) grades. The students reflected the racial/ethnic composition of the region, which was overwhelmingly Finnish and Caucasian in origin, with less than 5% other ethnicities.

Measures
Three measures were used in the study. Student-teacher relationships were measured with the School Well Being Profile Questionnaire (SWBPQ; Konu, Alanen, Lintonen, & Rimpelä, 2002;Konu & Lintonen, 2006), student behavioural and emotional strengths were assessed by the Behavioural and Emotional Rating Scale-2 (BERS-2; Epstein, 2004) and academic achievement was measured by computing student grade point average (GPA).
The SWBPQ is a student-rated measure of overall well-being in four categories: school conditions, social relationships, self-fulfilment and health status (Konu et al., 2002;Konu & Lintonen, 2006). The SWBPQ includes 56 items that ask students to respond on a 3-point Likert-type scale (1 = disagree, 2 = neither agree nor disagree, 3 = agree). Researchers have reported adequate internal consistency and validity with this measure (Konu et al., 2002;Konu & Lintonen, 2006). in the present study, we were interested in assessing student perception of student-teacher relationships rather than broader social relationships. Based on exploratory factor analyses, we extracted a five-item factor that measured student-teacher (ST) relationships. The five items were 'our school's teachers are friendly' , 'the teacher listens to my opinions' , 'the teacher treats us students fairly and equitably' , 'students' opinion are taken into account regarding school matters' and 'it is easy to get along with our teachers' . The ST relationship subscale had adequate Cronbach's alphas across the 3 years (5th grade α = .79; 6th α = .85; 7th α = .85).
Parents completed the Finnish-translated BERS-2 (Epstein, 2004;Sointu, Savolainen, Lambert, Lappalainen, & Epstein, 2014), in which 52 items are rated on a 4-point Likert-type scale (0 = not at all like your child, 1 = not much like your child, 2 = like your child, 3 = very much like your child). The BERS-2 includes five subscales (interpersonal strengths [iS], family involvement [Fi], intrapersonal strengths [iaS], school functioning [SF] and affective strengths [AS]), and these subscales form the Strength index score (strength). iS measures a child's ability to control his or her emotions or behaviours in social situations, Fi measures a child's participation in and relationship with his or her family, iaS measures in a broad sense a child's outlook on his or her competence and accomplishments, SF focuses on the child's competence in school and classroom tasks and AS assesses a child's ability to accept affection from others and express feelings towards others (Epstein, 2004). items in different subscales include statements such as 'Expresses remorse for behaviour that hurts or upsets others' (iS), 'interacts positively with parents' (Fi), 'identifies own feelings' (iaS), 'Completes school tasks on time' (SF) and 'Shows concern for the feelings of others' (AS). Prior research with Finnish samples found the BERS-2 to have adequate internal reliability, construct validity (Lappalainen, Savolainen, Kuorelahti, & Epstein, 2009; and convergent validity (Savolainen, nordness, Sointu, Lappalainen, & Epstein, 2013;Sointu, Savolainen, Lappalainen, & Epstein, 2012). For the present study, only the Strength index (strength) score was used. The reliability for the strength measure was acceptable (5th grade α = .94; 6th α = .95; 7th α = .95). Both the SWBPQ and BERS-2 were administered between April and May each year; thus, the interval between measurement points was approximately 1 year. notes: si = parent rated behavioural and emotional strengths, stR = student-teacher relationship. gPa = grade point average. number represents the grade of students.
Seventh-grade GPA was used as the outcome variable. GPA constituted four different academic school subjects (Finnish language and literature, English, Swedish and Mathematics). These data were obtained from official school records from participating school administration offices at the end of the school year. Students receive grades at the end of the school year for each subject they have studied, and these grades are given by subject-specific teachers (e.g. math or English) usually on the basis of several examinations. The grading of school subject in Finland ranges from 4 (lowest possible score) to 10 (highest possible score), where a score above 5 is considered a passing grade, and the modal grade is typically 8.

Procedures
The data were collected on all students with assent and consent between the years 2010 and 2012 as part of the Finnish national Board of Education funded study Eastern Finland Education Development Project. in the first year, the project coordinator team contacted participating municipalities and asked whether teachers and school administrator were willing to volunteer for the project. Then, parent/ caregiver participation was sought, and parents were asked to discuss the study with their child to secure the child's assent and to return the forms to the researchers. The student data collection was group-administered with paper-pencil questionnaire each year, and students completed the SWBPQ during a 1-week period of time within regular school hours. The parent BERS-2 questionnaire was given to students along with a response letter, return envelope to the researchers and detailed instructions for completing the BERS-2.

Data analysis
SPSS v19 was also used to generate descriptive statistics and reliability estimates for each scale used in this study. Mplus v 7.11 (Muthén & Muthén, 1998 was used to fit a cross-lagged panel model (Campbell & Kenny, 1999;Campbell & Stanley, 1963) to assess the reciprocal relationships between behavioural and emotional strengths and student-teacher relationships over time and the relationship between these two variables and academic achievement. We also tested the indirect effect of strengths on academic achievement through how strengths influence ST relationships. The model was built in four phases. The first phase included autoregressive paths between the three measurement points for the Strength index (strength) and ST relationship scores, and the GPA was regressed on the strength and ST relationship. in addition, the strength and ST relationship variables were allowed to correlate for each measurement point. Second, 1-year cross-lagged paths between the strength and ST relationship were added to the model. Third, non-significant cross-lagged paths and correlations were removed from the final model. Only the parameters from the final model are presented in the results section. Fourth, two indirect effects were specified and tested with bootstrapping method (the default bootstrap of Mplus), with maximum likelihood (ML) estimator and 1000 draws. We had to use to ML for bootstrapping due to the fact that MLR does not allow bootstrapping analysis. The first indirect effect was strength on 5th grade to strength on 6th grade to ST relationship on 7th grade on GPA. The second one was strength on 5th grade to strength on 6th grade to ST relationship on 7th grade on GPA.
The model was estimated using full information ML with robust standard errors (MLR). We used the Mplus default method to account for missing data, which assumes that data are missing at random and includes the missingnesss in the estimation process using a model-based algorithm (Muthén & Muthén, 1998. The goodness-of-fit was assessed using the comparative fit index (CFi; Bentler, 1990), the root mean square of the approximation (RMSEA; Steiger & Lind, 1980) and the standardised root mean square residual (SRMR; Hu & Bentler, 1999). We reported the χ 2 for the model; however, this indicator of fit was not interpreted because, with large sample sizes, the test is overly conservative and may be misleading (Brown, 2006). Models with CFi values above .95 and RMSEA and SRMR values less than .05 are generally considered 'close fitting' (Brown, 2006;Hu & Bentler, 1999). Fully standardised coefficients were used to interpret model parameters for all the direct and indirect effects.

Results
The correlation matrix along with means (M) and standard deviations (SD) is presented in Table 1. As the first-order correlations indicate, the Strength index (strength) score was highly correlated across time, suggesting that strengths are very stable for students at this age level. Similarly, the correlations were moderate to large in magnitude for ST relationship across time, indicating a moderate degree of stability in these relationships from year to year and from teacher to teacher.
The stability of the strength and ST relationship indicates that inter-individual change is relatively limited over time but also that there is greater change in the students' perception of their relationships with their teachers. The level of a child's strength indicates that parents see their children as possessing quite high levels of behavioural and emotional strengths, and the level of ST relationship indicates that student see their relationships with teachers as adequate. in addition, the mean GPA (7.96) indicates a relatively high level of academic achievement and follows the general structure of academic achievement in Finnish schools. The final cross-lagged model (with parameter estimates) is presented graphically in Figure 2. All the model fit indicators suggest that the data fit the model closely ( 2 12 = 33.47, p < .001; CFi = .97; RMSEA = .05, SRMR = .04).

Prediction of academic achievement
GPA in 7th grade was regressed on the strength and ST relationship in 7th grade to assess the ability of strength and ST relationship to predict academic achievement. Both strength (β = .39, p < .01) and ST relationship (β = .12, p < .01) were significantly related to GPA, indicating that students with greater strengths and students with better relationships with teachers achieved higher GPAs in 7th grade. in addition, strengths demonstrated a significant indirect effect with GPA through both mediating effects. The first indirect effect was statistically significant (b = .035, p < .05 [95% Ci = .003, .066]), indicating that higher strengths in 5th grade have a positive influence on GPA via 6th-and 7th-grade ST relationship, because they set a foundation for developing better ST relationship. in addition, the second indirect effect was statistically significant (b = .039, p < .05 [95% Ci = .001, .072]), indicating that higher strengths in 5th grade and 6th grade also have a positive influence on GPA via 7th-grade ST relationship, because they set a foundation for developing better ST relationship in the 6th grade. Overall, the model predicted 18% of the variance of GPA, which can be considered a rather large proportion of variance, especially when considering the use of different informants in measuring different variables (i.e. parents-strength, students-ST relationship, teachers-GPA).

Discussion
The purposes of this study were to investigate the relationships between student-teacher relationships and behavioural and emotional strengths over time and to assess the efficacy of these two variables in predicting academic achievement. With respect to stability, both constructs demonstrated adequate stability over 1-and 2-year time periods. These results were expected, as previous long-term (6-month) test-retest reliability has indicated adequate stability (Epstein, Hertzog, & Reid, 2001;Lambert et al., 2015); however, this study extended the period over a 2-year period. The students' perceptions of student-teacher relationships were stable over time and were congruent with the findings reported by others using Finnish samples (see Kämppi et al., 2012).
The study demonstrates that both behavioural and emotional strength and student-teacher relationship predicted academic achievement. in addition, greater behavioural and emotional strengths are related to the development of student-teacher relationship, which, in turn, lead to greater academic achievement. in other words, strengths have positive effects on academic achievement for two reasons: (1) strengths directly influence achievement and (2) strengths seem to set a foundation upon which student-teacher relationships are built. There are a number of reasons why strengths would be directly related to academic achievement. For example, students with higher interpersonal strengths (i.e. student's ability to control emotions and behaviours in social situations), affective strengths (i.e. student's ability to accept affection from and express affection to others) and school functioning (i.e. completing tasks on time, completing homework, etc.) may be related to self-regulatory behaviour in the classroom and at home (i.e. sit still in class and focus on teacher) (see also, Liew, 2012;Spinard & Eisenberg, 2009). Students who show more intrapersonal strength (i.e. is seen as having higher view of own competencies and accomplishments) may have a higher academic-related self-concept (see, e.g. Lappalainen & Hotulainen, 2012). Lastly, students who have greater strengths related to family involvement may get more school-related support at home and have families that are more engaged with schools, which numerous studies have shown to be related to positive school outcomes (see, e.g. Collins & Laursen, 2004or Collins, Maccoby, Steinberg, Hetherington, & Bornstein, 2000. The results of this study highlight the need for teachers to be aware of how their behaviour influences student outcomes (either positively or negatively) and how positive relationships can function as resiliency factors against school-related difficulties. in fact, Sutherland et al. (2010) argued that using a strength-based approach can improve teacher-student interaction and ameliorate challenging behaviour. These results seem to confirm their position and furthermore suggest that if teachers, parents and other educational stakeholders focus on improving and bolstering student strengths, these improvements may manifest in improved relationships and greater academic achievement. This can be also viewed from the fact that a positive student-teacher relationship is strongly associated with positive school outcomes (e.g. Liew, 2012). What seems to be the case is that home-identified strengths contribute to student-teacher relationships and that actual academic achievement may be particularly important to note for teachers' practical work. For example, a negative student-teacher relationship may have a long-lasting effect on a student via problematic behaviour, low self-esteem and poor academic performance. Therefore, teachers should acknowledge the importance of the student perspective of the relationship and its impact on their work. in addition, it is reasonable to assume that the paradigm shift towards a strength-based perspective in teaching and also with home-school collaboration may improve the quality of teacher work, as strengths have a significant (albeit small) influence on students' academic achievement and how students view their relationships with teachers. We are quite sure that teachers have the ability to identify and highlight the strengths of their students, but what seems to be also very interesting is that teachers may be able influence students' academic achievement and the relationships with students by conveying the importance of parental strength identification to parents and perhaps using multi-informant sources of strength-based assessment for this purpose (see, e.g. Sointu et al., 2014). Therefore, we suggest that educational professionals consider strength identification as part of their teaching work and also as an important part of collaboration between homes and schools. However, these suggestions must be studied more thoroughly before any further discussion can take place.
The importance of viewing behavioural and cognitive processes as complementary instead of separate factors in teaching and learning has garnered attention in the psychoeducational sciences (e.g. Liew, 2012). Educational professionals should consider the whole student in their work, including students' behavioural, emotional and learning skills, competencies, strengths and assets in teaching and learning. The results of the current study seem to give some credence to the complementary whole child perspective. in addition, Hotulainen et al. (2014) have suggested that, particularly for students with behavioural and/or learning challenges, a sole focus on academic learning skills and strengths may be problematic, as these students may not have possibilities to show other aspects of their individual strengths. The results of this study highlight the importance of behavioural and emotional strengths as important factors for student-teacher relationships and student academic achievement.

Limitations and future research
Several limitations of the current study should be noted. First, the phenomena investigated involved a sample of typical 5th grade students over a 3-year period. Student status, such as special education designation, high risk status or clinical diagnosis, was not separately investigated in the study, and this needs to be addressed in future research. Second, even though the sample was sufficient for the analysis, its national representativeness can be questioned, because the sample came from only one region of Finland. However, as the between-school variance of academic outcomes in Finnish schools is the lowest in the world (approximately, 2-5% of total variance) (Hautamäki, Kupiainen, Marjanen, Vainikainen, & Hotulainen, 2013), we can assume that the results of this study are generalisable to the national level. Third, as often is the case with longitudinal studies with several informants, the samples were unequal in size, leading to issues with missing data; however, we addressed these issues using full information ML as the estimation process for the model.
Additionally, as with all measures, further research studies need to be conducted on the psychometric properties of the BERS and SWBPQ measures and how these instruments predict other school-related behaviours. For example, the stability of emotional and behavioural strengths needs to be investigated with younger school samples, particularly kindergarten and first-grade samples, to determine when these strengths emerge and how stable they are in the early grades. More specifically, researchers should investigate intra-individual change over the entire range of school grades with a particular focus on the early grades. Researchers might consider using accelerated growth modelling techniques to assess change over long periods of time (e.g. 1st grade through 12th grade). in addition, the current study needs to be replicated with other informants. For example, in assessing student-teacher relationships, it would be important to collect data from the teacher perspective; and in measuring student strengths, it would be important to collect ratings from teachers and students. Future research might include latent variables instead of manifest variables, where each latent variable is measured by indicators from multiple informants. Finally, this study should also be expanded to investigate other academically related measures, including standardised tests, special education status and attendance.

Funding
This work was supported by Finnish national Board of Education.