Abstract
Even when accounting for past performance, academic achievement can be influenced by teacher expectations, which are lower for disadvantaged and visible ethnic minority children. We use a Quebec (Canada) population-based sample (N = 1,311) to examine whether ethnicity and teacher-perceived signs of disadvantage in kindergarten predict child reports of their relationship with teachers in fourth grade. Results suggest that visible minority children were 50% less likely and perceived disadvantaged children were 32% less likely to report having a positive relationship with their teacher. The findings are discussed in terms of directing efforts toward reducing teacher prejudice and improving child academic success.
How well a child does in school predicts eventual income, well-being, and health outcomes in adulthood (Alexander, Entwisle, & Dauber, 1993; Card, 1999; Heckman, 2007). Consequently, it remains important to identify modifiable factors that can be targeted to bring about improvements in child personal success. Much attention has been dedicated to identifying child skills that are predictive of later academic success (Duncan et al., 2007; Pagani, Fitzpatrick, Archambault, & Janosz, 2010). However, evidence suggests that classroom-level factors can also shape children’s learning and achievement trajectories (Hughes & Kwok, 2007; Hughes, Luo, Kwok, & Loyd, 2008). When children form close and warm relationships with teachers, they perform better in school, show higher levels of motivation and engagement with classroom tasks, and develop more positive relations with peers (Sabol & Pianta, 2012).
Attachment theory predicts that relationships with teachers are likely to mirror the patterns of attachment children form early in life with caretakers. Nevertheless, a positive relationship with a teacher is believed to moderate the risks associated with negative childhood experiences by helping children revise their internal working models of attachment or expectations vis à vis interpersonal relationships (Buyse, Verschueren, & Doumen, 2011; Sabol & Pianta, 2012). Indeed, teachers who develop relations with students that are warm in terms of closeness and support tend to promote positive development in children (Pianta, Steinberg, & Rollins, 2009). By contrast, habitual conflict with a teacher tends to undermine child motivation and emotional adjustment at school (Baker, 2006; Brendgen, Wanner, & Vitaro, 2006; Hamre & Pianta, 2001).
The direction of influence in classroom relations is likely to run not only from teacher to student but also from student to teacher (Houts, Caspi, Pianta, Arseneault, & Moffitt, 2010). For instance, children may evoke differential treatment from their teachers on the basis of their in-class behavior. Challenging students are likely to require more effort from teachers, which may lead to strained relationships. Disruptive students are also likely to have an adverse effect on classroom climate by increasing teacher stress and reducing time available for educational activities.
Using developmental systems theory, Pianta (1999) developed a model that advances our understanding of the relational dynamics between child sociodemographic characteristics and teacher interactions. In particular, this theory specifies that teacher–child interactions are influenced by the characteristics of each individual, the nature of their interaction, and the social environmental context in which both individuals find themselves. Furthermore, a developmental systems approach predicts that in addition to being influenced by the characteristics of children and teachers themselves, exchanges and interactions between teachers and students can also be shaped by distal systems such as community relations or social stereotypes and perceptions (Bronfenbrenner & Morris, 1998; Sameroff, 2010).
Previous research has found that across the elementary school years, racial minority students appear to experience more conflict with teachers regardless of gender, level of academic problems, behavior problems, and family relations (Jerome, Hamre, & Pianta, 2009). Other research has found that disadvantaged children are likely to experience more academic risk, though this may not include problems with teachers (Hamre & Pianta, 2005). Children facing sociodemographic risk are likely to be at greater academic risk because they begin school with higher levels of academic and behavior problems. Nevertheless, teacher beliefs and expectations regarding student ability and performance can also influence the relations they develop with at-risk children. Eventually, the expectations teachers hold of students can have consequences for their academic performance (Jussim & Harber, 2005).
In one study, which combined four separate meta-analyses, teachers rated their expectations of students in terms of talent, performance, and exertion of effort. The results meta-analyzed across 39 samples indicated that teachers were more likely to hold higher expectations of European American students than Black or Latino students (Tenenbaum & Ruck, 2007). Furthermore, another study examined teacher expectations of 1,800 European American, Latino, and African American students using two independent data sets. The authors found that regardless of student ability level, as measured by academic records, teacher expectations of European American students were between 0.75 and 1.00 standard deviations higher than expectations for Latino and African American students (McKown & Weinstein, 2008; Tenenbaum & Ruck, 2007).
Sociologists and social psychologists have long argued that social stratification processes begin as early as elementary school due in part to the cumulative effects of prejudice and social stereotypes. According to sociological theory, children perceived as belonging to the middle class are more likely to approximate teacher judgments of an “ideal” student (Rist, 1970). Consequently, children from high and low social classes tend to elicit differential treatment from their teachers. For example, teachers are likely to value compliance in working-class students and mastery and curiosity from middle-class students. These differences in skills are then believed to each reflect skills desirable for high and low skills employment (Farkas, 2003; Rist, 2000). Teachers are likely to infer middle-class status from visible cues such as student clothing and general appearance. Consequently, we conceptualize perceived disadvantage in terms of teacher assessments of children’s material status, including access to adequate clothing and food.
Developing a close relationship with adults is important for children’s adjustment to school. In the present study, we use a large population-based sample of children born in the province of Quebec, Canada, to examine how visible minority status and signs of disadvantage are associated with student self-reported relations with teachers. One limitation of prior studies on teacher–child relations is that they have generally considered teacher–child relations from the perspective of teachers (Sabol & Pianta, 2012). Consequently, to better understand teacher–child relations from the perspective of minority groups at risk of academic problems and to remove variance associated with teacher social desirability bias, we use child ratings of relationship quality with teachers. Based on Pianta’s model, we expect that ethnic minority children and children showing signs of disadvantage will perceive their relationships with fourth-grade teachers less positively.
We include a measure of children’s objective socioeconomic status to better isolate the contributions of children’s appearance in terms of ethnicity and disadvantage, to the quality of teacher–child relations. Finally, because child gender, behavior problems, and academic competence have been shown to represent individual characteristics of children that are closely linked to the quality of teacher–child relations, we examine our hypothesized associations while taking into account the contribution of sex, kindergarten classroom engagement, and math skills (Houts et al., 2010). Both classroom engagement and math skills represent consistent predictors of academic achievement over time (Duncan et al., 2007; Fitzpatrick & Pagani, 2012; Pagani, Fitzpatrick, Archambault, et al., 2010). Furthermore, we measure these skills in kindergarten to reduce the possibility that teacher bias in grade 4 confounds both perceptions of child ability and the relationships developed with grade 4 teachers.
Methods
Sample
Participants were 1,311 children from the Québec Longitudinal Study of Child Development (QLSCD) with complete data on parent reports of ethnicity and fourth grade teacher-rated signs of student disadvantage. This sample is the 11th wave of a randomly selected, stratified sample of 2,837 infants born between 1997 and 1998 in the province of Quebec, Canada. At the inception of this longitudinal study, 93 children were deemed ineligible and 172 were untraceable due to incorrect coordinates. Of the 2,572 remaining children, 14 were unreachable and 438 refused participation. For the early childhood phase, 2,120 five-month-old infants were thus deemed eligible for follow-up, representing 82% of the eligible target population defined by the birth cohort. This sampling procedure resulted in 2,120 children once active informed parental consent was obtained. Of these, 39% were firstborn. Reflecting the demographics of the province of Quebec, the majority of children were ethnically Canadian (71%) while 9% reported being European, 6% reported being African/Haitian or Native, respectively, and 14% reported belonging to the category “Other.” From school entry onward, informed consent was obtained from parents and teachers. The QLSCD did not have a school-level multilevel design. As such, the majority of the children were the only ones being followed in their respective classrooms and schools, precluding school-level analyses.
Procedure
For the preschool collection waves, data were collected from children and their families around the date of the child’s birthday. During the school-aged waves, data collection occurred in the spring. Families, teachers, and school principals received informed consent forms by mail at each data collection wave. Teachers and families also received and returned questionnaires by mail. Parents were also interviewed by phone or in person by a research assistant trained by the government-operated Direction of Social Services. Parents were compensated in the amount of $25 and were informed that their child would receive a small present.
Measure: Outcome Variable
Teacher–child relations
At the end of the fourth grade, children self-reported the extent to which they perceived their relationship with teachers as supportive and warm. Items include the following: You feel at ease to ask your teacher questions when there is something you don’t understand; your teacher gets mad at you easily (reverse coded); Your teacher congratulates you when you do well in something; You like your teacher; You can talk to your teacher, he/she listens and answers nicely; You are afraid of your teacher (reverse coded); Your teacher yells or swears at you (reverse coded); Your teacher makes fun of you in a mean way (reverse coded; Pianta, 1992) from 1 (never or always true) to 5 (often or always true) (α = .76). A rescaled mean of these nine items, ranging from values of 1 to 10, was used as a global measure of the quality of teacher–child relations (see Table 1). The sample mean for this variable was 8.61 (SD = 1.44), and the median for the sample was 8.89. This variable was dichotomized to reflect poor teacher–child relations (scores below the 26th percentile) and good teacher–child relations (scores at or above the 26th percentile). This cut-off point was selected to reflect a natural separation in the distribution which was negatively skewed, (Skewness = −1.58, SE = .068), reflecting a general tendency to report a positive relationship with one’s teacher. We were interested in explaining the distinction between this large group of children who report positive relations with teachers and those reporting negative relations with teachers.
Descriptive Statistics for Outcome, Predictor, and Control Variables.
Note. Child ethnicity was coded as 1 (Black or native ethnicity) and 0 (European descent Canadian or other). Teacher–child relationships were coded as 1 (for children who were at or above the 26th percentile) and 0 (for children below the 26th percentile). Child sex was coded as 1 (males) and 2 (females).
Measures: Independent Variables
Signs of disadvantage
According to sociological theory, teachers are likely to perceive as “ideal” students who show signs of membership to the middle class. As such, children who show signs of material disadvantage, through a lack of access to adequate clothing or to food, are likely to be perceived as being of lower socioeconomic standing. At the end of the fourth grade, teachers reported how often the child had attended class: Over or underdressed for school activities; Too tired for schoolwork; Late; and Hungry. Items were rated from 1 (never) to 5 (always). Means were computed across all four variables, for each child, α = .58. This measure was moderately correlated with actual socioeconomic status (see Table 2) suggesting some external validity.
Correlations Between Continuous Variables Including Signs of Disadvantage, Classroom Engagement, Number Knowledge, and Socioeconomic Status.
p < .001.
Socioeconomic status
Socioeconomic status was derived from mother and father reports of income, occupational prestige, and level of education (Whilms & Shields, 1996). Scores were standardized to a mean of zero. Socioeconomic status scores represent the mean across all available time points between 5- and 74 months.
Minority status
A dichotomous variable was created to reflect whether children belonged to a stigmatized minority group or not. We merged Black and Aboriginal into “ethnic minority” variable because of the relatively small ethnic diversity of our sample and previous research, which suggests that these two groups are particularly likely to suffer from stigmatization and stereotypes. In particular, prior Canadian census data has revealed that Blacks and Aboriginals achieve below the Canadian average (Davies & Guppy, 1998; Ruck & Wortley, 2002).
Measures: Control Variables
Classroom engagement
Kindergarten teachers rated seven items of classroom engagement behaviors from 1 (never) to 3 (always): Follows rules and instructions; Follows directions; Listens attentively; Completes work on time; Works autonomously; Works and plays cooperatively with other children; and Works neatly and carefully (α = .94; Fitzpatrick & Pagani, 2012; Pagani, Fitzpatrick, Archambault, et al., 2010; Pagani, Fitzpatrick, Barnett, & Dubow, 2010). From these items, a mean classroom engagement score (ranging from 1 to 3) was then computed for each participant. Scores were then standardized using z-scores.
Number Knowledge Test (NKT)
This test was administered by trained examiners to assess basic number knowledge (Okamoto & Case, 1996). These measures were collected at age five for all children: Knowledge of the number sequence from 1 to 10; Knowledge of the one-to-one correspondence in which a sequence is mapped onto objects being counted; Understanding the cardinal value of each number; Understanding the generative rule which relates adjacent cardinal values; and Understanding that each successive number represents a set, which contains more objects. Scores on the NKT are highly correlated with general intellectual ability and have been shown to be robust predictors of elementary school achievement (Duncan et al., 2007; Pagani, Fitzpatrick, Archambault, et al., 2010).
Data Analytic Strategy
In the present study, we conduct secondary analysis of data previously collected by the Quebec institute of Statistics to verify our research hypothesis that indicators of disadvantage and ethnic minority status will predict poorer relationship quality with teachers. The objective is to employ logistic regression to estimate whether child ethnicity and signs of disadvantage contribute to children’s report of the quality of their relationship with their elementary school teacher, above and beyond other personal characteristics such as early classroom behavior and academic achievement. We first estimate the unique contribution of child visible minority status and sex to teacher–child relations. Next, we examine whether any observed associations persist after the model is re-estimated while taking into account children’s school-entry math competence and classroom engagement. Finally, the model is re-estimated taking into account an objective measure of child socioeconomic status and visible signs of disadvantage.
Results
Missing Data
As in most longitudinal studies, incomplete data were a concern. To retain statistical power and reduce bias occasioned by selective attrition, we followed the recommendations of Graham (2009). Incomplete data were imputed for covariates and the outcome variable using NORM multiple imputations program (Schafer, 1999). The chief advantage of NORM is that it draws values from the conditional distribution of the variables. It does this using an iterative method based on EM algorithm to impute incomplete data. This iterative process depends on the available and valid observations from the original data set.
Logistic Regression Analysis
Table 3 shows results from the logistic regression models predicting the odds of having a poor relationship with teachers from child visible minority status and visible signs of disadvantage. According to the final model (Model 3), visible minority children were 50% less likely to report having a positive relationship with teachers. Furthermore, children who were rated by their teachers as showing more signs of disadvantage were 32% less likely to report having a positive relationship with teachers. These associations were observed after controlling for actual socioeconomic status, kindergarten classroom engagement, sex, and school-entry math ability. A one unit decrease in classroom engagement was associated with a 40% decrease in the odds of reporting a high-quality relationship with teachers, although this effect was weakened after controlling for signs of disadvantage.
Odd Ratios for the Logistic Regression Depicting the Association Between Child Visible Minority Status and Indicators of Disadvantage and Teacher–Child Relationships by Grade 4.
Visible minority children were rated by teachers as showing more signs of disadvantage (x = 1.66 vs. 1.46, p ≤ .001) and as being less engaged than other children (x = −.24 vs. −.03, p ≤ .001). Actual socioeconomic status and number knowledge did not predict the odds of children reporting a negative relationship with teachers.
Finally, in post hoc analyses, we examined the possibility that the observed associations might vary on the basis of gender and socioeconomic status. We therefore computed interactions for disadvantage and ethnicity with both gender and socioeconomic status. These interactions were then included in the regression model. None of the interactions achieved statistical significance, suggesting that the results did not differ on the basis of gender or socioeconomic status.
Discussion
Cross-disciplinary perspectives converge in highlighting the importance of social interactions in human learning (Bodrova & Leong, 2007; Meltzoff, Kuhl, Movellan, & Sejnowski, 2009; Shonkoff & Phillips, 2000). However, relatively little research has examined differential quality of learning environments due to teacher biases and prejudice. In the present study, we found support for the hypothesis that Black or Native children, as well as children who were perceived by teachers as disadvantaged (e.g., by being inadequately dressed or coming to school hungry and tired), would report having less supportive relations with teachers.
The observed associations persisted regardless of measures of student academic performance, actual socioeconomic status, gender, and classroom engagement behavior. Though we do not directly assess classroom quality, controlling for classroom engagement is likely to partially reflect the interaction between classroom quality and children’s ability to navigate the classroom environment. The inclusion of classroom engagement therefore helps reduce the possibility that poor general classroom quality may be accounting for the observed association between indicators of disadvantage and teacher–child relations.
Our findings are consistent with those of Jerome et al. (2009), who found that elementary school teachers are more likely to report experiencing conflict with Black students regardless of academic risk. Similar to Jerome et al., we did not find an association between socioeconomic status and relationship quality with teachers. However, we did find that visible signs of social class, which were not assessed in the aforementioned study, were predictive of greater odds of poor relationship quality with teachers, regardless of their actual social status or level of competence. This finding suggests that some children may face unforeseen risks not accounted for by their level of ability or family background characteristics.
The present findings are novel in that they address classroom prejudice based on children’s apparent social status, as reported by teachers, and children’s own reports of how well they got along with their teacher. By taking into consideration teachers’ subjective perceptions of student disadvantage, as opposed to relying strictly on objective measures of socioeconomic status, this study considers variance in perceived disadvantage, and how this variance accounts for social interactions in the classroom. The present study is strengthened by the use of these different sources for the measurement of predictor, outcome, and control variables, which reduces the possibility that social desirability and shared measurement error account for the observed associations. The findings therefore suggest that it is not that poor children are categorically treated differently by teachers. Rather, it is children who appear disadvantaged by showing up to class hungry, inappropriately dressed, tired, or late and who are of a minority status who experience poorer teacher relations. Our results are in line with predictions made on the basis of Pianta’s (1999) model. That is, we found evidence that social perceptions, shaped by larger social contexts and community relations, may influence teacher–child relations.
The quality of the relationship students develop with grade-school teachers can have important repercussions on children’s subsequent academic and life trajectories (Baker, 2006; Brendgen et al., 2006; Hamre & Pianta, 2001; Hughes & Kwok, 2007; Hughes et al., 2008; Pianta et al., 2009). Furthermore, a warm and positive relationship with teachers can represent a protective factor for the healthy development of children with less-secure attachment patterns with caretakers in the home (Buyse et al., 2011). For children, teachers also represent a first encounter with an adult authority figure in a social institution outside of their more informal relations in the home. Experiences in the classroom therefore help prepare children for their eventual future experiences in the adult workplace (Bowles, Gintis, & Osborne, 2001). If their initial experience with teachers in the classroom is conflictual and unsupportive, this may influence their self-perceptions as an individual, as well as their development of mental models and schemas for interacting with authority figures in society (Henricsson & Rydell, 2004).
Research in the area of interpersonal perception suggests that perceived low social status can engender perceptions of lesser warmth and competence (Fiske, Xu, Cuddy, & Glick, 1999). Perceived warmth and competence are fundamental dimensions of social perception that elicit functionally relevant affective and behavioral responses (Cuddy, Fiske, & Glick, 2008; Fiske, Cuddy, & Glick, 2007). For instance, perceiving others as callous and as being less competent can elicit emotions of anger and disgust, and motivate aggressive interpersonal behavior such as excluding or verbal assaults. This body of research may help explain why teachers appear to treat children differentially based on their ethnicity and perceived disadvantage.
Children may not consciously link the quality of their relationship with their teachers to their ethnicity or signs of disadvantage. Nevertheless, children who are of a minority status or who appear to be disadvantaged are aware of their poor relations with their teacher. They may come to draw comparisons between how they and other students are treated and develop poorer academic self-concept. This discrepancy may have important negative consequences on children’s academic trajectories, self-perception, and motivation.
This study is limited in that children’s ratings of the quality of their relationship with their teacher could reflect a range of variables particular to the child’s own experience. However, the aim of the present study was to explain the quality of teacher–child relations based on the particular dynamics shaping these relations, such as children’s academic skills, classroom engagement, and appearance in terms of ethnicity and social status. This study was also limited in that the observed alpha for our measure of signs of disadvantage was relatively low. The present study was based on the secondary analyses of a large-scale longitudinal study. As such, refined measures were not available for all constructs of interest. Although a higher alpha would have been preferable, it is unlikely that the employed measure is responsible for creating an illusory relationship between our predictor and outcome variable. Rather, it is more likely that the low reliability of this measure might have led to underestimating the true magnitude of the observed association. Furthermore, this measure of signs of disadvantage showed some external validity as it was correlated with an objective measure of disadvantage. Attrition might also have affected the results of the study. In order to reduce the possibility that self-selection into attrition might bias our observed associations, we followed the most recent effective practices for dealing with missing data in longitudinal research (Graham, 2009; Widaman, 2006). In the present study, it was not possible to examine how the full range of visible minority statuses might influence interactions with students, consequently the present results are applicable only to Black and Aboriginal children. Finally, the sample in the present study was from the province of Quebec, Canada, which differs from other Canadian provinces and the United States in terms of investment in social programming. Nevertheless, two large-scale studies provide evidence that the long-term consequences of disadvantage for child academic achievement tend to be the same in Quebec and the United States, which suggests that our results are generalizable to the North American context (Pagani, Boulerice, & Tremblay, 1997; Pagani, Boulerice, Tremblay, & Vitaro, 1999).
Future research in the area of teacher–child relations should examine whether school composition in terms of ethnicity and socioeconomic status is also associated with student perceptions of their relations with teachers. In addition, it would be informative to take into account the potential moderating role of teacher ethnicity and socioeconomic status on student–teacher interactions. These areas of investigation are likely to be especially important in countries such as the United States where socioeconomic racial disparities are especially large and have distinct historical origins.
In North America, low birth rates, rising immigration, and a growing gap between the rich and poor ensure that teachers will face increasingly diverse and disadvantaged students. Teacher biases against low social status children appear to influence interactions between students and teachers in the elementary school classroom. Consequently, teachers are not immune from the use of social stereotypes, which routinely operate in impression formation and interpersonal relations.
An important practical implication of the present research concerns the development of teacher training geared at combating stereotyping and prejudice in the classroom. Training can be provided to teachers to help them overcome stereotypes through willful processing and special training exercises (Lillis & Hayes, 2007; Walker & Crogan, 1998). Providing ongoing training to teachers that encourages them to attend to individuating counter-stereotype information about stigmatized group members may help reduce implicit prejudice. Teachers can also help reduce their own as well as student stereotypes by pointing out individual who possesses counter-stereotypic characteristics during classroom lessons. Finally, teachers can invite members of stigmatized groups to share their experiences of discrimination in the classroom to help provide minority children with strategies to cope with the negative consequences of bias (McKown, 2005).
The present study is an important catalyst for addressing the quality of teacher–child relations as it relates to children’s appearance and disadvantage. Ensuring that all children regardless of race and social class have equal chances to achieve success and positive adjustment in the classroom is an important task. Given the findings of the present study, we propose that policies be developed to provide teachers and students with opportunities to develop an awareness of social stereotypes and their potential influence in the classroom.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.