Faculty of Social and Behavioral Sciences
Graduate School of Child Development and Education
Running head: TEACHER-STUDENT RELATIONSHIP TRAJECTORIES
The importance of teacher-student relationships: Developmental
trajectories from kindergarten to sixth grade predicting motivation
and academic adjustment
Research Master Child Development and Education Research Master Thesis
Rianne Bosman (10002035)
Supervisors: dr. D. L. Roorda and dr. H. M. Y. Koomen Reviewers: dr. E. H. de Bree and prof. dr. T. T. D. Peetsma June 22nd, 2016
Abstract
In the present study, trajectories of teacher-student relationships were investigated from kindergarten to sixth grade. Furthermore, both predictors and outcomes of teacher-student relationship trajectories were examined. The sample consisted of 1300 Dutch students.
Teachers reported about their relationship with individual students in kindergarten, grade 3, and grade 6 (i.e., closeness, conflict, and dependency), and they provided information about
externalizing problem behavior in kindergarten. Students reported about motivation and
performed tests of verbal ability in kindergarten, and they did tests of reading and math in sixth grade. Latent class growth analyses demonstrated three trajectories for closeness: high and stable (largest), very high and decreasing, and moderate and increasing. For conflict, also three trajectories were identified: low and stable (largest), low and increasing, and high and
decreasing. Two trajectories were found for dependency: low and decreasing (largest) and low and increasing. In addition, boys, students with lower verbal abilities, and students with more externalizing problem behavior had higher chances of unfavourable relationship trajectories. Furthermore, very high and decreasing levels of closeness were associated with higher self-efficacy and reading comprehension. Low and increasing levels of conflict were related to lower task motivation and reading comprehension. Most important, low and increasing levels of dependency predicted lower self-efficacy, task motivation, reading achievement, and math achievement. The findings imply that low teacher-student closeness in kindergarten and longer exposure to negative teacher-student relationships during elementary school increase the probability of lower motivation and academic adjustment in sixth grade.
Keywords: Teacher-student relationships, academic adjustment, motivation, externalizing
Preface
Hereby I will clarify this thesis process and in particular my own role and the role of the co-authors. First, I started reading about teacher-student relationships and the possibilities of expanding previous research that could make valuable contributions to the current research field. Because the data of the COOL-cohort study were already collected (funded by the Netherlands Organization for Scientific Research (NWO-PROO)), I began investigating which particular data I could use. With their expertise in this research area, dr. Helma Koomen and dr. Debora Roorda helped me with constructing the research questions. With the support of dr. Ineke van der Veen of the Kohnstamm Intstitute (University of Amsterdam), questions
regarding the dataset and data collection were clarified. She also participated in the brainstorm sessions to finalize the appropriate design of the study. In order to finish my introduction and method section, I read multiple chapters and articles regarding growth mixture modeling and latent class growth analysis, and familiarized myself with the computer program Mplus. I performed all of the analyses, and after careful checks of my supervisors and dr. Marjolein Zee, we reflected about the interpretations of the results. Finally, I want to address that I wrote this thesis myself, but I could not have done this without the extensive feedback and supervision of dr. Debora Roorda and dr. Helma Koomen. They provided me with their thoughts regarding academic writing in general, and shared their exceptional knowledge about teacher-student relationships, motivation, and academic adjustment. Therefore, I would like to thank them for their support and guidance during this process.
The importance of teacher-student relationships: Developmental trajectories from kindergarten to sixth grade predicting motivation and academic adjustment
Research has shown that affective teacher-student relationships, characterized by high levels of closeness and low levels of conflict, contribute to students’ social-emotional,
behavioral, and academic adjustment (Hamre & Pianta, 2001; Ladd, Birch, & Buhs, 1999; Roorda, Koomen, Spilt, & Oort, 2011). Although most studies used cross-sectional designs (e.g., Baker, 2006; Baker, Grant, & Morlock, 2008), there are also longitudinal studies that provide insight in how teacher-student relationships influence students’ outcomes over time. For example, Hamre and Pianta (2001) found that early teacher-student relationships predicted academic and behavioral outcomes until upper elementary school for girls and even until middle school for boys. These results are supported by other longitudinal studies investigating the influence of teacher-student relationships on students’ academic, motivational, and
behavioral outcomes measured both within (Ladd et al., 1999; Pianta & Nimetz, 1991) and across school years (Hughes, Luo, Kwok, & Loyd, 2008; Jerome, Hamre, & Pianta, 2009).
However, longitudinal studies investigating the effects of teacher-student relationships on students’ outcomes usually do not provide insight in how relationships with different teachers change over time. These studies mainly examined how relationship quality at one moment influenced children’s adjustment at another moment. Still, attention for changes or stability in relationship quality over time seems to be warranted, because continued exposure to interpersonal adversity appears to be an important predictor of children’s maladjustment (Ladd & Burgess, 2001; Ladd, Herald-Brown, & Reiser, 2008), indicating that cumulative
experiences may have a longer lasting impact on children’s development in comparison to relationship quality measured on a specific moment in time.
These increased theoretical interests in the development of teacher-student relationship over time has inspired new research (e.g., Spilt et al., 2012). However, there is still relatively
little knowledge about the change or continuity in teacher-child relationships for individual children over the entire elementary school period. Therefore, trajectories of teacher-student relationships from kindergarten to sixth grade were examined in the present study.
Furthermore, as previous studies have found that relationship quality depends on certain child characteristics (McGrath & Van Bergen, 2015; Nurmi, 2012), it was also investigated whether child and family characteristics (e.g., gender, ethnicity, SES, externalizing behavior, and verbal ability) contributed to the probability of having specific trajectories of relationships with teachers. Finally, only a few studies investigated the influence of teacher-student relationships trajectories on children’s outcomes (O’Connor, Collins, & Supplee, 2012; O’Connor, Dearing, & Collins, 2011; O’Connor & McCartney, 2007; Spilt et al., 2012). Therefore we also
examined whether teacher-student relationship trajectories affected students’ motivation and academic adjustment in sixth grade.
Theoretical framework
One of the leading theories in teacher-student relationship research has been the extended attachment theory. According to attachment theory, poor relationships with teachers evoke feelings of insecurity and distress, which limit children in their academic and social growth (Pianta, 1999). On the contrary, sensitive teachers serve as a secure base, which makes children more inclined to explore the school environment and become motivated for school work (Birch & Ladd, 1997; Pianta, 1999). Studies based on the extended attachment theory often examine teacher-student relationship quality in terms of closeness, conflict, and dependency (Pianta, 2001). Closeness can be defined as the degree of warmth and open communication in the relationship, which helps to facilitate children’s learning and school performance. Conflict refers to dysfunctional interactions and negativity, which is hypothesized to be related to poor motivation and impaired academic achievement. Dependency can be defined as the degree of possessive child behaviors that indicate an overreliance on the teacher
as a source of support. High levels of dependency characterize the failure to use the teacher as a secure base (Verschueren & Koomen, 2012).
With regard to changes and stability in teacher-student relationship quality over time, theories of psychological risks and stress state that probabilities of maladjustment increase when children are for a longer time exposed to interpersonal stressors (Ladd et al., 2008). According to these theories, this continued exposure to interpersonal stressors may result in poor students’ outcomes (Ladd & Burgess, 2001). Exposure to, and especially, sustained exposure to interpersonal constraints is hypothesized to discourage productive forms of classroom engagement and to reduce children’s motivation for learning activities (Buhs & Ladd, 2001; Ladd et al., 2008). The longer children are exposed, the greater the anticipated influence on children’s maladjustment (Ladd, 2003). Thus, in contrast to experiences of temporary difficulties on a certain moment, cumulative experiences of interpersonal stressors may have a long lasting impact on children’s development. Based on this theory, it was expected that children with long lasting problematic relationships with their teachers have higher risks of maladjustment. Relatedly, it is possible that continuous support from teachers across school years could be more beneficial for students than support in one particular school year, as was suggested by Spilt et al. (2012).
Stability and change in teacher-student relationships trajectories
Some previous studies investigated changes and stability in teacher-student
relationships over time. Jerome, Hamre, and Pianta (2009) found that the degree of closeness in the teacher-student relationship decreased from kindergarten to sixth grade and conflict
increased over time (Jerome et al., 2009). Similarly, Pianta and Stuhlman (2004) found a trend of teachers reporting decreasing levels of closeness. In contrast, they reported decreases in conflict over time, which is exactly the opposite of what Jerome et al. (2009) found. These contradicting results with regard to conflict in teacher-student relationships may suggest that
different children experience different relationship trajectories with their teachers. For example, some children may experience increasing levels of conflict in their relationships with teachers, whereas others may experience decreasing or stable levels of conflict. Therefore, the present study focused on the identification of different trajectories in teacher-student relationships.
To our knowledge, only four studies investigated different trajectories of student relationships over time. First, O’Connor and McCartney (2007) examined teacher-student relationships trajectories from preschool to third grade in a demographically diverse sample. They examined teacher-student relationship as one construct, without taking into account positive or negative aspects of the relationship. Three different trajectories of teacher-student relationship quality were identified. More than half of the included sample had high and inclining levels of teacher-student relationship quality, whereas a quarter of the sample
experienced moderate and stable levels. The smallest group consisted of students with low and decreasing levels of relationship quality. Second, in a follow-up study O’Connor et al. (2011) found four different trajectories of teacher-student relationship quality from first to fifth grade. In contrast to their previous study, the largest group consisted of teacher-student relationships characterized by a high and stable quality. A quarter of the sample had a high and decreasing level of teacher-student relationship quality, and the smallest group consisted of students with a moderate and decreasing trajectory. Finally, there was a small group which had a moderate and increasing teacher-student relationship quality over time, whereas this was the largest group in the study of O’Connor and McCartney (2007).
A third follow-up study examined closeness and conflict as separate constructs (O’Connor, Collins, and Supplee, 2012). Previous research showed the relevance of making this distinction, as conflict appeared to have a stronger influence on primary school children’s academic adjustment than closeness (Roorda et al., 2011). O’Connor et al. (2012) found four different trajectories of closeness from pre-kindergarten to fifth grade. More than half of the
children had high and stable trajectories of closeness, one group had a moderate and increasing level, one group had a high and decreasing trajectory and the smallest group showed a low and stable trajectory. For conflict, six different trajectories were found, with the largest group of children having low and relatively stable levels of conflict over time in relationships with teachers. Approximately 10% of the children showed a moderate and decreasing trajectory. Furthermore, two small groups of children showed a low or moderate increase of conflict. In addition, two small groups had either a low or moderate level in pre-kindergarten, which increased over time, with peaks of conflict around third grade (O’Connor et al., 2012).
Spilt et al. (2012) also made the distinction between closeness and conflict in teacher-student relationships in their sample of children with below-average literacy skills. The authors found two distinct trajectories of teacher-student closeness from first grade to fifth grade. Approximately 85% of the children had a high and decreasing trajectory, and a smaller group consisted of children with moderate and increasing levels of closeness (Spilt et al., 2012). No high and stable trajectories of closeness were found, whereas this was by far the largest group in the study of O’Connor et al. (2012). Additionally, four different trajectories were
distinguished for conflict. Similar to O’Connor et al. (2012), half of the sample had a low and stable trajectory, one small group showed low and increasing levels of conflict and one small group had a high and declining trajectory. In contrast to findings of O’Connor et al. (2012), the smallest group consisted of children with high and stable levels of conflict (Spilt et al., 2012). Additional groups of children with peaks of conflict around third grade were not found in the study of Spilt et al. (2012), again in contrast other research (O’Connor et al., 2012).
To conclude, studies investigating relationship trajectories over time are scarce and include several inconsistencies. In the present study, trajectories of teacher-student
relationships from kindergarten to sixth grade were examined. In the Netherlands, kindergarten is included in primary school and lasts two years. Children leave primary school when they
finish grade six. Therefore, this study covered almost the entire period that children are in primary school. We also made the distinction between closeness and conflict (O’Connor et al., 2012, Spilt et al., 2012), and additionally we examined differences in growth trajectories of teacher-student dependency.
Predictors of teacher-student relationship trajectories over time
According to the developmental systems theory (Pianta, Hamre, & Stuhlman, 2003), the development of teacher-student relationships will be influenced by characteristics of the
relationship partners, such as children’s gender, ethnicity, SES, externalizing behavior, and verbal ability. Previous research provided ample evidence that children’s demographic characteristics and problem behaviors were indeed associated with the three dimensions of teacher-student relationship quality (McGrath & Van Bergen, 2015; Nurmi, 2012).
With respect to gender, teachers tend to rate their relationships with boys as higher in conflict and lower in closeness than their relationships with girls (Birch & Ladd, 1997; Ewing & Taylor, 2009; Saft & Pianta, 2001). When assessing relationship quality over time, Jerome et al. (2009) reported that boys experienced greater year-to-year stability for conflict in
comparison with girls, and they found a stronger decrease in teacher-reported closeness for boys during elementary school (Jerome et al., 2009). With regard to growth trajectories of teacher-student relationships, Spilt et al. (2012) found highly similar trajectories of conflict and closeness for boys and girls. However, there was no group of girls with high and stable levels of conflict, whereas about 5% of the boys could be identified as such. Furthermore, slightly more boys had high and declining levels of closeness than girls.
With regard to ethnicity, research in the United States has shown that teachers reported less positive relationships with African American students in comparison with Hispanic and White students (Hughes & Kwok, 2007; Murray & Murray, 2004; Saft & Pianta, 2001), and with students from a lower socio-economic background (Ladd et al., 1999; Rudasill, Reio,
Stipanovic, & Taylor, 2010; Wyrick & Rudasill, 2009). Jerome et al. (2009) found that ethnic minority children experienced more conflict in the relationship with their teachers over time than ethnic majority children. They did not find differences of SES in teacher-student
relationship quality over time. With respect to differences in relationship trajectories, Spilt et al. (2012) found that ethnic minority students had more often low and increasing, and high and decreasing levels of conflict. Furthermore, ethnic minority students experienced more often high and declining levels of closeness. In addition, they examined whether children from families with low SES were over- or underrepresented in certain teacher-student relationships trajectories. They found that boys from lower SES families experienced more often low and increasing levels of closeness (Spilt et al., 2012).
Externalizing behaviors in kindergarten also appeared to influence teacher-student relationships across elementary school. Externalizing behaviors are characterized by
undercontrolled and outwardly directed behavior that is disruptive for others, like hyperactivity, impulsivity, and aggression (Merrell, 1999). In studies investigating relationship quality over time, higher levels of externalizing behavior predicted more teacher-student conflict across school years (Henricsson & Rydell, 2004; Jerome et al., 2009). With respect to growth
trajectories, children with externalizing behavior were overrepresented in all other trajectories compared to the low and stable group (O’Connor et al., 2012). However, Spilt et al. (2012) found that children who experienced high and stable trajectories of conflict did not differ in externalizing behavior from children experiencing high and declining trajectories of conflict. With regard to teacher-student dependency, higher degrees of externalizing problem behavior predicted more dependency in teacher-student relationships over time (Birch & Ladd, 1997; Henricsson & Rydell, 2004). There seems to be less strong evidence for an association between externalizing behavior and teacher-student closeness across school years (Henricsson & Rydell, 2004; Jerome et al., 2009), although significant associations were found in cross-sectional
studies (e.g., Buyse et al., 2009; Thijs, Westhof, & Koomen, 2012). With respect to studies investigating growth trajectories, O’Connor et al. (2012) did not find significant differences between trajectories of closeness with regard to externalizing behavior, whereas Spilt et al. (2012) found that externalizing problem behavior was more prevalent in children who had high and declining closeness.
More recently, it has been argued that children with better language skills develop more close relationships with their teachers (Justice, Cottone, Mashburn, & Rimm-Kaufman, 2008; Moritz Rudasill, Rimm-Kaufman, Justice, & Pence, 2006), because they are more able to engage in elaborate conversations (Spilt, Koomen, & Harrison, 2015) and have less difficulties in understanding others than children with poor language abilities (Menting, Van Lier, & Koot, 2011). Incidental studies have found evidence that teachers developed more close relationships with children with better language abilities (Justice et al., 2008; Spilt et al., 2015). Furthermore, in one study an association between language ability and teacher-student conflict was found (Moritz Rudasill, 2006). However, others did not find evidence for a link between language abilities and the development of good teacher-student relationship quality (Howes et al., 2008; Pianta & Stuhlman, 2004). As far as we know, no prior research has focused on the effect of language ability on growth trajectories over time.
In the present study we examined whether child characteristics and behaviors predicted different teacher-student relationship trajectories over time. No research was conducted about the effect of child characteristics on trajectories of teacher-student dependency, and thus this study was the first to take this into account.
Effects of teacher-student relationships on motivation and academic adjustment As mentioned before, teacher-student relationships appeared to be associated with students’ learning outcomes, that is, school engagement and academic adjustment (see Roorda et al., 2011 for a meta-analysis). Empirical evidence has also been found for associations
between relationship quality and students’ outcomes over time. For example, researchers have found that negativity in the teacher-student relationship in kindergarten and first grade, that is, high levels of conflict and overdependency, were related to lower engagement and academic adjustment up to third grade (Buyse et al., 2009; Hughes et al., 2008) or even to eighth grade (Hamre & Pianta, 2001).
However, as far as we know, only two studies have investigated whether different growth trajectories of teacher-student relationships contributed to academic achievement. O’Connor and McCartney (2007) found that low and declining relationship trajectories
predicted low achievement in third grade, whereas no differences existed between the effects of moderate and stable or high and inclining trajectories on achievement. Spilt et al. (2012)
showed that high and stable levels of conflict as well as low and increasing conflict were most strongly associated with underachievement, whereas both high and declining and low and stable trajectories were not predictive of underachievement. Furthermore, they found that low and increasing closeness for boys was related to underachievement in fifth grade, whereas no effect of the different trajectories on underachievement was found for girls. To our knowledge, no previous research examined the effects of relationship trajectories on engagement or
motivation. Therefore, in the present study, effects of relationship trajectories on task
motivation (i.e., the positively valued experiences that individuals derive directly from a task (Thomas & Velthouse, 1990)) and self-efficacy (i.e., the feeling of competency; engaging in a task to get a sense of efficacy (Spinath & Spinath, 2005)) were also examined. In addition to students’ motivation, this study elaborated on previous studies that investigated the influence of different teacher-student relationship trajectories on academic achievement.
Present study
The present study examined growth trajectories of teacher-student relationships in conflict, closeness, and dependency from kindergarten to sixth grade as predictors of
task-motivation, self-efficacy, and academic adjustment. More specifically, it was investigated (1) whether different trajectories of teacher-student conflict, closeness, and dependency could be identified from kindergarten to sixth grade, (2) if these trajectories could be predicted by specific child characteristics (e.g., gender, ethnicity, SES, externalizing behavior, and verbal ability), and (3) how these specific trajectories of teacher-student relationships were related to motivation and academic adjustment in sixth grade.
Based on previous research on trajectories of teacher-student relationships, several hypotheses were formulated. First, the largest group of students was expected to have relatively high and stable or high and declining levels of closeness. In addition, trajectories of low and increasing or low and stable trajectories of closeness were anticipated. Second, it was expected that the largest group consisted of students with low and stable levels of conflict over time, and smaller groups had either increasing, declining, or stable trajectories. No prior research was carried out about trajectories of dependency in teacher-student relationships, and therefore we did not formulate specific hypotheses.
With regard to predictors of teacher-student relationship trajectories over time, it was expected that boys, ethnic minority students, students from low SES families, and students with higher levels of externalizing behavior would be overrepresented in groups that showed more unfavourable trajectories (Spilt et al., 2012), such as low and decreasing levels of closeness and high and stable or increasing levels of conflict over time. Although no prior research has
focused on the predictive effect of verbal ability on relationship trajectories over time, it was also expected that students with less verbal ability were more inclined to have less favourable relationship trajectories (Spilt et al., 2015).
Finally, we expected that unfavourable relationship trajectories, such as increasing or high and stable levels of conflict and dependency, were predictive of lower adjustment in sixth grade (O’Connor & McCartney, 2007; Spilt et al., 2012). Based on Spilt et al. (2012), low and
increasing trajectories of closeness were probably related to lower adjustment scores (Spilt et al., 2012). To our knowledge, no prior studies have investigated the influence of teacher-student relationship trajectories on teacher-students’ motivational beliefs, and therefore no specific hypotheses were formulated.
Method Participants
The current study was conducted using data from three waves of the Dutch COOL-cohort study (e.g., Hornstra, Van der Veen, Peetsma, & Volman, 2013; Zee, Koomen, & Van der Veen, 2013). The COOL-cohort study started in the academic year 2007-2008 when children were in the second year of kindergarten. Follow-up measurements took place when children were in third grade (2010-2011) and sixth grade (2013-2014). The final sample consisted of a total of 1300 children and their teachers from 109 schools in the Netherlands. During the first COOL-measurement, 1300 children participated (49.3% girls), who were on average 5.6 years (SD = 0.43). In the second wave, 1248 children participated (49.1% girls) and they were on average 9.6 years (SD = 0.43). In the third wave, 1097 children participated (50.1% girls), with a mean age of 11.6 years (SD = 0.43). Information about family
composition was available for 97.6% of the students. The majority of children were raised by both of their parents (92.7%), other children were raised only by their mother (4.3%) or their father (0.4%). Furthermore, most children in the sample had fathers (88.4%) and mothers (87.8%) born in the Netherlands. Other common birth countries of parents were Morocco (fathers = 2.7%; mothers = 2.6%) and Turkey (fathers = 1.2%; mothers = 1.7%). Half of the mothers (52.2%) had completed senior secondary vocational education, and approximately a quarter of children’s mothers (24.2%) had completed higher education. Others completed no more than primary education (3.8%) or secondary prevocational education (16.9%). No specific information was available of the teachers in this sample.
Procedure
All schools from a previous cohort study (PRIMA; Driessen, Van Langen, & Vierke, 2004) were asked to participate in the COOL study. In addition, other Dutch schools were also invited to participate. Data collection in the first wave occurred in three phases. First, schools received an invitation to participate in the COOL-cohort study. When schools agreed to
participate, informed consent was obtained from children’s parents. Second, when both schools and parents agreed to participate, school administrators provided data on students’ background characteristics. Third, research assistants visited schools to obtain information from teachers and students. Teachers reported about their relationship with individual students in
kindergarten, grade 3, and grade 6. Additionally, kindergarten teachers were asked to rate children’s externalizing behavior problems and students performed a test measuring verbal ability. Students themselves answered questionnaires about their motivational beliefs when they were in grade 6. Finally, sixth grade students completed tests of reading and math achievement (Driessen, Mulder, Ledoux, Roeleveld, & Van der Veen, 2007).
At the start, 7389 students were included in the COOL study. However, 82% of the children did not participate in follow-up measurements. Schools no longer agreeing to participate, children repeating grades, and children transferring to other schools were among the most common reasons for dropout. Because too much data needed to be estimated if students who only participated at one moment were included (Dong & Peng, 2013), the children of which data was available for at least 2 waves of data collection were used in the present study. The children that were included did not differ significantly from the original sample with regard to all variables measuring relationship quality, motivation, and academic adjustment at each measurement wave.
Measurements
At each measurement occasion, the Dutch version of the Student-Teacher Relationship Scale (STRS) was used to assess teachers’ perception of their relationship with individual students (Koomen, Verschueren, Van Schooten, Jak, & Pianta, 2012; Koomen, Verschueren, & Pianta, 2007). Teacher-student relationship quality was represented by three factors: Closeness, Conflict, and Dependency. Closeness represents the extent to which teachers experience their relationship with a student to be warm, open, and secure. An example of an item measuring Closeness is: “I share an affectionate and warm relationship with this child”. Conflict refers to relationships between teachers and students that are discorded and coercive. An item is: “This child and I always seem to be struggling with each other”. Finally, the factor Dependency measures the degree to which children show age inappropriate demanding and claiming behavior. An example of an item is: “This child reacts strongly to separation from me”. In the present study, a short form of the STRS was used, which consisted of 5 items per dimension (Zee, Koomen, & Van Der Veen, 2013). Items were selected based on the highest factor loadings found in previous research (Koomen et al., 2012). Items were rated on a 5-point scale ranging from 1 (definitely does not apply) to 5 (definitely applies). The Dutch version of the STRS has demonstrated adequate psychometric properties, such as a satisfactory internal consistency, test-retest reliability, and construct validity from preschool to upper elementary school (Doumen et al., 2009; Koomen et al., 2012).
In a validation study of Koomen et al. (2012), the strongest non-invariance over time was found for the item: “If upset, this child will seek comfort from me”. Therefore, in the present study, this item was deleted. The fit of the adjusted three factor model was adequate, χ2(74) = 709.225, p < .001, RMSEA = 0.081, 90% CI = [.076, .087], CFI = .947, TLI = .935, SRMR = .044. After deletion of this item, the three factor model had the best fitting factor-structure, and the model was measurement invariant over time. Cronbach’s alpha coefficients were .92, .92, .95 for Conflict, .83, .82, .84 for Closeness, and .90, .92, .92 for Dependency at
time 1, 2, and 3, respectively. These values indicate good to excellent internal consistency, and were comparable with coefficients found by Koomen et al. (2012).
Externalizing problem behaviors. Kindergarten teachers reported about children’s externalizing behaviors on four items that were selected from other questionnaires for the COOL-cohort study (Driessen et al., 2007; Jungbluth, Roede, & Roeleveld, 2001): “This student is often rude”, “This student always complies to the rules”, “This student always tries to get what he/she wants”, and “This student never fights”. Items were rated on a scale from 1 (completely untrue) to 5 (completely true). After recoding items 2 and 4, higher scores represented more externalizing behavior problems. Cronbach’s alpha for this scale was acceptable (e.g., .77).
Task motivation. The Task Motivation Scale was used to measure children’s task motivation in sixth grade. This scale consisted of items measuring the focus on mastering tasks and the degree of learning opportunities in school (Seegers, Van Putten, & de Brabander, 2002).This questionnaire consisted of 5 items (e.g., “If I do not understand something at school immediately, I will put in more effort”), which were rated on a 5 point scale ranging from 1 (definitely not true) to 5 (definitely true). Previous research found support for the construct validity of the Task Motivation Scale (Hornstra, Van der Veen, Peetsma, & Volman, 2013). Cronbach’s alpha was acceptable (e.g., .74) in the current study.
Self-efficacy. Children reported about their self-efficacy in sixth grade. Self-efficacy was measured by the Academic Efficacy subscale of the Patterns of Adaptive Learning Survey (PALS; Midgley et al., 2000), which assessed children’s expectancies about their capability to perform academic tasks in the classroom. This questionnaire consisted of 6 items (e.g., “I can also learn difficult things at school”) which were answered on a 5-point scale, ranging from 1 (definitively not true) to 5 (definitely true). The Academic Efficacy subscale displayed
sufficient reliability and construct validity in previous research (Midgley et al., 2000). In this study, Cronbach’s alpha was .79, which was similar to the findings of Midgley et al. (2000).
Math achievement. Students’ math achievement was measured by their performance in nationally normed tests that were developed by the Dutch assessment institute CITO
(Hollenberg & Van Der Lubbe, 2011). Each year, approximately 80% of the Dutch schools use the CITO test (Hollenberg & Van Der Lubbe, 2011). Research showed that several tests
developed by CITO had sufficient construct validity across different test versions (Van Boxtel, Engelen, & de Wijs, 2011). Children performed a math test consisting of 96 multiple-choice questions. Exercises included geometry, multiplication, and addition. Its aim was to screen and determine the current level of mathematics for children in grade 6. The total amount of correct answers provides a general score of math ability of children (Driessen et al., 2007). Raw scores of math achievement were transformed to age-appropriate ability scores. These ability scores were standardized on a continuous scale ranging from 0 to 150. Reliability of the math achievement test was excellent (α = .92; Janssen, Verhelst, Engelen, & Scheltens, 2010).
Reading achievement. The score on reading achievement consisted of the scores of three subtests designed by CITO: reading comprehension, vocabulary, and technical reading (Driessen et al., 2007). The test for reading comprehension measured the proficiency in conceptual reasoning and practical reading ability, and consisted of 55 multiple-choice items. Reliability of the reading comprehension test was adequate (α = .88; Weekers, Groenen, Kleintjes, & Feenstra, 2011). The vocabulary test consisted of 70 items that aimed to measure children’s receptive vocabulary, which refers to the extent to which children comprehend and use language. Although no information about reliability of the vocabulary test in sixth grade is known, researchers found excellent reliability of vocabulary tests for children in third grade (α = .92; Van Berkel et al., 2010). For the tests of reading comprehension and vocabulary, age-appropriate and standardized ability scores were used. Technical reading was assessed by a test
in which children were asked to read aloud as many words as possible in three minutes (CITO, 2009). The extent to which children were able to decode words accurately and quickly was measured. Reliability of the technical reading test was excellent (α = .94; Krom, Jongen, Verhelst, Kamphuis, & Kleintjes, 2010).
Verbal ability. In kindergarten, children completed a test of language ability in which different aspects of language development and emerging literacy were examined. This test was also nationally normed and developed by CITO (Lansink, 2009). Passive vocabulary, critical listening, rhyming, and auditory synthesis were assessed. The total score represented the extent to which children were capable of understanding and using language. In the present study, again standardized ability scores were used. Research showed good reliability for the test of verbal ability (α = .89; Lansink & Hemker, 2010).
Demographics. School administrators provided background information of students and their families, such as gender, ethnicity, and SES. Gender was dummy coded, with 0 representing girls and 1 representing boys. Ethnicity was based on both fathers’ and mothers’ country of origin. Given the small amount of ethnic groups other than Dutch, a dichotomy was made between ethnic majority and ethnic minority students. Students who had at least one parent with a birth country other than the Netherlands were characterized as ethnic minority students, represented by 1. Ethnic majority students were represented by 0. This resulted in 86.4% ethnic majority students and 11.1% ethnic minority students. Maternal education was used as a proxy for socio-economic status (SES), and consisted of four categories: no more than primary education, secondary prevocational education, senior secondary vocational education, and higher education.
Statistical analyses
To answer the first research question, latent class growth analysis (LCGA) was
growth mixture modelling (GMM), and its aim is to classify individuals into specific subgroups or classes based on their responses during multiple measurements over time. Individuals within these subgroups have more similar response patterns than individuals between groups (Jung & Wickrama, 2008). LCGA is thus a person-centered approach rather than a variable-centered approach. Because ordinary GMM did not reach model convergence and within class parameter estimates were generally uninterpretable (Bauer & Curran, 2003), LCGA was used in the present study. Due to the nested structure of the data (i.e., students were nested in schools), the cluster option TYPE=COMPLEX in Mplus was used, with schools being the cluster variables. Decisions about the number of trajectories that should be retained were based on several guidelines. First, BIC was used as an indication of model fit, because simulation studies
suggested that BIC is superior to other IC statistics (Yang, 2006). BIC should be as low as possible. Second, class selection was guided by a high entropy, which should be near 1, and high posterior probabilities of class membership, which should also be near 1 (Jung & Wickrama, 2008). In addition, model parsimony and theoretical interpretability of the found classes were taken into account.
To answer the second research question, child characteristics (i.e., gender, externalizing problem behavior, verbal ability, ethnicity, and SES) were added simultaneously to the best fitting model. Finally, to investigate the third research question, the outcome variables that were measured in grade six were added to the model (i.e., self-efficacy, task motivation, reading achievement, and math achievement). In these final models of Closeness, Conflict, and Dependency, significant predictors were also included as covariates.
After excluding participants with missing data on two measurement occasions, there was still missing data. For Closeness, 10.5% of the data were missing, 10.5% with regard to the model of Conflict, and 10.4% for the model of Dependency. As data were not missing
measurement occasion. With regard to all of the included variables in this study, no differences in means were found between the sample with missing data and the sample with EM
imputation. Therefore missing data was treated using full information maximum likelihood estimation in Mplus.
Results
Table 1 presents the correlations and descriptive statistics for all continuous variables. Cross-year correlations for Conflict were slightly larger than for Closeness and Dependency. Correlations between Dependency and Closeness across school years were mostly not significant, whereas there were significant correlations at almost all measurement occasions between Conflict and Dependency and between Conflict and Closeness. Furthermore, all variables had a relatively normal distribution (skewness < 1.5, kurtosis < 3.1) and no outliers were detected.
Relationship trajectories
Closeness. Although BIC statistics for the three and four class model were highly similar (three-class BIC = 6671, four-class BIC = 6662), the three-class model had a higher entropy (i.e., .87) and higher posterior probabilities (i.e., .91 - .95). The majority of students (n = 947, 73%) showed a high and relatively stable level of Closeness (I = 3.87; S = −0.07, p = .001). Additionally, 175 students (13%) had moderate and increasing trajectories of Closeness (I = 2.91; S = 0.36, p < .001), and 178 students (14%) showed a very high and decreasing trajectory of Closeness (I = 4.79; S = −0.59, p < .001; see Figure 1). The largest difference in mean scores between different trajectories was found in Kindergarten. Mean levels of
trajectories were more similar in Grade 3 and Grade 6 (see Table 2).
Conflict. Although fit of the four-class model was somewhat better than the fit of the three-class model (three-class BIC = 7404, four-class BIC = 6799), the three-class model had higher entropy (i.e., .92), and higher posterior probabilities (i.e., .85 - .89).With model
parsimony and theoretical interpretability taken into account, the three-class model was most preferable. The largest group (n = 1139, 88%) consisted of students with low and relatively stable levels of Conflict (I = 1.55; S = −0.10, p = .003). A small group of students (n = 65, 5%) started with high and decreasing levels of Conflict (I = 3.40, S = −0.79, p < .001). In addition, a group of students (n = 96, 7%) had low and increasing levels of Conflict (I = 2.05, S = 0.65, p < .001; see Figure 2). Mean scores of the found trajectories are presented in Table 2.
Dependency. With regard to teacher-student dependency, a two-class model was most favourable. Although BIC statistics and entropy were approximately equal for either a two-class model (BIC = 8433, entropy = .82) or a three-two-class model (BIC = 8379, entropy = .83), posterior probabilities of the two-class model were much better (.84 - .96). The largest group (n = 1155, 89%) consisted of students with a low and decreasing level of Dependency (I = 2.11; S = −0.23, p < .001). Furthermore, a smaller group of students (n = 145, 11%) had relatively low and increasing levels of Dependency (I = 2.68; S = 0.27, p = .01; see Figure 3). Table 2
presents the means of the two trajectories of Dependency at each wave of measurement. Predictors of class membership
Closeness. With regard to teacher-student Closeness, the high stable trajectory was used as a reference category, because this class consisted of the largest group of students. Significant predictors of class membership were gender, externalizing problem behavior, and verbal ability (Table 3). Boys appeared to have a higher chance than girls to be in the moderate and increasing class of Closeness than in the high and stable class (B = 0.44, p = .023, OR = 1.56). In addition, students with higher levels of externalizing problem behavior in
Kindergarten were more often found in the moderate and increasing class of Closeness than in the high and stable class (B = 0.66, p < .001, OR = 1.93). Students with lower levels of
externalizing problem behavior were more inclined to have a very high and decreasing trajectory of Closeness compared to having a high and stable trajectory of Closeness (B =
−0,61, p < .001, OR = 1.84). Furthermore, students with higher scores on verbal ability had a higher chance of having a very high and decreasing trajectory of Closeness compared to students with a high and stable trajectory (B = 0.03, p = .001, OR = 1.03).
Conflict. For Conflict, the low and stable trajectory was used as a reference category. Gender and externalizing problem behavior were significant predictors of class membership (Table 3). Boys appeared to have a higher chance than girls of a high and decreasing trajectory (B = 0.80, p = .001, OR = 2.22) or a low and increasing trajectory (B = 1.02, p = .001, OR = 2.77) compared to the low and stable trajectory. Furthermore, students with higher levels of externalizing problem behavior were more often found in the high and decreasing trajectory (B = 2.83, p < .001, OR = 16.93) or the low and increasing trajectory (B = 1.23, p < .001, OR = 3.43), compared to the low and stable trajectory of Conflict.
Dependency. With regard to the two-class model of Dependency, the low decreasing trajectory was used a reference category. Both verbal ability and externalizing problem behavior in Kindergarten had a significant association with class membership (Table 3). Students with lower levels of verbal ability appeared to have higher chances of showing a low and increasing trajectory of Dependency (B = −0.03, p = .014, OR = 1.034) than a low and decreasing trajectory. In addition, the analyses revealed that students with higher levels of externalizing problem behavior were more often found in the group with a low and increasing trajectory of Dependency (B = 0.59, p < .001, OR = 1.81), as compared to a low and decreasing trajectory.
Associations between relationship trajectories and school outcomes
Closeness. It was investigated whether trajectories of teacher-student Closeness could predict motivation and academic adjustment of students in sixth grade. Students with a very high and decreasing trajectory showed higher levels of self-efficacy compared to students with a relatively high and stable level of Closeness, χ2(1) = 5.27, p = .022 (Table 4). In addition, it
was found that students with a very high and decreasing trajectory had higher levels of reading comprehension in comparison with students with a moderate and increasing trajectory of Closeness, χ2(1) = 4.86, p = .028. No differences were found regarding other outcome variables.
Conflict. Table 4 shows mean differences of students’ outcomes in sixth grade with regard to specific trajectories of teacher-student Conflict. Students with low and increasing levels of Conflict obtained significantly lower task motivation in Grade 6 compared to students with low and stable levels of teacher-student Conflict, χ2(1) = 7.238,p = .007. Students with a
low and increasing trajectory also performed significantly lower on reading comprehension than students who displayed low and stable levels of Conflict, χ2(1) = 6.301, p = .012. With regard to other outcome variables, no differences between trajectories of teacher-student Conflict were found.
Dependency. Significant differences between the two trajectories of teacher-student Dependency were found for all students’ outcomes in Grade 6. Students with a low and increasing level of Dependency obtained significantly lower scores in self-efficacy (χ2(1) = 9.892, p = .002), task-motivation (χ2(1) = 9.290, p = .002), vocabulary (χ2(1) = 14.730, p < .001), technical reading (χ2(1) = 10.974, p = .001), reading comprehension (χ2(1) = 40.518, p < .001), and math achievement (χ2(1) = 51.499, p < .001) in comparison to students with a low and decreasing trajectory of Dependency.
Discussion
The present study aimed to identify teacher-student relationship trajectories from kindergarten to sixth grade. In addition, we investigated whether relationship trajectories could be predicted by child characteristics (i.e., gender, SES, ethnicity, verbal ability, and
externalizing problem behavior), and whether students’ outcomes in sixth grade (i.e., self-efficacy, task motivation, reading achievement, and math achievement) differed between
relationship trajectories. The results of this study contribute to the existing literature, but also extend previous research about relationship trajectories of teacher-student closeness, conflict, and dependency.
Teacher-student closeness
In this study we identified several growth trajectories of teacher-student closeness by using a person-centered approach. We found that most of the students had high and relatively stable levels of closeness from kindergarten to sixth grade. In addition, we found a small group with a very high level of closeness in kindergarten, which decreased significantly in third grade and remained relatively stable until sixth grade. Another small group of students had a
moderate and increasing trajectory. An important finding of this study is that the mean level of teacher-student closeness for each of the identified trajectories was highly similar in sixth grade, whereas larger differences between the trajectories were found in kindergarten. These results are comparable with findings of Spilt et al. (2012), who found that differences between closeness trajectories were smaller in fifth grade compared to first grade.
The findings of this person-centered study are in contrast with studies that have used variable-centered approaches. Variable-centered studies try to identify a general trend over time for all participants without specifying different trajectories for subgroups of the total sample. These studies found decreases of closeness from preschool to first grade (Pianta & Stuhlman, 2004) and from kindergarten to sixth grade (Jerome et al., 2009). In contrast, the majority of students in the present study had high and relatively stable levels of closeness, which is similar to findings of other person-centered studies (O’Connor et al., 2012; Spilt et al., 2012).
Although the high and stable trajectory had a significant negative slope (−0.07), in practice this decrease can be considered negligible. An identical small negative slope was found for the largest group of students in the study of Spilt et al. (2012). Therefore, most students appear to have high and stable levels of closeness during elementary school.
With regard to predictors of relationship trajectories, boys and students with higher levels of externalizing problem behavior had more often moderate and increasing trajectories. Students with higher scores on verbal ability and lower levels of externalizing problem
behavior had higher chances of a very high and decreasing trajectory of closeness. Because all students had similar levels of closeness in sixth grade, child characteristics contribute to initial levels of closeness in kindergarten rather than to changes in closeness during elementary school.
In addition, no prior person-centered study has examined the association of verbal ability on relationship trajectories. With regard to variable-centered studies, students with better verbal ability had higher levels of closeness in kindergarten, indicating that young children with more advanced language skills are inclined to develop more close relationships with their teachers (cf., Justice et al., 2008; Moritz Rudasill et al., 2006; Spilt et al., 2015). The ability to have elaborate conversations with teachers may thus contribute to an affectionate teacher-student relationship. Although Spilt et al. (2015) concluded that teacher-students with lower verbal ability scores in kindergarten are also at risk for developing less close relationships with their teachers, this was not supported in this study because no trajectories with low levels of
closeness were found. Therefore, future studies should further examine the association between verbal ability and unfavourable trajectories of teacher-student closeness.
With regard to family characteristics as predictors of growth trajectories, we could not confirm the findings of Spilt et al. (2012), who found that ethnic minority students were more prevalent in moderate and increasing trajectories of closeness. An important difference is that the present study investigated a community sample of students, whereas Spilt et al. (2012) used an at-risk sample of students with lower literacy skills. Variable-centered studies using non-risk samples also found that the level of closeness in kindergarten was not associated with SES and ethnicity (e.g., Jerome et al., 2009; Pianta & Stuhlman, 2004). It can therefore be concluded
that SES and ethnicity do not predict the level of closeness in kindergarten or changes in closeness over time in community samples.
As far as students’ outcomes were concerned, we concluded that students with very high and decreasing levels of closeness had higher scores in self-efficacy compared to students with a high and stable trajectory. Furthermore, students with moderate and increasing levels of closeness scored lower on reading comprehension in sixth grade compared to students with a very high and decreasing trajectory, even when controlling for significant predictors such as verbal ability. The findings indicate that students with the lowest levels of teacher-student closeness in kindergarten have lower academic adjustment in sixth grade, even when they have relatively similar levels of closeness in grade six compared to students with other closeness trajectories. Based on these results, it can be argued that the initial quality of teacher-student closeness in kindergarten is more important for students’ outcomes than changes in relationship quality. This is in line with the study of Hamre and Pianta (2001), who found that the quality of teacher-student relationships in kindergarten was predictive of students’ outcomes at the end of elementary school. In addition, the findings are comparable to attachment theory, which
suggests that internal working models about relationships are formed in early years and are less inclined to change over time (McGrath & Van Bergen, 2015). Given that kindergarten is an introduction to school, it is especially important that relationships in kindergarten are a high in closeness. We can thus conclude that support of teachers in kindergarten is more important for students’ outcomes than continuous support of teachers across elementary school years. Teacher-student conflict
With regard to teacher-student conflict, we found that the largest group of students had a low and relatively stable trajectory of conflict. In addition, two smaller groups of students were identified with either a low and increasing trajectory or a high and decreasing trajectory of conflict. The identification of the three trajectories in this person-centered-study imply that
different teacher-student dyads exist, which explains the contradictory findings of variable-centered studies that either identified increases (Jerome et al., 2009) or decreases of teacher-student conflict (Pianta & Stuhlman, 2004).
The findings of the present study are similar to previous findings of teacher-student conflict trajectories. Previous studies also found that the majority of students had low and stable levels of conflict during elementary school (O’Connor et al., 2012; Spilt et al., 2012), and in addition, these studies identified multiple other trajectories, such as a high and decreasing trajectory and a low and increasing trajectory. However, the increasing and decreasing trajectories of Spilt et al. (2012) consisted of approximately 15 to 20% of the total sample, whereas smaller groups of 7 to 8% were found in this study. A possible explanation for this difference in group size is the diversity in sample selection. Spilt et al. (2012) examined an at-risk group of students, whereas this study consisted of a community sample of students. O’Connor et al. (2012) also investigated a community sample, and the group size of a low and increasing trajectory (7%) was similar to this study. Another important difference is that we were not able to identify a subgroup of students with high and stable levels of conflict over time, which can be explained by the fact that we did not differentiate our models with respect to boys and girls (Spilt et al., 2012). Spilt et al. (2012) investigated teacher-student trajectories separately for boys and girls, and concluded that only boys had a high and stable trajectory of conflict.
Several child characteristics were related to specific trajectories of teacher-student conflict. We found that boys and students with higher levels of externalizing problem behavior were overrepresented in a low and increasing trajectory compared to a low and stable
trajectory. We already know from variable-centered studies that boys and children with more externalizing problems have higher risks of developing more problematic teacher-student relationships (McGrath & Van Bergen, 2015; Nurmi, 2012). Taken into account results of other
person-centered studies (O’Connor et al., 2012; Spilt et al., 2012), we can argue that boys and students with more externalizing problem behavior have a higher probability of more negative relationship trajectories over time. Similar to the results of Spilt et al. (2012), boys and students with more externalizing problems were more prevalent in high and decreasing trajectories of teacher-student conflict. Although they developed more positive relationships with their teachers during elementary school, these students were still overrepresented in the high and decreasing trajectory compared to students in the low and stable trajectory. It is possible that boys and students with more behavior problems were less prepared for the learning
environment in kindergarten and early years of elementary school, but that they learned how to regulate their behavior after they spent more years in school. Another explanation is that externalizing problem behavior was only measured in kindergarten, and the development of externalizing problem behavior over time was not taken into account.
With respect to associations between trajectories of conflict and students’ outcomes, we found that students with low and increasing trajectories of conflict had lower scores on both task motivation and reading comprehension. This confirms previous research, in which
increasing levels of conflict resulted in underachievement of students in sixth grade (Spilt et al., 2012). Furthermore, no differences in students’ outcomes were found between high and
decreasing trajectories and low and stable trajectories, which was also one of the conclusions of Spilt et al. (2012). This suggests that even though children have more behavior problems and high levels of conflict in kindergarten, they develop less conflict with their teachers in later school years, and in turn they are more likely to avoid difficulties in both motivation and academic adjustment in sixth grade. This is in contrast to the results with respect to teacher-student closeness, in which relationship quality in kindergarten was specifically important for children’s motivation and academic adjustment. In line with the risk and stress hypothesis, cumulative experiences of interpersonal stressors are thus more important for children’s
development (Ladd et al., 2008), because children with a low and increasing trajectory had worse students’ outcomes compared to children without this increase in teacher-student conflict over time.
Teacher-student dependency
This was the first study that identified trajectories of teacher-student dependency over time. Researchers have argued that during elementary school years, students become more independent of their teachers (Ang, Chong, Huan, Quek, & Yeo, 2008), because older children rely more on their own emotional needs (Ang, 2005). In this study, a large majority of students had a relatively low level of dependency in kindergarten, which decreased further during elementary school. Although a decrease in dependency between teachers and students might be the norm, a small group of students had a significant increase in dependency during elementary school. These students started with relatively low levels of dependency in kindergarten,
however, teacher-student dependency increased to a moderate level in sixth grade. Furthermore, several child characteristics predicted trajectories of teacher-student dependency. Students with higher levels of externalizing problem behavior and lower levels of verbal ability in kindergarten had more often a low and increasing trajectory of dependency. A possible explanation for this finding is that teachers of students with more behavior problems and lower literacy skills may be more inclined to regulate behavior and to reassure students. Consecutively, these students might rely more on this reassurance of their teacher, leading to increasing levels of dependency over time. Previous research indicated that teachers reported more support and behavior regulation for students with whom they had a more dependent relationship, especially when children had higher levels of hyperactive behavior (Thijs,
Koomen, & Van Der Leij, 2008). Therefore, it is possible that teacher-student dependency is a more reciprocal construct in which behavior of both the teacher and the student has an effect on teacher-student dependency over time, leading to increasing levels of dependency during
elementary school. Nonetheless, in a cross-lagged study, no evidence was found between a reciprocal influence of externalizing problem behavior and teacher-student dependency over time (Roorda, Verschueren, Vancraeyveldt, Van Craeyevelt, & Colpin., 2014). However, this study did not make a distinction between different trajectories of teacher-student dependency, and therefore it is possible that in a more problematic trajectory (e.g., low and increasing levels of dependency), students and teachers do influence each other reciprocally. Future research should combine cross-lagged models with latent class growth analysis to increase knowledge on associations between child characteristics and teacher-student dependency trajectories over time.
Most importantly, significant differences were found between teacher-student
dependency trajectories with regard to students’ outcomes. Students with a low and increasing trajectory had lower scores on all outcomes variables in comparison to students with a low and decreasing trajectory. This indicates that students who rely more on their teachers during elementary school both have lower levels of motivation and lower scores of academic
adjustment in sixth grade. These findings are in line with the study of Hamre and Pianta (2001), who concluded that negative teacher-student relationships predicted less positive work habits and more disciplinary infractions in eighth grade. However, these researchers did not focus specifically on teacher-student dependency, but combined measures of conflict and
dependency. The present study extends their research in a way that not the initial level of dependency in kindergarten, but rather increases in teacher-student dependency results in less beneficial outcomes for students. Similar to trajectories of teacher-student conflict, increases in teacher-student dependency over time increase the probability of poor student outcomes (Ladd et al., 2008). However, it must be noted that increases in teacher-student dependency were related to lower scores on all of the measured outcomes, whereas increases in teacher-student conflict were only related to lower scores of task motivation and reading comprehension. Thus,
in contrast to previous beliefs (Ang, 2005), increasing levels of teacher-student dependency during elementary school appeared to be more important in predicting motivation and academic adjustment than increasing levels of teacher-conflict over time. Previous research also found stronger associations between teacher-student dependency and students’ adjustment in comparison to teacher-student conflict (Zee et al., 2013).
Limitations and future research
This study provides empirical support for the importance of teacher-student relationship trajectories during elementary school. However, some limitations should be taken into account when interpreting the results. First, the quality of the dyadic relationship between teachers and students was only assessed by using teacher reports. Nevertheless, it would be interesting to also include child reports in future research, because researchers found different clusters based on similarities and differences between reports of teachers and children (Wu, Hughes, & Kwok, 2010). Although a large group had similar views of teacher-student relationship quality, others had incongruent views. Because of these incongruent views, it is recommended to also
investigate developmental trajectories over time using child reports. However, by using only child reports, it is important to take into consideration the shared informant bias with all of the outcome variables. Furthermore, information about characteristics of teachers can give more insight in the factors that contribute to relationship trajectories, because previous research has indicated that teacher characteristics, such as ethnicity, gender, and educational level, also contribute to teacher-student relationship quality (Mashburn & Henry, 2004; Saft & Pianta, 2001).
Second, in this study we controlled for significant predictors of teacher-student relationship quality in kindergarten. However, we did not take into account time-varying covariates that influenced the quality of teacher-student relationship during elementary school. For instance, research has shown that there is a reciprocal influence of teacher-student conflict
and externalizing problem behaviors over time (Doumen et al., 2008; Roorda et al., 2014). Therefore, it is informative to investigate whether the same trajectories would be found when time-varying covariates were included at multiple measurement occasions. In addition, other covariates as internalizing behavior and prosocial behavior were not included in this study, although they have been found to influence the quality of teacher-student relationships over time (Roorda et al., 2014).
Third, only three waves of data collection were included in this study. Future research should increase the number of measurements to each school year, to account for year-to-year stability or differences in teacher-student relationship quality. Especially with regard to closeness (Figure 1), it can be seen that for two out of three trajectories, the largest change arises from kindergarten to third grade, whereas smaller to no change was found for the trajectories from third grade to sixth grade. More insight in the rate of change in early school years gives valuable information that helps to determine when best to implement interventions aimed at improving teacher-student relationships. Although we were able to include three waves of measurements and model the different trajectories for each dimension of the teacher-student relationship, examining relationship quality at each school year gives more insight in the development of relationship quality over time.
Furthermore, it is important to take in mind the following aspects. First, the results should not be interpreted as causal because of the non-experimental nature of the data. In addition, effect sizes of the predictive effect of specific trajectories on students’ outcomes were relatively small (ranging from 0.06 to 0.20). An explanation for these small effect sizes is that associations between variables were found in a community sample of students, whereas at-risk samples would possibly generate larger effects. Moreover, the analyses were to some extent data driven, because model selection mostly relied upon statistical criteria. Limited research
about teacher-student relationship trajectories, especially for teacher-student dependency, made it necessary to base model selection on statistical criteria rather than on theoretical assumptions. Conclusion and implications
Despite the aforementioned limitations, the current study provides strong evidence for the existence of different trajectories of teacher-student relationship quality from kindergarten to sixth grade. With regard to positive aspects of the teacher-student relationship, the initial level was most important for children’s development, whereas changes in negative aspects of the relationship had more influence on motivation and academic adjustment. Especially the increase in teacher-student dependency during elementary school was a predictor for less motivation and lower scores in academic adjustment. This finding is important considering that no previous research was carried out about the development of teacher-student dependency over time and its influence on students’ outcomes, mainly because earlier studies found insufficient support for the validity and reliability of the dependency subscale of the STRS (Doumen et al., 2009; Ewing & Taylor, 2009). Future research should use the psychometrically valid scale of dependency (Koomen et al., 2007; Koomen et al., 2012) to complement the emerging body of research that has used dependency as one of the dimensions of teacher-student relationships (e.g., Thijs et al., 2008; Zee et al., 2013). In the present study, no effects are found regarding ethnicity and SES, indicating that mainly child characteristics are related to trajectories of teacher-student relationship quality, in contrast to the familial background of children.
These conclusions have several implications for adjusting intervention programs to improve teacher-student dyads. Teacher-student closeness is of vital importance in
kindergarten, which implies that interventions aimed at improving teacher-student closeness should be implemented in or even before kindergarten. Although it has been argued that internal working models of teachers are resistant to change, it might be important to consider a
