Case, teacher and school characteristics influencing teachers' detection and reporting of child physical abuse and neglect: results from an Australian survey.

Abstract

To identify the influence of multiple case, teacher and school characteristics on Australian primary school teachers' propensity to detect and report child physical abuse and neglect using vignettes as short hypothetical cases.
A sample of 254 teachers completed a self-report questionnaire. They responded to a series of 32 hypothetical physical abuse and neglect scenarios by rating each of the vignettes on a 5-point scale for likelihood of abuse/neglect (detection) and likelihood to report (reporting). Teacher and school characteristics were also captured.
Multivariable multilevel analysis was used because of the hierarchical structure of the data with teachers nested within schools. A modest proportion of the variance in teachers' detecting and reporting scores was attributable to school membership. In the full model, case characteristics were found to exert the strongest influence on detecting and reporting tendency, in particular the type, frequency and severity of child physical abuse or neglect were the most important predictors of detection and reporting. At the teacher level, attention to legal reporting obligations was found to be the strongest and most significant predictor of reporting. The effect of teachers' training on both detecting and reporting emerged as a counter-intuitive finding. At the school level, characteristic effects were not as strong.
Teachers detecting and reporting CAN is a complex decision-making process. The most important determinants of teacher decision making are case characteristics. These characteristics impact upon both detection and reporting. Future research should be directed towards identifying and testing the influence of other teacher and, to a lesser extent, school characteristics that were not included in the current study. Further research is also required to identify the components, nature and duration of appropriate training for teachers and the links between these features and reporting outcomes.
Findings highlight the need for ongoing evaluation and enhancement of teacher education in CAN. The study underlines the importance of educating teachers about: (a) the warning signs and indicators of different types of CAN; (b) the differential effects of CAN; (c) responding to child victims including responses to direct disclosures; and (d) accurate and timely reporting.

Full text

Page 1

QUT Digital Repository:
http://eprints.qut.edu.au/
Walsh, Kerryann M. and Bridgstock, Ruth S. and Farrell, Ann M. and Rassafiani, Mehdi
and Schweitzer, Robert (2008) Case, teacher and school characteristics influencing
teachers' detection and reporting of child physical abuse and neglect: Results from an
Australian survey. Child Abuse and Neglect, 32(10). pp. 983-993.

© Copyright 2008 Elsevier

Page 2

Teachers detecting and reporting child physical abuse and neglect
1
Title
Case, teacher and school characteristics influencing teachers’ detection and reporting of child
physical abuse and neglect: Results from an Australian survey.
Authors
Kerryann Walsh, School of Early Childhood, Queensland University of Technology, Brisbane,
Australia
Ruth Bridgstock, Centre for Learning Innovation, Queensland University of Technology, Brisbane,
Australia
Ann Farrell, School of Early Childhood, Queensland University of Technology, Brisbane, Australia
Mehdi Rassafiani, Centre for Learning Innovation, Queensland University of Technology, Brisbane,
Australia
Robert Schweitzer, School of Psychology and Counselling, Queensland University of Technology,
Brisbane, Australia
Corresponding author/ Reprint requests
Kerryann Walsh
School of Early Childhood
Queensland University of Technology
Victoria Park Road
Kelvin Grove 4059
Queensland, Australia
k.walsh@qut.edu.au
Phone +61 7 3138 3174
Fax +61 7 3138 3989
Keywords
Child abuse, teachers, detection, reporting
Acknowledgements
This study was funded by the Abused Child Trust (Grant No RM2003000890). The Abused Child
Trust is the leading provider and advocate of recognised, quality services for the prevention and
treatment of child abuse and neglect in Queensland. We acknowledge in-kind support from the
Queensland University of Technology’s School of Early Childhood and the Faculty of Education’s
Centre for Learning Innovation. Special thanks to Ray Duplock for survey design and analysis
advice, and Vincent Conway and Rebecca Mathews who provided research assistance in the design
phase of the study. Ethical approval was obtained for this research from the Queensland University
of Technology’s Human Research Ethics Committee (Ref No QUT 2740H). We thank the
reviewers and editor for the many helpful remarks that led to significant improvements in this
paper.

Page 3

Teachers detecting and reporting child physical abuse and neglect
2
Abstract
Objective: To identify the influence of multiple case, teacher and school characteristics on
Australian primary school teachers’ propensity to detect and report child physical abuse and neglect
using vignettes as short hypothetical cases.
Methods: A sample of 254 teachers completed a self-report questionnaire. They responded to a
series of 32 hypothetical physical abuse and neglect scenarios by rating each of the vignettes on a 5-
point scale for likelihood of abuse/neglect (detection) and likelihood to report (reporting). Teacher
and school characteristics were also captured.
Results: Multivariable multilevel analysis was used because of the hierarchical structure of the data
with teachers nested within schools. A modest proportion of the variance in teachers’ detecting and
reporting scores was attributable to school membership. In the full model, case characteristics were
found to exert the strongest influence on detecting and reporting tendency, in particular the type,
frequency and severity of child physical abuse or neglect were the most important predictors of
detection and reporting. At the teacher level, attention to legal reporting obligations was found to be
the strongest and most significant predictor of reporting. The effect of teachers’ training on both
detecting and reporting emerged as a counter-intuitive finding. At the school level, characteristic
effects were not as strong.
Conclusions: Teachers detecting and reporting CAN is a complex decision-making process. The
most important determinants of teacher decision making are case characteristics. These
characteristics impact upon both detection and reporting. Future research should be directed
towards identifying and testing the influence of other teacher and, to a lesser extent, school
characteristics that were not included in the current study. Further research is also required to
identify the components, nature and duration of appropriate training for teachers and the links
between these features and reporting outcomes.
Practice implications: Findings highlight the need for ongoing evaluation and enhancement of
teacher education in CAN. The study underlines the importance of educating teachers about: a) the

Page 4

Teachers detecting and reporting child physical abuse and neglect
3
warning signs and indicators of different types of CAN; b) the differential effects of CAN; c)
responding to child victims including responses to direct disclosures; and d) accurate and timely
reporting.

Page 5

Teachers detecting and reporting child physical abuse and neglect
4
Introduction
Australian data on the sources of notifications for finalised child protection investigations show that
the most common sources of these notifications in the year spanning mid 2004 to mid 2005 were
school personnel (including teachers), police, and parents or guardians (Australian Institute of
Health and Welfare [AIHW], 2006). Rates of teacher reports vary but generally, those States or
Territories with mandatory reporting laws for teachers have higher rates of teacher reports (AIHW,
2006). Most Australian States and Territories have legislation compelling teachers to report
knowledge or suspicion of child abuse and neglect (CAN). However, these mandatory reporting
laws have differences such that teachers in different States and Territories have different legislative
duties to report (Mathews & Walsh, 2004a, 2004b; Mathews, Walsh, Butler, & Farrell, 2006).
There were no legislative reporting obligations for teachers in the state of Queensland before 2004,
at which time a narrow duty was introduced in the form of the Education and Other Legislation
(Student Protection) Amendment Act 2003, requiring Queensland teachers to report known or
suspected child sexual abuse perpetrated by school staff. Reporting obligations for Queensland
teachers are, therefore, extremely limited compared to the obligations for teachers in other
Australian jurisdictions. Most of the CAN likely to come to the attention of Queensland teachers,
such as physical abuse and neglect, they are not required by law to report (Bromfield & Higgins,
2005; Mathews et al., 2006).

Although Queensland teachers do not have mandatory reporting obligations to report all forms of
CAN, they have strong, but less binding, institutional policies regarding reporting of all forms of
suspected CAN. In Queensland, the policy for government (or State) school teachers requires their
reporting of harm to school students via the school principal (Education Queensland, 2004). Non-
government schools must have policies and procedures in place for teacher reporting of all forms of
CAN.

Page 6

Teachers detecting and reporting child physical abuse and neglect
5
A further context for this study is the nature and extent of teacher training in CAN. Noting the
critical role which teachers can play in detecting and reporting CAN, Queensland teachers have
been compulsorily trained to recognise and respond to CAN since 1999. Training comprises a short
three-hour school-based interactive workshop using a standard package comprising audiovisual
segments, activities, question and answer clarification, and small-group discussions on case
scenarios (Education Queensland, 1998, 2003, 2004). Training is delivered by school leaders,
generally principals and guidance officers (or school counsellors) rather than child protection
specialists. The most recent version of the training package, the one in place at the time of the
study, supports institutional policy and instructs school staff to respond to 4 categories of harm: i)
harm caused by an education employee; ii) harm caused by other students; iii) harm caused by
forces outside the state educational institution environment; and iv) student self harm.

The present study comprises analysis of a vignette-based component of a larger study designed to
provide a cross-sectional snapshot of Queensland primary (or elementary) teachers’ CAN detecting
and reporting practices. In the vignette component of the study, we focused on teachers detecting
and reporting two forms of CAN, child physical abuse and child neglect. We did this for two
methodological reasons. First, to minimise chances of measurement bias because the topic of child
sexual abuse by school staff was extremely sensitive in Queensland in 2004 after the Report of the
Board of Inquiry into Past Handling of Complaints of Sexual Abuse in the Anglican Church
Diocese of Brisbane (O’Callaghan & Briggs, 2003) and teachers were, understandably, preoccupied
with child sexual abuse. Second, restricting the vignettes to child physical abuse and neglect
enabled us to present brief and practical scenarios which teachers would be most likely to encounter
in their normal work.

Page 7

Teachers detecting and reporting child physical abuse and neglect
6
Literature
Notifying child protection authorities of known or suspected CAN, whether mandatory or not, has
long been a contested issue for teachers (Mathews & Walsh, 2004a, 2004b; Walsh, Farrell,
Bridgstock, & Schweitzer, 2006) and there is little from the field of educational research to inform
understandings of teachers’ decision making in cases of CAN. To address this gap in the research,
at a conceptual level we adapted Dalgeish’s (1998, 2003) General Judgement and Decision Making
(GJDM) model to investigate teachers’ detecting and reporting CAN. Within this model, teachers’
decisions to notify were conceptualised as a two-part process in which teachers must first detect
CAN, and then report it (Egu & Weiss, 2003). At the judgement stage, it was proposed that teachers
ask themselves “Is this abuse?” - this is a detection question. At the action stage, teachers ask
themselves “Will I report it?” - this is a reporting question. Moreover, using the GJDM model
teachers’ complex professional judgements and actions were viewed as being influenced by
different factors and considerations at each stage of the two-part process. At the judgement
(detection) stage, it is proposed that teachers attend primarily to characteristics of the case, that is,
the distinguishing signs and symptoms of CAN: its seriousness, frequency and impact on the child.
At the action (reporting) stage, it is anticipated that teachers respond (or not respond) to suspicions
of CAN based on their own personal and school characteristics: their knowledge of laws, policies
and procedures for reporting CAN, their personal values and attitudes, their beliefs about whether
reporting is likely to lead to a positive outcome for the child, and the features of their school
environment that facilitate or impede reporting. Using the GJDM model, therefore, aided our
understanding of how teachers arrived at the point of notification. Studying factors influencing
teachers’ detecting and reporting practices, holds the potential to advance understanding of teacher
decision making in cases of CAN and inform enhancements to policy and training.

Decision-making researchers in the field of reporting of CAN have highlighted the influence of
multiple variables on teachers’ detecting and reporting practices. Specifically, three groups of

Page 8

Teachers detecting and reporting child physical abuse and neglect
7
characteristics: case, teacher and school characteristics, have been identified in empirical research
as influencing teachers’ propensity to detect and report CAN (see for example Hansen, Bumby,
Lundquist, Chandler, Le, & Futa, 1997; O’Toole, Webster, O’Toole, & Lucal, 1999; Turbett &
O’Toole, 1983; Warner-Rogers, Hansen & Speith, 1996; Webster, O’Toole, O’Toole, & Lucal,
2005; Zellman & Bell, 1990). Of these three groups of influences, Zellman and Bell (1990), and
O’Toole et al. (1999) found case characteristics to be most influential in guiding decisions. For
example, teachers are more likely to recognise and report cases of physical abuse over other types
of CAN (Crenshaw, Crenshaw, & Lichtenberg, 1995) and to report physical abuse or neglect when
they have observed and documented it over time or when there is a serious event that provokes an
immediate response (Hawkins & McCallum, 2001a, b). They are also more likely to report when
they are cognisant of the impact of the abuse on the child (Shor, 1997). Other case characteristics
such as the socioeconomic status of the child’s family have been found to have a mixed effect on
detecting and reporting tendency (Turbett & O’Toole, 1983; O’Toole et al., 1999; Zellman & Bell,
1990). Although teachers consider the quality of their relationship with a child’s parents, parent
cooperativeness has been found to have inconsistent effects on reporting tendency (Crenshaw et al.,
1995).

Studies of the effects of teacher characteristics on detecting and reporting CAN have yielded mixed
results for gender (Kenny, 2001; O’Toole et al., 1999), parental status (Nightingale & Walker,
1986; O’Toole et al., 1999), years of experience (Crenshaw et al., 1995; Kenny, 2001; O’Toole et
al., 1999), teaching context (Beck, Ogloff, & Corbishley, 1994; O’Toole et al., 1999) and amount of
training (Nightingale & Walker, 1986; Zellman & Bell, 1990). Interestingly, higher teacher
education levels do not appear to predict CAN detection or reporting (O’Toole et al., 1999) while,
conversely, teachers’ desires to fulfil mandatory reporting obligations have been found to exert a
strong positive influence on reporting tendency (Hawkins & McCallum, 2001a, b). Confidence
levels in relation to accurately identifying CAN have been found to impact reporting decisions

Page 9

Teachers detecting and reporting child physical abuse and neglect
8
(Crenshaw et al., 1995; Kenny, 2001, 2004). Further, past experiences of detecting and reporting
have been associated with current detecting and reporting tendencies (O’Toole et al., 1999), but
studies have also shown that teachers who have reported also sometimes fail to report (Kenny,
2001), highlighting the complex discretionary nature of this phenomenon.

The effect of school characteristics on teachers detecting and reporting has received little attention
in the research literature. There is mixed evidence that school size influences reporting tendency
with both larger schools and schools with lower child-teacher ratios tending to make more reports
(O’Toole et al., 1999). The effects of other important school characteristics such as school
socioeconomic status, the perceived frequency of CAN in the school community, and the overall
school climate as measured by open discussion of CAN at the school, are yet to be tested in
empirical research.

The purpose of the present study was to expand this literature base on the effects of case, teacher
and school characteristics on teachers’ detecting and reporting CAN by addressing shortcomings in
the existing research. Most notably, although associations have been made at the bivariate level,
only O’Toole and colleagues (1999) have previously attempted to model the combined effects of
case, teacher and school characteristics on detecting and reporting tendency. Further no studies have
incorporated into their analyses the hierarchical structure of data, wherein teachers and, arguably,
their reporting practices, are nested within schools (Ciarleglio & Makuch, 2007).

The main research questions addressed by the vignette study were: i) to what extent do case, teacher
and school characteristics influence teachers’ propensity to detect and report hypothetical cases of
child physical abuse and neglect (CPN); ii) which group of variables has the greatest effect; and iii)
specifically, how much variance can be accounted for at the school level? Therefore, consistent with
previous research (e.g., O’Toole et al., 1999; Zellman & Bell, 1990), it was hypothesised that

Page 10

Teachers detecting and reporting child physical abuse and neglect
9
detection and reporting of CPN would vary according to the characteristics of the case, the teacher
and the school. It was predicted that case characteristics would exert the strongest influence on
detecting and reporting tendency, and that a proportion of the variance in detecting and reporting
tendency would be attributable to the nesting of teachers within their schools.

Method
Survey instrument
Data were collected using a survey instrument, the Child Abuse and Neglect Teacher Questionnaire
2004 (CANTQ2004), developed for teachers by the research team based upon findings of previous
research in Table 1. The CANTQ2004 comprised initial sections measuring teacher and school
characteristics. In the final section of the survey, vignettes were used to measure case characteristics
influencing teachers’ propensity to detect and report hypothetical cases of child physical abuse and
neglect. Vignettes have been used widely as a research tool for analysing professionals’ judgements
and actions, and as a means of eliciting group similarities and differences in responses (see for
example Crenshaw et al., 1995; Dalgleish, 2003; O’Toole et al., 1999; Taylor, 2006; Williams &
Soydan, 2005; Zellman & Bell, 1990). A vignette design has clear merits when it is carefully
planned to elicit decisions closely resembling those made in normal work situations. Taylor (2006)
argues such a design might even produce more valid results than decision-makers’ own accounts of
their actual practice. It must be acknowledged, however, that because this was a vignette-based
study, it did not assess actual decisions; although teachers’ personal histories of detecting and
reporting were collected as part of the demographic information and included as one of the teacher
variables listed below.

Vignette features were based on the findings of empirical research. Five case characteristics were
selected as independent (explanatory) variables: i) type of CAN (physical abuse/neglect); ii)
frequency of CAN (high repeat/low repeat); iii) impact of CAN on the child (high impact/low

Page 11

Teachers detecting and reporting child physical abuse and neglect
10
impact); iv) socio-economic status of the child’s family (high SES/low SES); and v)
cooperativeness of the child’s parent(s) (high cooperation/low cooperation). These variables were
measured dichotomously (e.g., high repeat/low repeat; high impact/low impact). Each vignette
comprised a series of statements in a fixed order containing five case variables relevant to teachers
detecting and reporting. Thirty-two vignettes were generated so that each level of each case variable
was manipulated with each level of each other case variable. For example:
A child comes to school with a badly bruised arm. You have noticed this on a number of
occasions previously. The child can perform usual activities. The family appear to be
struggling financially. The child’s parents are seldom seen at school and have been
defensive in interactions with you.
Participants were asked two key questions after each vignette: a detection question (How likely is it
that this child has been abused or neglected?); and a reporting question (How likely is it that you
will report this case?). Responses were rated on a 5-point likert type scale with 1 indicating most
unlikely and 5 indicating most likely.

Eleven teacher characteristics were selected as independent (explanatory) variables and questions
were completed to determine each of the following: i) gender (female/male); ii) age (in years); iii)
parental status (parent/non-parent); iv) qualification level (undergraduate/postgraduate); v) years of
teaching experience; vi) teaching level (lower primary years P-3/upper primary years 4-7/others);
vii) formal training in CAN (yes/no); viii) belief in legal obligation to report CAN (yes/no); ix)
confidence in identifying child physical abuse and neglect; x) previously suspected CAN (yes/no);
and xi) previously reported CAN (yes/no).

Four school characteristics were selected as independent (explanatory) variables. Teachers
completed questions eliciting information about: i) school size (less than 200 students/200-500
students/more than 500 students; ii) socioeconomic status of children at the school

Page 12

Teachers detecting and reporting child physical abuse and neglect
11
(low/average/high); iii) discussion about CAN at the school (yes/no); and iv) perceived frequency
of CAN in the school community (more/equal/less than other communities).

Applying a multivariable multilevel design enabled us to assess, simultaneously, the influence of
multiple explanatory variables and groups of variables on the two key outcome (dependent)
variables: i) teachers’ propensity to detect CPN (detection) as measured by teachers’ responses to
the vignette questions asking ‘how likely is it that this child has been abused or neglected?’ and ii)
teachers’ propensity to report CPN (reporting) as measured by teachers’ responses to the vignette
questions asking ‘how likely is it that you will report this case?’. This approach is particularly
helpful for identifying relationships among diverse influences in episodes involving considerable
complexity such as decision making in research contexts wherein data have a hierarchical or nested
structure situations (Ciarleglio & Makuch, 2007; Hair, Anderson, Tatham, & Black, 1998).

Participants and procedure
Approval to conduct the study was obtained from the University Human Research Ethics
Committee (UHREC) and from Education Queensland’s Strategic Policy and Education Futures
Branch. Registered teachers working in Queensland government (or State) primary schools,
preschool to seventh grade (teaching children aged 5-12 years) in 11 administrative districts in
South East Queensland (N=302) were invited to participate in the study via a letter to their school
principals (or head teachers). One hundred and nineteen principals (39.4%) responded. Of these, 35
(11.6%) accepted the invitation to participate and 84 (27.8%) declined. This sampling method was
(and still is) the only approved way to recruit teachers working in government schools for research
purposes. Participating schools had slightly higher student enrolments, that is, they were larger
schools (Mean = 617.35, SD 534.01) compared with schools overall (Mean = 464.39, SD 303.47).
Across most of the 11 districts, sample schools were proportionately distributed, but there was
evidence of over sampling from one district (10.6% of schools overall, response rate of 23.53%)

Page 13

Teachers detecting and reporting child physical abuse and neglect
12
and undersampling from another district (9.93% of schools overall, response rate of 2.94%). To
contextualise these differences, if only 4 less schools in the first district and 2 more schools in the
second district had participated the results would have been proportionate.

A total of 1,077 surveys were distributed to teachers in 35 participating schools. Completed
questionnaires were returned by 254 teachers resulting in a final response rate of 23.6%. Reflecting
the overall profile of teachers in Queensland government primary schools, participating teachers
were predominantly female (86.2%), and the largest proportion of teachers was in the age-range 41-
50 (36.9%) (Department of Education and the Arts, 2005). Teachers in the sample had an average
of approximately 15 years teaching experience (M = 14.68 years, SD = 9.3) which is comparable
with that of teachers overall and with sample data reported in previous studies (for example
Crenshaw et al., 1995; O’Toole et al., 1999; Webster et al., 2005). There was a good spread of
teachers from all year levels within the schools and from schools’ administration, with 40.6% of
respondents working in the upper school (years 4-7), 36.2% in the lower school (preschool to year
3), and 23.2% in other school administration or specialist teaching roles. We believe this study is
the largest of Australian primary teachers’ detecting reporting practices to date and it is certainly the
largest Australian study of teachers as non-mandated reporters.

Data analysis
Multivariate multilevel modelling was used to estimate the explanatory power of independent
variables and to estimate the proportion of variance in the dependent variables that could be
explained at the case, teacher and school levels. Multivariate approaches are required when there is
a need to analyse more than one dependent variable at one time (Twisk, 2006), and multilevel
modelling is appropriate for explaining relationships between variables measured at different levels
in a hierarchical or nested structure (Ciarleglio & Makuch, 2007). A multivariate multilevel
modelling approach assumes that responses from teachers within each school are not independent of

Page 14

Teachers detecting and reporting child physical abuse and neglect
13
one another. Multilevel modelling can account for such dependencies by estimating variance
associated with school membership (Tabachnick & Fidell, 2007).

Data analyses were performed using the statistical program MLwiN 2.02 (Rasbash, Steele, Browne,
& Prosser, 2005). To explain the variation in teachers’ detecting and reporting, and to test the
hypothesis that teachers’ detecting and reporting of CPN would vary according to the characteristics
of the case, the teacher and the school, we developed three models with three levels. The response
indicators - the two dependent variables, detection and reporting - at the first level were nested
within teachers at level two, and nested within schools at level three. Model 1 included only the
explanatory variables related to the case variables. Model 2 included case variables and teacher
variables. Model 3 included all explanatory variables together - case variables, teacher variables and
school variables.

The fit of each nested model was compared with the previous model using the likelihood ratio test
(-2 log-likelihood), wherein each successive model was compared with the previous model: the first
model was compared with the empty (base) model without explanatory variables, the second model
was compared with the first model and so on. The differences between the models having a chi-
squared distribution were compared to determine significant changes between models. The
proportion of variance explained by each of the models, and the unexplained variance remaining
after all predictor variables were included in Model 3 was calculated using the R2 statistic. A
variance partition coefficient (VCP) was also calculated for the base model to identify the
proportion of the total residual variation solely attributable to differences between schools (Rasbash
et al., 2005).

Page 15

Teachers detecting and reporting child physical abuse and neglect
14
Results
From the 254 completed surveys, a total of 8128 vignette responses were obtained (254 teachers x
32 vignettes) for each of detecting and reporting. The mean vignette detection rating was 2.79 (SD
= 0.56), and the mean vignette reporting rating was 2.82 (SD = 0.83). These results (Table 1) show
that, generally, if teachers in this sample detected a case of child physical abuse or neglect, they
tended to report it. A strong relationship between detection and reporting scores was confirmed at
the bivariate level (Pearson’s r = 0.72, p<.0001).

Insert Table 1 about here

Table 2 shows the results of three fitted models: Model 1 containing only case variables; Model 2
containing case and teacher variables; and Model 3 containing case, teacher and school variables.
The numbers in the columns show the estimated effect (β coefficient), the standard error of the
effect, and the significance (p-value) of the tests for each model. The -2 log likelihood value for
each model illustrates a significant difference between each successive model (p< .001) confirming
the better fit of the more elaborate model (Model 3) to the data.

The R2 value in Model 1 at the bottom of the first two columns of Table 2, shows that the greatest
proportion of variance is distributed at the case level for both detecting (20%) and reporting
(16.7%). Hence, case characteristics alone accounted for exactly one-fifth of the variation in
teachers’ responses for detecting and slightly less than one-fifth of the variation in their reporting
responses. This remains consistent across the three models indicating the stability and importance of
case level factors as predictors even when other variables are added.

In Model 3, the R2 value indicates that a significant yet moderate proportion of the variance in
teachers’ detecting (31%) and reporting (20.1%) scores was explained by the variables measured