Introduction

Classroom management is a challenging task for teachers (Emmer & Sabornie, 2015; St-Amand et al., 2021b), a multifaceted endeavor that is far more complex than just establishing classroom rules. Overall, it is about time, space, material, codes, procedures and routines, curriculum, effective planning and assessment, observation, self-evaluation, and, more importantly, warm, stimulating, and supportive interactions (Emmer & Sabornie, 2015). Practically speaking, classroom management should seek to remove barriers between students and learning (Debbag & Fidan, 2020; Emmer et al., 2003) and favor their motivation, engagement, and learning through effective supervision (Archambault & Chouinard, 2016), while seeking to elicit student cooperation (Evertson et al., 1983). Therefore, it requires effective practices that create a classroom atmosphere that nurtures interest in learning (Rambe & Harahap, 2023; Snyder, 1998). The present study is interested in teachers’ use of humor (e.g., course-related humor, course-unrelated humor, self-disparaging humor, other-disparaging humor) and its association with students’ emotional well-being, sense of school belonging, and school engagement.

Humor in classroom management

As a well-established pedagogical practice that has been recognized for several decades in classroom management (Banas et al., 2011; St-Amand et al., 2021a), teacher humor seems to favor positive social interactions in the classroom, while evolving in terms of understanding as the children mature (Martin, 2010); in fact, students of all ages identify humor as one of the characteristics they value most in a teacher (Archambault & Chouinard, 2016; Cefai & Cooper, 2010). Researchers suggested that humor can harm (if misused) or strengthen social relationships between students and teacher; in addition, humor is thought to help students and teachers feel good and closer together (Friedman & Kuipers, 2013; Ho, 2016). This strategy is extensively used to establish or restore authority, as well as to help students adopt the values of the school code of conduct (Garner, 2006). Research has indicated that teachers’ use of humor is significantly associated with students’ motivation (Conkell et al., 1999; Luo et al., 2023), classroom and school climate (Kosiczky & Mullen, 2013), and the quality of social relations (AbdAli et al., 2016). Teachers’ humor makes it easier for students to learn difficult course content (Abdulmajeed & Hameed, 2017; Özdemir, 2017) and to succeed (Abdulmajeed & Hamed, 2017; Al-Duleimi & Aziz, 2016). Researchers have also noted an increase in students’ school engagement when teachers use humor judiciously (Hoad et al., 2013). On the teacher side, humor is reported to provide increased job satisfaction while also being a strategy that can be used to reduce teacher stress (Booth-Butterfield et al., 2007; Mawhinney, 2008). In their systematic review of the literature, Banas et al. (2011) noted that teacher humor is positively associated with the quality of teacher feedback, positive student emotions, positive perceptions of the school environment, and positive teacher perceptions. In other fields of research, meta-analyses have indicated that humor has a positive influence on workplaces (Mesmer-Magnus et al., 2012), communication (Walter et al., 2018), romantic relationships (Hall, 2017), media (Eisend, 2009), and within the field of psychology in general (Mendiburo-Seguel et al., 2015). Although a considerable amount of literary, linguistic, philosophical, communication, educational, and psychological work has been devoted to it, humor eludes any precise definition (Derouesné, 2016; Tanay et al., 2013) but generally involves a perceptual dimension (Banas et al., 2011; Martin, 2010). For example, Martin (2010) argued that humor constitutes a generic term with a generally positive, socially desirable connotation that refers to the perception of what people say or do as funny, and in a context where humor elicits joy and laughter. Although humor is a fundamental element of human communication, the concept of humor is often used inappropriately. As Tanay et al. (2013) pointed out, humor is often confused with concepts such as “joke,“ “laughter,“ or even “wit.“ As for the way it is defined: “Humor is not a unidimensional concept; instead, there are a variety of types that instructors may employ in their classrooms including: nonverbal humor, jokes, unplanned humor, self-disparaging humor, and aggressive humor (Goodboy & Bolkan, 2015)” (p. 46). To the best of our knowledge, the only conceptual study on the concept of humor in classroom management was conducted by St-Amand et al. (2021a), who highlighted the complex nature of this practice by proposing a general definition of teacher humor in classroom management contexts:

Whether planned or spontaneous, humor in the classroom is a skill displayed by the teacher and a form of communication containing very specific objectives, such as wanting to improve the classroom climate, to reduce students’ anxiety, to resolve conflicts, to improve and maintain social ties, to encourage school engagement, and, ultimately, to support students’ academic achievement. Much more than a unique pedagogical approach that can be modeled from one individual to another, teachers’ humor is characterized by a subjective and personal character that is specific to each teacher. When effectively conducted, teachers’ humor elicits positive emotional and behavioral responses from students. (St-Amand et al., 2021a, p. 120)

Theoretical framework and literature review

Drawing on several theoretical perspectives from the field of psychology (Fave et al., 1996; Zillmann & Cantor, 1996), Wanzer et al. (2010) developed a theory adapted to the educational context (instructional humor processing theory) that attempts to explain the complex processes occurring in the relationship between teacher humor and student learning. This theory allows us to examine and better interpret how teacher humor is processed and perceived by students. According to this theory, students must first recognize an incongruity in the teacher’s message and then interpret and resolve it. If the incongruity is not resolved, the student will not perceive the humorous message and as a result may be distracted or confused. However, if the student resolves the incongruity, he or she may perceive a humorous message and laughter may result; in this context, the nature of the humorous message and how it is interpreted emotionally determines whether or not humor facilitates motivation and learning.

Teachers’ humor and students’ emotional well-being

Several studies have used the instructional humor processing theory in recent years. The majority of these have been conducted at the university level (Bolkan et al., 2018; Bolkan & Goodboy, 2015; Goodboy et al., 2015; Imlawi et al., 2015; Tsukawaki & Imura, 2020; Wanzer et al., 2010) while fewer have been conducted in high school or elementary settings (Bieg et al., 2017, 2019; Ziyaeemehr, 2011). These studies have attempted to better understand the relationship between teacher humor and student emotions (Bieg et al., 2017, 2019; Wanzer et al., 2010), teacher humor and student learning (Bolkan et al., 2018; Tsukawaki & Imura, 2020), and teacher humor and school engagement (Bolkan & Goodboy, 2015; Goodboy et al., 2015; Imlawi et al., 2015), and the reasons why teachers refrain from using humor in the classroom (Ziyaeemehr, 2011). From these studies, humor related to course content was found to have a significant and positive association with enjoyment of being in class, while being negatively associated with boredom and anger (Bieg et al., 2019). Among high school students, Bieg et al. (2017) also reported a significant and positive association between humor related to course content and enjoyment of being in class, as well as a negative association with boredom. Furthermore, these results suggested that the more course content-related humor the teacher uses, the less anxiety students experience. According to Bieg et al. (2017), humor unrelated to course content and self-disparaging humor are not predictors of these positive emotions. Among college students, Wanzer et al. (2010) also determined that humor related to course content displays a positive relationship with learning-related emotions. In light of these studies conducted specifically with high school, elementary school, and college students, it can be concluded that humor related to course content appears to be closely linked to students’ overall emotional well-being.

Students’ emotional well-being and engagement

Although there is no consensus regarding the definition of emotional well-being, it does include psychological dimensions such as the presence of positive emotions (Liddle & Carter, 2015). Lately, many studies have, therefore, increased the importance of the mediating role of emotional well-being (including emotions such as enjoyment, happiness, or a general measure of positive emotions) as well as its direct influence on school engagement. At the elementary level, students’ feelings such as happiness (Kwon et al., 2017) and enjoyment (Reschly et al., 2008) have been significantly and positively associated with school engagement. Enjoyment has also mediated the relationship between teacher support and school engagement (Liu et al., 2018). Chen et al. (2020) found that student enjoyment in the classroom mediated the relationship between the teachers’ communication skills and students’ engagement. At the high school level, McKeering et al. (2021) highlighted a significant and positive association between emotional well-being and engagement. Gong and Bergey (2020) found that positive emotions completely mediate the relationship between student efficacy and engagement, while Liu et al. (2021), for their part, indicated that teacher support can be positively and significantly associated with behavioral engagement via enjoyment. As for age groups, older students (12 to 14 years old) showed lower emotional well-being than younger groups (10 years old) (McKeering et al., 2021). Among college students, positive emotions (time 1) were shown to influence engagement (time 2) significantly and positively. The association between teacher enthusiasm and student engagement was found to be mediated by emotional well-being (boredom and enjoyment) (Dewaele & Li, 2021). With regard to the relationship between student self-control and behavioral and emotional engagement, positive emotion was identified as a mediator (King & Gaerlan, 2014).

Teachers’ humor and students’ sense of school belonging

The sense of school belonging is a fundamental component of student engagement and is generally stronger for females than for males, while varying from one grade to another (Faircloth & Hamm, 2005; Finn, 1989; Goodenow & Grady, 1993; Korpershoek et al., 2020; Sari, 2013; Smith et al., 2020; St-Amand et al., 2020a). It is also a multidimensional and complex concept comprising several definitional attributes. Students must: (1) feel a positive emotion toward school; (2) maintain positive social relationships with their peers and teachers; (3) perceive a synergy (harmonization) and a certain similarity with the members of the group; and (4) become actively involved in the school environment (St-Amand et al., 2017b). Several factors contribute to creating and structuring students’ sense of school belonging (Ahmadi et al., 2020; Allen et al., 2018; Janosz et al., 1998). Allen et al. (2018) categorized the determinants of the sense of school belonging into factors at the individual level (e.g., personality, self-esteem, social skills, motivation, optimism), the micro level (e.g., social relationships, parents, peers, teacher support, presence of friends), and the meso level (e.g., extracurricular activities, discipline in the classroom, the climate of justice, the climate of security). Teacher humor was identified as a very important pedagogical strategy that builds students’ sense of school belonging (Certo et al., 2003; Cothran & Ennis, 2000; FitzSimmons, 2006; Glaser & Bingham, 2009; Hillman, 2011; LoVerde, 2007; Ozer et al., 2008) and makes students want to attend class (Seaman, 2017). Teacher humor is a pedagogical strategy that positively changes the classroom atmosphere (Cooper et al., 2018) and helps in developing a good social network (Kibler et al., 2019); Cooper et al. (2018) showed that, on average, teacher humor slightly increases classroom belonging for 37.8% of respondents, and a great deal in 42.2% of respondents (sample = 1637 college students). Other researchers reported that teacher humor at T1 is a significant predictor of students’ sense of belonging at T2 and T3 (sample = 335 college students) (Sidelinger et al., 2012). On the flip side, a teacher who never uses humor could also contribute to students’ sense of belonging. As Stuart and Rosenfeld (1994, p. 93) explained:

The one benefit of teachers being perceived as humorless or as using hostile humor exclusively relates to affiliation in the classroom. Students with these two types of teachers are more likely to increase interaction among themselves, perhaps either to relieve boredom or to unite and share perceptions of a common “enemy.“

Students’ sense of school belonging and engagement

Developing student engagement in school and preventing disengagement is a major concern for teachers (Archambault et al., 2019; Mbikayi & St-Amand, 2017; St-Amand, 2016, 2018; St-Amand et al., 2017a, b, c). One of the overriding elements that increases school disengagement is a low sense of school belonging (Allen et al., 2023; Archambault et al., 2022; Christenson & Thurlow, 2004). Conversely, scientists have suggested that a high sense of school belonging facilitates school engagement (Fong Lam et al., 2015; Hughes et al., 2015; Janosz et al., 1998). Seminal work by Wehlage et al. (1989) and Goodenow (1993a, 1993b) showed that when students feel supported by peers and adults in their learning environment (school) and identify with (or feel included in) that environment, they tend to place more value on learning and achieving. For Goodenow (1993b), these elements form the basis of what is known as “school climate” or a “sense of school belonging.” Previous studies clearly indicated a positive association between the sense of school belonging/climate and the quality of school engagement (Janosz et al., 1998; St-Amand et al., 2020b). Indeed, researchers have demonstrated the association between the sense of school belonging and several motivational variables (Korpershoek et al., 2020). Some studies found a positive and significant association between the sense of school belonging and the three forms of school engagement (cognitive, affective, and behavioral) (Korpershoek et al., 2020; St-Amand et al., 2020b, 2021a). Other theoretical studies highlighted the positive association between students’ sense of school belonging and school engagement (Connell et al., 1994; Newmann et al., 1992). Pioneers in the study of the sense of school belonging highlighted that such an association started in the late 1980s. Connell et al. (1994) suggested a direct relationship between a sense of school belonging and school engagement, which, in turn, positively influences academic achievement. In developing their theoretical model, Newmann et al. (1992) also pointed out the significant and positive relationship between the sense of belonging and school engagement, specifying that belonging directly influences school engagement. Finn (1989) developed a dynamic model explaining the links between the sense of school belonging and school engagement. This model states that participation in activities is a critical component of school success, which, in turn, contributes to the development and structuring of the sense of school belonging. Also, in the late 1980s, Wehlage et al. (1989) focused on pedagogical practices and the effectiveness of the school in promoting education. From this theoretical perspective, the relationship between the sense of school belonging and school engagement is bidirectional in nature, which is theoretically different from the theoretical evidence presented thus far.

Rationale of the study and hypotheses

As discussed in the previous sections, researchers have attempted to investigate the relationship between teacher humor and school engagement through the lens of the instructional humor processing theory (Bolkan & Goodboy, 2015; Goodboy et al., 2015; Imlawi et al., 2015). These studies were conducted among adults and college students, not with elementary and high school students, and, to the best of our knowledge, only Bolkan and Goodboy’s (2015) study considers the affects as a mediating variable in the relationship between teacher humor and school engagement. Using structural equation modeling, their work indicates that the relationship between teacher humor and school engagement is mediated by students’ affects related to the course and the teacher, as well as the recommended course behaviors. In addition to not taking into consideration other types of school engagement (cognitive, affective, behavioral) (Fredericks et al., 2004), these researchers do not account for emotions that underlie school engagement: students’ sense of school belonging and emotional well-being. The present study is intended to contribute to the work of Bolkan and Goodboy (2015) in that teacher humor – that is, the ability to make the learning environment more effective – may influence students’ emotional well-being and a basic need such as the sense of school belonging, which, in its definition, carries a great deal of emotional significance: “Belonging also refers to positive emotions, which could be described as emotional attachments, more precisely to a feeling of intimacy, feeling part of a supportive environment, and a sense of pride in the school” (St-Amand et al., 2017a, p. 14). The present study, therefore, seeks to determine how teachers’ use of humor (e.g., course-related humor, course-unrelated humor, self-disparaging humor, other-disparaging humor) relates to students’ sense of school belonging, emotional well-being, and engagement, and to test for invariance across elementary and high school students and across males and females. Derived from the instructional humor processing theory, Fig. 1 illustrates the determinants of school engagement from the different groups of predictor variables; one of these groups (sense of school belonging) is directly linked with the different forms of school engagement, while other variables (adequate and inadequate humor) indicate indirect links with school engagement.

Fig. 1
figure 1

Initial Model

Model inspired by the instructional humor processing theory describing the links between the different types of teacher humor, emotional well-being, the sense of school belonging, and school engagement.

Full size image

The organization of these associations within the model leads us to formulate six research hypotheses:

H1

The different types of teacher humor (course-related humor, course-unrelated humor, self-disparaging humor, other-disparaging humor) have a positive influence on the sense of school belonging.

H2

The sense of school belonging has a positive influence on the three types of school engagement (cognitive, affective, and behavioral).

H3

The different types of teacher humor (course-related humor, course-unrelated humor, self-disparaging humor, other-disparaging humor) have a positive influence on students’ emotional well-being.

H4

Students’ emotional well-being has a positive influence on the three types of school engagement (cognitive, affective, and behavioral).

H5

The association between the different types of teacher humor and the sense of school belonging is stronger for high school students than for elementary school students.

H6

The association between the different types of teacher humor and the sense of school belonging is stronger for females than for males.

Methodology

Sample

The participants comprised 395 students (boys = 106; girls = 270; other = 8; NA = 11) (secondary school students = 291; primary school students = 97, NA = 7) from five public school boards located in rural areas, and one private secondary school situated in an urban area (Mage = 14.11) with a proportion of 93% speaking French at home. The data collection took place during the months of April and May, 2021. Students were instructed to respond to all questions online and to keep their answers confidential. It took less than 15 minutes to complete the questionnaire. The questionnaire was comprised of 40 items, allowing for the measure of nine different latent constructs. Participants had to indicate their level of agreement regarding each item on a Likert scale. To measure the variables in this study and the quality of certain characteristics of the school environment, the authors used only part of the Questionnaire sur l’environnement socioéducatif (QES-secondaire) [Questionnaire on the socio-educational environment (QES-high school)], namely the sense of school belonging, and the cognitive, affective, and behavioral dimensions of school engagement (Janosz & Bouthillier, 2007). To measure teachers’ adequate and inadequate humor, we used the Teacher Humor Scale based on the work of Wanzer et al. (2006) and validated by Frymier et al. (2008). It allowed us to measure variables such as course-related humor, course-unrelated humor, self-disparaging humor, and other-disparaging humor. Finally, emotional well-being was measured using the General Well-Being Scale (Perreault, 1989).

Measures

To measure the emotional dimension of the sense of school belonging, we used a five-item subscale that assessed students’ sense of school belonging (items: “I feel proud to be a student at my school,” “I feel like I’m really part of my school,” “I like my school,” “I am happy to be back to school after a long school break,” and “I wish I were in a different school”) (Janosz & Bouthillier, 2007). The last item was reverse coded, and item scores were averaged to generate a score reflecting school belonging (M = 4.54, SD = 1.77, α = 0.86) (1 = strongly disagree to 6 = strongly agree).

Self-reported items were used to measure school engagement. These items represent three dimensions converging toward a more global concept measuring school engagement. In this study, the authors consider each of these three dimensions in a unique way, as suggested by most scholars in the field of school motivation (Fredricks et al., 2004). First, behavioral engagement measures positive behaviors such as the following of classroom rules and adherence to classroom norms, as well as the absence of disturbing behaviors (Fredricks et al., 2004; Martins et al., 2022). To measure behavioral engagement, participants responded to the four-item subscale that assessed this dimension (Janosz & Bouthillier, 2007) (items: “In the past 12 months, have you missed school without a valid excuse?”, “In the past 12 months, have you missed a class while you were in school?”, “In the past 12 months, have you disturbed your class on purpose?”, and “In the past 12 months, have you responded to a teacher by being impolite?”) (M = 5.25, SD = 0.96, α = 0.60) (1 = strongly disagree to 6 = strongly agree). Second, affective engagement in school tasks refers to feelings, interest, perceptions, and attitudes toward school (Fredricks et al., 2004). To measure affective engagement, participants responded to the five-item subscale that assessed this dimension (Janosz & Bouthillier, 2007) (items: “I like school,” “I have fun at school,” “What we learn in class is interesting,” “I am very enthusiastic when the job to be done is quite difficult,” and “Often I don’t feel like stopping work at the end of a course”) (M = 3.95, SD = 1.13, α = 0.88) (1 = strongly disagree to 6 = strongly agree). Third, the cognitive dimension of school engagement relates to the psychological investment in learning school subjects (Fredricks et al., 2004). To measure cognitive engagement, participants responded to the three-item subscale that assessed this dimension (Janosz & Bouthillier, 2007) (items: “I am willing to make efforts in mathematics,” “I am willing to devote time to mathematics,” and “I want to learn more about what we do in mathematics”) (M = 4.30, SD = 1.4, α = 0.92) (1 = strongly disagree to 6 = strongly agree).

To measure teacher humor, we used the Frymier et al. (2008) scale originally developed to measure teacher humor in college classes. We translated the instrument into French and adapted it for an elementary and secondary school context by removing a subscale (offensive humor) that was not appropriate for elementary and high school students (e.g., makes references to drinking or getting drunk in a humorous way, talks about drugs or other illegal activities in a humorous way). We preserved 21 items to measure four humor subscales: course-related humor (e.g., uses humor related to the course material) (M = 3.73, SD = 1.13, α = 0.86), course-unrelated humor (e.g., tells jokes unrelated to the course content) (M = 3.63, SD = 1.38, α = 0.84), self-disparaging humor (e.g., makes fun of him/herself in class) (M = 3.43, SD = 1.32, α = 0.78), and other-disparaging humor (e.g., picks on students in class for their intelligence) (M = 1.52, SD = 0.77, α = 0.86). We asked students to rate the degree of agreement regarding each item based on a six-point Likert-type scale (1 = strongly disagree to 6 = strongly agree).

To measure emotional well-being, psychological well-being scales are used, sometimes called “happiness scales.“ The Santé Québec survey used an abbreviated and considerably modified version of the General Well-Being Scale (Perreault, 1989), which has been referred to as the “Bien-Être Santé Québec” (BESQ) [Québec’s Health and Well-Being]. Seven dimensions of emotional well-being are explored: energy (e.g., I felt full of spirit and energy), control of emotions (e.g., it was easy for me to control my emotions), general mood (e.g., I felt in a good mood and light-hearted), interest in life (e.g., a lot of interesting things happened), stress (e.g., I felt sufficiently relaxed), health perception (e.g., I have not had any problems with my health), and emotional isolation (e.g., I felt loved and appreciated). For each of the dimensions, one item explores the positive affect in question; the instrument therefore has seven items (M = 3.88, SD = 1.10, α = 0.83). We asked students to rate the degree of agreement regarding each item based on a six-point Likert-type scale (1 = strongly disagree to 6 = strongly agree).

Preliminary analyses

First, we conducted preliminary analyses, which indicated an acceptable distribution of the data, homogeneity of variance, and the absence of multicollinearity. Following initial data processing, we removed a few outliers. We dealt with the missing data by proceeding with a technique called “maximum likelihood” (EM or expectation maximization). Since our data showed a very low percentage of missing data (5%), this technique correctly reflected the uncertainty of missing values ​​and preserved important aspects of distributions (Tabachnick & Fidell, 2013).

Main analyses

Structural equation modeling analyses (SEM) were performed to test the hypothesized associations presented in Fig. 1. A first hypothetical model is usually tested. To examine whether this model adequately fits the data, different fit indices are needed: chi-square (χ2), CFI, TLI, and RMSEA. As Hu and Bentler (1999) suggested, a good model should provide acceptable results on various fit tests. The global adjustment index used is χ2 (also called the chi-square likelihood ratio or generalized likelihood ratio). A nonsignificant value at the χ2 index generally reflects a good fit (Tabachnick & Fidell, 2013). Other indices have been used such as the CFI (comparative fit index) and the TLI (Tucker–Lewis index). Values ​​greater than or close to 0.95 for these two indices indicate an appropriate fit of the data (Hu & Bentler, 1999; Kline, 2016). The RMSEA (root mean square residual error of approximation) requires a value of 0.06 or less to be considered an adequate data fit (MacCallum et al., 1996).

From the various adjustment indices obtained while testing the hypothetical model, the modification indices (Lagrange multiplier) were used to improve the adjustment of the model; in modifying the hypothetical model, we made sure we respected the logic and consistency of the underlying theory (Perry et al., 2015). The preferred estimation technique in this research is maximum likelihood. Maximum likelihood is a commonly used estimation method for this type of analysis. According to Kline (2016), this method is unbiased in addition to being efficient and consistent.

In order to explore whether the associations under study varied according to the grade level of the students and their gender, we used a multigroup approach, as advocated by Byrne (2016), in a confirmatory approach to comparing models. This invariance procedure confirms the equality (or not) of the estimated parameters. To achieve this, we imposed equality constraints on the parameters of the models to check whether the models are equivalent according to the school grade of the students (primary school/high school) and their gender. These statistical procedures are clearly explained by Byrne (2016). Two indices are used to measure the invariance of the parameters: the chi-square difference and the CFI difference (Byrne, 2016). Since the use of both methods is still the subject of debate in the scientific community, and that “it is hoped that statisticians engaged in Monte Carlo simulation research related to structural equation modeling will develop more efficient and useful alternative approaches to this decision-making process in the near future” (Byrne, 2016, p. 307), we opted to report the X²-difference test knowing that more work needs to be conducted in this area (Byrne, 2016). To perform these statistical analyses, STATA software (version 17) was used.

Results

Descriptive and correlation statistics

Table 1 shows the descriptive statistics of the variables included in the model under study and the correlations between them. The means vary from 1.52 (other-disparaging humor) to 5.25 (behavioral engagement). All correlations were significant (p < .01), except for the associations between course-unrelated humor and the sense of school belonging (0.08, p = .11); course-unrelated humor and cognitive engagement (-0.06, p = .22); course-unrelated humor and affective engagement (-0.01, p = .89); self-disparaging humor and behavioral engagement (-0.01, p = .91); self-disparaging humor and affective engagement (0.06, p = .25); self-disparaging humor and cognitive engagement (0.03, p = .57); other-disparaging humor and cognitive engagement (-0.01, p = .88); and other-disparaging humor and course-related humor (0.03, p = .56). The significant correlations varied from weak (-0.08, p < .01) to strong (0.69, p < .01). Five variables displayed negative correlations with other-disparaging humor: well-being (-0.08, p < .01), the sense of school belonging (-0.21, p < .01), behavioral engagement (-0.21, p < .01), affective engagement (-0.18, p < .01), and cognitive engagement (-0.01, p < .01). Three variables displayed negative correlations with course-unrelated humor: behavioral engagement (-0.16, p < .01), affective engagement (-0.01, p < .01), and cognitive engagement (-0.06, p < .01). Our results showed that all other correlations were positive.

Table 1 Descriptive Statistics (Means and SDs) and Pearson Correlations among Study Variables
Full size table

Structural equation modeling

The results of the hypothetical framework (confirmatory factor analysis) (the latent factors under study) indicated a close acceptable fit with the data (X2[999] = 2361.850, p < .000; RMSEA = 0.059 (95%CI = [0.056, 0.062]); CFI = 0.865; AIC = 47641.951). Specifically, although the CFI value was slightly below the desired threshold of 0.90 (CFI = 0.865), the RMSEA test was below the threshold of 0.08, which is acceptable for some researchers (RMSEA = 0.059) (Hooper et al., 2008). Kenny & McCoach (2003) pointed out that the CFI fit index tends to deteriorate as the number of variables increases. This issue is quite common when validating complex models with many variables. Since the present structural equation model has nine latent variables and 57 observed variables, the model fit was considered adequate in the circumstances.

In Fig. 2, we illustrate the basic hypothetical model for examining the associations between our latent variables. More precisely, the latent variables, the sense of school belonging (Belonging) and well-being (WB), mediate the relationships made up of the different types of teacher humor (Related = humor related to course content, Unrelated = humor unrelated to course content, Selfdisp = self-disparaging humor, Otherdisp = other-disparaging humor) in order to explain the three types of school engagement (CogEng = cognitive engagement, AffEng = affective engagement, BehEng = behavioral engagement). Because our initial hypothetical model (Model 1) did not fit well according to the criteria mentioned above, we carried out a certain number of modifications to improve the model fit.

Fig. 2
figure 2

Hypothetical Model (Model 1)

Full size image

Subsequent models

With regard to the modification indices, five links were removed (well-being and behavioral engagement; self-disparaging humor and the sense of school belonging; self-disparaging humor and well-being; other-disparaging humor and well-being; unrelated humor and the sense of school belonging) because they were not significant. One association was added between the sense of school belonging and well-being because it was significant. In addition, a few error terms were correlated (57 and 59; 59 and 60; 57 and 65; 65 and 66; 64 and 65; 46 and 47; 43 and 44; 40 and 41; 84 and 85; 85 and 89; 89 and 91). In Model 2, the fit indices were all satisfactory (see Table 2). In the final model (Model 2), five nonsignificant associations were removed; all the other links of Model 1 were preserved, and one was added, because they were significant. Hence, the results suggest that course-related humor has a significant and positive association with the sense of school belonging (SE = 0.05, p < .001, β = 0.47, 95%CI (0.38 – 0.57)) and well-being (SE = 0.09, p < .001, β = 0.41, 95%CI (0.30 – 0.53)), that unrelated humor has a significant and negative association with well-being (SE = 0.05, p < .001, β = - 0.14, 95%CI (-0.24 – - 0.04)), and that other-disparaging humor has a significant and negative association with the sense of school belonging (SE = 0.05, p < .001, β = - 0.24, 95%CI (-0.33 – - 0.14)). For its part, the sense of school belonging has a significant and positive association with well-being (SE = 0.06, p < .001, β = 0.28, 95%CI (0.16 – 0.39)), and the three types of school engagement, namely cognitive engagement (SE = 0.06, p < .001, β = 0.26, 95%CI (0.17 – 0.40)), affective engagement (SE = 0.04, p < .001, β = 0.64, 95%CI (0.56 – 0.72)), and behavioral engagement (SE = 0.06, p < .001, β = 0.48, 95%CI (0.35 – 0.60)). Compared to the sense of school belonging, well-being has a significant and positive association with only two types of school engagement, namely affective engagement (SE = 0.04, p < .001, β = 0.35, 95%CI (0.26 – 0.43)) and cognitive engagement (SE = 0.06, p < .001, β = 0.29, 95%CI (0.17 – 0.40)). Additionally, the indirect association path coefficient between other-disparaging humor and the different types of school engagement and well-being through the sense of school belonging mediation was significant (p < .001), indicating that belonging fully mediates the relationships under study. Second, the indirect association path coefficient between unrelated humor and the different types of school engagement (affective and cognitive engagement) through well-being mediation was significant (p < .001), indicating that well-being fully mediates the relationships under study. Finally, the indirect association path coefficient between course-related humor and the three types of school engagement (behavioral, affective, and cognitive engagement) through belonging mediation and well-being mediation was significant (p < .001), indicating that belonging and well-being fully mediate the relationships under study except for the association between related humor, well-being, and behavioral engagement.

Fig. 3
figure 3

Final Model (Model 2)

Full size image
Table 2 Results for Models 1 and 2
Full size table

Invariance analysis

A multigoup analysis was conducted with an X²-difference test to determine whether the associations between course-related humor and the sense of school belonging, and other-disparaging humor and the sense of school belonging, were invariant regarding gender. The X²-difference test was statistically nonsignificant between other-disparaging humor and the sense of school belonging (p = .867, p < .05), meaning that there is no difference between males and females. However, the X²-difference test was significant between course-related humor and the sense of school belonging (p = .04, p < .05), meaning that there is a difference concerning males and females. Further analysis determined that it is significantly stronger for females (β = 0.32) than for males (β = 0.24) (Hypothesis 6). In regard to students attending elementary schools and high schools, the X²-difference test determined that the association between course-related humor and the sense of school belonging was not significant (p = .32, p < .05), meaning that there is no difference between high school students and elementary school students. The X²-difference test also indicated that there is no difference between elementary school students and high school students regarding the association between other-disparaging humor and the sense of school belonging (p = .09, p < .05), making this association invariant regarding grades (Hypothesis 5).

Discussion

The objective of this study was to determine how teachers’ use of humor (e.g., course-related humor, course-unrelated humor, self-disparaging humor, other-disparaging humor) relates to students’ sense of school belonging, engagement, and emotional well-being, and to test for invariance across elementary and high school students as well as across males and females. Our hypotheses were based on the instructional humor processing theory (IHPT) derived from the work of Wanzer et al. (2010), as well as on many studies exploring teacher humor, students’ sense of school belonging, and school engagement (Bolkan & Goodboy, 2015; Goodboy et al., 2015; Imlawi et al., 2015). Although Bolkan and Goodboy (2015) considered students’ affects as a mediator of the relationship between teacher humor and engagement, no study has explored jointly the sense of school belonging and emotional well-being. Our results are a contribution to the IHTP (Wanzer et al., 2010) and indicate, for the first time, how different types of humor can influence the sense of school belonging, engagement, and emotional well-being, and thus answer research questions that have not been empirically studied.

The first hypothesis assumed that the different types of teacher humor (course-related humor, course-unrelated humor, self-disparaging humor, other-disparaging humor) were positively associated with the sense of school belonging. This hypothesis was partially supported as not all types of humor were associated with the sense of belonging. Our results indicated that, among the four types of teacher humor that we measured, only humor related to course content (positive significant) and other-disparaging humor (negative significant) had a significant association with the sense of school belonging, which can be categorized as adequate and inadequate humor, respectively (Frymier et al., 2008). The relation between course-related humor and the sense of school belonging may be due to the emotional component of the latter. Researchers have suggested that belonging may refer to having emotional attachments, feeling intimacy, feeling needed, feeling useful and supported, feeling proud to attend the institution, and, finally, feeling good (St-Amand et al., 2017c). The work of Baumeister and Leary (1995) and the more recent work of St-Amand et al. (2017b) support the idea that the presence of a sense of school belonging involves positive emotions such as happiness, satisfaction, enthusiasm, and a state of calm. While course-related humor can trigger positive emotions, it is possible that other-disparaging humor (e.g., picks on students in class for their intelligence) has the opposite effect. As demonstrated by Bieg et al. (2019), teachers using humor associated with course content contributes to weakening the decrease in enjoyment and the increase in boredom and anger.

Our results also showed that the sense of school belonging was positively associated with the three types of school engagement (cognitive, affective, and behavioral) (H2), as well as with emotional well-being. These relationships can be explained by the fact that the sense of school belonging is a generator of several emotions that underlie students’ motivation to learn (Baumeister & Leary, 1995). St-Amand et al. (2020b) empirically demonstrated that the positive emotions derived from an emotion such as the sense of school belonging can partially explain the relation between belongingness and school engagement. However, it is also possible that achievement emotions may play a central role in this relationship. In recent years, researchers have documented the notion of “achievement emotions,” which are defined as “emotions that are directly linked to achievement activities or achievement outcomes”. Pekrun et al. (2002) suggested that emotions can be generated while attending school, simply by studying or completing exams. Thus, it is possible that the sense of school belonging influences achievement emotions as well as students’ emotional well-being depending on the context in which it all takes place (e.g., class-related, learning-related, and test-related emotions) and that these emotions drive school engagement.

While it is tempting to believe that all types of humor can be effective and supportive in a classroom setting (St-Amand et al., 2021a), our results clearly indicate that only humor associated with course content is effective in developing students’ sense of school belonging and emotional well-being (H3). While it appears detrimental to use other-disparaging humor, we do not consider the other two types of humor to be completely inappropriate (self-disparaging humor and course-unrelated humor). As Bieg et al. (2019) noted, the use of these two types of humor is probably not advantageous to consider in one’s instructional planning, especially for providing emotional experiences that contribute to the development of students’ sense of school belonging and emotional well-being. On the other hand, both types of humor can have other potential roles to play, such as building students’ resilience in their educational journey (Bondy et al., 2007). Clearly, more research is needed to investigate these relationships.

Except for behavioral engagement, our results showed that emotional well-being is positively and significantly associated with affective and cognitive engagement (H5). These results refer to the numerous studies that have shown that positive emotions (e.g., enjoyment) play a key role in students’ school engagement (Kwon et al., 2017; McKeering et al., 2021). The nonsignificant association between emotional well-being and behavioral engagement is likely due to the nature of the variable itself. Indeed, behavioral engagement was essentially measured from a perspective related to the adoption of norms, values, and respect toward the school’s code of life. To get a more complete picture of the phenomenon, it would be desirable to measure behavioral engagement by taking into account behaviors directly related to the learners’ effectiveness (e.g., raising a hand to ask a question, completing homework, etc.) (Nguyen et al., 2018).

Finally, we hypothesized that the association between the different types of teacher humor and the sense of school belonging is stronger for high school students than for elementary school students (H5), and stronger for females than for males (H6). Hypothesis 5 could not be validated by our data, although the understanding of teacher humor evolves as the children get older (Martin, 2010). Indeed, developmental stages do exist in the production and understanding of humor. Beginning at age 7, for instance, children discover that words can have double meanings and their jokes become more complex and varied (Nwokah et al., 2013; Sahayu et al., 2022). Most of our youngest participants were in sixth-grade classrooms, so older than seven, which is not a big difference from much of our older sample, in this case middle and high school students. The invariance of these results can therefore be explained by the nature of our sample. As for Hypothesis 6, our results indicated that the association between course-related humor and the sense of school belonging was stronger for females than for males. Across elementary, middle, and high school levels, girls are generally more engaged than boys in school (Marks, 2000) and likely more responsive to the determinants of school belonging, which may explain these results.

Conclusion

As in all studies, this one has limitations and other research avenues that we would like to highlight. First, our sample is made up of students from public schools situated in rural Canada, and one private secondary school situated in an urban area. This situation prevents us from generalizing our results in major cities across the country and among indigenous communities. Studying these relationships across different cultures would provide a more comprehensive look at the phenomenon under study (Davies, 2003; Jiang et al., 2020). Second, even though the internal consistency of our scales appeared adequate, it would have been useful to use a scale such as the PSSM (Psychological Sense of School Membership scale), which measures other dimensions of the sense of school belonging, such as the quality of social relations (St-Amand et al., 2020a). The fact that the scales were all self-reported and based on self-observations limited our view of the phenomenon. In this sense, it would be beneficial to add analytical perspectives such as observation, classroom recordings, or critical incident methods. On the question of causality, and to validate the results of the present study, it would be appropriate to conduct a study in a controlled environment with an experimental design. Third, we measured this phenomenon at a single measurement time. A longitudinal design would allow us to study the evolution of these relationships, while seeking to identify the most at-risk periods during the school year or the periods when humor has the greatest impact on students’ sense of school belonging, engagement, and emotional well-being. Fourth, as researchers have found that the level of emotional well-being varies as students age, testing for invariance regarding emotional well-being and its determinants such as teacher humor would be appropriate in future studies (McKeering et al., 2021). Finally, researchers have demonstrated that school belonging can be triggered by specific instructor characteristics, such as encouragement of student participation and quality interaction (Freeman et al., 2007). Understanding where these practices fall in relation to teacher humor would be beneficial in maximizing teaching effectiveness.