Vous êtes sur la page 1sur 11

Journal of Research in Special Educational Needs  Volume 16  Number 1  2016 2–12

doi: 10.1111/1471-3802.12047

The teacher efficacy for inclusive practices (TEIP)


scale: dimensionality and factor structure
Mi-Hwa Park,1 Dimiter M. Dimitrov,2 Ajay Das1 and Margaret Gichuru3
1
Murray State University; 2George Mason University; 3Western Kentucky University

Key words: self-efficacy, inclusion, pre-service teachers.

also been a growing interest in the way pre-service teachers


The Teacher Efficacy for Inclusive Practices (TEIP) perceive and respond to children with disabilities. In addi-
scale is designed to measure teacher-self efficacy tion, there has been a focus on whether pre-service teachers
to teach in inclusive classrooms. The original study are adequately prepared to provide effective instruction and
identified three scale factors: efficacy in using care to these young children with disabilities (Malone,
inclusive instruction (EII), efficacy in collaboration
Straka and Logan, 2000; Sexton, Snyder and Wolfe et al.,
(EC), and efficacy in managing behavior (EMB)
(Sharma et al., 2012). The purpose of our study was
1996). Studies in different specialisation areas show that
to examine the TEIP scale for dimensionality and to self-perceptions on individuals’ confidence and competence
cross-validate its factor structure for pre-service facilitate the connection between the training received and
teachers in the context of early childhood educa- how the practitioners utilise the currently attained knowl-
tion. A bifactor model fit to the data revealed that edge and skills (Das, Gichuru and Singh, 2013; Leyser,
the TEIP scale is essentially unidimensional, that is, Zeiger and Romi, 2011). Previous research also indicates
there is one dominant latent factor and the origi- that teachers’ self-efficacy towards inclusion shapes stu-
nally found three scale factors (EII, EC, and EMB) dents’ achievement and behaviours as well as teachers’
represent specific aspects of the general factor of attitudes and classroom management skills (Ahsan,
teacher self-efficacy to teach in inclusive class- Sharma and Deppeler, 2012; Gibson and Dembo, 1984;
rooms. Along with providing validation evidence,
Tschannen-Moran and Hoy, 2001). For example, Woolfolk
these findings have important implications for the
scoring on the TEIP scale using classical test analy-
(2007) argued that teachers’ perceived teaching efficacy has
sis or unidimensional item response theory models. a direct correlation with students’ academic achievement. In
addition, those teachers who demonstrate greater perceived
teaching efficacy apply a number of behaviour management
strategies, and they use more effective teaching methods
(Guskey, 1988; Woolfolk, Rosoff and Hoy, 1990).
Introduction
The movement to provide free and appropriate public edu- Some researchers have questioned whether pre-service
cation for young children with special needs has come teachers are adequately prepared to face the challenges of
about due to the Individuals and Disabilities Education including special needs children with disabilities in their
Improvement Act (2004) and the position paper of Division future classrooms (e.g., Druckman and Bjork, 1994). Fur-
of Early Childhood and National Association for Education thermore, pre-service teachers themselves have indicated
of Young Children (DEC/NAEYC, 2009). As a result, that they do not consider themselves to be adequately pre-
young children with special needs are increasingly receiv- pared to work with these children. For example, Mastropieri
ing services in their neighbourhood early childhood educa- and Scruggs (2000) reported that only about 25% of the
tion centres. Appropriate service delivery to these children, teachers believe that they possess the skills to implement
therefore, becomes critical. effective inclusion. McConkey and Bhlirgri (2003) con-
ducted a survey of 56 teachers working in 38 preschool in
The field of early childhood education is committed to Ireland and reported that even though most early childhood
supporting the inclusion of children with special needs in educators were devoted to including children with autism in
high-quality inclusive educational environments. The ser- their classrooms, a large number felt inadequately prepared
vices that young children with special needs receive will not and lacked knowledge to meet the needs of these children.
only depend on the severity of their disability, but also on In another study, Buell, Gamel-McCormick and Hallam
the knowledge and skills of the early childhood educators. (1999) surveyed 189 family childcare providers and
This notion, therefore, has brought about greater attention reported that the providers expressed a number of concerns
on how early childhood educators are prepared to serve when working with children with disabilities. The concerns
children that have special needs (Abbitt and Klett, 2004; included a lack of knowledge about disabilities, an addi-
Jarvis and Pell, 2004). Commensurate with this, there has tional burden of caring for a child with a disability while at

2 ª 2014 NASEN
Journal of Research in Special Educational Needs, 16 2–12

the same time working with typically developing children teachers because of the nature of their position. For example,
and the necessity of purchasing assistive technology. early childhood educators are required to collaborate with
speech, physical or occupational therapists to meet the needs
Within the context of teacher preparedness for inclusive of children with disabilities in their classrooms (Friend and
education, a number of researchers have explored the Cook, 2013). The value of encouraging collaboration among
construct of teacher self-efficacy (e.g., Loreman, Sharma teachers, school professionals and parents is a growing
and Forlin, 2013; Sharma, Loreman and Forlin, 2012; element in literature (Guo, Justice and Sawyer, 2011; Hoy
Weisel and Dror, 2006). Teachers’ self-efficacy beliefs are and Spero, 2005; McGinty, Justice and Rimm-Kaufman,
commonly examined through the theoretical lens of 2008). Increased collaboration among school personnel is
Bandura’s social cognitive theory (Labone, 2004; viewed as strong predictors of successful classroom instruc-
Tschannen-Moran and Hoy, 2001). Within this theoretical tion (Marks and Louis, 1997; McGinty, Justice and
framework, Bandura (1997) acknowledged the role that Rimm-Kaufman, 2008). Guo, Justice and Sawyer (2011)
cognitive factors play in the regulation of an individual’s note that collaboration is one of the main factors that may
behaviour. He suggested that ‘cognitive processes mediate enable teachers to improve their management of difficult
change, but that cognitive events are induced and altered situations connected to teaching children with low level of
most readily by experiences of mastery arising from effec- engagement, therefore supporting teachers’ self-efficacy. To
tive performance’ (Bandura, 1997, p. 191). Thus, in order to improve preschool teachers’ self-efficacy, it is critical to
successfully complete a task, an individual needs certain promote collaboration among preschool teachers because
knowledge and skills in addition to a sense of confidence so teachers’ higher level of collaboration is positively con-
that one’s efforts will be successful. In other words, accord- nected to self-efficacy (Guo, Justice and Sawyer, 2011).
ing to Bandura (1997), self-efficacy refers to one’s beliefs
in their ‘capabilities to organize and execute the courses of Contemporary approaches to measuring teacher efficacy are
action required to produce given attainments’ (p. 3). based on Bandura’s (1977; 1997) social cognitive theory
concepts to teachers. For example, the teacher efficacy scale
In the context of teaching, Bandura’s theory of self-efficacy (TES; Gibson and Dembo, 1984) is based on Bandura’s
implies that self-efficacy is related to the amount of effort self-efficacy and outcome expectancy dimensions of social
that teachers put into their performance and their persever- cognitive theory. The Gibson and Dembo’s TES consists of
ance in challenging tasks (Pajares, 1996). Teachers with a 16 items in Likert format, with nine items taping on per-
sense of low teaching efficacy may put little effort into their sonal teaching efficacy and seven items taping on general
work because they do not think they have the skills or teaching efficacy (GTE). However, some researchers have
resources to make a difference in the lives of the students. challenged the validity of the TES, particularly in regard
Furthermore, teachers with a sense of high teaching efficacy to its GTE scale (e.g., Guskey and Passaro, 1994;
would put forth an extensive amount of effort in the class- Tschannen-Moran, Hoy and Hoy, 1998).
room and would apply persistent efforts (Lamorey and
Wilcox, 2005). Bandura (1977; 1997) also argued that self-efficacy is best
conceptualised and measured as a multidimensional con-
While providing a strong support for the principle of inclu- struct and that researchers would find the most utility from
sion, previous research also focused on constructs related to self-efficacy by focusing on a specific context and activity
teachers’ instructional skills and behaviour management domain. That is, researchers should align a given activity
strategies (Romi and Leyser, 2006). A number of specific with self-efficacy for that activity rather than examining a
instructional strategies that early childhood educators would global assessment of self-efficacy. Following this logic,
particularly need in inclusive classrooms include embedded Bandura (1990) developed a set of scales, titled multidi-
instruction, differentiated instruction, and activity-based and mensional scales of perceived self-efficacy (MSPSE), to tap
experiential learning. In spite of challenges experienced by self-efficacy in nine domains: enlisting social resources,
pre-service teachers in managing student behaviour in inclu- academic achievement, self-regulated learning, leisure-time
sive settings, pre-service teachers that reported higher levels skills and extracurricular activities, self-regulatory efficacy
of self-efficacy had more success in managing student (to resist peer pressure for high-risk behaviours), self-
behaviour in inclusive classrooms (Main and Hammond, efficacy to meet others’ expectations, social self-efficacy,
2008; Mergler and Tangen, 2010). In addition, the pre- self-assertive efficacy and enlisting parental and community
service teachers expressed higher level of perceived teaching support. In line with the logic behind the MSPSE and based
efficacy in using inclusive instructions and the lowest level of on an analysis of role expectations in Norwegian schools,
perceived teaching-efficacy in collaborating with others. Skaalvik and Skaalvik (2007) developed a multidimen-
sional 24-item Norwegian teacher self-efficacy scale
Another important factor in this regard is the need for (NTSES) measuring six dimensions by four items each. The
increased collaboration with student families and profes- dimensions were self-efficacy for instruction, adapting edu-
sionals (Bruder, Dunst and Mogro-Wilson, 2011; Odom, cation to individual students’ needs, motivating students,
Buysse and Soukakou, 2011). Collaboration among teachers keeping discipline, cooperating with colleagues and
and other professionals is especially needed for pre-school parents, and coping with changes and challenges. Skaalvik

ª 2014 NASEN 3
Journal of Research in Special Educational Needs, 16 2–12

and Skaalvik (2010) used the NTSES to study relations importantly, there is no psychometric assessment of the
among perceived school context, teacher self-efficacy, dimensionality of the scale and the nature of the three scale
teacher burnout and teacher job satisfaction. factors identified in the original study, EII, EC and EMB.
Valid applications of the TEIP scale should be based on
Other scales have also been developed to reflect the speci- information that reveals whether (1) the scale data are
ficity of teacher efficacy in the context of a given subject essentially unidimensional, that is, whether there is one
matter. For example, Riggs and Enochs (1990) developed a dominant latent factor that underlies the examinees’ scores
subject matter instrument to measure efficacy for teaching on test items, and (2) the role of the three scale factors, EII,
science, referred to as the science teaching efficacy belief EC and EMB, are they different dimensions or do they just
instrument (STEBI). This instrument is based on the TES represent different aspects of one general factor of the tar-
and also consists of two largely uncorrelated subscales: geted teacher self-efficacy? Essential unidimensionality of
personal science teaching efficacy and science teaching scale data is a key assumption in using either classical test
outcome expectancy. In most applications, the STEBI con- analysis or item response theory (IRT) for calibration and
sists of 25 items with a five-point Likert-type scale. The scoring the responses on scale items (e.g., Nandakumar and
STEBI has been modified several times to address subject- Yu, 1996). The search for answers of these key questions is
specific teaching. For example, one such modification the motivation behind the efforts in this study.
resulted in the mathematics teaching efficacy belief instru-
ment (MTEBI) for pre-service teachers (Enochs, Smith and The main purpose of this study is to cross-validate the
Huinker, 2000). The MTEBI consists of 21 items related to three-factor structure of the TEIP scale (Sharma, Loreman
two independent subscales, namely: 13 items on the personal and Forlin, 2012) in the context of interdisciplinary early
mathematics reaching efficacy subscale and eight items on childhood education (IECE). It is important to note that the
the mathematics teaching outcome expectancy subscale. TEIP scale and its previous practical applications were used
in the context of elementary, secondary and college educa-
In the context of inclusive education of special needs chil- tion across different countries, but not in early childhood
dren, Sharma, Loreman and Forlin (2012) developed the education in the USA. Of particular interest in this study is
teacher efficacy for inclusive practices (TEIP) scale to to determine whether the TEIP scale data are essentially
measure teachers’ self-efficacy for inclusive education and unidimensional and whether the tree scale factors (EII, EC,
analysed TEIP data collected in four different countries. and EMB) represent different dimensions or they are just
Based on results from an exploratory factor analysis (EFA), specific aspects of the teacher self-efficacy in implementing
Sharma, Loreman and Forlin (2012) identified three factors inclusive practices. Related to this purpose are the follow-
that underlie the data for 18 items in the TEIP scale, namely: ing two research questions (RQs) addressed in the study.
efficacy to use inclusive instructions (EII), efficacy in col-
laboration (EC) and efficacy in managing behaviour (EMB). RQ1. Do the study data support the three-factor struc-
In another study, the TEIP scale was used to examine differ- ture of the TEIP scale?
ences in self-efficacy to teach in inclusive classrooms across
demographic characteristics of pre-service teachers from RQ2. Do the scale factors EII, EC, and EMB represent
Canada, Australia, Hong Kong and Indonesia (Loreman, independent latent factors or they are just specific
Sharma and Forlin, 2013). The TEIP scale was also used to aspects of a general factor of teacher self-efficacy in
identify variables that impact the perceived teaching efficacy implementing inclusive practices?
of pre-service teachers towards inclusive education in Ban-
gladesh (Ahsan, Sharma and Deppeler, 2012). Ahsan and his Method
colleagues reported an overall mean score of the perceived Sample
teaching-efficacy of the pre-service teachers on the TEIP This study was conducted at a regional university in Ken-
scale equal to 4.84 (standard deviation = 0.52). Therefore, tucky, USA. Participants were 134 undergraduate students
they concluded that ‘the pre-service teachers had relatively who registered for the IECE programme. IECE prepares all
high level of perceived teaching-efficacy towards inclusive teachers who graduated from the university for their future
education’ (Ahsan, Sharma and Deppeler, 2012, p. 10). work on providing young children (infants, toddlers, pre-
schoolers and kindergarten children with and without dis-
Purpose of the study abilities) and their families with an early intervention, care
Given the increasing trend of using the TEIP scale in studies and educational services. The data were collected at the
on the readiness of pre-service teachers to teach in inclusive beginning of the fall 2012 semester and the spring 2013
classrooms (e.g., Loreman, Sharma and Forlin, 2013), it is semester. The breakdown of the sample by background
important to provide more rigorous cross-validation of this characteristics of the participants is (1) 67.2% with less than
scale. The original study on the scale (Sharma, Loreman 10 hours of classroom observation of a child with disabili-
and Forlin, 2012) suggested a three-factor structure of the ties, (2) 77.6% with less than 10 hours of experience in
scale data based on EFA. However, there are no cross- teaching children with disabilities, (3) 38.1% at the begin-
validation studies of the proposed structure in the frame- ning of the teacher training programme and (4) 32.1% major-
work of confirmatory factor analysis (CFA). Most ing IECE, 53.% majoring elementary education and 14.2%

4 ª 2014 NASEN
Journal of Research in Special Educational Needs, 16 2–12

majoring learning and behaviour disorder. By gender and cally, item 20 (see Appendix A) is used here instead of item
ethnicity, the participants were 94% females and 91% White. 8 in the TEIP scale: ‘I am confident in my ability to get
students to work together in pairs or in small group’
Instrumentation (Sharma, Loreman and Forlin, 2012, p. 19). As both items
As described earlier, the TEIP scale is designed to measure are indicators of the same scale factor (EII), their ‘switch’ is
perceived teacher efficacy to teach in inclusive classrooms not problematic in this regard. In this study, the grouping of
(Sharma, Loreman and Forlin, 2012). The scale consists of items by latent factors (EII, EC or EMB) is based on the
18 items scored on a six-point Likert scale (1 = strongly EFA three-factor solution proposed in the original study on
disagree, 2 = disagree, 3 = disagree somewhat, 4 = agree the TEIP scale (Sharma, Loreman and Forlin, 2012, p. 16).
somewhat, 5 = agree and 6 = strongly agree). The middle For the data in our sample (N = 134), the Cronbach’s
category (‘neither agree nor disagree’) is omitted, most coefficient alpha for internal consistency reliability of the
likely with the purpose to eliminate the central tendency bias scores was found to be (1) 0.977, for the entire scale of 18
that occurs when respondents avoid using extreme response
categories (e.g., Dimitrov, 2012, p. 10). An EFA conducted in
the original study on the scale (Sharma, Loreman and Forlin, Figure 1: A hypothesised confirmatory factor analysis
(CFA) model with three correlated latent factors for the
2012) resulted in three factors labelled EII, EC and EMB. teacher efficacy for inclusive practices (TEIP) structure.
The reader may find the description of the items and their
grouping into EII, EC and EMB factors in the paper with the
original study (Sharma, Loreman and Forlin, 2012, p. 16).
The Cronbach’s alpha coefficient for internal consistency
reliability of the data for the norming sample of pre-service
teachers with the original study (N = 609) is reported to be
0.89 for the total scale score and ranging from 0.85 to 0.93
for the scores on the three scale factors, EII, EC and EMB.

In this study, we used an earlier survey version of the TEIP


scale that includes two additional items (items 6 and 20, see
Appendix A). Items 6 and 14 from the survey were
excluded from the analysis, after a preliminary test on data
fit, so 18 items were retained to address the RQs (RQ1 and
RQ2). With this, the 18 items used for the analysis in this
study are the same as those in the TEIP scale (Sharma,
Loreman and Forlin, 2012), with one exception. Specifi-

Figure 2: A bifactor model for the factor structure of the teacher efficacy for inclusive practices (TEIP) scale (see
Appendix A).

ª 2014 NASEN 5
Journal of Research in Special Educational Needs, 16 2–12

items, (2) 0.929, for the EII factor, (3) 0.953, for the EC Table 1: Descriptive statistics for teacher efficacy for
factor and (4) 0.944, for the EMB factor (each scale factor inclusive practices (TEIP) items by specific domains
is measured by six items). Item/domain Item/total
Domain/item* Mean SD correlation correlation
Statistical data analysis Efficacy to use inclusive instruction
As the two RQs (RQ1 and RQ2) involve testing for
Item 1 (1) 3.89 1.10 0.84 0.79
hypothesised factor models of the TEIP scale, a CFA was
used to address both the RQs. First, the three-factor model Item 2 (2) 3.98 1.02 0.87 0.85
depicted in Figure 1 was used to address RQ1. Item 3 (5) 3.73 1.12 0.86 0.85
Item 4 (6) 3.92 1.01 0.87 0.85
The CFA model depicted in Figure 2 was tested for data fit Item 5 (7) 3.81 1.15 0.85 0.80
to address RQ2. Under this model, referred to as bifactor Item 20** 3.87 1.10 0.86 0.88
model, there is one general latent factor that underlies the
Efficacy in collaboration
scores on the respondents on the construct being measured
(here, teacher self-efficacy to teach in inclusive classrooms) Item 13 (19) 4.01 1.20 0.905 0.861
and three specific latent factors (EII, EC and EMB) that Item 15 (23) 4.23 1.18 0.885 0.856
capture unique aspects of the construct. The general factor Item 16 (24) 3.95 1.18 0.932 0.912
is loading on all items of the scale, whereas each specific Item 17 (25) 4.23 1.25 0.927 0.880
factor is loading on a cluster of six items. All four factors Item 18 (26) 4.18 1.18 0.912 0.882
(one general and three specific) are uncorrelated because
Item 19 (28) 3.54 1.09 0.840 0.818
each specific factor accounts for the unique part of the
variance among the respective cluster of six items that is Efficacy in managing behaviour
unexplained by the general factor. An acceptable data fit of Item 7 (10) 3.82 1.05 0.915 0.851
the bifactor model would indicate that the TEIP scale data Item 8 (11) 3.84 1.01 0.911 854
are essentially unidimensional, with one general factor and Item 9 (12) 3.94 1.01 0.901 0.864
three specific latent factors (EII, EC and EMB) that repre- Item 10 (13) 4.11 1.04 0.924 0.911
sent unique aspects of teacher self-efficacy on teaching in
Item 11 (14) 3.53 1.16 0.828 0.744
inclusive classrooms. If this is the case, an important prac-
tical implication is that it is appropriate to analyse TEIP Item 12 (9) 4.18 1.11 0.845 0.853
scale data using classical or IRT procedures that assume Note: Given in parentheses are the item numbers used in Sharma, Loreman
essential unidimensionality and to score the performance of and Forlin (2012, p. 16).
the respondents on the general factor and on each of the *Items 6 and 14 from the 20-item teacher efficacy for inclusive practices
(TEIP) survey used in this study (see Appendix A) are excluded from the
specific factors (e.g., Chen, West and Sousa, 2006; Reise, analysis because of poor data fit under the confirmatory factor analysis
Morizot and Hays, 2007). (CFA) models depicted in Figures 1 and 2 [item 6 corresponds to item 8 in
Sharma, Loreman and Forlin, (2012, p. 16)].
**Item 20 is not included in the 18-item scale provided by Sharma,
The CFA analysis under the models in Figures 1 and 2 was Loreman and Forlin (2012, p. 16).
performed through the use of the computer program Mplus SD, standard deviation.
(Muthén and Muthén, 2010). It should be noted that the
CFA analyses were conducted by declaring in the Mplus for each item. Also, each item highly correlates with the
source code all dependent variables (18 items) as ‘categori- score on the corresponding domain-specific factor (EII, EC
cal’, to take into account that these variables are not treated or EMB) and with the total score on the entire TEIP scale.
as continuous but, instead, as ordered categorical variables. This classical psychometric information about the scale
Technically, this means that the CFA analysis is based on items, coupled with the Cronbach’s alpha coefficients
the polychoric correlations among the items, which are reported earlier, provide evidence of high internal consis-
more appropriate for categorical variables, instead of the tency of the measures for the TEIP scale and each of the three
Pearson correlations, which are appropriate for continuous domain-specific factors (EII, EC and EMB).
variables (e.g., Dimitrov, 2012, p. 104; Finney and
DiStefano, 2006; Muthén and Muthén, 2010). As described earlier, the testing for data fit under the factor
models depicted in Figures 1 and 2 was performed in the
Results framework of CFA using the computer program Mplus.
Basic descriptive statistics for the respondents’ scores on
individual items of the TEIP scale are provided in Table 1. As The results for data fit under the CFA models depicted in
described earlier, each item is scored on a six-point Likert Figures 1 and 2 are summarised in Table 2. The goodness-
scale, from 1 = strongly disagree to 6 = strongly agree, with of-fit indices reported in Mplus for CFA with categorical
a theoretical average of 3.50 for the item score. The exami- variables, are (1) chi-square (χ2) test, (2) comparative fit
nation of the mean item scores in Table 1 shows that the index (CFI), (3) Tucker–Lewis Index (TLI), (4) root mean
pre-service teachers in the study sample tend to score some- square error of approximation (RMSEA) and (5) weighted
what higher than the theoretical average (3.50) on the scale root mean square residual (WRMR). A statistically

6 ª 2014 NASEN
Journal of Research in Special Educational Needs, 16 2–12

Table 2: Testing for data fit of the three-factor and bifactor models for the structure of the teacher efficacy for
inclusive practices (TEIP) scale
90% CI for RMSEA
Model χ2 df CFI TLI WRMR RMSEA LL UL
Three factors 260.58 132 0.994 0.993 0.886 0.085 0.070 0.100
Bifactor model 181.56 117 0.997 0.996 0.651 0.064 0.045 0.082

CFI, comparative fit index; CI, confidence interval; df, degrees of freedom; LL, lower limit; RMSEA, root mean square error of approximation;
TLI, Tucker–Lewis Index; UL, upper limit; WRMR, weighted root mean square residual.

significant χ2 (p < 0.05) does not support data fit, but this rule Table 3: Standardised factor loadings under the
does not play a critical role in the decision on data fit as the bifactor model of the teacher efficacy for inclusive
χ2 test is very sensitive to sample size (e.g., Dimitrov, 2012, practices (TEIP) scale
p. 104). Rather, the assessment of model fit is based on the Domain-specific factors
joint evaluation of the fit indices, with cutting scores as Item TEIP general factor EII EC EMB
follows: (1) CFI > 0.95 for an excellent fit and CFI > 0.90 for
1 0.787 0.471
an acceptable fit; (2) TLI > 0.95 for an excellent fit and TLI
> 0.90 for an acceptable fit; (3) WRMR = 0.00 indicates a 2 0.874 0.286
perfect data fit, but in practice WRMR < 1.00 is used to 3 0.851 0.046
indicate an adequate fit; and (4) RMSEA = 0.00 indicates a 4 0.872 0.015
perfect fit, but in practice RMSEA ≤ 0.05 is used to indicate 5 0.818 0.021
an adequate data fit; using the 90% confidence interval for
20 0.907 0.111
RMSEA, an excellent data fit is indicated when the lower
13 0.849 0.301
value of this interval is close to (or includes) zero and its
upper value is smaller than 0.08 (e.g., Hu and Bentler, 1999; 15 0.874 0.236
Yu and Muthén, 2002). 16 0.919 0.214
17 0.860 0.417
Under the guidelines for model fit, the examination of 18 0.900 0.216
the results in Table 2 indicates that the three-factor
19 0.820 0.170
model, depicted in Figure 1, provides an acceptable data
7 0.822 0.464
fit. Also, the three latent factors were found to be highly
correlated, namely (1) 0.95 between EII an EC, (2) 0.94 8 0.858 0.440
between EII and EMB, and (3) 0.90 between EC and 9 0.877 0.267
EMB. These high correlations suggest that it is meaning- 10 0.945 0.174
ful to test for a bifactor model with one general latent 11 0.738 0.336
factor and three domain-specific latent factors (EII, EC
12 0.876 0.018
and EMB). The results in Table 2 indicate that the bifactor
model in Figure 2 provides better data fit compared Note: In bold are statistically significant factor loadings (P < 0.05).
with the three-factor model in Figure 1 by the criteria for EC, efficacy in collaboration; EII, efficacy to use inclusive instruction;
EMB, efficacy in managing behaviour.
all goodness-of-fit indices. For example, the WRMR,
which is particularly relevant for CFA models with cat-
egorical variables, is smaller under the bifactor model
(WRMR = 0.651) compared with the three-factor model of the variance in item 1, which is not explained by the
(WRMR = 0.886). general factor. Likewise, the squared factor loading for
item 2, (0.286)2 = 0.0818, shows that the EII factor
The standardised factor loadings under the bifactor model accounts uniquely for 8.18% of the item variance. The
are provided in Table 3. As can be seen, all factor loadings loadings for the remaining four items on the EII factor are
on the general factor are statistically significant and sub- not statistically significant, which indicates that the
stantial in magnitude, with a range from 0.738 to 0.945; residual variance in these items, after taking into account
(typically, loadings higher than 0.40 are considered the variance explained by the general factor, does not
salient). For proper interpretation of the loadings on the quite relate to the specificity that is captured by the EII
domain-specific factors, it is important to reiterate that factor. The factor loadings on the other two specific latent
these factors account for unique variation in their items, factors, EC and EMB, are all statistically significant, with
after controlling for the item variation explained by the the exception of the loading of item 12 on EMB. Thus, the
general factor. For example, the squared factor loading for items associated with the EC and EMB latent factors are
item 1 on the domain-specific factor EII, (0.471)2 = significant measures of the specific aspects of the targeted
0.2218, indicates that the EII factor accounts for 22.18% construct (TEIP) represented by these two factors.

ª 2014 NASEN 7
Journal of Research in Special Educational Needs, 16 2–12

Discussion it was found that these factors are highly correlated, thus
The TEIP scale is developed to measure self-efficacy of suggesting the presence of a general factor of teacher self-
school teachers to teach in inclusive classrooms (Sharma, efficacy to teach in inclusive classrooms. It should be
Loreman and Forlin, 2012). The importance of using valid reminded that one item in the TEIP scale reported in
TEIP measures stems from the conception that teacher’s Sharma, Loreman and Forlin (2012) was replaced by an
self-efficacy is critical in providing high-quality preschool item from an earlier survey version of the scale. Specifi-
programmes and increasing development and learning out- cally, item 20 in the survey (see Appendix A) was used here
comes of young children (Mashburn, Pianta and Barbarin instead of one item in the TEIP scale (see item 8 in Sharma,
et al., 2008). In spite of government investment to provide Loreman and Forlin, 2012, p. 19). This was done for con-
high-quality preschool programmes, reports suggest that sideration of data fit in testing the two CFA models in this
one third of US preschool programmes do not meet the set study (see Figures 1 and 2), and it is not considered as
standards and are of low quality (LoCasale-Crouch, Konold problematic to the factorial structure of the scale as both
and Pianta et al., 2007), and yet one factor that is connected items serve as indicators of the same latent factor, EII.
to teacher’s effectiveness to classroom instruction is the
sense of self-efficacy (Guo, Piasta and Justice et al., 2010; The second finding in this study is that the TEIP scale data
Justice, Mashburn and Hamre et al., 2008). is essentially unidimensional, with one dominating general
dimension and three specific latent factors that represent
A teacher’s ability to promote inclusion in early childhood unique aspects of the targeted construct, namely: EII, EC
classrooms is a critical component that directly contributes and EMB. An important practical implication of this finding
to the teacher’s effectiveness, even though many teachers is that it is appropriate to score the teachers’ performance on
view themselves as inadequately prepared to support chil- the entire TEIP scale and on each of its specific aspects (EII,
dren’s learning in inclusive settings (Darling-Hammond, EC and EMB) using classical test analysis or IRT models.
2006; Dozier and Berlotti, 2000). With the TEIP scale
(Sharma, Loreman and Forlin, 2012), this aspect of teacher- A brief comparison of the factorial structure of the TEIP, as
self efficacy is measured by items associated with the scale identified in this study, with the factorial structures of other
factor EII. Another factor measured by the TIEP scale scales on teacher self-efficacy also deserves attention. As
(Sharma, Loreman and Forlin, 2012) is the sense of collabo- described in the introduction, the Bandura’s conception of
ration in fostering teachers’ self-efficacy. Collaboration self-efficacy as a multidimensional construct has been
among preschool teachers validates and expands their reflected in the development of self-efficacy scales, which
knowledge of content and pedagogy, factors that increase the consist of two or more independent subscales. For example,
teacher’s sense of efficacy (Guo, Justice and Sawyer, 2011). the MSPSE (Bandura, 1990) consists of nine subscales, the
A third scale factor in the TEIP scale relates to EMB. This NTSES consists of six subscales (Skaalvik and Skaalvik,
factor reflects the conception that teacher’s self-efficacy 2007), the TES (Gibson and Dembo, 1984) consists of two
towards inclusion shapes students’ achievement and subscales and the STEBI (Riggs and Enochs, 1990) also
behaviours as well as teachers’ attitudes and classroom man- consists of two subscales. In contrast, the factorial valida-
agement skills (Ahsan, Sharma and Deppeler, 2012; Gibson tion of the TEIP scale in this study resulted in essential
and Dembo, 1984; Tschannen-Moran and Hoy, 2001). unidimensionality, with one dominating general dimension
and three specific latent factors that represent unique
The original study on the TEIP scale of 18 items suggested aspects of the targeted construct (self-efficacy of school
a three-factor structure of the scale based on results from an teachers to teach in inclusive classrooms). The essential
EFA. The main purpose of this study was to cross-validate unidimensionality of the TEIP scale can be explained by the
the proposed three-factor structure of the TEIP scale for the relatively limited context of teacher self-efficacy targeted
study population of pre-service teachers. The focus was to by the scale, whereas scales that measure broader aspects of
test whether (1) the TEIP data are essentially unidimen- self-efficacy are multidimensional.
sional, that is, there is one dominant factor that underlies the
scale scores, and (2) the latent factors EII, EC and EMB are Limitations of the study
different dimensions or they represent specific aspects of a One limitation of this study is the relatively small sample
unidimensional TEIP construct. The answer to these ques- size (N = 134), which typically results in larger standard
tions is of critical importance for the choice of appropriate errors of parameter estimates under the confirmatory model
approaches to the analysis of TEIP scale data (e.g., both of data analysis. Although the stability of relationship
classical and IRT-based test analyses assume essentially between the latent factors and their observed indicators
unidimensional data). under the CFA model used in this study prevented serious
inflation in parameter estimates, the results should be inter-
There are several findings related to the purpose of this preted with caution. Another limitation is the restricted
study. First, the cross-validation of the TEIP scale in the demographic make-up of the sample, with 94% females and
context of CFA supports the grouping of items on the EII, 91% White participants, which does not allow for
EC and EMB factors as proposed in the original study on generalisation of the study findings over males and non-
the scale (Sharma, Loreman and Forlin, 2012). In addition, White ethnic groups.

8 ª 2014 NASEN
Journal of Research in Special Educational Needs, 16 2–12

In conclusion, we trust that the findings in this study


provide both researchers and educators with valuable infor- Address for correspondence
mation about the nature of the TEIP scale (Sharma, Dimiter M. Dimitrov,
Loreman and Forlin, 2012) and its psychometric depend- Graduate School of Education,
ability in measuring teacher self-efficacy to teach in inclu- George Mason University,
4400 University Drive,
sive classrooms as a key factor for quality education of MS 6D2, Fairfax,
children with disabilities. Additional studies on using the VA 22030-4444,
TEIP scale with larger samples and diverse populations of USA.
pre-service teachers in the context of early childhood edu- Email: ddimitro@gmu.edu.
cation are recommended in an ongoing process of the vali-
dation of this scale.

Appendix A: Self-efficacy in implementing inclusive practices scale

This survey is designed to help us understand the nature of factors influencing the success of routine classroom activities in
creating an inclusive classroom environment. Please circle the number that best represents your opinion about each of the
statements. Please attempt to answer each question
1 2 3 4 5 6
Strongly disagree Disagree Disagree somewhat Agree somewhat Agree Strongly agree

SD D DS AS A SA
1 I can use a variety of assessment strategies (for example, 1 2 3 4 5 6
portfolio assessment, modified tests, performance-based
assessment, etc.).
2 I am able to provide an alternate explanation or example when 1 2 3 4 5 6
students are confused.
3 I am confident in designing learning tasks so that the individual 1 2 3 4 5 6
needs of students with disabilities are accommodated.
4 I can accurately gauge student comprehension of what I have 1 2 3 4 5 6
taught.
5 I can provide appropriate challenges for very capable students. 1 2 3 4 5 6
6* I am confident in my ability to get students to work together in 1 2 3 4 5 6
pairs or in small groups.
7 I am confident in my ability to prevent disruptive behaviour in 1 2 3 4 5 6
the classroom before it occurs.
8 I can control disruptive behaviour in the classroom. 1 2 3 4 5 6
9 I am able to calm a student who is disruptive or noisy. 1 2 3 4 5 6
10 I am able to get children to follow classroom rules. 1 2 3 4 5 6
11 I am confident when dealing with students who are physically 1 2 3 4 5 6
aggressive.
12 I can make my expectations clear about student behaviour. 1 2 3 4 5 6
13 I can assist families in helping their children do well in school. 1 2 3 4 5 6
14* I can improve the learning of a student who is failing. 1 2 3 4 5 6
15 I am able to work jointly with other professionals and staff 1 2 3 4 5 6
(e.g., aides, other teachers) to teach students with disabilities
in the classroom.
16 I am confident in my ability to get parents involved in school 1 2 3 4 5 6
activities of their children with disabilities.
17 I can make parents feel comfortable coming to school. 1 2 3 4 5 6

ª 2014 NASEN 9
Journal of Research in Special Educational Needs, 16 2–12

SD D DS AS A SA
18 I can collaborate with other professionals (e.g., itinerant 1 2 3 4 5 6
teachers or speech pathologists) in designing educational
plans for students with disabilities.
19 I am confident in informing others who know little about laws 1 2 3 4 5 6
and policies relating to the inclusion of students with
disabilities.
20 I am confident in adapting school-wide or state-wide 1 2 3 4 5 6
assessment so that students with all disabilities can be
assessed.
Note. This survey is an earlier version of the teacher efficacy for inclusive practices (TEIP) scale that does not include items 6 and 20 (Sharma, Loreman and Forlin, 2012, p. 16).
In the present study, items 6 and 14 were excluded after a preliminary test for data fit, so 18 items were retained for the confirmatory factor analysis (CFA) under the models depicted
in Figures 1 and 2.

References DEC/NAEYC (2009) Early Childhood Inclusion: A Joint


Abbitt, J. T. & Klett, M. D. (2004) ‘Identifying influences Position Statement of the Division for Early
on attitudes and self-efficacy beliefs towards Childhood (DEC) and the National Association for the
technology integration among pre-service educators.’ Education of Young Children (NAEYC). Chapel Hill:
Electronic Journal for the Integration of Technology The University of North Carolina, FPG Child
in Education, 6, pp. 28–42. Development Institute.
Ahsan, M. T., Sharma, U. & Deppeler, J. M. (2012) Dimitrov, D. M. (2012) Statistical Methods for Validation
‘Exploring pre-service teachers’ perceived of Assessment Scales in Counseling and Related
teaching-efficacy, attitudes and concerns about Fields. Alexandria, VA: American Counseling
inclusive education in Bangladesh.’ International Association.
Journal of Whole Schooling, 8 (2), pp. 1–20. Dozier, T. & Berlotti, C. (2000) Teacher Quality
Bandura, A. (1977) ‘Self-efficacy: toward a unifying Initiative. Washington, DC: US Department of
theory of behavioral change.’ Psychological Review, Education.
84, pp. 191–215. Druckman, D. & Bjork, R. A. (eds) (1994) Learning,
Bandura, A. (1990) Multidimensional Scales of Perceived Remembering, Believing: Enhanced Human
Academic Efficacy. Stanford, CA: Stanford University. Performance. Washington, DC: National Academy
Bandura, A. (1997) Self-Efficacy: The Exercise of Press.
Control. New York: W. H. Freeman and Company. Enochs, L., Smith, P. L. & Huinker, D. (2000)
Bruder, M. B., Dunst, C. J. & Mogro-Wilson, C. (2011) ‘Establishing factorial validity of the mathematics
‘Confidence and competence appraisals of early teaching efficacy beliefs instrument.’ School Science
intervention and preschool special education and Mathematics, 100, pp. 194–202.
practitioners.’ International Journal of Early Finney, S. J. & DiStefano, C. (2006) ‘Non-normal and
Childhood Special Education, 3 (1), pp. 13–37. categorical data in structural equation modeling.’ In
Buell, M. J., Gamel-McCormick, M. & Hallam, R. A. G. R. Hancock & R. O. Mueller (eds), Structural
(1999) ‘Inclusion in a childcare context: experiences Equation Modeling: A Second Course,
and attitudes of family childcare provider.’ Topics pp. 269–314. Greenwich, CT: Information Age
in Early Childhood Special Education, 19, Publishing.
pp. 217–24. Friend, M. & Cook, L. (2013) Interactions: Collaboration
Chen, F. F., West, S. G. & Sousa, K. H. (2006) ‘A Skills for School Professionals. (7th edn). Boston,
comparison of bifactor and second-order models of MA: Pearson Education.
quality of life.’ Multivariate Behavioral Research, 41, Gibson, S. & Dembo, M. H. (1984) ‘Teacher efficacy:
pp. 189–225. a construct validation.’ Journal of Educational
Darling-Hammond, L. (2006) Powerful Teacher Psychology, 76, pp. 569–82.
Education: Lessons from Exemplary Programs. Guo, Y., Justice, M. & Sawyer, B. (2011) ‘Exploring
San Francisco, CA: Jossey-Bass. factors related to preschool teachers’ self-efficacy.’
Das, A. K., Gichuru, M. & Singh, A. (2013) Teaching and Teacher Education, 27, pp. 961–8.
‘Implementing inclusive education in Delhi, India: Guo, Y., Piasta, S. B., Justice, L. M. & Kaderavek, J.
regular school teachers’ preferences for professional (2010) ‘Relations among preschool teachers’
development delivery modes.’ Professional self-efficacy, classroom quality and children’s
Development in Education, 39 (5), pp. 698–711. language and literacy gains.’ Teaching and Teacher
doi:10.1080/19415257.2012.747979. Education, 26, pp. 1094–103.

10 ª 2014 NASEN
Journal of Research in Special Educational Needs, 16 2–12

Guskey, T. R. (1988) Improving Student Learning in Marks, H. M. & Louis, K. S. (1997) ‘Does teacher
College Classrooms. Springfield, IL: Charles C. empowerment affect the classroom? The implication
Thomas Publishers. of teacher empowerment for instructional practice and
Guskey, T. R. & Passaro, P. D. (1994) ‘Teacher efficacy: student academic performance.’ Educational
a study of construct dimensions.’ American Evaluation and Policy Analysis, 19, pp. 245–75.
Educational Research Journal, 31, pp. 627–43. Mashburn, A. J., Pianta, R. C., Hamre, B. K., Downer,
Hoy, A. W. & Spero, R. B. (2005) ‘Changes in teacher J. T., Barbarin, O. A., Bryant, D., Burchinal, M.,
efficacy during the early years of teaching: a Early, D., & Howes, C. (2008). Measures of
comparison of four measures.’ Teaching and Teacher classroom quality in prekindergarten and children’s
Education, 21, pp. 343–56. development of academic, language, and social skills.
Hu, L. T. & Bentler, P. M. (1999) ‘Cutoff criteria for fit Child Development, 79 (3), pp. 732–49.
indexes in covariance structure analysis: conventional Mastropieri, M. A. & Scruggs, T. E. (2000) The Inclusive
criteria versus new alternatives.’ Structural Equation Classroom: Strategies for Effective Instruction.
Modeling, 6, pp. 1–55. Columbus, OH: Prentice-Hall/Merrill.
Individuals with Disabilities Education Improvement Act McConkey, R. & Bhlirgri, S. (2003) ‘Children with
of 2004, Pub. L. No. 108-446, § 602, 118 Stat. 2647 autism attending preschool facilities: the experience
(2004). and perceptions of staff.’ Early Child Development
Jarvis, T. & Pell, A. (2004) ‘Primary teachers’ changing and Care, 173 (4), pp. 445–52.
attitudes and cognition during a two-year science McGinty, A. S., Justice, L. M. & Rimm-Kaufman, S. E.
in-service programme and their effect on pupils.’ (2008) ‘Sense of school community for preschool
International Journal of Science Education, 26, teachers serving at-risk children.’ Early Education &
pp. 1787–811. Development, 19, pp. 361–84.
Justice, L. M., Mashburn, A., Hamre, B. & Pianta, R. Mergler, A. G. & Tangen, D. (2010) ‘Using
(2008) ‘Quality of language and literacy instruction microteaching to enhance teacher efficacy in
in preschool classrooms serving at-risk pupils.’ Early pre-service teachers.’ Teaching Education, 21 (2),
Childhood Research Quarterly, 23, pp. 51–68. pp. 199–210.
Labone, E. (2004) ‘Teacher efficacy: maturing the Muthén, L. K. & Muthén, B. O. (2010) Mplus User’s
construct through research in alternative paradigms.’ Guide. Los Angeles, CA: Muthén & Muthén.
Teaching and Teacher Education, 20, pp. 341–59. Nandakumar, R. & Yu, F. (1996) ‘Empirical validation of
Lamorey, S. & Wilcox, J. (2005) ‘Early Interventionist DIMTEST on nonnormal ability distributions.’ Journal
Self-Efficacy Scale: a measure and its applications.’ of Educational Measurement, 33, pp. 355–68.
Early Childhood Research Quarterly, 20 (1), Odom, S. L., Buysse, V. & Soukakou, E. (2011)
pp. 69–85. ‘Inclusion for young children with disabilities: a
Leyser, Y., Zeiger, T. & Romi, S. (2011) ‘Changes in quarter century of research perspectives.’ Journal of
self-efficacy of prospective special and general Early Intervention, 33 (4), pp. 344–56.
education teachers: implication for inclusive Pajares, F. (1996) ‘Self-efficacy beliefs in academic
education.’ International Journal of Disability, settings.’ Review of Educational Research, 66,
Development and Education, 58 (3), pp. 241–55. pp. 533–78.
LoCasale-Crouch, J., Konold, T., Pianta, R., Howes, C., Reise, S. P., Morizot, J. & Hays, R. D. (2007) ‘The role
Burchinal, M., Bryant, D., Clifford, R., Early, D., & of the bifactor model in resolving dimensionality
Barbarin, O. (2007) ‘Observed classroom quality issues in health outcomes measures.’ Journal of
profiles in state-funded pre-kindergarten programs and Quality of Life Research, 16, pp. 19–31.
associations with teacher, program, and classroom Riggs, I. & Enochs, L. (1990) ‘Toward the development
characteristics.’ Early Childhood Research Quarterly, of an elementary teacher’s science teaching efficacy
22, pp. 3–17. belief instrument.’ Science Education, 74,
Loreman, T., Sharma, U. & Forlin, C. (2013) ‘Do pp. 625–38.
pre-service teachers feel ready to teach in inclusive Romi, S. & Leyser, Y. (2006) ‘Exploring inclusion
classrooms?’ Australian Journal of Teacher pre-service training needs: a study of variables
Education, 38, pp. 27–44. associated with attitudes and self-efficacy beliefs.’
Main, S. & Hammond, L. (2008) ‘Best practice or most European Journal of Special Needs Education, 21,
practiced? Pre-service teachers’ beliefs about effective pp. 85–105.
behaviour management strategies and reported Sexton, D., Snyder, P., Wolfe, B., Lobman, M.,
self-efficacy.’ Australian Journal of Teacher Stricklin, S. & Akers, P. (1996) ‘Early intervention
Education, 33 (4), pp. 28–39. inservice training strategies: perceptions and
Malone, D. M., Straka, E. & Logan, K. R. (2000) suggestions from the field.’ Exceptional Children,
‘Professional development in early intervention: 62, pp. 485–95.
creating effective inservice training opportunities.’ Sharma, U., Loreman, T. & Forlin, C. (2012) ‘Measuring
Infants and Young Children, 12, pp. 53–62. teaching efficacy to implement inclusive practices.’

ª 2014 NASEN 11
Journal of Research in Special Educational Needs, 16 2–12

Journal of Research in Special Education Needs, 12, Weisel, A. & Dror, O. (2006) ‘School climate,
pp. 12–21. sense of efficacy and Israeli teachers’ attitudes
Skaalvik, E. M. & Skaalvik, S. (2007) ‘Dimensions of toward inclusion of students with special needs.’
teacher self-efficacy and relations with strain factors, Education, Citizenship and Social Justice, 1 (2),
perceived collective teacher efficacy, and teacher pp. 157–74.
burnout.’ Journal of Educational Psychology, 99, Woolfolk, A. E. (2007) Educational Psychology.
pp. 611–25. (10th edn). Boston, MA: Pearson Education, Inc.
Skaalvik, E. M. & Skaalvik, S. (2010) ‘Teacher Woolfolk, A. E., Rosoff, B. & Hoy, W. K. (1990)
self-efficacy and teacher burnout: a study of relations.’ ‘Teachers’ sense of efficacy and their beliefs about
Teaching and Teacher Education, 26, pp. 1059–69. managing students.’ Teaching and Teacher Education,
Tschannen-Moran, M. & Hoy, A. W. (2001) ‘Teacher 6, pp. 137–48.
efficacy: capturing an elusive construct.’ Teaching and Yu, C. Y. & Muthén, B. (2002) Evaluation of Model Fit
Teacher Education, 17, pp. 783–805. Indices for Latent Variable Models with Categorical
Tschannen-Moran, M., Hoy, A. W. & Hoy, W. K. (1998) and Continuous Outcomes. Paper presented at the
‘Teacher efficacy: its meaning and measure.’ Review of annual meeting of the American Educational Research
Educational Research, 68, pp. 202–48. Association, New Orleans, LA.

12 ª 2014 NASEN

Vous aimerez peut-être aussi