Abstract
Teachers’ attributions of student behavior are associated with the decisions they make about how to respond to behavior problems and support their students. This exploratory study investigates teachers’ attributions of student externalizing behavior, how attributions vary as a function of teacher and school characteristics, and the association of those attributions with teachers’ perceived likelihood of referring students to mental health services or implementing punitive discipline. We provided a sample of U.S. secondary school teachers (N = 426) a vignette depicting a student with oppositional defiant disorder and asked them to describe what was happening with the student. Attributions were categorized as internal to the student (e.g., mental health), external (e.g., problems at home), both (internal and external), and neutral (e.g., “normal teen behavior”). Results indicated the majority of teachers attributed vignette behavior to external factors. Results suggest that teachers indicating both internal and external attributions had increased likelihood of providing a mental health referral. Teacher and school characteristics were also associated with attributions. Results provide support for the importance of considering teachers’ attributions of student behavior and the association of these attributions to pathways into mental health services.
Responding to student behavioral problems is one of the most challenging aspects of teachers’ jobs, and a key reason for leaving the profession (Feng, 2009; Skaalvik & Skaalvik, 2017). This may be, in part, because teachers report a lack of confidence, knowledge, and training in how to provide behavioral supports (Reinke et al., 2011; Walter et al., 2006). In particular, teachers report being uncertain about how to understand the cause of student behavioral problems (Rothì et al., 2008). This uncertainty may result in the under-utilization of school-based support services and misapplication of punitive discipline (e.g., detentions, demerits, suspensions; Marzano et al., 2003). Prior studies have found teachers’ attributions about the causes of student behavioral problems may be linked to their motivation to act to support students (Bertrand & Marsh, 2015), and may influence whether teachers recognize behaviors as indicative of a need for mental health services (Christenson et al., 1983; Loades & Mastroyannopoulou, 2010).
Extant research has found that teachers attribute student misbehavior to a variety of factors, both internal (Guttman et al., 1982; Zakaria et al., 2013) and external (Paramita et al., 2020; Savina et al., 2014) to the student. Although prior research provides important insights into how teachers attribute behavior and how attributions may relate to their subsequent responses, most prior research has focused on elementary students (Gibbs & Gardiner, 2008; Nemer et al., 2019; Wang & Hall, 2018). More research is needed on teacher attributions regarding adolescent behavior, especially because the prevalence of mental health problems increases in adolescence (Merikangas et al., 2010). As such, it may be difficult for teachers to identify whether behaviors are reflective of “normal” variation in teenage behavior, or are a sign of needed support. Teachers play key roles in referring adolescents for mental health services (Farmer et al., 2003; Stiffman et al., 2000) and in disciplinary referrals. Additionally, many contextual factors differ between elementary school settings and that of higher grade levels, such as class size, student-to-teacher ratios, student workload, and developmental needs, that likely affect the way teachers view behavior. Some prior research has indicated that the use of student supports and services declines at higher grade levels (Green et al., 2018; Wagner et al., 2006). Given these differences, an explicit focus on middle and high school teacher attributions of student behavior is needed. A deeper understanding of secondary teachers’ attributions of adolescent misbehavior could provide insight into one pathway through which students may be more or less likely to receive school mental health services or punitive discipline (Bibou-Nakou et al., 2000).
Weiner’s (1972, 1985) Attribution Theory serves as a framework with which to understand teachers’ perceptions of student behavior and how causal attributions may contribute to their likelihood of employing various intervention strategies. Attribution Theory describes how people perceive seemingly ambiguous behaviors of others, and then make meaning or causal inference about those behaviors. Attribution Theory posits there are three dimensions of causal attributions: (a) locus of control (internal or external), (b) stability (unstable/temporary or stable/permanent), and (c) controllability (controllable or uncontrollable; Weiner, 1979). Education researchers have often used Attribution Theory to conceptualize teacher attributions of student misbehavior (e.g., Andreou & Rapti, 2010; Bibou-Nakou et al., 2000; Cothran et al., 2009; Hughes et al., 1993). In a systematic review of teacher attribution studies, Wang and Hall (2018) found teachers tended to attribute student misbehavior to internal student characteristics (e.g., effort, personality) or external student characteristics (e.g., home dysfunction, lack of parental interest), instead of teacher- or school-based influences (e.g., instruction, teacher attitudes).
Attribution Theory does not position some attributions as “better” than others, rather it provides a framework for understanding a range of teacher perceptions of student behavior. These causal attributions may have important implications for how teachers respond to student behavior. For example, Hart and DiPerna (2017) found that when teachers were provided with context for the student’s misbehavior, they were more likely to have positive emotional responses, as compared to teachers who were not provided context. Weiner (1980) theorized that positive emotional responses from teachers could lead to a greater willingness to intercede or seek outside help; conversely other scholars have found teachers may look outside the classroom to professional support for students they perceive to be difficult to teach (Soodak & Podell, 1994). Ultimately, teacher’s attributions may serve as a mediator between student behavior and access to supportive responses, including referrals to mental health services (Zhou & Urhahne, 2013).
Teachers’ attributions of student behavior and decisions about whether to refer students for mental health services may also be influenced by their own perceptions and knowledge of mental health and services, and by factors related to the student and school context (Stiffman et al., 2004). According to the Gateway Provider Model (Stiffman et al., 2004), teacher decision-making is influenced by factors including (a) students’ need for services, (b) student predisposing characteristics (e.g., gender, race/ethnicity, risk factors), (c) enabling factors (e.g., service availability and acceptability), and (d) structural characteristics (e.g., characteristics of the organization). Prior studies have focused on students’ need for services and student predisposing characteristics, finding that both have important associations with whether teachers provide mental health referrals (Green et al., 2018; Headley & Campbell, 2011; Loades & Mastroyannopoulou, 2010; Splett et al., 2019). Less well known is how organizational structures of schools may be associated with whether teachers decide to respond supportively or punitively. These organizational characteristics might provide important information about systematic variation in teachers’ perceptions of and responses to youth behaviors.
In this study, we examine attributions teachers assign to externalizing student behavior by presenting teachers with vignettes describing a student with behaviors that are characteristic of oppositional defiant disorder (ODD). ODD is one of the most common externalizing behavior disorders identified among adolescents and often has a first onset in adolescent years (Merikangas et al., 2010; Nock et al., 2007). Studies document that ODD is comorbid with internalizing disorders and can also “mask” the presence of internalizing problems in children (Boylan et al., 2007). As such, middle and high school teachers have the potential to identify the behavioral challenges associated with ODD early on in their course and provide students with mental health supports.
Due to many confounding classroom- and student-specific variables, it can be difficult to isolate teachers’ attributions in real-life contexts. We selected a vignette method, because it has been used in prior studies of student behavior and allows researchers to test select variables, while holding other variables constant (Loades & Mastroyannopoulou, 2010). In this way, student-specific variables (e.g., gender, severity of behavior) can be isolated to observe their impact on teacher’s causal attributions. Teachers in the current study were asked to read a vignette and complete a survey regarding their attributions of the behaviors presented. We specifically focus on whether teachers attribute student behavior to internal factors, external factors, both internal and external factors, or “normal” adolescent behavior. We anticipate that an attribution to internal and/or external factors is more likely to activate a response than attributions to “normal” behavior, as the Gateway Provider Model (Stiffman et al., 2004) suggests that teacher response is partially reliant on their identification of student need. However, we also anticipate that there may be important differences between internal attributions (situating the locus of control as within the child) and external attributions (situating control with external factors), as studies indicate that teachers are more punitive when they believe that students are responsible for their behaviors (Wang & Hall, 2018). Furthermore, we hypothesize that teachers who make attributions to factors that are both internal and external may be open to a wider range of interpretations of student behavior and perhaps more likely than their colleagues to seek support for students.
Next, we study the association between organizational characteristics and attributions to internal causes (vs. all others) and external causes (vs. all others). We focus specifically on internal and external factors because of prior research underscoring the implications of these two attributions (Wang & Hall, 2018), yet limited research on contextual and organizational characteristics that may be associated with their use. We therefore aim to answer the following research questions:
Method
Participants
Participants were middle and high school teachers recruited from a sample of schools across the United States (Green et al., 2022). To identify teachers, we downloaded a list of all U.S. public middle and high schools from the Department of Education website and stratified by region of the country and urbanicity. Within each stratum, we randomly selected schools using a random number generator, with over-selection in large cities. Once a school was selected, a research assistant searched for a publicly-available list of teacher email addresses through the school website. If there was no publicly available list of email addresses, the school was excluded. If an email list was available, we used a random number generator to select one general education teacher. We sent selected teachers a postcard informing them of the study and that they would receive an invitation via email. We then sent teachers an email invitation to complete a one-time online survey. Non-respondents were sent up to two follow-up emails and a survey by mail. If there was no response, we randomly selected a second person from the same school to contact. If they did not respond to the same series of contacts, we randomly selected a replacement school. Participants received a $15 gift card for participation.
A total of 1,386 teachers were invited to participate. Overall, 501 teachers completed the survey online (n = 322, 64.3%) or by mail (n = 179, 35.7%) representing a 36.1% response rate. Analyses included only teachers who provided complete responses to the externalizing vignette and all variables that were the focus of analysis (N = 426). Consistent with U.S. national data (Taie & Goldring, 2017), teachers primarily identified as female (68.3%) and non-Latinx White (91.3%; see Table 1). The university Institutional Review Board approved all study procedures.
Individual Teacher Characteristics and School Characteristics (N = 426).
Multiple responses could be selected, so percentages add to more than 100%.
Vignette Measure
Vignette construction
The vignette used in the current study was adopted from research conducted by the Center for Multicultural Mental Health Research (Chavez et al., 2010; Lapatin et al., 2012). As the original vignette was designed to study parent perceptions of child need for mental health services, we adapted the vignette to be relevant to school contexts. Nineteen middle and high school teachers from school districts in the Northeast reviewed the vignette in a series of focus groups; teachers provided feedback on the text and description of students (e.g., recommending that text focus on behaviors observable in a classroom; Green et al., 2018). We developed two versions of the vignette: one reflecting moderate and one reflecting severe behavior problems (see Appendix Table 1). Additionally, the vignettes were purposely designed to include descriptions of behavior, but to not provide information that may indicate a specific attribution for student behavior. These final vignettes were 100–130 words with symptom descriptions based on ODD criteria from the Diagnostic and Statistical Manual of Mental Disorders-Fifth Edition (DSM-5; American Psychiatric Association [APA], 2013). Subsequently, we validated the vignettes using data from 14 clinical and research experts in child and adolescent mental health and data from 100 school psychologists recruited from the National Association of School Psychologists mailing list (Green et al., 2022). This validation process was used to confirm that the “severe” vignettes reflected greater impairment than the “moderate” vignettes and that the vignettes were considered realistic.
In the current study, teachers were randomly assigned to read a vignette with a female (Anna) or male name (David). These two names were selected because were previously identified as names that are commonly used in both English and Spanish and are not typically associated with a single race or ethnicity but are typically associated with female and male gender (Lapatin et al., 2012). No other information about student demographic characteristics was provided. Teachers were asked to imagine the student being in their school; schools varied considerably in terms of their age-range and racial/ethnic composition and this method was selected to make the vignette seem relevant to individual respondents in their school context.
Attribution question
After reading the vignette, teachers were asked “What, if anything, is going on with [Anna/David]?” Teachers were given an open box for a free-form response. This is similar to a question used by Jorm et al. (2010) to assess teacher recognition of depression following presentation of a vignette describing a student with depressive symptoms.
Coding
Our research team created a codebook to code responses to the attribution question described above. First, three team members independently read 20 responses and met to discuss common categories that we identified among respondents. Once we established a common set of categories, we read 20 more responses independently and reconvened to discuss how this set of responses fit into our existing categories, what categories might need modification, and what new categories we needed. The codebook ultimately included 12 codes, described and defined in Table 2.
Codes, Their Definitions, and Frequency.
Once we established our codebook, we recoded all responses coded previously. Each response was independently coded by two team members who met weekly to discuss codes and questions. Independent codes were compared until a reliability kappa of .90 was reached (at which point we had coded 80 responses). We divided the remaining responses and each coder independently coded responses, meeting weekly to discuss codes and questions. Responses could be assigned multiple codes (e.g. both an internal and an external attribution).
Mental Health Service Referral and Punitive Disciplinary Response
To assess two common responses to student behavioral problems, we asked teachers whether they would refer the student in the vignette for mental health services and whether they would use a punitive disciplinary response. Mental health service referral was assessed by asking teachers “Using a scale of 1–10, where 1 is very unlikely and 10 is very likely, how likely would you be to refer [Anna/David] to a school counselor, social worker, or psychologist for mental health services?” Teachers rated their response on a scale of 1 to 10. Punitive disciplinary responses were assessed by asking teachers how likely they would be to “Give [Anna/David] a detention or demerit” and teachers rated themselves as “extremely unlikely,” “unlikely,” “likely,” and “extremely likely.”
Teacher Demographics
Teachers indicated their gender (male, female, another gender identity), race, ethnicity, and number of years that they had been teaching.
School Characteristics
We extracted data from the U.S. Department of Education Common Core of Data (https://nces.ed.gov/ccd/) to measure school demographic characteristics. These included region of the country based on U.S. Census regions (Northeast, Midwest, South, West), school urbanicity (City, Suburb, Town, Rural), Title I funding status, school level (middle, high), and student racial/ethnic composition (percent of students identifying as Black, Hispanic/Latino(a), and White). For descriptive analyses, racial/ethnic composition was further categorized as majority Black, majority Hispanic/Latino(a), and majority White, when ≥50% of students were identified as being of that race/ethnicity. Schools in which there was no majority were categorized as non-majority schools.
Analysis
To identify the most common teacher attributions for behaviors characteristic of ODD, we first reported the number and percent of teacher respondents whose responses to the attributions question were consistent with each of the 12 categories. We collapsed these 12 categories into three overarching categories: internal attributions, external attributions, and neutral attributions (e.g., attributions to “normal teen behavior”). There were 33 participants whose attributions were coded only into the internal category (e.g., “David is experiencing anger management issues”), 195 participants whose attributions were coded only into the external category (e.g., “problems at home. . .perhaps divorce of parents”), 84 with responses that were coded into both the internal and external category (e.g., “something emotional—maybe changes at home, possibly drug or alcohol abuse”), and 21 with responses that were coded in the neutral category, with or without other codes (e.g., “puberty/possibly problems at home causing anger/possible eating disorder”). In addition, there were 93 participants whose responses were not coded into any of these categories (e.g., “something in her life has changed,” “something is going on with Anna,” “not sure”). The internal and external attribution categories are consistent with prior research on teacher attributions of student behavior (Nemer et al., 2019). Next, to examine the association of teacher attributions with the likelihood teachers would refer a student for mental health services or give a detention/demerit, we created binary variables indicating whether each response was coded as an internal attribution only, external attribution only, both internal and external attribution, or neutral (“normal” teenage behavior). We entered three categories of attributions (internal, external, and both) into a series of regression models, with the neutral category as the reference group. Responses not included in one of these categories were excluded from the analysis (leaving a sample size of n = 333). First, a linear regression model allowed us to test the association of attributions with rated likelihood of mental health referral (rated on a scale of 1–10), controlling for vignette severity and gender. Second, using an ordinal regression, we estimated the association of internal only, external only, and internal and external attributions (using “neutral” as the reference group) with likelihood of giving a detention/demerit (rated on a scale of one to four). Given the small number of ordered categories in the detention/demerit variable (four categories), we selected an ordinal regression procedure, controlling for vignette severity and gender.
Next, to test the association of student vignette demographics, teacher characteristics, and school organizational and structural factors with teacher attributions, we constructed two multivariate logistic regression models in which these factors predicted whether teachers’ responses included attributions coded as (a) internal and (b) external. As teacher responses could be coded into multiple categories, these logistic regression models were estimated as separate analyses. For this analysis, we used the full sample of 426 participants.
Results
Attributions
We identified teachers’ attributions of student behaviors as reflecting four internal attributions, six external attributions, and two neutral attributions. All of the identified attributions can be found in Table 2. The most commonly coded attribution was Home, an external attribution coded in 62.2% of teacher responses. The next most frequently coded attributions were Peer Relationships and Drug and Alcohol Use, which were coded in 13.1% and 15.5% of teacher responses, respectively. Less common were attributions to ODD (0.7%), Mental Health (10.1%), and related terms, such as Anger (6.3%).
Association of Attributions With Teacher Reported Likelihood of Mental Health Referral and Punitive Discipline
On a scale of 1 to 10, teachers reported a high likelihood of referring the student in the vignette for mental health services (M = 8.12, standard deviation [SD] = 2.00). Adjusting for vignette gender and severity, results of a linear regression model indicated that participants whose responses were coded as both internal and external attributions had significantly higher reported likelihood of mental health service referral (B = 1.03, standard error [SE] = 0.48, b = 0.23, t = 2.17, p = .03; see Table 3), as compared to those with neutral attributions. Specifically, participants whose responses were coded as both internal and external had a mean rating for likelihood of mental health referral of 8.5 (SD = 1.8), compared to participants with responses coded as a neutral attribution who had a mean rating for likelihood of mental health referral of 7.5 (SD = 1.8; Cohen’s d = -0.6 medium effect size). Codes of internal only and external only attributions were not significantly associated with teacher-reported likelihood of referral for mental health services. A sensitivity analysis using ordinal regression produced results consistent with these findings.
Association of Codes With Likelihood of Referral for Mental Health Services or a Detention/Demerit.
Analyses only include participants whose responses were coded as internal only, external only, both internal and external, or neutral. bNeutral is the reference group.
p < .05.
On a scale of one to four, teachers reported a low likelihood of giving the student in the vignette a detention or demerit. Over two-thirds of teachers (67.6%) said that they were unlikely or extremely unlikely to give a detention/demerit. Internal and external attributions, when compared to neutral attributions, were not significantly associated with likelihood of giving a detention or demerit (see Table 3).
Association of Attributions With Teacher and School Factors
A series of logistic regression models tested the association of teacher responses (internal and external attributions) with teacher and school characteristics, adjusting for vignette gender and severity. When holding other factors constant, teachers who taught for 10 or more years had almost three times greater odds of providing internal attributions than teachers who taught for fewer than 10 years (odds ratio [OR] = 2.97, confidence interval [CI] = 1.67–5.25; see Table 4). The only school characteristics associated with internal attributions were school level and Title I status. High school teachers had about two times greater odds of providing responses coded as internal attributions than middle school teachers (OR = 1.96, CI = 1.22–3.17). Teachers in Title I schools also had almost two times greater odds of providing responses coded as internal attributions, as compared to teachers from non-Title I schools (OR = 1.89, CI = 1.11–3.21).
Association of Vignette Factors, Individual Teacher Characteristics, and School Characteristics With Teacher Use of Internal and External Attributions (n = 426).
Note. OR = odds ratio; CI = confidence interval.
p < .05. **p < .01. ***p < .001.
Teacher responses coded as external attributions were significantly associated with two teacher racial/ethnic demographic variables. As compared to teachers identifying as non-Latino(a) White, Hispanic or Latino(a) teachers had more than three times greater odds of providing a response coded as an external attribution (OR = 3.13, CI = 1.15–8.46). In contrast, Black teachers were less likely to provide an external attribution than non-Latino(a) White teachers (OR = 0.27, CI = 0.10–0.72), which is equivalent to White teachers having about 3.7 times greater odds of providing an external attribution. In terms of school characteristics, teachers in rural schools were less likely to provide external attributions than teachers in urban schools (OR = 0.37, CI = 0.18–0.76), which is equivalent to teachers in urban schools having about 2.7 times greater odds of providing an external attribution.
Discussion
Schools are uniquely situated to address the mental health needs of adolescents, as they have the ability to circumvent barriers associated with community-based mental health service access (Duong et al., 2020). Teachers can play a crucial role in referring students to school-based mental health service providers when behaviors indicate a mental health problem (Farmer et al., 2003; Stiffman et al., 2000) and teachers’ attributions about student behaviors may contribute to how they fulfill this role (Christenson et al., 1983; Splett et al., 2019). Within this study, we used vignettes to examine secondary teachers’ attributions of behaviors that were consistent with a DSM-5 (APA, 2013) diagnosis of ODD. We found teachers seldom identified these behaviors as symptomatic of ODD, or even a mental health problem, generally. More often, teachers attributed these behaviors to factors external to the student, specifically problems at home. These findings are consistent with extant literature that highlights teachers’ attributions of external factors, including home, family, and parental influence on adolescent externalizing behavior (Croll & Moses, 1985; Paramita et al., 2020; Savina et al., 2014; Vernberg & Medway, 1981). Failure to recognize a mental health problem may contribute to insufficient teacher response and fewer pathways to mental health support for adolescent externalizing behaviors.
Furthermore, teachers’ attributions of behavior may be linked to their choices of how to respond to student behavior (Hart & DiPerna, 2017). In the current study, our results provide preliminary evidence that teachers who attributed behaviors to both internal causes (e.g., mental health, anger, substance use) and external cause (e.g., home problems, peer relationships) were significantly more likely to say they would make a mental health service referral compared to teachers who provided a neutral attribution. Unlike prior studies (as described by Wang & Hall, 2018), we did not find a distinction in likelihood of referral between teachers who made only internal or external attributions, suggesting that teachers in the current sample may not be specifically attuned to the locus of control of student behaviors when making decisions about referrals. Instead, these results suggest teachers who hold more open views of the causes of student problems or who are more open to the wide range of possible causes may be more likely to refer students to mental health providers. As such, our findings suggest there is no clear “ideal” attribution of student behavior, but teachers who understand that student behavior can be influenced by a myriad of factors may be more likely to connect them with resources. These results lead us to suggest that teachers be prepared to consider multiple potential attributions underlying student behavior and, ideally, make mental health referrals that are based on the behavior itself, regardless of those attributions. Additionally, the possibility that teachers have a range of attributions that may influence their decisions points to the need for schools to implement systematic mental health screenings (Briesch et al., 2018) to ensure that all students in need of mental health services receive support.
Results also indicate that teacher attributions of student externalizing behaviors varied in relation to teacher- and school-level characteristics. Teachers were more likely to make internal attributions than all other attributions when they taught in a high school, as compared to a middle school. These results suggest teachers may perceive acting out behaviors among older students as more likely rooted in mental health problems. Furthermore, teachers with more years of experience were more likely to assign an internal attribution to student behavior, perhaps reflecting increased knowledge of mental health with experience. In examining external attributions, compared to non-Latino(a) White teachers, teachers who identified as Hispanic or Latino(a) were more likely to provide external attributions, while those who identified as Black were less likely to make external attributions. Extant research does not clearly indicate why this is the case, but there may be cultural differences in attributions regarding behavior, an important area for future research. Teachers were also less likely to make an external attribution when they worked in rural vs. urban schools. Rural schools are more likely to hire teachers who grew up locally (Boyd et al., 2005), and thus teachers may be more knowledgeable about students’ lives outside of school, and less likely to make external attributions without specific information. Another possibility is that cultural attitudes toward student behavior may systematically vary by urbanicity (Kulinna et al., 2006) and teacher perceptions of their school environment (Pas & Bradshaw, 2014). Finally, teachers were more likely to make external attributions when they worked in Title I vs. non-Title I schools. It is possible that students in Title I schools experience more challenges in their lives outside of school (Evans & Cassells, 2014), resulting in a higher propensity for teachers to make external attributions for student behavior.
Without a strong understanding of mental health problems, middle and high school teachers may not regularly refer students in need of mental health support to school psychologists or other school-based services (Stiffman et al., 2004). The finding that few teachers identified externalizing behaviors as symptomatic of a mental health problem is a concern, indicating a need for increased mental health literacy among school staff. School leaders and policymakers may consider investing in targeted professional learning experiences to ensure staff can recognize signs of mental health problems. This is especially true during and following the COVID-19 pandemic, as mental health problems among school-age children have increased (Magson et al., 2021).
School-based mental health service providers are uniquely positioned to address the mental health service delivery gap, given their expertise in effective mental health services, and their collaborative work with teachers. Future studies could focus on the role of school-based mental health service providers in supporting teacher responses to student behavior.
Limitations
Despite multiple attempts to engage non-responders, only 36.1% of teachers who we contacted completed the survey. Although this is higher than many other web-based surveys (Van Mol, 2017), teachers who chose to participate may have been those most interested in supporting student mental health.
To assess teacher attributions, we provided participants an open-response text box to describe their perceptions of student behavior. This approach has been used by others (Jorm et al., 2010); however, the length and quality of teacher responses varied considerably, and it is possible that teachers may have felt limited by the size of the box. Future research could use interviews, which would allow for follow-up questions, clarification, and elaboration. In addition, the relatively low R-squared value, particularly for the analysis of likelihood of providing mental health referrals and detentions/demerits, suggests that the variables analyzed in the current study account for a relatively small amount of variance in the outcomes. For example, our current dataset provides limited context regarding mental health services available in teachers’ school communities, and it is likely that access to services within the school or community may play a role in how teachers think about and respond to mental health problems. Additionally, our study does not explicitly address how teacher’s own skills, self-efficacy, and knowledge of behavior may impact their decision to refer to mental health services. More robust information about communities in which teachers work and teacher-level factors could help explain some of the differences we noted.
Furthermore, our data required us to make a number of analytic decisions. For example, in the analysis of teacher and school characteristics we compared internal attributions to all other attributions, and external attributions to all other attributions. However, there may be teacher and school characteristics associated with the provision of different combinations of internal and external attributions (e.g., internal compared to external, both internal and external compared to neutral) that could be the focus of future study. As our approach to this research was largely exploratory, given the lack of previous work on this topic, we acknowledge that different analytic decisions would yield different results. There are many ways that we could have elected to combine variables and design an analysis plan. Future research can build on and consider replicating the preliminary results from the current study, to continue to test the association of internal, external, and neutral attributions with teachers’ referral decisions and teacher and school characteristics.
Vignettes are the most common method used to measure teachers’ attributions (Nemer et al., 2019). Our use of vignettes allowed us to systematically vary the gender of students and the severity of behavior. Nevertheless, results are limited by the fact that teachers were presented with hypothetical scenarios and asked to report on their potential responses. In their study of staff attributions of challenging behavior, Lucas et al. (2009) found participants’ attributions of real-life behavior differed from the same behavior later assessed using a vignette-based method. It is likely that, among other factors, the nature of specific incidents and student-teacher interactions, teacher and student level characteristics, and the intersection of student and teacher characteristics play a role in how teachers attribute and respond to behavior. Future research on secondary teachers’ attributions of externalizing behavior could use real-life situations to understand how context contributes to identifying mental health problems. Furthermore, we focused on attributions of locus of control, the most common attribution assessed in research (Nemer et al., 2019), but future research could also study attributions of stability and controllability.
Conclusion
The current study contributes to a small body of literature assessing teachers’ attributions of student behavior in their classrooms (Nemer et al., 2019; Wang & Hall, 2018). We specifically add to prior research by introducing an analysis of teacher and school characteristics associated with teachers’ attributions. Results suggest there may be important systematic differences in how teachers make attributions about behavior. These characteristics potentially affect decision-making about how teachers will (or will not) support students, consistent with the Gateway Provider Model (Stiffman et al., 2000). In the future, it will be important to examine contextual factors that influence pathways to care or punitive disciplinary responses, to understand the interaction of teachers’ personal attributions and the schools and communities in which they serve students.
Footnotes
Appendix
Oppositional Defiant Disorder Vignettes. a
| Oppositional defiant disorder: Moderate (female) | Oppositional defiant disorder: Severe (male) |
|---|---|
| Anna is a student in your class. She has a longstanding group of friends, of which she is the leader. Anna has above-average intelligence, is very articulate, and has maintained good grades in school without working hard. Over the past six months, Anna has started to lose her temper a lot. She often appears angry, and she frequently blames others for her mistakes. Lately Anna’s friends have begun to stay away from her because she is overly sensitive, becomes annoyed easily, and at times she seems to try to annoy her friends on purpose. You have noticed that Anna argues with you more than other students her age. | David is a student in your class. He has a longstanding group of friends, of which he is the leader. David is very articulate and has maintained good grades in school without working hard. In the last few months, David has increasingly argued with his friends and put them down. He often appears angry, and he frequently blames others for his mistakes. David’s friends have begun to stay away from him because he has been overly sensitive, becomes annoyed easily, and at times he seems to try to annoy his friends on purpose. David defies his teachers, and he refuses to comply with adults’ requests and rules. He argues loudly with his teachers, more than other students in his classes. Recently, his teachers have often had to place David in detention. |
Names were systematically varied to indicate student gender.
Acknowledgements
We are grateful for assistance from the National Association of School Psychologists to conduct this research. We are also thankful for the contributions of members of our research team, including Javier Guzmán, Shannon Gribben, Melissa Holt, Rachel Oblath, and Suzanne Vinnes.
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
This study was supported by a National Institute of Mental Health grant (K01MH085710) to J. Green.