Teachers’ Responses to Child Maltreatment

Abstract

Purpose

Teachers play an important role in detecting and reporting child maltreatment by virtue of their extensive contact with children. Current research, while limited, shows a tendency among teachers to under-detect and under-report various forms of child abuse and neglect. Methods: Using data from a pilot evaluation of child maltreatment training for teachers, we examined teachers’ self-reported behaviours for different forms of maltreatment. Participants included 45 experienced teachers from a large urban centre in the province of Ontario (Canada) who rated their likelihood of, and rationale for, detecting and reporting maltreatment based on four case vignettes.

Results

Accuracy rates for detection and reporting were highest for sexual abuse (95% and 93%), followed by neglect (87% and 75%), emotional abuse (86% and 70%), and physical abuse (58% and 27%). Differences in detection and reporting rates were only significant between physical abuse and other maltreatment types (sexual abuse and neglect for detection; sexual abuse, neglect, and emotional abuse for reporting), with accurate detection and reporting rates being significantly lower for physical abuse. Teachers stated that the physical abuse vignette lacked convincing evidence of child maltreatment. Further, many teachers indicated that additional information was needed to determine whether the vignette represented physical abuse and, as such, was reportable.

Conclusion

Our findings suggest the need for more targeted interventions to improve teachers’ decision-making with potential child maltreatment scenarios, especially for physical abuse, so that children who are experiencing maltreatment may be more accurately detected and reported to child protection.

Currently, only a small proportion of children who experience child maltreatment receive professional help (Burczycka & Dyna, 2016; Cotter & Beaupré, 2014). Child maltreatment experiences are notably under-reported (Stoltenborgh et al., 2014), with several studies finding that self-reports of maltreatment from children and parents are 10 times higher than the rate of maltreatment confirmed by child protection agencies (Gilbert et al., 2009; Finkelhor, 2008). Without timely identification and intervention, many children experience negative short- and long-term effects of child maltreatment on their physical and mental health, interpersonal relationships, and more general life experiences (Afifi et al, 2014; Afifi et al., 2016; Doyle & Cicchetti, 2017; Fresno et. al., 2017; Proskynitopoulos et al., 2021; (Romano et al., 2015). By virtue of teachers’ ongoing, frequent contact with children, they are well positioned to play an important role in ensuring that maltreated children are identified and that the appropriate authorities are contacted, such as child welfare or the police (Hobbs et al., 2019). However, many professionals, including teachers, have been found to under-detect and under-report child maltreatment (Ben Natan et al., 2012; Greco et al., 2017, 2020; Tufford & Lee, 2020). As such, continued research is needed to understand how teachers’ beliefs, knowledge, and skills influence the recognition and reporting of child maltreatment (Greco et al., 2017). The current study explored a small group of Canadian teachers’ accuracy rates and decision-making intentions for detection and reporting across various forms of child maltreatment to better understand possible knowledge gaps and inaccurate beliefs.

Rates of Child Maltreatment

Worldwide, it is predicted that 1 in 15 children and adolescents experience maltreatment annually (Svevo-Cianci et al., 2010). In a nationally representative Canadian study (Afifi et al., 2014), 32% of adults reported a history of child maltreatment, defined as physical abuse, sexual abuse, and/or exposure to intimate partner violence. These findings were corroborated by a slightly more recent retrospective self-report study, wherein three in ten Canadians reported experiencing child maltreatment (i.e., intimate partner violence, sexual abuse, physical abuse) at least once during their childhood (Burczycka & Dyna, 2016). Within the province of Ontario (Canada), the annual incidence rate of child abuse and neglect for 2018 was 3.1 investigations per 1,000 children. Forty percent of child maltreatment reports (37,922/94,476) were substantiated across child welfare agencies. An additional four percent of cases lacked sufficient evidence yet were still suspected to be maltreatment by case workers (Fallon et al., 2020). The primary concern among substantiated cases was exposure to intimate partner violence (45%), followed by neglect (21%), physical abuse (19%), emotional abuse (12%), and sexual abuse (3%). Evidently, self-report rates are substantially higher than investigated and confirmed maltreatment cases (Cyr et al., 2013; Kim et al., 2017). Although retrospective adult surveys are subject to recall bias (Stoltenborgh et al., 2014), self-reported prevalence of child maltreatment has been deemed an essential measurement tool (Hillis et al., 2016), as many instances of maltreatment remain undisclosed or unreported (Cater et al., 2016; Gilbert et al., 2009; Hillis et al., 2016). Therefore, it is important to use these measurements to help compare prevalence versus reported rates.

Teachers’ Role in Detecting and Reporting Child Maltreatment

Early intervention is imperative to promote lifelong well-being and healthy development among children who have experienced maltreatment (Mathews et al., 2017; Goebbels et al., 2008). School personnel are critical in effective and timely identification in cases of maltreatment (Cater et al., 2016; King, 2011; King & Scott, 2014; McGarry & Buckley, 2013). Due to their legal obligation to report suspected maltreatment (Child, Youth and Family Services Act (2017)) and frequent interactions with children over extended periods of times, teachers are well-positioned to identify and report child maltreatment (Gubbels et al., 2021; King, 2011; Mathews et al., 2015). In Ontario, in 2018, the largest portion of all child welfare referrals (32%) were initiated by school staff, though the substantiation rate specific to school staff referrals was not described (Fallon et al., 2020). However, studies with teachers in Canada (Beck et al., 1994; King, 2011; Tite, 1993) and other countries (Dinehart & Kenny, 2015; Feng et al., 2010; Goebbels et al., 2008; Greco et al., 2017, 2020; Hupe & Stevenson, 2019; Sainz et al., 2020) have found that teachers are not dutifully reporting child maltreatment in a timely manner, if at all. These studies found that between 4–73% of teachers who suspected maltreatment did not report it. This may suggest that teachers’ referrals to child protection agencies should be higher than what has been found in existing incidence rate studies (e.g., Fallon et al., 2020). More recent studies with Canadian teachers are needed to explore this hypothesis.

Factors Influencing Teachers’ Detecting and Reporting Behaviours

For teachers to accurately recognize signs of maltreatment and make a report to child protection authorities, they need opportunities to acquire the necessary knowledge, skills, and confidence in child maltreatment identification and reporting procedures (Baweja et al., 2016; Falkiner et al., 2017; Weegar & Romano, 2019). However, many teachers have been found to receive inadequate training (Baweja et al., 2016; Falkiner et al., 2017; Feng et al., 2015). Consequently, teachers often have limited knowledge about maltreatment detection and reporting, which contributes to under-reporting and/or inaccurate reporting (Greco et al., 2020; King & Scott, 2014).

Factors Influencing Teachers’ Detection of Maltreatment

Various beliefs and attitudes frequently guide teachers’ awareness of child maltreatment signs. For example, under-detection has been associated with teachers’ belief that they are not equipped to determine if maltreatment has occurred without obvious signs, such as bruising or a child’s disclosure of sexual abuse (de Haan et al., 2019; Falkiner et al., 2017; Gubbels et al., 2021). Other factors that have been found to influence teachers’ detection of child maltreatment include the type of maltreatment, frequency of incidents, and perceived level of impact on the child (Christodoulou et al., 2019; Gubbels et al., 2021; Vanderfaeillie et al., 2018; Walsh et al., 2008). A perceived lack of adequate information also impacts teachers’ abilities to identify abuse and neglect (Greco et al., 2017).

With regard to maltreatment type, there is conflicting literature on detection based on form. Research performed in Greece (Christodoulou et al., 2019), New Zealand (Rodriguez, 2002), Estonia (Toros & Tiirik, 2016), and the United States (Webster et al., 2005) found teachers were most likely to identify sexual abuse. However, Australian research (Falkiner et al., 2017) suggests that physical abuse is most likely to be detected. Further, in Belgium, Vanderfaeillie et al. (2018) found that teachers identify physical and sexual abuse at roughly similar rates. To add to the differing findings, sexual abuse has also been suggested to be least likely be detected by Australian researchers (Falkiner et al., 2017), and physical abuse was found least likely to be detected in New Zealand (Rodriguez, 2002) and U.S. (Webster et al., 2005) studies. Neglect seems to frequently be the second form of maltreatment most easily identified in Australian (Falkiner et al., 2017; Walsh et al., 2008) and New Zealand (Rodriguez, 2002) teacher samples. However, it appeared to be the most challenging for Greek teachers to identify (Christodoulou et al., 2019). Emotional abuse was the form that fluctuated the most in detection; it was the second easiest maltreatment form to identify for U.S. (Webster et al., 2005), the third easiest for Greek teachers (Christodoulou et al., 2019), and the most challenging for Belgium teachers (Vanderfaeillie et al., 2018).

The only similar Canadian study to our knowledge was by King (2011), who found that fewer than half of the participating teachers believed that they would be able to recognize signs of intimate partner violence exposure, sexual abuse, or emotional abuse. Additionally, 29% and 38% of teachers did not think that they had sufficient knowledge to identify physical abuse and neglect, respectively. It should be noted that this study examined teachers’ perception of their abilities rather than detection behaviours. As such, research is needed to better understand how child maltreatment type influences Canadian teachers’ detection accuracy.

Factors Influencing Teachers’ Reporting of Maltreatment

Even when teachers suspect child maltreatment, studies show that they often do not report their suspicions to the appropriate authorities (Greco et al., 2017, 2020). Factors that may impede reporting include a desire for proof that maltreatment has occurred prior to making a report, a lack of knowledge around reporting procedures, a belief that reporting within their school was sufficient, and uncertainty around the relationship between their school and social services (Alazri & Hanna, 2020; Falkiner et al., 2017; Greco et al., 2017). Further, teachers have fears around reporting, such as worries about making an inaccurate report, negative repercussions to the child and/or teacher once the report is made (e.g., that the child will be automatically removed from their home), and damage to the parent-teacher relationship (Falkiner et al., 2017; Forsner et al., 2020; Greco et al., 2020; Gubbels et al., 2021; Toros & Tiirik, 2016). Recent studies have also found that teacher’s certainty surrounding reporting decreases when accounting for culturally diverse parenting practices (de Haan et al., 2019; Gubbels et al., 2021; Nagy et al., 2019). Gubbels et al. (2021) found that awareness of differences in cultural norms for parenting practices added to a teacher’s challenge in assessing risk of maltreatment. Further, de Haan et al., (2019) noted that teachers misidentified common parenting practices as child maltreatment and vice versa for certain ethnic populations. Teachers have also been found to have difference in reporting rates based on ethnicity. In Belgium, teachers were found to report differently, viewing cases of potential maltreatment more serious and in more need for intervention if the parents were not of Western European origin (Vanderfaeillie et al., 2018). In the United States, education personnel were found to disproportionally report African American and Hispanic children to child protection services relative to white children, especially in low-income contexts (Krase, 2015; Sinanan, 2011). These findings indicate that teachers reporting behaviours vary based on culture of the child. Potentially some of this variation can be accounted for by the arbitrary cut-off line between physical abuse and corporal punishment. With some ethnic groups having higher acceptance of corporal punishment (Friedson, 2016), this may increase the risk of being suspected of child maltreatment. Factors that have been found to promote a teacher’s decision to report are prior experience with identification and reporting, legal requirements to report maltreatment, a high level of certainty about their suspicions, severe cases that require immediate attention, and ongoing occurrences of potential maltreatment events (Falkiner et al., 2017; Greco et al., 2020; Walsh et al., 2008).

Teachers’ reporting behaviours are also found to be influenced by the form of maltreatment. However, the growing body of literature lacks consensus on which form is most likely to be reported. Research performed with Greek (Bibou-Nakou & Markos, 2017), Turkish (Karadag et al., 2015), and Australian (Walsh et al., 2008) teachers found physical abuse to be the most likely reported. In contrast, research implemented with Estonian (Toros & Tiirik, 2016), U.S. (Webster et al., 2005), and Canadian (Beck et al., 1994) teachers found sexual abuse to be most likely reported. Further, studies with Belgium (Vanderfaeillie et al., 2018) and Greek teachers (Christodoulou et al., 2019) found that sexual and physical abuse were equally most likely to be reported. Lastly, an Australian study found neglect most frequently brought to the attention of child protection services (Goebbels et al., 2008). To add to the mixed findings, physical abuse has also been found to be least likely to be reported in studies with Taiwanese (Feng et al., 2010) and U.S. teachers (Webster et al., 2005), and sexual abuse has been found to be least reported in a study with Australian teachers (Goebbels et al., 2008). The only form of maltreatment that seems to have some consistency is emotional abuse, with it typically being least reported (Beck et al., 1994; Christodoulou et al., 2019; Toros & Tiirik, 2016; Vanderfaeillie et al., 2018).

The most recent research performed to examine teachers’ intent to report in Canada is quite dated (Beck et al., 1994). This study provided participating teachers with four reportable vignettes focused on physical abuse, sexual abuse, neglect, and emotional maltreatment. Teachers responded on a 7-point Likert-scale to indicate the likelihood that the vignette represented maltreatment and that they would report the kind of situations outlined in the vignette. Results showed that sexual abuse (M = 6.45, SD = 0.97) was the most likely type of maltreatment to be reported, followed by neglect (M = 6.36, SD = 1.07), physical abuse (M = 5.38, SD = 1.70), and emotional abuse (M = 5.21, SD = 1.67). Due to how old this study is, there is a desperate need for further examination of reporting differences among Canadian teachers.

Study Objectives and Hypotheses

To reduce the discrepancy between actual versus reported rates of child maltreatment, more research is needed to understand the factors that may influence teachers’ decisions around potential instances of child abuse and neglect. To help address this gap, we explored detection and reporting behaviours across various types of maltreatment using a sample of teachers from a large urban region in Ontario (Canada). Our first objective was to determine teachers' detection and reporting rates for different types of child maltreatment, as portrayed in case vignettes. Based on past, albeit limited research, it was hypothesized that sexual abuse, neglect, and physical abuse would be detected and reported above chance levels, whereas detection and reporting of emotional abuse would be below chance. The second objective was to compare the detection and reporting rates across different forms of maltreatment. Based on past research, we expected that sexual abuse would be the most accurately detected and reported maltreatment type, followed by neglect, physical abuse, and emotional abuse. We anticipated that physical abuse would be less detected and reported than neglect by Canadian teachers because of the added uncertainty created by Sect. 43 of Canada's Criminal Code (Legislative Service Branch, 2021). Specifically, Sect. 43 gives parents the right to use "reasonable physical force" for disciplinary purposes. This reduces clarity for teachers by establishing a grey zone where parental use of physical punishment is sometimes permissible by law and as such, not reportable to child protection authorities or to the police. Finally, this study aimed to identify common factors influencing teachers’ decision-making for child maltreatment detection and reporting (or lack thereof).

Methods

Participants and Procedures

Participants included 45 English-speaking teachers from a major urban region in Ontario (Canada) who had two or more years of experience teaching in elementary school (kindergarten to Grade 8). The majority of teachers were female (86.7%), with an average age of 35.18 years (SD = 10.10) and 10.56 years (SD = 8.30) of teaching experience. Slightly more than half of the teachers held a permanent position (55.6%). Teachers who worked with kindergarten and grade 7–8 students each made up 22.2% of the sample. Of the remaining teachers, 35.6% taught grades 1–3 and 42.2% taught students in grades 4–6. Out of the 45 teachers, 22% reported having had previous experience working with children who experienced maltreatment, and 37.8% indicated having participated in prior child maltreatment training.

The teachers in the current exploratory study based on already-collected data were originally part of a quasi-experimental evaluation of a training workshop aimed at improving knowledge of child maltreatment within a school setting (Weegar & Romano, 2019). Ethics approval was received from our University’s Office of Research Ethics and Integrity. The original study used a nonprobability, convenience sampling method (Dillman et al., 2014) whereby teachers were recruited through graduate and continuing education classes, social media platforms, and snowball sampling. Teachers who participated in the study completed an online consent form and questionnaire that included socio-demographics, training background, and child maltreatment vignettes. While data were also collected following the training workshop, this study relied exclusively on data from pre-training measures.

Measure

Child Maltreatment Vignettes

Self-report measures have been typically used to investigate teachers’ detection- and reporting-related behaviours because it can be challenging to assess teachers’ actual abilities in this area (Falkiner et al., 2017; Greco et al., 2017, 2020; King, 2011). However, a growing number of studies are using case vignettes (Christodoulou et al., 2019; Dinehart & Kenny, 2015; Toros & Tiirik, 2016; Vanderfaeillie et al., 2018; Walsh et al., 2008; Webster et al., 2005) to further understand teachers’ decision-making processes. In the current study, we used four reportable child maltreatment vignettes (i.e., sexual, physical, and emotional abuse, as well as neglect) selected from the Recognition of Child Maltreatment Test (Alvarez et al., 2010; Donohue et al., 2015). This measure has been used in past research, and it generates both quantitative (ratings) and qualitative (rationale for ratings) data.

In developing the Recognition of Child Maltreatment Test, Alvarez et al. (2010) created 19 situations from a literature review that focused on child maltreatment identifiers and then recruited two child welfare practitioners to validate which scenarios represented a reportable instance of maltreatment within the United States context. A finalized vignette list of eight vignettes was determined based on scenarios that had 100% agreement between the two practitioners. In the current study, we relied on three child welfare practitioners (i.e., intake workers at a local child welfare agency) and one researcher with expertise in child protection independently to review the eight vignettes from Alvarez et al. (2010) and decided whether the situation was reportable based on Ontario child protection standards. There was agreement across most vignettes. Where there was a discrepancy, the second and third author of the current study jointly made a final decision. Due to the rigorous development of the vignettes by Alvarez et al. (2010), no alterations were made to the child- and parent-related content in the cases. However, the context of the vignettes were slightly modified so that they were school based (e.g., changing the setting from a clinic to a classroom) because the original vignettes were written for mental health professionals.

Of the eight vignettes, we focused on the four scenarios that represented a reportable instance of child maltreatment. Each of the four vignettes focused on one form of maltreatment, namely sexual abuse, emotional abuse, physical abuse, and neglect. The sexual abuse vignette presents a scenario of a parent-teacher interview with a mother of a 10-year-old girl. The teacher tells the parent about the odd behaviours and low mood exhibited by the child. The mother responds by discussing her husband’s controlling relationship with their child and how she overheard, through a closed door, her child telling her father that he should not put his hands ‘there’. The neglect vignette depicts a 9-year-old boy arriving at school with a burn on his hand. When asked what transpired, he states he grabbed a hot pan while making dinner and then mentions that his mother is rarely home due to work or gambling with friends. The emotional abuse vignette portrays parents frequently meeting with a teacher because of their son’s outbursts in class. During the meetings, the teacher notices that the parents frequently make derogatory comments about their sons (e.g., calling them an idiot or stupid) and blame him for their family’s difficulties. Lastly, the physical abuse vignette is about a 6-year-old girl who has a long linear bruise on the back of her arm and leg. The teacher had noticed similar bruising on a previous occasion. When asked what happened, both the parent and child recount the same story (i.e., child fell on the sidewalk), and the mother further mentions the child’s clumsiness. Teachers read each vignette and used a 7-point Likert-scale from (1) ‘highly unlikely’ to (7) ‘highly likely’ to rate whether the vignette represented a form of child maltreatment (i.e., detection) and whether they would report the incident to a child protection agency. After rating each vignette for detection and reporting, teachers were also asked to provide a written rationale for their ratings.

Data Analysis

We relied on non-parametric univariate tests to examine child maltreatment detection and reporting accuracy because of the dichotomous nature of the data (Green & Salkind, 2013). All quantitative analyses were conducted using SPSS 25.0. To address the first research objective, a binomial test evaluated whether the proportion of accurate responses for each maltreatment vignette was greater than chance. To determine accurate versus inaccurate responses, a cut-off score of 5 was used from our 7-point scale with 4 representing a neutral response. This cut-off was chosen for ease of analysis. Chance expectations were chosen as a comparison for each vignette because there is a lack of established research to provide guidance about expected rates for teachers’ ability to correctly detect and report child maltreatment. To control for Type 1 error, the Bonferroni correction was applied so we used a p-value cut-off of 0.013 (original p-value of 0.05 divided by 4 for each form of maltreatment = 0.013).

To address the second research objective (i.e., if accuracy rates differed significantly across the maltreatment vignettes), we conducted a Cochran test, which is the nonparametric equivalent to a one-way analysis of variance. This test was used because it accounts for the dependency among scores of a repeated-measures design where participants are assessed more than once on a dichotomous variable (Green & Salkind, 2013). Kendall’s W coefficient was used to assess the level of consensus among teachers’ detection and reporting rates. If the test statistic W = 0, there is no overall trend of agreement for detection and reporting between different forms of maltreatment. In contrast, W = 1 indicates that teachers have unanimous agreement, with the exact same detection and reporting rates for different forms of maltreatment (Marcinkiewicz, 2017). The McNemar test was then used as a follow-up to make pairwise comparisons between accurate detection for different maltreatment types and between accurate reporting for different maltreatment types. This test was selected because it determines response consistency between two repeated measurements (i.e., responding to each of the four vignettes) of a dichotomous variable (e.g. accurate versus inaccurate detection, accurate versus inaccurate reporting; Green & Salkind, 2013). To control for type 1 error, a Bonferroni correction was used with a p-value cut-off of 0.008 (original p-value of 0.05 divided by 6 for a comparison between each form of maltreatment = 0.008).

For the third study objective of understanding teachers’ rationales for their detection and reporting ratings, we conducted a content analysis with QDA miner. Each vignette was analyzed separately using an inductive approach to create themes within accurate and inaccurate teacher responses for each form of maltreatment. Two reviewers (first author and a research assistant) independently coded teachers’ decision-making rationales and then engaged in consensus coding to reach agreement on common themes. Any conflicting designations were reviewed with the third author. Theme frequencies were calculated to highlight more common beliefs among teachers (Hsieh & Shannon, 2005). Only themes discussed by more than one participant were included.

Results

Detection and Reporting Rates

For child maltreatment detection, Table 1 shows that the percentage of teachers who correctly identified physical abuse from the related vignette did not differ significantly from chance. For the remaining vignettes, teachers’ detection rates were significantly better than chance and ranged from 86% (emotional abuse) to 95% (sexual abuse). For reporting of maltreatment based on the vignette responses, emotional abuse (70%) did not differ significantly from chance. For the remaining maltreatment types, teachers were found to report neglect (75%) and sexual abuse (93%) at significantly higher-than-chance expectations. In contrast, teachers significantly under-reported physical abuse (27%).

Table 1.

Teachers’ rates of child maltreatment detection and reporting (N = 45)

Type of Maltreatment	Accurate Detection	p-value =	Accurate Reporting	p-value =
	% (N)		% (N)
Sexual Abuse	95 (43)	< .001	93 (42)	< .001
Neglect	87 (39)	< .001	75 (34)	.001
Emotional Abuse	86 (38)	< .001	70 (31)	.014
Physical Abuse	58 (26)	.371	27 (12)	.004

The p-value of .013 (Bonferroni correction) was used to determine statistical significance

Bolded p-values indicate that detection/reporting rates were significantly different than chance expectations

Detection and Reporting Rate Differences Between Maltreatment Types

In the next set of analyses, we compared differences in teachers’ rates of detection between different types of child maltreatment. The Cochran test was statistically significant ${\chi }^2\left(4,,,,\mathrm{N},=,44\right)=20.09,p<.001$) indicating that accurate versus inaccurate detection rates differed significantly across the various types of child maltreatment. The Kendall W effect size was 0.152, which indicates a low consensus among teachers in their detection of different child maltreatment types. Based on the McNemar pairwise comparisons, Table 2 shows that the detection rates for physical abuse (58%) were significantly lower than those for sexual abuse (95%) and neglect (87%). All remaining comparisons were not statistically different.

Table 2.

Detection rate differences between different forms of maltreatment (N = 45)

Type of Maltreatment	Sexual Abuse	Neglect	Emotional Abuse				Physical Abuse
Sexual Abuse	-	.289		.289		< .001
Neglect		-	1.00				.007
Emotional Abuse			-				.017
Physical Abuse							-

The p-value of .008 (Bonferroni correction) was used to determine statistical significance

Bolded p-values indicate that detection rate differences between the two maltreatment types were statistically significant

For differences in teachers’ reporting rates across the various types of child maltreatment, the Cochran test was once again significant (${\chi }^2\left(4,,,\mathrm{N},=,43\right)=43.35,p<.001)$. Thus, accurate versus inaccurate reporting differed significantly across child maltreatment types. The Kendall W effect size was 0.336, suggesting that reporting agreement was relatively low. Results from the McNemar test (Table 3) indicate that reporting rates for the physical abuse vignette (27%) were significantly lower than rates for the sexual abuse (93%), neglect (75%), and emotional abuse (70%) vignettes. All remaining comparisons were not statistically significant.

Table 3.

Reporting rate differences between different forms of maltreatment (N = 45)

Type of Maltreatment	Sexual Abuse	Neglect	Emotional Abuse	Physical Abuse
Sexual Abuse	-	.039	.013	< .001
Neglect		-	.754	< .001
Emotional Abuse			-	.001
Physical Abuse				-

The p-value of .008 (Bonferroni correction) was used to determine statistical significance

Bolded p-values indicate that reporting rate differences between the two maltreatment types were statistically significant

Teachers’ Detection and Reporting Rationales

Content analysis was conducted to explore themes from teachers’ rationales about whether child maltreatment would be detected and/or reported, based on the vignettes.

Accurate Detection

Table 4 indicates that four main themes emerged from accurate detection responses to the vignettes, specifically child risk factors (e.g., atypical behaviour), parent risk factors (e.g., inappropriate touching, lack of supervision, derogatory comments), informant credibility (e.g., concerned mother), and contextual factors (e.g., frequency of past similar occurrences, different cultural parenting practices). Child risk factors were most commonly mentioned as the basis for identifying neglect (n = 29 of 61 child risk factor comments, 48%), followed by physical abuse (n = 15, 25%), sexual abuse (n = 11, 18%), and emotional abuse (n = 6, 10%). For the neglect vignette, the child risk factors commonly mentioned by teachers involved the child being home alone and the risk to the child’s physical safety because of the lack of parental presence and supervision.

Table 4.

Vignette Response Themes

Themes from Teachers Responses Across the Four Child Maltreatment Vignettes
Accurate Detection	Child Risk Factors					Parent Risk Factors					Informant Credibility					Contextual Factors
Type of Maltreatment	Count Total	S	N	E	P	Count Total	S	N	E	P	Count Total	S	N	E	P	Count Total	S	N	E	P
n	61	11	28	6	15	79	21	26	32	0	22	16	3	N/A	3	28	1	9	3	15
%		18	48	10	25		27	33	41	0		73	14	N/A	14		4	32	11	54
Inaccurate Detection	Not Enough Information					Information Does Not Suggest Maltreatment
Type of Maltreatment	Count Total	S	N	E	P	Count Total	S	N	E	P
n	17	1	4	1	11	18	1	2	3	12
%		6	24	6	65		6	11	17	67
Accurate Reporting	Inappropriate Circumstances					Against the Law/ Requirement to Report					Obtain Help for Child/Family					Informant Credibility
Type of Maltreatment	Count Total	S	N	E	P	Count Total	S	N	E	P	Count Total	S	N	E	P	Count Total	S	N	E	P
n	50	20	14	14	2	12	5	5	1	1	28	5	8	10	5	15	15	0	0	0
%		40	28	28	4		42	42	8	8		18	29	36	18		100	0	0	0
Inaccurate Detection	Not Enough Information					Information Does Not Suggest Maltreatment
Type of Maltreatment	Count Total	S	N	E	P	Count Total	S	N	E	P
n	36	2	9	7	18	15	0	1	3	11
%		6	25	19	50		0	7	20	73

S sexual abuse, N neglect, E emotional abuse, P physical abuse, n total number of teachers who commented on that theme in their rationale, Count Total number of counts from all forms of maltreatment in that theme, % n/Count Total

Risk factors at the parent level were most frequently considered by teachers in the identification of emotional abuse (n = 32 of the 79 parent risk factor comments, 41%), followed by neglect (n = 26, 33%) and sexual abuse (n = 21, 27%). Teachers made no such comments around identification of child maltreatment in the vignette depicting physical abuse. For emotional abuse, one teacher commented: Parents appear verbally/emotional abusive – calling their children names, being surprised that they have positive qualities. Also, the fact that they blame the boys for the “problems of the family” is concerning.

Another theme among accurate detectors was around the credibility of the informant disclosing the maltreatment. This theme was most frequent among teachers with regard to the vignette around sexual abuse (N = 16 of the 22 informant credibility comments, 73%), relative to physical abuse (n = 3, 14%), and neglect (n = 3, 14%). No comments about informant credibility were made by teachers in the identification of emotional abuse. For the sexual abuse vignette, many teachers commented that their identification of this maltreatment type was influenced to a large extent by the concern of the mother, which suggested to teachers that they needed to take the matter seriously and have it investigated.

For the final theme of contextual factors, teachers most frequently made such comments to justify their detection of physical abuse (n = 15 of the 28 contextual comments, 54%), followed by neglect (n = 9, 32%), emotional abuse (n = 3, 11%), and sexual abuse (n = 1, 4%). For physical abuse, one teacher commented: The fact that I noticed similar bruising on at least one other occasion makes me suspect child maltreatment. Without that piece I probably wouldn’t suspect.

Inaccurate Detection

Table 4 indicates that two themes emerged among teachers who did not accurately detect the occurrence of child maltreatment from the vignettes. Specifically, teachers either noted there was not enough information to decide or the information provided did not suggest that maltreatment had occurred. In some instances, not having enough information included teachers mentioning that they would want to speak with others, such as colleagues or administrators, for feedback and support to help determine how to interpret the situation. This theme of not having enough information was particularly notable for the physical abuse vignette as it represented 65% (n = 11 of 17) of the comments made, in comparison with neglect (n = 4, 24%), sexual abuse (n = 1, 6%) and emotional abuse (n = 1, 6%). One teacher commented: I would need to know the student and gross motor capabilities, truthful/untruthful interactions, and see the body language/hear tone of both student and parent. Another teacher mentioned: [I] would need to look into the matter further, like how did the student react to my probing questions. For the neglect vignette, one teacher described increased uncertainty due to not knowing if the level of independence the child was given by the parent was a reflection of acceptable cultural practice.

With regard to information not meeting the threshold to constitute maltreatment, this theme also was most notable for physical abuse, where 67% (n = 12 of 18) of the comments were for this type of maltreatment. This was followed by the vignettes for emotional abuse (n = 3, 17%), neglect (n = 2, 11%), and sexual abuse (n = 1, 6%). For physical abuse, many teachers noted that the scenario did not indicate maltreatment because children often experience bruising from various activities. Several also commented that the consistency between the parent and child reports about what happened suggested the scenario was not physically abusive in nature.

Accurate Reporting

Comments from teachers who were accurate reporters of maltreatment fell into four themes: inappropriate circumstances; legal reporting requirements; desire to obtain help for the family; and informant credibility. Inappropriate circumstances was the most frequent theme for the sexual abuse vignette (n = 20 of the 50 comments about inappropriate circumstances, 40%), followed by neglect (n = 14, 28%), emotional abuse (n = 14, 28%), and physical abuse (n = 2, 4%). For the sexual abuse vignette, many teachers noted that the father touched his daughter inappropriately. Another teacher commented: Anything sexual in nature should be reported.

The theme of mandated reporting requirements was equally common for sexual abuse (n = 5 of the 12 requirement to report comments, 42%) and neglect (n = 5, 42%), followed equally by emotional abuse (n = 1, 8%) and physical abuse (n = 1, 8%). One teacher who accurately reported sexual abuse noted: I would need to follow procedures. I am almost certain my concerns need to be at least reported (even if they are not correct). Across all four vignettes, teachers mentioned that they would report the incident to authorities to obtain help for the family. Emotional abuse had the most comments (n = 10 of the 28 comments, 36%), followed by neglect (n = 8, 29%), sexual abuse (n = 5, 18%), and physical abuse (n = 5, 18%). For emotional abuse, one teacher commented: The children need to have caring relationships with the adults in their lives. The adults need support to properly parent their children. Finally, for the informant credibility theme, sexual abuse was the only maltreatment type that received comments of this nature (n = 15 of all 15 comments, 100%). Similar comments were made for reporting sexual abuse as they were for detection, namely that the nature of the mother’s concerns indicated to teachers that they needed to take the disclosure seriously and follow up with a child protection report.

Inaccurate Reporting

Teachers who inaccurately indicated that they would not report the maltreatment (as described in the vignette) provided rationales that fit within two main themes, namely not having enough information, or having insufficient evidence to confirm that maltreatment had occurred. Most of these comments were in regard to physical abuse, where 50% (n = 18 of the 36 comments) stated that more information was needed, followed by neglect (n = 9, 25%), emotional abuse (n = 7, 19%), and sexual abuse (n = 2, 6%). For the physical abuse vignette, one teacher noted: Until I knew more, I wouldn’t likely make a report. I would perhaps speak to the child’s previous teacher, or administration to see if there was any history. Another teacher stated: There’s just not enough for me to feel comfortable making a report. I imagine causing the little girl more trauma by reporting. Further, the physical abuse vignette also had the greatest number of comments pertaining to insufficient evidence to warrant a child protection report (n = 11 out of the 15 comments, 73%), followed by emotional abuse (n = 3, 20%) and neglect (n = 1, 7%). It should be noted that no such comments were made for the sexual abuse vignette. For physical abuse, several teachers commented that the scenario seemed like an accident or that the student was just clumsy.

Discussion

Few researchers have sought to empirically understand child maltreatment detection and reporting behaviours of Canadian teachers. Using data from a Canadian sample of elementary teachers, we helped address this gap by exploring teachers’ accuracy and decision-making process in their intent to detect and report child maltreatment. To examine this issue, we relied on empirically-developed case vignettes representing sexual abuse, physical abuse, emotional abuse, and neglect.

Detection and Reporting Relative to Chance

Our findings indicated partial support for the hypothesis that sexual abuse, physical abuse, and neglect would be detected and reported at above chance levels, whereas emotional abuse would have below-chance detection and reporting rates. As expected, teachers detected sexual abuse and neglect at above-chance levels. For the sexual abuse vignette, teachers tended to rely on informant variables (e.g., a concerned parent) as the basis for determining that maltreatment had occurred, whereas for neglect, there was a greater reliance on child risk factors (e.g., child alone, physical safety). Teachers’ responses to the emotional abuse and physical abuse vignettes were not in line with expectations. Specifically, teachers detected emotional abuse at above-chance levels but physical abuse at chance levels. It is difficult to draw any solid conclusions about these findings, although the qualitative data provides several possible insights. For instance, teachers’ level of confidence about whether inappropriate or abusive parental behaviour was occurring seemed to influence their maltreatment detection responses. Teachers appeared confident about their decision making with regard to emotional abuse, citing a number of factors related to parental behaviour that guided their detection (e.g., parents making derogatory comments and placing blame on their children for family issues). In contrast, teachers made no comments at the level of parental behaviours as the basis for detecting physical abuse. Moreover, the physical abuse scenario also had the highest number of comments about the information not suggesting physical abuse and/or the information being insufficient to make a decision.

Reporting responses were relatively similar to detection in that teachers reported sexual abuse and neglect at above-chance levels, in line with our hypotheses. Interestingly, comments around the reporting requirements were most frequently made by teachers for these two types of maltreatment. Contrary to expectations, teachers reported emotional abuse at chance level and physical abuse at below-chance levels. While no causal inferences can be drawn, the higher-than-expected reporting responses for emotional abuse may be partially explained by the qualitative findings that indicated many teachers wanted to protect the child and support the family. These findings also suggest that teachers seem quite aware of emotional abuse in terms of potential indicators and associated negative consequences. For physical abuse, teachers rarely mentioned that the circumstances were inappropriate (about 4% in comparison to the other vignettes, which ranged from 28–40%).

The low levels of accurate detection and reporting for the physical abuse vignette is surprising, based on past research by King (2011). In their findings, Ontario teachers self-reported their ability to detect sexual abuse, emotional abuse, and neglect at much lower levels relative to physical abuse. Further, results from a study by Falkiner et al. (2017) showed that teachers believe they are not prepared to identify and make a report of a maltreatment without obvious signs, such as bruising, yet our vignette depicted a child with long linear bruising. Lastly, the findings for physical abuse are also unexpected given that King and Scott (2014) noted that Ontario teachers’ reporting of physical abuse tended to be over-represented relative to other professionals (e.g., police, social service providers), which the authors inferred to suggest that teachers were more likely to report physical abuse.

The differences in findings may be explained at a methodological level, as both King (2011) and Falkiner et al. (2017) focused on teachers’ perceived capabilities in detecting signs of child maltreatment, whereas we relied on vignettes to examine behavioural intentions. As such, teachers’ beliefs about their knowledge and abilities in detecting maltreatment might differ from their actual application of skills through consideration of case scenarios. At a content level, it may be that the detection and reporting of physical abuse was more ambiguous than anticipated. This would align with previous studies that have also found a perceived lack of sufficient information and uncertainty influence teachers’ ability to detect and report maltreatment, in particular for physical abuse (Falkiner et al., 2017; Greco et al., 2017, 2020). Certainly, in Canada, the fact that Sect. 43 of our Criminal Code (Legislative Service Branch, 2021) gives parents the right to use “reasonable force” for disciplinary purposes muddies the waters by establishing a grey zone where parental use of physical punishment is within the law and as such, not reportable to child protection authorities or to the police. In addition, based on teachers’ responses, the uncertainty may also stem from children frequently having bruises and other physical injuries from regular activities, such as riding a bike or playing sports, or from clumsiness due to ongoing motor skill development. Given the dearth of research in this area, additional studies are needed to explore possible discrepancies between teachers perceived and actual abilities for physical abuse detection and reporting.

Differences in Detection and Reporting Between Child Maltreatment Types

The second objective of this study was to compare detection and reporting rates as a function of maltreatment type. We hypothesized that sexual abuse would be the most accurately detected and reported type while emotional abuse was expected to be least accurately detected and reported. Our hypotheses were partially supported in that sexual abuse was the most frequently detected and reported form of maltreatment among the four vignettes. However, contrary to expectations, physical abuse was the least accurately detected and reported by teachers, not emotional abuse.

For the sexual abuse vignette, it should be noted that the rates of detection and reporting were only statistically significant in comparison with physical abuse. These findings suggest that detection and reporting rates are relatively similar among sexual abuse, emotional abuse, and neglect. It is difficult to interpret these results because of limited past research. However, one possible explanation for the detection findings is that all three forms of child maltreatment had comments from teachers about inappropriate parental behaviours, whereas there were no such comments about the physical abuse vignette. As such, teachers appeared to focus on the same theme of parental behaviours as an important basis for the detection of sexual and emotional abuse, as well as neglect.

As previously discussed, one potential reason for reporting behaviours being comparable among the three maltreatment types may also be related to frequent comments about inappropriate circumstances (e.g., father inappropriately touching child, absent mother gambling, parents emotionally abusing child). In contrast, only a few teachers made comments related to inappropriate circumstances for the reporting (and detection) of physical abuse. In general, our findings align with those from studies in Canada (Beck et al., 1994) and other countries (Christodoulou et al., 2019; Toros & Tiirik, 2016; Webster et al., 2005) that found teachers to be most accurate at reporting sexual abuse. Also similar to Beck et al. (1994), it was noted that neglect was reported at rates that were similar to sexual abuse. Inconsistent with our findings, Beck et al. (1994) found that physical abuse was reported more frequently than emotional abuse. Further, results with Greek (Christodoulou et al., 2019), Estonian (Toros & Tiirik, 2016), and Belgian (Vanderfaeillie et al., 2018) teachers have indicated that physical abuse is more accurately identified and reported. Potential differences may be related to different vignette characteristics between the current study and those in Beck et al. (1994), Christodoulou et al., (2019), Toros and Tiirik (2016), and Vanderfaeillie et al., (2018). Further, parenting practices surrounding discipline vary across cultures (e.g., no form of corporal punishment is prohibited by law in Greece; Government of Greece, 2009) and, as such, these difference in reporting may be reflected in teachers reporting behaviours. Also, Beck et al.’s (1994) study is dated, and it is highly likely that teachers’ knowledge of child maltreatment has changed over the years, especially for maltreatment types that have garnered more recent attention (i.e., neglect and emotional abuse). In contrast, Webster et al.’s (2005) study with U.S. teachers aligned with our results in that under-reporting was less frequent in cases of emotional abuse relative to physical abuse. Webster et al. (2005) suggested these findings could be due to teachers perceiving signs of emotional abuse as less ambiguous than those for physical abuse. This aligns with our findings as there were fewer comments for the emotional abuse vignette related to uncertainty and a need for more information.

Limitations and Future Research Directions

Well-designed vignettes have been found to help researchers examine intention and decision-making processes for child maltreatment (Walsh et al., 2008; Tufford & Lee, 2020). Although we adopted a rigorous approach in adapting our vignettes, this study is limited by the fact that vignettes are not able to evaluate real-life decision-making and cannot account for all nuances in individual cases (Walsh et al., 2008). Further, while the vignette measure used was the most recent and rigorous measure at the time of our study, the vignettes were not equivalent in terms of controlling for child, parent, and maltreatment characteristics and failed to capture teachers’ intent to detect and report intimate partner violence. Future research with larger samples might benefit from a factorial survey design to control for such factors (e.g., type of abuse, severity of abuse, child age, ethnicity and gender, informant characteristics). This would allow for a more rigorous assessment of the relationship between each individual factor and the decision made by teachers. Further, randomization of factors within the vignettes, as well as the order in which vignettes are presented, would make for a more robust experimental design and the possibility of making conclusions about causal relationships. This possibility would address a second study limitation around order effects since the order of presentation of the vignettes was not random. Third, for the question about whether teachers were likely to make a report of suspected maltreatment, it was not clearly stipulated as to who the teachers were to report even though the inference was that the report would be made to a child protection agency. Lastly, this study was based on a convenience sample of teachers from a large urban region in Ontario (Canada). It is likely that the teachers who participated viewed the training more positively and saw benefits from learning about issues related to child maltreatment. Subsequently, it is unclear if the results can be generalized to teachers in other settings. However, given that much of the existing literature is based on self-report data, and few studies have explored Canadian teachers’ behaviour, we believed this was an important step in guiding future research.

Applied Implications

Prior research suggests that teachers under-detect and under-report child maltreatment, though significant gaps remain in our understanding of influencing factors. To examine this issue further in a Canadian context, we explored a small group of Ontario teachers’ responses to different types of child maltreatment. We found variability in detection and reporting rates, not only with respect to chance expectations but also with respect to one another. The rates were especially low for accurate detection and reporting of physical abuse; therefore, exploration of why physical abuse can be challenging to identify, and report may be an important avenue for future studies in Canada. Considering the dearth of research in this area with Canadian teachers, additional studies are needed to evaluate these findings and more acutely discern differences in teachers’ detection and reporting behaviours. Further, better understanding is required around the saliency of and interactions between case, teacher, and system factors in the decision-making of educators (Alazri & Hanna, 2020).

For other professionals who may be supporting teachers with maltreatment reporting (e.g., child welfare practitioners, school administrators), our findings highlight potential areas for improved education and tailored guidance. For instance, other professionals could reassure teachers that they do not need to be certain about their suspicions in order to report their concerns about child maltreatment, as well as continue to engage in discussions alongside teachers about what might constitute “reasonable grounds” for reporting. Further, our results suggest that many teachers only shared their suspicions within their school system. As such, additional child maltreatment training initiatives would benefit from the inclusion of all school personnel so that there is a shared understanding of reporting obligations and a supportive school environment.

Current findings, while preliminary, suggest that future teacher training (and other interventions aimed at improving teachers’ detection and reporting behaviours) would benefit from including the following: (1) similarities and differences in child symptoms and parent risk factors across maltreatment types; (2) common concerns among teachers that can lead to under-detection and/or under-reporting (e.g., fears around informant credibility and repercussions for making a mistake when reporting); (3) what constitutes enough information to meet “reasonable grounds” for reporting and how this may look different across different forms of maltreatment. Also, this study illustrated that particular attention may need to be paid to physical abuse and the ambiguities surrounding it (e.g., Sect. 43 of Canada’s Criminal Code, which permits corporal punishment, bruising from age-appropriate activities). Other contextual factors identified in this study, though not a primary focus, would appear important to address in teacher training. For instance, diverse literature has identified cultural factors related to reporting behaviours (e.g., de Haan et al., 2019; Gubbels et al., 2021; Nagy et al., 2019; Vanderfaeillie et al., 2018) so it would be important to introduce this topic in future child maltreatment teacher training (e.g., how to understand various parenting practices within a culturally sensitive perspective while also balancing reporting responsibilities). Experiential exercises whereby teachers could discuss various case scenarios and their complexities would seem to be a particularly effective approach to helping improve child maltreatment detection and reporting behaviours.

Author Contributions

All authors contributed to the study conception and design. Material preparation and data collection were performed by Kelly Weegar. Analysis were performed by Amanda Glouchkow. The first draft of the manuscript was written by Amanda Glouchkow and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding

No funding, grants or other support was received for conducting this study.

Availability of Data Material

Not applicable.

Code Availability

Not applicable.

Declarations

Ethics Approval

Approval was obtained from the Office of Research Ethics and Integrity Board at the University of Ottawa. The procedures used in this study are in accordance with the Tri-Council Policy Statement (2014). (Date: September 2019 No: H-09–19-4902).

Consent to Participate

Informed consent was obtained from all individual participants included in the study.

Consent for Publication

Not applicable.

Conflicts of Interest/ Competing Interests

The authors have no conflicts of interest/competing interests to declare for this article.