Professional Documents
Culture Documents
Reproduction in any form (including the internet) is prohibited without prior permission from the Society
527
The
British
Psychological
Society
www.bpsjournals.co.uk
There is contradictory evidence about which factors impact on teachers attitudes and
beliefs about the teaching of children with learning support needs in mainstream schools.
While some studies suggest that experience of teaching children with special needs results
in more positive teachers attitudes towards inclusion (Avramidis, Bayliss, & Burden, 2000;
Cook, Tankersley, Cook, & Landrum, 2000; LeRoy & Simpson, 1996; Romi & Leyser, 2006;
Villa, Thousand, Meyers, & Nevin, 1996), others provide evidence of the opposite position
(Forlin, 1995; Harvey, 1985; Leyser, Kapperman, & Keller, 1994; Soodak, Podell, &
Lehman, 1998; Wilczenski, 1993), and yet others have found the length of teaching
experience has no influence at all (Avramidis et al., 2000; Villa et al., 1996). Alternatively, it
may be teacher training rather than experience per se, which exerts a positive impact on
teachers beliefs about teaching learners with special needs in mainstream ( Bender, Vail, &
Scott, 1995; Ben-Pajooh, 1992; Smith, 1995). Although Romi and Leyser (2006) disputed
this finding that as students advanced in their initial teacher-training programmes, they
actually became less supportive of inclusion. Indeed, Avramidis et al. (2000) suggested
that it was specifically postgraduate, university-based education that made an impact and
that other forms of teacher education made no difference. These conflicting findings
suggest that it would be useful to try to separate the effects of postgraduate training from
that of teaching experience.
Furthermore, teachers role may also be an important factor in this. Leyser (2002)
found that general class teachers were less likely to modify teaching strategies for
children with learning difficulties than special education teachers. With regards to
children with autistic spectrum disorders who were included in mainstream,
Mavropoulou and Padeliadu (2000) found that regular education teachers were more
concerned with social and psychological well-being while special education teachers
were more educationally goal oriented. When both groups of teachers were asked about
their current success in including students with disabilities in their classrooms, general
teachers rated their understanding of inclusion and their ability to motivate students
lower than special education teachers (Buell, Hallam, Gamel-McCormick, & Scheer,
1999). Ward, Center, and Bochner (1994) compared attitudes towards inclusion among
principal teachers, general mainstream teachers, resource teachers (learning support
teachers), school psychologists/counsellors, and nursery school heads in Britain,
Australia and Canada, and found that the general mainstream teachers group held the
most negative attitudes towards inclusion. This concept is a cause for concern, as
negative attitudes towards children with disabilities are likely to have a harmful impact
on the outcome of inclusive educational practices (Tait & Purdie, 2000).
How comfortable a teacher feels around people with disabilities in general is also
likely to have some impact on his/her attitude towards teaching children with learning
support needs. Studies by Leyser et al. (1994) and Parasuraman (2006) have both
suggested there may be a relationship between experience of disabled people and
teachers attitudes. Using Gethings (1991) Interaction with Disabled Persons Scale
(IDP), Forlin, Tait, Carroll, and Jobling (1999b) found that the studentteachers who felt
most comfortable in their interactions with people with disabilities were those who had
regular contact, i.e. at least once a week.
Teaching efficacy is another potentially important variable with regard to teaching
learners with learning support needs. Teaching efficacy relates to a teachers feelings of
his/her own capacity to successfully facilitate learning. It has been found to be related to
student outcomes such as achievement (Ross, 1992), and to motivation (Midgley,
Feldlaufer, & Eccles, 1989). Stein and Wang (1988) reported that teachers with a strong
sense of efficacy are more willing to modify teaching methods to accommodate student
needs. Soodak and Podell (1993) found that regular and special educators with a high
sense of teaching efficacy were most likely to be supportive of inclusive placements.
Moreover, teachers evidencing high efficacy were found to be more willing to take
responsibility for meeting the needs of students with learning difficulties in their own
classrooms.
529
Teaching efficacy was thought to have two distinct components, general teaching
efficacy (the belief that external influences can be overcome by teaching) and personal
teaching efficacy (the teachers belief about his or her own capability to bring
about changes in students). However, this conceptualization has been challenged
on the grounds that teaching efficacy is context specific and varies across
participants (e.g. Ross, Cousins, & Gadalla, 1996) and across different student groups
(e.g. Raudenbuch, Rowen, & Cheong, 1992). Furthermore, Tschannen-Moran and
Woolfolk Hoy (2001) argued that the general teaching efficacy component of the
construct is not concerned with an individual teachers sense of efficacy at all. In
response to these criticisms, Tschannen-Moran, Woolfolk Hoy, and Hoy (1998)
developed a new definition of teaching efficacy that recognizes its context-specific
nature. Tschannen-Moran and Woolfolk Hoy (2001) subsequently devised a
measurement of teaching efficacy based on items that represent specific teaching
tasks in contextual classroom situations.
Attribution theory offers a useful conceptual framework for examining teachers
beliefs about childrens difficulties in learning. Bar-Tal (1978) defined attributions as the
inferences that observers make about the causes of behaviour, either their own or those of
other people. Weiner (1985) identified that causal attributions can be classified in terms
of three dimensions: locus of causality; stability; and controllability. Locus of causality
refers to the source of the attribution: an internal locus of causality indicates that the
behaviour is caused by a characteristic of the individual; an external locus of causality is
one that is external to the individual; for example, behaviour attributed to factors in the
environment. Stability refers to how enduring the cause appears, and controllability
describes the degree of control an individual has over the cause of the behaviour.
Attributions that teachers make about their pupils could have important
consequences in the classroom for teachers behaviours and pupil outcomes. For
example, a teacher who attributes a pupils failure in a test to factors external to the child
may modify teaching practices more compared with a teacher who attributes the failure
to the pupils disposition (Jordan, Lindsay, & Stanovich, 1997). Furthermore, Weiner
(1984) argued that attributing failure to factors internal to the child can result in the
child feeling guilt or shame and experiencing lower self-esteem. In terms of stability
attributions, if a teacher views the causes of a childs difficulty as unstable, s/he may have
more of an expectation of change and progress for the child than a teacher who sees
them as stable. Weiner (1979) found that attributing failure to a stable factor can result in
students themselves giving up and not trying. Finally, with regards to the controllability
dimension, a teacher who sees a pupil as being in control of his or her own progress may
be less accepting of failure. Clarke and Artilles (2000) argued that if a teacher shows
anger or gives punishment following a negative outcome, the child is likely to feel as
though s/he is in control. Similarly, giving help to a child on an easy task may negatively
impact on the childs self-perception as s/he might interpret the help as a low ability cue.
Bibou-Nakou, Kiosseoglou, and Stogiannidou (2000) and Brophy (1996) found that
learners were more likely to be disciplined if they were seen as being in control of the
problem and were more likely to be treated sympathetically if they were seen as victims
of circumstances. Clark (1997) found that children with identified learning difficulties
were more likely to be rewarded after failure than their peers with no identified learning
difficulties. Teachers expressed less anger and more pity towards the children with
learning difficulties and held lower expectations of their future success.
A number of attribution studies have focused on the dimension of locus of causality.
Brown and Rogers (1991) and Campbell and Sedikides (1999) argued that teachers
evidence a self-serving bias as they tend to attribute problems within the child or within
the childs family, rather than to teaching-related factors. For example, in Mavropoulou
and Padeliadus (2002) study, teachers rejected school-related factors as causing
behaviour problems, viewing them as caused by family- and pupil-related factors.
Similarly, Hos (2004) study found that teachers attributed misbehaviours most to
student effort and least to teachers factors, and Jordan et al. (1997) reported that
teachers holding an interventionist perspective (attributing problems to an interaction
between the student and the environment) persevered more with pupils who failed to
grasp concepts, than those holding a pathognomonic perspective (where difficulties in
learning are viewed as inherent in the pupil).
Stability and controllability attributions have been less well researched with teachers.
Clark (1998) found that teachers attributed the cause of learning difficulties as internal
to the child, and also as stable and as uncontrollable. Woolfson, Grant, and Campbell
(2007) compared stability and controllability attributions in mainstream teachers and
special education teachers and found that mainstream teachers viewed childrens
difficulties as less amenable to change than the special education teachers. In terms of
controllability, the mainstream teachers saw children with learning difficulties as having
less control over their difficulties with schoolwork than those with no identified support
needs, while special school teachers saw no difference between the groups.
The present study then aimed to examine the influence of general teaching
experience, specific experience of teaching children with learning difficulties, and
postgraduate qualifications on teachers role, teaching efficacy, and attitudes towards
disabled persons on teachers attributions about childrens difficulties in learning.
Method
Participants
Participants were recruited from three school districts across central Scotland. Taking
part were 10 mainstream schools, 1 mainstream learning support service, and 9 special
schools that covered a range of special educational support needs similar to those that
general mainstream teachers might experience in their classrooms: learning difficulties
(six schools, including a school that catered for children who also had sensory
impairments); social, emotional, and behavioural difficulties (two schools); and
language difficulties (one school). All schools were primary schools for children aged
512 years. Scottish primary school teachers are trained to teach over the age range
312 years. The mainstream learning support service operated on a peripatetic basis and
was made up of small teams of learning support teachers who each provided
educational support to learners with special needs included in mainstream schools
within a particular geographical locale. Each team of teachers was based at one of the
mainstream primary schools to which it delivered its services.
Questionnaires were completed and returned by 125 teachers. Seven were omitted
because of missing data. The final 118 questionnaires thus came from 44 general
mainstream teachers, 33 mainstream learning support teachers, and 41 special
education teachers. Tabachnick and Fidell (2001) recommend a sample $ 104 the
number of predictor variables for testing individual predictors in multiple regression,
assuming a medium-sized relationship between predictors and the dependent variable.
For the present study with six predictors (see page 15), a sample of 110 would therefore
be sufficient. When first-order cross-product multiplicative interaction variables were
531
subsequently added into the analysis as predictors (see page 16), the power of the
analysis, calculated using G*Power 3 (Faul, Erdfelder, & Lang, 2007) was reduced but
was nevertheless still satisfactory at .84
Materials
Interaction with Disabled Persons Scale (Gething, 1991)
The Interaction with Disabled Persons Scale (IDP) is a list of 20 statements that have
been used to describe how some people feel when they are in contact with a disabled
person. Participants are asked to respond to each item on a forced-choice 6-point Likert
scale indicating to what extent they agree or disagree with the statement. High scores
indicate negative attitudes towards disabled people while low scores indicate positive
attitudes. Gething and Wheeler (1992) reported that testretest reliability coefficients
ranged between .51 for a 1-year period and .82 over a 2-week period. Internal
consistency across 15 studies ranged from .74 to .86 (Gething, 1991). Gething et al.
(1997) international validation across nine countries, including the United States,
Canada, England, and Scotland, also concluded that the IDP scale is a valid measure,
with mean scores falling within 10 points across countries and moderate-to-high internal
consistency. While Gething (1991) reported a six-factor solution to the IDP scale,
supported by Forlin, Fogarty, and Carroll (1999a), who labelled the factors as
discomfort, sympathy, uncertainty, fear, coping, and vulnerability, this structure has
been criticized because some factors have only two items (Tait & Purdie, 2000), making
interpretation difficult when the two items are not obviously related. Also, Tait and
Purdie argued that the coping and vulnerability subscales are conceptually very similar.
Both MacLean and Gannon (1995) and Tait and Purdie (2000) proposed two orthogonal
subscales (discomfort and sympathy) derived from ten IDP items. However, Tate and
Purdie also presented a four-factor solution (discomfort, sympathy, embarrassment, and
vulnerability) derived from 16 items, which they suggested better represented the ideas
contained in the scale. Because of this lack of consensus as to factor structure, factor
analysis was carried out in the present study to determine the appropriate factor
solution for our sample (see Results section).
adaptation of Clarks (1997) vignettes but in the present study, as teachers were required
to complete several questionnaires in this study, we reduced the amount of information
in the vignettes to take account of time constraints. This was done by controlling for the
variables of effort and ability by presenting all pupils in vignettes as of lower ability than
the class average and all pupils high in effort and then reducing the number of vignettes
presented. A two-stage pilot study was carried out to validate this revision of the Teacher
Attribution Scale. Firstly, 10 graduate students answered questions about the level of
ability and effort represented in each vignette as well as whether or not each of the
children had identified learning difficulties. All participants answered as anticipated,
signalling that the information in the vignettes was clear. Secondly, a pilot study was
carried out with 19 teachers showing that internal consistency for each scale exceeded
.9 (locus of causality r :95, controllability r :95, and stability r :94), which
suggested that we did not need to retain all eight vignettes, so the number of vignettes
was reduced to four, leaving the scales with reliabilities of .86 (locus of causality), .87
(stability), and .91 (controllability). Using the revised scale, participants were asked to
make attributions as in the Woolfson et al. (2007) study, on a 5-point Likert scale, with 1,
strongly agree to 5, strongly disagree. For example, after reading the following
vignette,
Christopher has ability below that of most children in his class. He always tries his best,
participants were asked to rate their agreement with the following statements on a
5-point Likert scale:
(1)
(2)
(3)
Christophers poor performance on the test is due to factors within himself such as
ability, effort etc, rather than due to external factors such as his teacher or the
curriculum.
Christophers performance in class is likely to continue more or less the same.
It is within Christophers own control to improve his performance further.
Procedure
Questionnaire packs were distributed to teachers at staff meetings and participants
were invited to return their completed questionnaires in the accompanying envelopes
and leave them in the box, which would be collected at a specified time/date, generally
34 working days after distribution. The order of presentation of questionnaires in the
packs was varied to minimize the impact of possible order effects. Teachers were also
asked to provide information on any postgraduate qualifications in special educational
needs, their teaching role, teaching experience, and experience of teaching children
with additional learning support needs.
Results
Table 1 shows the alpha coefficients obtained in this study for each of the instruments
used. While there is some debate in the literature concerning what constitutes
acceptable levels of reliability (e.g. Schmitt, 1996), alpha coefficients greater than .7 are
generally accepted as indicating good reliability.
533
Scale
Interaction with Disabled Persons Scale (IDP)
Teachers Sense of Efficacy Scale
Teacher Attribution Scale (TAS)
TAS: causality
TAS: stability
TAS: controllability
.82
.91
.86
.97
.92
.93
Item no.
9
11
12
17
18
1
2
3
5
13
.738
.633
.710
.718
.704
.468
.695
.582
.433
.450
.736
.688
.704
.763
.775
.629
.778
.530
.654
.679
load on each factor. Internal reliability was good for the Discomfort scale (a :79) and
acceptable for the Sympathy scale (a :68). Also, as Gannon and MacLean and Tait and
Purdie found, the two scales were not highly correlated (r :24, p :008), indicating
that the subscales are measuring two independent attitude dimensions. These are the
two subscales that will be used in the analyses.
Three multiple regression analyses were first carried out to investigate the impact of
the teacher predictor factors teachers role, years of teaching experience, experience
working with pupils with additional learning support needs, postgraduate qualifications
in special educational needs, teaching efficacy, and attitudes towards disabled people on
locus of causality, stability and controllability attributions. Higher scores for causality
reflect an external locus of causality and higher scores for stability reflect more unstable
attributions i.e. that recognize the possibilities of change in learner performance. Scores
for controllability were reversed so that higher scores reflect the child being attributed
with control. This meant that across all three types of attribution, higher scores
indicated more positive attributions. The support needs variable could not be entered in
these regression analyses as a predictor, as scores for the two conditions of the support
needs variable (needs identified and needs not identified) were themselves teachers
attribution scores on the relevant vignettes and therefore would obviously be highly
correlated with the dependent attribution total scores. Analysis of the attributions on
the two sets of support need vignettes and their interaction with teachers role was
therefore carried out separately using multivariate analysis of variance.
Of the participants, 15% reported having 05 years of teaching experience,
14% reported 610 years of experience, 15% reported 1115 years of experience, and
56% reported more than 15 years of experience. In addition, 16% of teachers reported
having no experience of children with special needs, and another 16% reported having
less than 5 years of such experience. A further 23% of participants reported having
between 5 and 10 years, while 44% had more than 10 years of experience with
additional support needs. Postgraduate qualifications were recorded in terms of the
highest postgraduate qualification gained in special educational needs. 50% of
participants had undertaken no postgraduate training, 3% were currently studying for a
qualification, 8.5% had a postgraduate certificate, 1.7% had a postgraduate certificate
and were currently undertaking a further qualification, 32% had gained a diploma, 0.8%
(one participant) had two diplomas, and 2.5% had a Masters degree. One participant
reported having a different, unspecified qualification. Table 3 shows how mean
attribution scores differed by each of these participant characteristics.
The predictor of teachers role was entered using dummy coding. This requires one
category to be excluded to prevent perfect multicollinearity in the model and so in this
case, special education was the excluded category. Thus two dichotomous variables
were created; mainstream teachers (1 for mainstream and 0 for special education and
learning support) and learning support teachers (1 for learning support and 0 for
mainstream and special education). This meant that the special education group was the
reference group, allowing the two groups of teachers working in mainstream settings to
be compared with those working in special education environments. Years of teaching
experience also used dummy coding with 0 representing less than 115 years of
teaching and 1 representing more than 15 years of teaching experience. Experience of
working with pupils with additional support needs again used dummy coding with 0
representing less than 5 years experience and 1 representing 5 or more years. Training
was examined in a similar way, with 0 representing no postgraduate training in special
education and 1 representing some kind of postgraduate training. Teaching efficacy
Teachers role
General mainstream
Mainstream learning support
Special education
Years of teaching experience
0 5
6 10
1115
15
Years including any work with special needs
None
Less than 5
Between 5 and 10
10
Highest postgraduate qualification gained in SEN
No qualification
Currently studying for a qualification
Postgraduate certificate
Postgraduate certificate and currently studying for other
Postgraduate diploma
Two diplomas
Masters degree
Others
Characteristic
12.55 (4.09)
14.45 (4.56)
13.49 (4.65)
12.22 (2.51)
14.13 (4.33)
15.39 (3.88)
13.02 (4.90)
11.95 (4.02)
12.05 (4.44)
14.82 (3.39)
13.67 (4.91)
12.63 (4.52)
15.75 (4.35)
14.40 (3.75)
12.00 (0.00)
14.26 (4.67)
16.00 (0.00)
12.00 (4.00)
12.00 (0.00)
(15.3)
(13.6)
(15.3)
(55.9)
(16.1)
(16.1)
(23.7)
(44.1)
18
16
18
66
19
19
28
52
59 (50.0)
4 (3.4)
10 (8.5)
2 (1.7)
38 (32.2)
1 (0.8)
3 (2.5)
1 (0.8)
44 (37.3)
33 (28.0)
41 (34.7)
Frequency (%)
12.66 (3.43)
18.00 (1.63)
14.40 (4.27)
12.00 (0.00)
14.18 (4.64)
20.00 (0.00)
14.67 (6.11)
12.00 (0.00)
12.05 (3.64)
13.74 (3.54)
14.43 (3.35)
13.62 (4.64)
13.72 (2.78)
13.88 (3.46)
14.94 (4.62)
13.09 (4.30)
12.57 (3.45)
13.36 (4.08)
14.83 (4.39)
9.88 (3.80)
13.00 (5.03)
13.10 (3.90)
12.00 (5.66)
10.39 (4.85)
20.00 (0.00)
16.00 (4.00)
16.00 (0.00)
10.79 (4.30)
11.58 (3.75)
10.54 (4.03)
10.56 (4.99)
11.89 (2.68)
10.44 (4.68)
11.33 (4.69)
10.36 (4.72)
10.00 (3.46)
9.55 (3.87)
12.54 (5.27)
Reproduction in any form (including the internet) is prohibited without prior permission from the Society
was analysed using the total scores from the Teachers Sense of Efficacy Scale.
Finally, attitudes towards disabled people were investigated using the Discomfort and
Sympathy subscales of the IDP scale.
Hierarchical regression analysis was employed with the teachers role dummy
variables entered first, followed by all main effects and then first-order interactions
which were explored by creating cross-product multiplicative terms. Following Aitken
and Wests (1991) recommendation, the predictor variables were centred prior to
construction of the interaction terms to reduce potential multicollinearity problems.
Using the variance inflation factor ( VIF ) to test for multicollinearity, VIF values of
between 1.09 and 2.22 were obtained, showing multicollinearity was not a problem as
high multicollinearity is represented by VIF values above 4 (Hair, Anderson, Tatham, &
Black, 1995).
The results of the regression analyses are summarized in Table 4. No significant
interactions were found in any of the analyses so the third models containing these
interactions have not been included in this table. The first model was not significant,
(F 2; 115 1:77, MS residual 19:58), so teachers role was not found to predict locus
of causality attributions. Years of teaching experience was found to be a negative
predictor of external locus of causality attributions (b 20:28, t 22:77, p :007),
while experience with additional support needs was found to be a positive predictor of
external locus of causality attributions (b 0:28, t 2:53, p :013). Also, teaching
efficacy was found to positively predict external causality attributions (b 0:19,
t 2:08, p :040). This second model explained 9% (adjusted R2) of the variance,
(F 8; 109 2:46, MS residual 18:03, p :017), indicating that teachers with more
experience attribute the childrens failure in the test more to factors internal to the child
than those with less experience. Conversely, participants with more experience of
Stability
Controllability
b
20.10
0.10
t
2 0.98
0.93
b
2 0.27
2 0.16
t
2 2.62*
2 1.58
b
2 0.28
2 0.30
t
22.73**
22.99**
Model 1: Summary
R2(adj )
0.01
F
1.77
R2(adj )
0.04
F
3.50*
R2(adj )
0.07
F
5.56**
b
0.05
0.17
20.28
0.28
0.05
0.19
0.10
20.07
t
0.40
1.56
2 2.77**
2.53*
0.38
2.08*
1.08
2 0.75
b
2 0.17
2 0.13
2 0.16
0.09
0.03
0.11
0.01
2 0.33
t
2 1.30
2 1.24
2 1.60
0.89
0.24
1.24
0.05
2 3.51**
b
2 0.24
2 0.31
2 0.02
2 0.16
2 0.24
2 0.09
0.11
2 0.06
t
21.85
22.91**
20.19
21.48
1.95
21.01
1.16
20.64
Model 2: Summary
R2(adj )
0.09
F
2.46
R2(adj )
0.15
F
3.59
R2(adj )
0.17
F
2.69
*p , :05.
**p , :01.
537
working specifically with pupils with additional learning support needs attributed to
external classroom factors the pupils failure in the class test more than those with less
experience. Moreover, teachers with a higher sense of teaching efficacy also attributed
the failure to external factors more than those with a lower sense of teaching efficacy.
Teachers role was found to predict stability attributions as Model 1 in the stability
regression was significant (F 2; 115 3:50, MS residual 15:81, p :034), with the
mainstream teachers dummy variable a negative predictor of stability attributions,
(b 20:27, t 22:62, p :010). When the other factors were included in Model 2,
this now explained 15% (adjusted R 2) of the variance (F 8; 109 3:59,
MS residual 14:01, p :001), and teachers role no longer emerged as a predictor
of stability attributions. Instead the IDP subscale of Sympathy was found to be the best
predictor of stability attributions (b 20:33, t 23:51, p :001). More positive
attitudes on the IDP scale are represented by lower values, so teachers with more
positive attitudes not overwhelmed by sympathy saw childrens difficulties in learning as
being more amenable to change.
Teachers role was also found to be predictive of controllability attributions
(F 2; 115 5:56, MS residual 18:32, p :005; Table 4). Both the mainstream
(b 20:28, t 22:73, p :007) and the learning support (b 20:30, t 22:99,
p :003) dummy variables were found to be negative predictors. However, in Model 2
where all the factors are included, it was only the mainstream learning support variable
that emerged as a predictor (b 20:31, t 22:91, p :004). This model explained
17% (adjusted R2) of the variance (F 8; 109 2:69, MS residual 17:69, p :010).
Therefore, teachers role was found to be the only predictor of controllability
attributions. The relationship between teachers role and controllability attributions will
be referred to again in the subsequent multivariate analysis of variance.
The postgraduate qualifications variable did not emerge as a predictor of attributions.
There were originally eight categories of postgraduate qualifications (no postgraduate
qualifications, studying for a postgraduate qualification, postgraduate certificate, two
postgraduate certificates, postgraduate certificate and studying for a diploma, two
diplomas, diploma and studying for a masters degree, and masters degree) but no effects
emerged when these categories were investigated. Therefore the dichotomous variable
of postgraduate qualification or no postgraduate qualification was devised to further
investigate potential effects, as we had done for teaching experience and additional
support needs. For teaching experience there were initially four categories: (a) 05
years; (b) 610 years; (c) 1115 years; and (d) more than 15 years. Using this breakdown
of experience, this variable did not emerge as a predictor of attributions. These four
categories were then merged into two categories less than 10 years and more than 10
years of experience. Again, this breakdown of experience did not appear as a predictor
of attributions. It was only when experience was broken down into the categories of 15
or less years or more than 15 years that it emerged as a predictor. It should also be noted
that when additional support needs was investigated with its original four categories of
(a) none, (b) less than 5 years, (c) between 5 and 10 years, and (d) more than 10 years, it
did not come out as a predictor. However, postgraduate qualifications did not emerge as
a predictor, regardless of how it was categorized.
A mixed 3 2 (general mainstream, learning support and special education
identified learning support needs or no identified support needs) multivariate analysis of
variance (MANOVA) was carried out to analyse the effects of support needs on the
dependent teachers attribution variables of locus of causality, stability, and controllability
as well as interaction with teachers role. There appeared to be problems with
homogeneity associated with the MANOVA, as Boxs M test was significant (p , :001).
For this reason, Pillais Trace was used in place of Wilks Lambda (Tabachnick & Fidell,
2001). Mean ratings and standard errors for each of the attributions by teachers group
and support needs conditions are shown in Table 5. A multivariate effect was found for
the within-participants factor of support needs (Pillais trace F 3; 113 3:07, p :031).
Univariate analysis identified a significant main effect of support needs for the
controllability attributions (F 1; 115 5:87, MS error 1:37, p :017; h2 :01; small,
but not trivial, effect size). Teachers made significantly lower controllability ratings for the
learners with identified support needs (mean 5:16, SE 0:20) compared with those
without identified needs (mean 5:53, SE 0:22). Therefore, learners without
identified support needs were seen as being more in control over their own progress.
No main effects of support needs were, however, found for locus of causality or stability
attributions.
Table 5. Mean (SE) attribution ratings by teachers role and support needs condition
Causality
Stability
Controllability
Identified
support needs
General mainstream
teachers mean (SE)
Yes
No
Yes
No
Yes
No
6.23 (0.34)
6.32 (0.35)
6.18 (0.32)
6.39 (0.33)
4.43 (0.33)
5.57 (0.36)
Learning support
teachers mean (SE)
7.21
7.24
6.76
6.61
4.76
4.79
(0.39)
(0.40)
(0.37)
(0.38)
(0.38)
(0.42)
Special education
teachers mean (SE)
6.68
6.81
7.61
7.22
6.29
6.24
(0.35)
(0.36)
(0.33)
(0.34)
(0.34)
(0.37)
As the influence of teachers role on attributions has already been reported in the
regression analyses results, MANOVA results for this are not reported again here.
A multivariate teachers role support needs interaction effect was however obtained
(Pillais trace F 6; 228 2:22, p :042). Univariate tests revealed the interaction to be
significant for the controllability attributions (F 2; 115 6:66, MS error 1:37,
p :002; h2 :01, a small effect size), but not for locus of causality or stability
attributions. Follow-up analysis was carried out using Tukey HSD to compare the
controllability means for the teachers groups and support needs conditions. The
harmonic mean was used due to the unequal group sizes. This analysis revealed that
while the learning support and the special education teachers saw no difference
between the pupils with identified support needs and those without, the general
mainstream teachers saw the pupils without any identified support needs
(mean 4:43, SE 0:33) as having more control than the pupils with identified
needs (mean 5:57, SE 0:36), p , :01. These findings are illustrated in Figure 1.
Discussion
The study found that teachers with more than 15 years of experience viewed childrens
difficulties as more internally attributable than those with less than 15 years of
experience. One explanation is that perhaps pre-service training courses now focus
more on inclusive practices compared with 15 years ago when the more experienced,
longer-serving teachers in our sample were trained, making it more likely that recently
trained teachers view learner difficulties as responsive to external, teacher-driven
539
influences. Alternatively, as Ward et al. (1994) found that enthusiasm for inclusion
increased with distance from regular classroom practice, it could be that teacher
educators and trainers enthusiasm about the effects teachers can have on learners with
difficulties has been transmitted to, and retained by, the more recently trained teachers.
As anticipated, teachers with more experience of working with children with
additional learning support needs attributed difficulties in learning more to external
factors such as teacher- and school-related factors. This echoes the findings of Avramidis
et al. (2000), LeRoy and Simpson (1996), and Cook et al. (2000) and suggests a
constructive willingness to accept responsibility for learner progress. Surprisingly,
postgraduate qualifications in additional learning support needs were not found to
impact on teachers attributions which contrasts with the findings of Avramidis et al.
(2000). One explanation for the apparent disparity may be that such qualifications may
only address surface attitudes and do not adequately deal with underlying attributional
beliefs about children with additional support needs. Perhaps postgraduate training in
additional support needs should be more hands on, offering teachers placement
opportunities to gain greater experience with learners with special needs. Both pre- and
post-service trainings should aim not only to provide teachers with the relevant skills
and knowledge to support inclusion, but also more fundamentally to tackle their beliefs
about learners. It should be noted that in-service training was not considered due to
Avramidis et al.s (2000) finding that only postgraduate university-based training made a
difference. Certainly, the findings of the present study suggest a need for further
research into the impact of current training approaches, whether pre-service, in-service
short courses or postgraduate qualifications, on teachers beliefs about learners with
difficulties. This could take the form of in-depth qualitative research which focuses on
what teachers find useful and what they feel is required.
Teachers with a higher sense of efficacy attributed the childrens difficulties more to
external factors than those with a lower sense of efficacy, suggesting that teachers who
feel more competent and have a greater belief in the power of their profession are more
541
Conclusion
Teachers with a higher sense of teaching efficacy and those with more experience of
working with learners with additional support needs both attributed learners failure in
class more to external factors than did those with less experience of teaching children
with additional support needs and with lower teaching efficacy. Teachers who did not
feel high levels of sympathy viewed learner difficulties as more amenable to change.
Special education teachers saw that the learners were having more control over their
own progress than mainstream learning support teachers. Mainstream teachers were
also less optimistic about learner progress than special education teachers, viewing the
children with learning support needs as having less control over their own progress
than the children without such needs. No relationship was found between postgraduate
qualifications and teachers beliefs about learners. These findings have important
implications for effective inclusion in terms of modification of teaching, expectations of
success, and setting of goals, and teachers pre- and post-service training.
References
Aitken, L., & West, S. (1991). Multiple regression: Testing and interpreting interactions. Newbury
Park, CA: Sage.
Avramidis, E., Bayliss, P., & Burden, R. (2000). A survey into mainstream teachers attitudes
towards the inclusion of children with special educational needs in the ordinary school in one
local educational authority. Educational Psychology, 20, 193213.
Bar-Tal, D. (1978). Attributional analysis of achievement-related behaviour. Review of Educational
Research, 48, 259271.
Ben-Pajooh, A. (1992). The effect of social contact on college teachers attitudes towards students
with severe mental handicaps and their educational integration. Journal of Special Needs
Education, 7, 231236.
Bender, W., Vail, C., & Scott, K. (1995). Teachers attitudes to increased mainstreaming:
Implementing effective insruction for students with learning disabilities. Journal of Learning
Disabilities, 28, 8794.
Bibou-Nakou, I., Kiosseoglou, G., & Sogiannidou, A. (2000). Elementary teachers perceptions
regarding school behavior problems: Implications for school psychological services.
Psychology in the Schools, 37(2), 123134.
Brophy, J. (1996). Teaching problem students. New York: The Guildford Press.
Brown, J. D., & Rogers, R. J. (1991). Self-serving attributions: The role of physiological arousal.
Personality and Social Psychology Bulletin, 17, 501506.
Buell, M., Hallam, R., Gamel-McCormick, M., & Scheer, S. (1999). A survey of general and special
education teachers perceptions and in-service needs concerning inclusion. International
Journal of Disability, Development and Education, 46, 143156.
543
Parasuraman, K. (2006). Variables that affect teachers attitudes towards disability and inclusive
education in Mumbai, India. Disability and Society, 21(3), 231242.
Raudenbush, S., Rowen, B., & Cheong, Y. (1992). Contextual effects on the self-perceived efficacy
of high school teachers. Sociology of Education, 65(2), 150167.
Romi, S., & Leyser, Y. (2006). Exploring inclusion preservice training needs: A study of variables
associated with attitudes and self-efficacy beliefs. European Journal of Special Needs
Education, 21(1), 85105.
Ross, J. A. (1992). Teacher efficacy and the effect of coaching on student achievement. Canadian
Journal of Education, 17(1), 5165.
Ross, J., Cousins, J., & Gadalla, T. (1996). Within-teacher predictors of teacher efficacy. Teaching
and Teacher Education, 12, 385400.
Schmitt, N. (1996). Uses and abuses of coefficient alpha. Psychological Assessment, 8(4),
350353.
Smith, L. (1995). The impact of early life history on teachers beliefs: In-school and out-of-school
experiences as learners and knowers of science. Teachers and Teaching: Theory and Practice,
11(1), 536.
Soodak, L., & Podell, D. (1993). Teacher efficacy and student problem as factors in special
education referral. Journal of Special Education, 27, 6681.
Soodak, L., Podell, D., & Lehman, L. (1998). Teacher, student and school attributes as predictors of
teachers responses to inclusion. Journal of Special Education, 31, 480497.
Stein, M., & Wang, M. C. (1988). Teacher development and school improvement: The process of
teacher change. Teaching and Teacher Education, 4, 171187.
Tabachnick, B., & Fidell, L. S. (2001). Using multivariate statistics (4th ed.). Needham Heights,
MA: Allyn and Bacon.
Tait, K., & Purdie, N. (2000). Attitudes towards disability: Teacher education for inclusive
environments in an Australian university. International Journal of Disability, Development
and Education, 47(1), 2538.
Tollefson, N., & Chen, J. (1988). Consequences of teacher attribution for student failure. Teaching
and Teacher Education, 4, 259265.
Tschannen-Moran, M., & Woolfolk Hoy, A. (2001). Teacher efficacy: Capturing an elusive
construct. Teaching and Teacher Education, 17, 783805.
Tschannen-Moran, M., Woolfolk Hoy, A., & Hoy, W. K. (1998), Teacher efficacy: Its meaning and
measure. Review of Educational Research, 68, 202248.
Villa, R., Thousand, J., Meyers, H., & Nevin, A. (1996). Teacher and administrator perceptions of
heterogeneous education. Exceptional Children, 63(1), 2945.
Ward, J., Center, Y., & Bochner, S. (1994). A question of attitudes: Integrating children with
disabilities into regular classrooms? British Journal of Special Education, 21, 3439.
Weiner, B. (1979). A theory of motivation for some classroom experiences. Journal of
Educational Psychology, 71, 325.
Weiner, B. (1984). Principles for a theory of student motivation and their application within an
attributional framework. In R. Ames & C. Ames (Eds.), Student motivation ( Vol. 1, pp. 1538).
New York: Academic Press.
Weiner, B. (1985). An attributional theory of achievement motivation and emotion. Psychological
Review, 92, 548573.
Wilczenski, F. L. (1993). Changes in attitudes towards mainstreaming among undergraduate
education students. Educational Research Quarterly, 17, 517.
Woolfson, L., Grant, E., & Campbell, L. (2007). A comparison of special, general and support
teachers controllability and stability attributions for childrens difficulties in learning.
Educational Psychology, 27(2), 293304.
Received 8 March 2007; revised version received 15 November 2007
(3)
(4)
Christopher has ability below that of most children in the class. He always tries his
best.
Andrew is considered to have lower ability for academic tasks than most children
in the class. He works hard in class. He receives services from learning support to
help him in language and maths.
Johns ability is considered to be below that of many children in his class. He works
hard in class.
Stevens ability is below that of most of his classmates. He has difficulty with
writing tasks which involve constructing sentences and spelling correctly. The
learning support teacher helps him with this. He always tries his best in class.