Attitudes, Behavior and Relations in the Early School Years

Abstract

In the early school years, the emphasis is more and more on cognitive output factors. Non-cognitive development is receiving less attention than before, though such factors are important determinants of academic success. This study aims at answering two questions: (1) How do young children perform on a number of non-cognitive characteristics, more specifically, attitudes, behavior, and relationships? (2) Are there any differences with regard to those characteristics according to the pupils’ social and ethnic/immigrant background? To answer the questions, data from the Dutch large-scale cohort study COOL5-18 were analyzed. The main sample included nearly 6500 grade 2 pupils (6-year-olds). Teachers answered questions about their pupils’ attitudes, behavior, and relationships. One- and two-way analyses of variance were employed, and effect sizes were computed. The results showed that the teachers rated their pupil’s work attitude as lower than their behavior and popularity. They were more positive regarding their relationship with the pupils. More important was that there were differences according to the pupils’ social and ethnic/immigrant backgrounds: ethnic minority/immigrant pupils scored less positive on all non-cognitive characteristics than native Dutch pupils, and the higher the parental educational level, the more favorable their children performed on the non-cognitive characteristics. These findings are discussed and possible solutions are presented.

1. Introduction

Traditionally, early childhood education and care (ECEC) institutions have focused on play, while primary schools focused on intentional learning, initially on learning to read, write, and count. Nowadays, however, in most countries, the political and social attention on early learning has increased considerably, which has led to a clear shift in the respective curricula. In some countries, both kinds of institutions are still part of separate organizations and are accommodated in separate buildings with differing curricula, while in other countries, both institutions have been fully integrated, both in terms of housing and curriculum [1]. The main change has been the expansion of planned and formalized learning and learning through play in the early years. The curriculum at the ECEC level is typically designed with a holistic approach to support children’s early cognitive, physical, social, and emotional development and introduce young children to organized instruction. Education at the primary level is devised to provide pupils with fundamental skills in reading, writing, and mathematics and establish a solid basis for learning and understanding core areas of knowledge and personal and social development [2]. This shift towards planned learning included a much stronger accent on goal-setting, planning, formalized instruction, acquiring knowledge and skills, and evaluation and testing on the one hand and less attention on free play and non-cognitive and social-emotional aspects on the other hand [3,4,5]. This development has not gone without controversy and tensions: The growing pressures towards the formalization of early year’s education to improve educational outcomes, i.e., academic achievement, has led to a decrease in time for children to engage in play [6].

However, it is not only in the preschool phase that academic achievement has gained more emphasis; this is also the case in the primary school years. Plentiful (monitoring) studies have been conducted to check for any developments [7,8]. In many studies, researchers have sought explanations for achievement differences between various categories of pupils. For several decades now, the focus in these studies has been on factors in a child’s home situation. Again and again, research findings point to a correlation between a child’s family background, educational opportunities, and educational and societal success [7,9,10]. The main indicators of family background are socio-economic status (SES) and ethnic/immigrant origin [11]. Many researchers have stressed the disturbing fact that the achievement gap always has been wide and, notwithstanding targeted policies and the longtime investment of enormous budgets, still is widening [12,13].

Because of the increasing emphasis on academic achievement, as also for pupils in the early school years, most studies concentrate on the development of language and numeracy skills [8,14,15]. The general tendency is that ethnic minority/immigrant children and low-SES children academically achieve (much) lower than their ethnic majority/non-immigrant and middle- and high-SES classmates. Particularly worrisome is the finding that these differences already exist when the children enter the schooling system at a very young age. Therefore, it is felt to be imperative to start taking measures to prevent and combat educational and developmental delays at an early age, preferable in the pre- and early school years [16,17]. The importance of cognitive as well as non-cognitive skills in producing economic and social success is emphasized, just as the relevance of both formal academic institutions and families as sources of learning [18].

Non-cognitive factors, such as attitudes and behavior, have an important but undervalued function in school careers of children [19]. Various correlational and longitudinal studies indicate that a positive development of children’s socio-emotional competencies contributes to psychosocial adjustment, adequate attitudes, and, ultimately, better academic and behavioral outcomes [20]. Teaching and learning in schools include social, emotional, behavioral, and academic components [21]. Pupils normally do not learn alone but rather in collaboration with teachers, in classes with peers, and with the support of parents at home. Because emotional processes and relationships affect how and what we learn, it is felt that teachers, schools, and families must actively and effectively address these non-cognitive aspects of the educational process [22].

Exactly how the association between socio-emotional competencies and academic achievement can be explained and what the underlying processes are remains unclear, however [19]. According to one perspective, individual characteristics and competences are central. Research has consistently found that emotional intelligence correlates with higher psychological and emotional wellbeing and with less anxiety and depressive symptoms, and through this, contributes to higher academic achievement. Another perspective focuses on close and intimate peer relations and the participation in and acceptance of a peer group [23,24]. Research has shown that peers play an important role with regard to making academic decisions. Yet another perspective emphasizes the importance of the classroom social climate, which integrates interpersonal relationships with peers and teachers and is a socioemotional environment in which pupils can feel secure, valued, and supported. To some degree, underlying these perspectives is the notion of attachment. Caring and secure relationships established within a classroom, especially those between teachers and pupils, play a decisive role regarding the children’s sense of wellbeing [19].

Positive attachment to parents and teachers directly and indirectly influences pupils’ school success, such as higher grades and test scores but also greater emotional regulation and social competence, each of which in turn is connected with more academic success. A relevant finding is that this effect tends to be stronger for socioeconomically and ethnic minority disadvantaged pupils. Enhancing secure teacher–pupil relationships therefore is fundamental to raising achievement of especially at-risk groups. Teachers must connect with their pupils with warmth, respect, and trust, and several studies suggest that it is probably easier to establish such attachment relationships in pre- and primary schools than in later years [25,26,27].

Recently, a new challenge has emerged: As a result of the COVID-19 pandemic, school closures have widened the already existing achievement gap even more. Reasons given are that disadvantaged pupils tend to have less access to technology (internet, laptops), have no physical space for studying at home, receive no or only little support from their parents, and spend less time learning compared with their more affluent peers [28,29,30]. There is some evidence that during the pandemonium of COVID-19 and the accompanying long periods of school closure, the level of socio-emotional well-being has gone down as well, as contact and face-to-face interaction with teachers and peers is imperative for a healthy development. This is especially the case for children from socioeconomic and ethnic disadvantaged backgrounds [31].

From the above, it can be concluded that for many decades now, differences exist between children from diverse social and ethnic/immigrant backgrounds regarding their level of academic achievement. The picture with respect to non-cognitive outcome measures, such as the pupils’ attitudes, behavior, and relationships, is less clear, and this is especially the case for the early school years. What in particular is lacking are representative large-scale quantitative data [32]. This hiatus is particularly worrisome, as it is more and more being recognized that non-cognitive factors, including social and emotional competences, play a decisive role in children’s successful school career. The present study aims at shedding more light on this issue. The national Dutch COOL5-18 cohort studies contain information on both pupils’ attitudes and behavior and teacher-pupil relationships, as well as their social and ethnic/immigrant backgrounds. Interesting is the fact that these data are available on children in their early school years, viz., grade 2 (6-year-olds). This study aims at answering the following questions: (1) How do young children perform on a number of non-cognitive characteristics, more specifically, attitudes, behavior, and relationships? (2) Are there any differences with regard to those characteristics according to the pupils’ social and ethnic/immigrant background?

2. Materials and Methods

2.1. Participants

The data for the present study come from the three waves of the large-scale Dutch cohort study COOL5-18 collected in the 2007/2008, 2010/2011, and 2013/2014 school years [33,34,35]. In the last wave, a total of 437 primary schools with 28,529 pupils in grades 2, 5, and 8 (6-, 9-, and 12-year-olds; Dutch primary schools cater for children from 4 to 12 in 8 grades) participated in this national study. The total sample included a so-called reference sample of 340 schools, which is representative of all primary schools (n ≈ 6800). In the present study, the focus is on the grade 2 pupils in this sample (n ≈ 6500); their average age was 5.7 years per 1 January. The main aim of the COOL study was to monitor the developments of pupils from diverse backgrounds, both in terms of cognitive and non-cognitive characteristics. To achieve this, various instruments were used to collect the relevant data, not only language, reading, and math tests but also several scales capturing social-emotional aspects. In addition to the (general) monitoring function, the COOL data were also analyzed to answer specific policy topics, such as effects of early childhood education and care programs; effects of class composition; gender differences; and denominational school differences.

2.2. Instruments

Three instruments/measures are relevant for this study. First of all, the pupils’ socio-ethnic background, which was constructed on the basis of two family characteristics available from the schools’ administration. (1) The parental educational level, with three levels: low (maximum of pre-vocational secondary education), medium (maximum of senior secondary vocational education), and high (higher professional and university education). (2) The pupils’ ethnic origin, with two categories: non-Western immigrant, and native Dutch and Western immigrant. Combining these two family characteristics resulted in six categories: low/immigrant (5.8%); low/native (8.5%); medium/immigrant (6.5%); medium/native (37.6%); high/immigrant (4.2%); high/native (37.4%).

Secondly, the so-called pupil profile comprised 10 questions to the teachers about their pupils’ attitudes and behavior [36]. The answer options for the 5-point Likert items were: 1 = definitely not true; 2 = not true; 3 = not true, true; 4 = true; 5 = definitely true. Factor analyses revealed three factors: behavior, work attitude, and popularity (73.4% of explained variance). All reliability coefficients were good, that is, 0.80 or higher; see

Table 1. Non-cognitive pupil characteristics (2014 data).

Scale	n Items	α	Example of Item
Behavior	4	0.82	This pupil often is cheeky.
Work attitude	3	0.83	This pupil gives up quickly.
Popularity	3	0.86	This child is popular with classmates.
Dependency	5	0.90	This child is overly dependent on me.
Conflict	5	0.93	This child and I always seem to be struggling with each other.
Closeness	5	0.88	I share an affectionate, warm relationship with this child.

n = number of items; α = Cronbach’s α reliability.

. Three scale scores were computed by averaging the scores of the constituent items, if necessary, after recoding negatively formulated items. A high score thus indicates favorable behavior, a good work attitude, and a popular pupil.

Third is the Teacher–Student Relationship Scale (TSRS) [37,38], which is based on attachment theory [25]. The underlying idea is that a supportive teacher–pupil relationship is a very important determinant of pupils’ emotional and behavioral adjustment and—ultimately—high academic achievement. For the COOL cohort study, a shortened version of the TSRS was used. The teachers had to answer a total of 15 questions about their relationship with the pupils. The answer options were the same as those for the pupil profile, that is: 1 = definitely not true; 2 = not true; 3 = not true, true; 4 = true; 5 = definitely true. Factor analyses resulted in three scales, with 74.1% of explained variance: dependency (the degree to which the pupil is clingy, overly dependent, and overly reliant on the teacher); conflict (the degree to which a teacher–pupil relationship is a negative, unpleasant, and conflictual one); and closeness (the degree to which a teacher–pupil relationship is a satisfactory and positive one, characterized by warmth, support, and affection). Reliability coefficients were good (0.80 or higher); see Table 1. As with the pupil profile, three scale scores were computed by averaging the scores of the constituent items.

2.3. Data Analysis

To answer the research questions, both descriptive analyses and one- and two-way analyses of variance were performed. In addition, effect sizes were computed to obtain an impression of possible specific differences between socio-ethnic background categories. In the analyses of variance, the eta coefficient was computed to obtain an impression of the magnitude of the (overall) differences between the six socio-ethnic background categories. This coefficient can be interpreted in the same way as the correlation coefficient r. To do so, the criteria of Cohen can be taken as a guideline: 0.10 = weak; 0.30 = medium; and 0.50 = strong [39]. To better map possible specific differences, the mean scores for each of the socio-ethnic categories were compared to the mean scores for a reference category, which was the modal category of medium/native background, that is, medium level educated native Dutch parents. To gain insight into the magnitude of these differences, a so-called effect size (ES) was calculated for each difference [40]. ES values have the advantage of not depending on the size of the samples, and because they involve a standardized coefficient, indicators from different domains can be compared to each other [41]. In its most simple form, an ES is the difference between the means for two categories divided by the pooled standard deviation; this ES is referred to as Cohen’s d. With regard to the interpretation of d, the rule of thumb provided by Cohen is usually followed: 0.20 = small, 0.50 = medium, and 0.80 = large (this interpretation of effect size d thus differs from that of eta or r) [39]. Although using the pooled standard deviation to calculate the effect size generally gives a better estimate than the reference standard deviation, it is still slightly biased upwards. Therefore, here, a correction was used, suggested by Hedges and Olkin [42], who called the resulting effect size g, which can be interpreted in the same way as d.

3. Results

As mentioned before, this study uses data from the three waves of the COOL5-18 cohort study.

Table 2. Non-cognitive pupil characteristics by school year.

	School Year
	2008			2011			2014
Scale	m	sd	n	m	sd	n	m	sd	n
Behavior	3.64	0.78	9324	3.67	0.80	8331	3.71	0.79	6469
Work attitude	3.40	0.88	9253	3.40	0.89	8236	3.45	0.88	6375
Popularity	3.76	0.71	9243	3.79	0.71	8220	3.81	0.71	6370
Dependency	2.16	0.75	9339	2.15	0.76	8323	2.10	0.74	6455
Conflict	1.74	0.75	9340	1.71	0.75	8327	1.66	0.74	6458
Closeness	3.87	0.59	9324	3.90	0.60	8312	3.97	0.61	6457

m = mean; sd = standard deviation; n = number of pupils.

provides descriptive statistics of the six scales discerned. The goal of these analyses is to check how well the pupils perform according to their teachers on each of the scales but also if there are any significant differences between the three years.

On the basis of a comparison of the school years, it can be concluded that there are only marginal differences in means and standard deviations between each of the three years. In general, differences amount to less than five hundredths and thus are negligible. The fact that the scores on the various scales are stable over the years means that in a period of six years, there has not been any development with regard to these non-cognitive pupil characteristics. Therefore, it is justified to restrict further analyses to just one school year, in this case, the most recent one was opted for, that of 2013/2014, which included nearly 6500 pupils.

Regarding their attitudes and behavior, in 2013/2014, the pupils achieved the highest on popularity (3.81), followed by behavior (3.71), and the lowest on work attitude (3.45). Teachers therefore rated their pupils’ work attitude as lower than the pupils’ behavior and popularity.

Regarding the teacher–pupil relations, the general picture looks favorable. As norms for interpretation that are often provided by test developers are absent here, and certainly for this shortened version, it is difficult to interpret the scores in an absolute sense. Nevertheless, it appears that teachers are of the opinion that they are rather close to their pupils (3.97), that their pupils are not overly dependent (2.10), and that they have a pleasant relationship with their pupils, i.e., no conflicted relationships (1.66). To help interpretation: in the English version of the TRSR, the Likert scale scores ranged from 1 = definitely does not apply; 2 = does not really apply; 3 = neutral, not sure; 4 = applies somewhat; to 5 = definitely applies [37].

More interesting than an absolute interpretation of the pupils’ attitudes, behavior and relationship is a relative interpretation according to their socio-ethnic background. An overview of these analyses is provided in

Table 3. Non-cognitive pupil characteristics by socio-ethnic background (means; 2014 data).

Scale	Socio-Ethnic Background
	Low/Im	Low/Nat	Med/Im	Med/Nat	Hi/Im	Hi/Nat	eta
Behavior	3.60	3.60	3.50	3.72	3.61	3.81	0.12
Work attitude	3.28	3.27	3.23	3.44	3.48	3.58	0.13
Popularity	3.54	3.66	3.59	3.84	3.68	3.92	0.17
Dependency	2.26	2.24	2.28	2.10	2.16	2.00	0.13
Conflict	1.87	1.74	1.94	1.63	1.82	1.55	0.16
Closeness	3.75	3.90	3.80	4.00	3.87	4.04	0.15

Low/Im = low educated/immigrant; Low/Nat = low educated/native Dutch; Med/Im = medium educated/immigrant; Med/Nat = medium educated/native Dutch; Hi/Im = high educated/immigrant; Hi/Nat = high educated/native Dutch.

.

From the eta coefficients, it can be inferred that the overall differences according to socio-ethnic background are weak. It is clear, though, that the high education/native category in all instances has the most favorable scores, while the category medium education/immigrant holds the most unfavorable position. The latter actually comes as a surprise, as there is reason to expect that the category low education/immigrant would have scored even more unfavorable.

The eta coefficients in Table 3 provide a general picture of the differences according to socio-ethnic background.

Table 4. Non-cognitive pupil characteristics by socio-ethnic background (effect sizes g; reference category = medium/native background; 2014 data).

Scale	Socio-Ethnic Background
	Low/Im	Low/Nat	Med/Im	Med/Nat	Hi/Im	Hi/Nat
Behavior	−0.15	−0.15	−0.27	0.00	−0.14	0.12
Work attitude	−0.18	−0.19	−0.24	0.00	0.05	0.16
Popularity	−0.44	−0.26	−0.36	0.00	−0.23	0.12
Dependency	0.21	0.19	0.24	0.00	0.08	−0.14
Conflict	0.33	0.15	0.42	0.00	0.26	−0.11
Closeness	−0.41	−0.17	−0.34	0.00	−0.22	0.07

Low/Im = low educated/immigrant; Low/Nat = low educated/native Dutch; Med/Im = medium educated/immigrant; Med/Nat = medium educated/native Dutch; Hi/Im = high educated/immigrant; Hi/Nat = high educated/native Dutch.

contains the effect sizes for possible specific differences between the mean scores presented in Table 3.

The effect sizes in Table 4 confirm that the high/native category holds the most favorable position, but that the differences with the reference category of medium/native are less than small (in terms of Cohen’s rule of thumb). On the other hand, the differences with the medium/immigrant category are more substantial, somewhere between small and medium. The pupils in this category especially stand out with regard to the following characteristics: Their relationship with the teacher is much more conflicted; they are less popular among their classmates; and they experience less warmth, support, and affection in their relationship with the teacher. It is not only the medium/immigrant category that scores unfavorably; for several of the characteristics, this also applies to the low/immigrant category. These pupils score even less favorable regarding popularity and closeness.

The coefficients in Table 4 in a certain sense point to an immigrant-native Dutch divide: immigrant pupils appear to hold a more unfavorable position. Therefore, it may be relevant to conduct separate analyses for immigrant pupils versus native Dutch pupils. The results of this analysis are presented in

Table 5. Non-cognitive pupil characteristics by parental education and immigrant background (means; 2014 data).

Scale	Parental Education				Ethnicity
	Low	Medium	High	eta *	Immigrant	Native Dutch	eta
Behavior	3.60	3.68	3.79	0.09	3.56	3.75	0.09
Work attitude	3.27	3.41	3.57	0.12	3.31	3.48	0.07
Popularity	3.61	3.80	3.90	0.13	3.60	3.96	0.14
Dependency	2.25	2.12	2.01	0.11	2.24	2.07	0.09
Conflict	1.79	1.68	1.58	0.10	1.89	1.61	0.14
Closeness	3.84	3.97	4.03	0.11	3.80	4.01	0.13

* All associations are linear.

. This table also contains separate analyses for parental educational level.

To start with parental educational level, although the eta coefficients are only weak, the mean scores show a gradual development, which in fact is linear: The higher the parental level, the more favorable the pupil characteristics. The mean scores according to ethnicity in all instances show a more favorable position for native Dutch children. However, in this case, too, in terms of eta, the differences are only small. In addition to these (one-way) analyses of variance, hierarchical two-way analyses of variance were performed to check which of the two background characteristics, parental education or ethnicity, had the greatest impact. Although it should be stressed that the differences (eta’s) already were very small, these analyses showed that the beta’s hardly differed. Regarding behavior and popularity, there were no differences at all; regarding work attitude and dependency, parental education was somewhat more important; and regarding conflict and closeness, ethnicity was slightly more important.

4. Discussion

For quite some time now, there has been a discussion regarding the fact that, in the early school years, the focus has shifted too much from (free) play to (planned) learning, and as a consequence, also from an emphasis on non-cognitive factors to that on cognitive factors [3]. It is being argued that non-cognitive factors are being undervalued, as non-cognitive skills and motivation are important determinants of academic success [18]. The results of the present study show that teachers rated their pupils’ work attitude as lower than their behavior and popularity. Teachers seem somewhat more positive about their relation with the pupils: They are of the opinion that they are rather close to their pupils, that their pupils are not overly dependent, and that they have a pleasant relationship with their pupils. Interesting are the differences according to the pupils’ social and ethnic background. Although the differences in a statistical sense are not really big, there clearly is a link to this background: ethnic minority/immigrant pupils score lower on all non-cognitive characteristics than native Dutch pupils, and the higher the parental educational level, the more favorable their children perform on the non-cognitive characteristics.

The focus in this study was on non-cognitive characteristics of young pupils. Many studies have shown there to be a correlation between cognitive (i.e., academic) achievement and socio-ethnic background [8]. This study demonstrates that this correlation also exists with non-cognitive characteristics. As a reaction to the finding that disadvantage already exists when the children enter the educational system and that, therefore, action is required at that stage (or even earlier), numerous pre- and early-school compensation programs have been developed and implemented. Many of them aim at improving both cognitive and non-cognitive development in an educational institution as well as at home. Though the budgets of such programs are plentiful, the results mostly are disappointing, however [16,43].

The results of the present study may give a too rosy picture as the data analyzed were collected before the COVID-19 pandemic. Meanwhile, several studies have shown that as a consequence of the school closures, the already existing disadvantage gap regarding academic achievement has even widened. Therefore, in various countries, educational recovery programs have been implemented. In The Netherlands, for instance, the Ministry of Education has introduced an unprecedented National Program Education (NPE) to combat COVID-19-related educational delays. The core of this program consists of providing schools with additional budgets which they can spend on implementing evidence-based interventions. The total budget amounts to EUR 8.5 billion. Primary and secondary schools receive at least EUR 700 per pupil per year; depending on their pupils’ social-ethnic backgrounds, schools may receive even more [44]. The focus in such programs is on language and math, however. While it is clear that many children also struggle with social-emotional and psychological problems, information regarding occurrence still is scarce. Nevertheless, it would be quite reasonable to expect that the gap in terms of non-cognitive outcomes has widened also, as contact and face-to-face interaction with teachers and peers is imperative for a healthy development. This probably is especially the case for children from socioeconomic and ethnically disadvantaged backgrounds. Therefore, to get a better understanding more large-scale studies focusing on social-emotional development are needed [32].

This study has a limitation, which is associated with the method of data collection. The questionnaires with the questions about the pupils’ attitudes, behavior, and relationship were filled in by their class teachers. An intriguing question now is whether the reported differences according to ethnic/immigrant background are real or (partly) influenced by teacher bias. It is a well-documented fact that teachers not always are adaptive to the needs of all of their pupils. To be really adaptive, they must make adequate, that is, unbiased and unprejudiced, assessments of their pupils’ needs. This is, however, not always the case. Assessments may be influenced by prejudice and stereotypes related to specific groups, such as low-SES and ethnic minorities/immigrant pupils. Teaching affected by such group associations may consequently lead to increased achievement gaps and educational inequalities [45,46]. Research into possibilities to reduce teacher bias are very scarce. Promising are some recent teacher interventions and education programs aimed at reducing the negative effects of teachers’ biased group associations and contributing to equal educational opportunities for all pupils [46]. The number of such interventions is very limited, however, and many more experiments are needed.

A specific approach aiming at improving teacher–pupil relationships of disadvantaged groups is culturally relevant teaching (CRT), which acknowledges the need for effective education for the entire classroom while also meeting the specific needs of ethnic and culturally diverse pupil groups [47]. CRT is more of an underlying pedagogy than an intervention. The main features are that it acknowledges the strengths of the pupils’ diverse backgrounds and employs cultural resources to teach knowledge, skills, values, and attitudes. Although CRT is considered as a powerful method for increasing pupil engagement and achievement and for reducing achievement gaps, research into its effectiveness is limited and mainly consists of case studies. This does not take away the fact that some quantitative studies provide support for the effectiveness of CRT in everyday classrooms [48].

Many studies confirm that social-emotional competencies can be taught [22,49]. The Social-Emotional Learning (SEL) approach considers that, as with academic skills, the development of social and emotional skills can be acquired through explicit instruction [50]. The acquisition of social and emotional competences in this approach takes place within and outside the classroom in the school context, but also at the family and community levels [20]. Though SEL in principle aims at pupils of all ages, research on its effectiveness for the pre- and early-school phase is mostly lacking and thus urgently needed.