Effects of Contact on Childrens Attitudes Toward

Disability: A Longitudinal Study1

PAM MAR AS^

RUPERTBROWN
University of Kent
Canterbury
Kent, England

School of Social Sciences

University of Greenwich
London, England

A quasi-experimental study was conducted on temporal effects of intergroup contact

on nondisabled (ND) childrens attitudes toward disability. Children from a
mainstream primary school were involved in an integration program with children
from a school for children with severe learning disabilities (SLD).3 Measures were
administered 3 times over a period of 3 months to 26 integrating (experimental) and
24 nonintegrating (control) children. Social orientations in the experimental group
became significantly more positive over time, while the control group showed little
change. The experimental and control children initially categorized on the basis of
gender and disability; subsequently the strategies of the experimental children were
more idiosyncratic while the control children still used the same two dimensions.

Such an outcome will not occur spontaneously. Nor will it be

achieved by legislation alone. It has to be contrived and patiently
nurtured. It means greater discrimination in favor of those children with special needs, in proportion to the severity of their
disabilities. (Department of Education & Science, 1978, p. 102)

In recent years, there have been important policy and legislative moves
aimed at the educational integration of children with disabilities (e.g., in the
U.K. see Department of Education & Science, 1981,1987). These moves have
important implications in terms of attitudes of nondisabled (ND) children
The research reported here was funded by grant (#F.2368) from The Leverhulme Trust.
We would like to thank two anonymous reviewers for their comments on an earlier draft of
this paper.
2Correspondence concerning this article should be addressed to Pam Maras, School of
Social Sciences, University of Greenwich, Avery Hill Campus, Avery Hill Road, Eltham, London
SE9 2HB, UK.
%evere learning disabilities (SLD) is the term currently used in the U.K. to describe people
who might formally have been described as having severe mental handicap or severe
retardation. This term is interchangeable with severe learning disabilities.

2113
Journal of Applied Social Psychology, 1996,26,23, pp. 21 13-2134.
Copyright 0 1996 by V. H. Winston & Son, Inc. All rights reserved.

21 14

MARAS AND BROWN

toward children with disabilities (DIS) as a consequence of increased contact

in schools.
There is growing debate about both the appropriateness of integration for
all and the way research into integration has been conducted (e.g., Hornby,
1992; Lindsay, 1989). It has been reported that attitudes toward disabled people
are frequently negatively biased (Wright, 1988), although there is some debate
about the origins of those attitudes. Some theorists have suggested that ambivalent attitudes toward people with disabilities are rooted in the same causes as
those toward other so-called disadvantaged groups such as ethnicity or gender
(e.g., Katz, Hass, & Bailey, 1988; Langer & Chanowitz, 1988). Research that
has specifically considered the attitudes of children and adolescents toward
disability reflect a general consensus that the lack of contact between people
with and without disabilities can result in negative attitudes and unrealistic
perceptions by the latter of the former (e.g., Brinker & Thorpe, 1984; Cavallaro
& Porter, 1980; Esposito & Reed, 1986; Furnham & Gibbs, 1984; McConkey,
McCormack, & Naughton, 1983; Strauch, 1970; Strohmer, Grand, & Purcell,
1984; Voeltz, 1980, 1984). However, Donaldson (1980) suggests that contact
per se may not be enough to reduce prejudice toward people with disabilities.
She argues that contact needs to be structured with planned experiences
between children with and without disabilities.
Addressing this latter issue, a small number of studies have focused on the
psychological processes underpinning attitudes arising out of integrated school
contact. For example, Fortini (1987) attempted to identify predictors of positive behavior toward children with disabilities, Lewis and Lewis (1987, 1988)
looked at cognitive mechanisms underlying attitude formation. In their schoolbased study, Lewis and Lewis (1987) found that normal children as young as
6 and 7 years acquired attitudes about children with severe learning difficulties
(SLD) after only a very limited amount of integrated contact. Previous research
by Hazzard (1983) had posited that childrens knowledge of disability was
directly related to their chronological age (the older the child, the greater the
knowledge). These authors all suggest that integrated school contact has an
effect on childrens attitudes. However, Hazzard suggests that attitude change
is an effect of age, while Lewis and Lewis (1987) conclude that it is the contact
per se that has the effect. Lewis and Lewis (1988) found in a follow-up study
that the children had generally maintained their positive attitudes shown
towards children with SLD (p. 161).
Lewis and Lewis (1988) utilized the intergroup perspective of Allport
(1954) as an explanation for their findings. Drawing mainly on research in
ethnic relations, Allport (1954/1979) suggested that the effects of the contact
would be greatly enhanced by the presence of a number of precursors including
institutional support, common goals, and the perception of communality between

EFFECTS OF CONTACT

21 15

groups. This view has grown in popularity and has come to be held by an
increasing number of researchers (e.g., Amir, 1969; Cook, 1978; Pettigrew,
1971). More recently, two seemingly contrasting models of intergroup contact
have emerged, both of which have their roots in social identity theory (Tajfel,
1978) and which offer very different strategies for optimizing the effects of
intergroup contact (Brewer & Miller, 1984; Hewstone & Brown, 1986).
In summary, Brewer and Miller (1984) see the blurring or breaking down
of group or category boundaries as essential to personalization (the reduction
of categorical biases); in contrast, Hewstone and Brown (1986) propose that
maintaining group boundaries can be beneficial for aiding positive generalization beyond the contact situation. There is some evidence which supports both
Brewer and Millers and Hewstone and Browns positions (e.g., see Vivian &
Brown, 1993; Vivian, Hewstone, & Brown, in press; but cf. Bettencourt,
Brewer, Rogers-Croak, & Miller, 1992; Miller, Brewer, & Edwards, 1985).
However, as yet, these models of contact are largely untested in real life
contexts. Furthermore, both models, and the research they stem from, have
been mainly concerned with contact between ethnic, national, or ad hoc groups.
It remains to be seen how applicable they are to social situations involving
children with and without disabilities.
Early work on attitudes toward disability showed trends toward contact
improving attitudes (e.g., Cheder, 1965). More recently, researchers have
begun to pay attention to how the contact experience should be structured. For
example, Acton and Zarbatany (1988) found that ND childrens preference for
mildly mentally retarded (MR) schoolmates increased significantly after
participating in structured cooperative activities with them. Further, Johnson,
Rynders, Johnson, Schmidt, and Haider (1979) compared the effects of integration that had cooperative, individual, or laissez-faire goal structures on mainstream childrens attitudes towards highly trainable MR peers. They found that
there were more interpersonal interactions and attraction between ND and MR
children in the cooperative situation than in the individual or laissez-faire
conditions. Although this study is a strong indicator that cooperative goals are
likely to enhance integration, it is not entirely clear that the findings are directly
relevant to all kinds of special educational needs (SEN), since the participants
were all teenagers and the integrated students were very much more able
than many children with SEN. Similar qualifications may apply to two further
studies in which Johnson and Johnson were involved, both of which found
beneficial attitudinal and achievement effects of cooperative (vs. individualistic) learning programs (Armstrong, Johnson, & Balow, 1981; Johnson &
Johnson, 1981).
Further studies lend weight to the efficacy of cooperative learning (e.g.,
Bryan, Donahue, &Pearl, 1981; Foot, Morgan, & Shute, 1990). One suggestion

21 16

MARAS AND BROWN

is that cooperative learning is optimal in a situation where participants have

important roles but where these roles are interdependent on each other (Cowie
& Rudduck, 1990). Here, there is the possibility of the promotion of the
capacity to negotiate meaning from the tasklgroup and for the acknowledgment of the existence of multiple perspectives on any issue (Cowie &
Rudduck, 1990, p. 154).
The research reported in this paper examined what happened to some mainstream childrens attitudes over a period of time when children with SLD are
integrated in a structured way into their mainstream school. A school was identified from which, as part of the curriculum, a number of children are randomly
selected each year to participate in an integrated program with children from a
special school for children with SLD at regular weekly intervals. This
integrated program offered a unique research opportunity to conduct an experimental study within a naturalistic context. The program favored Hewstone and
Browns (1986) model of categorized contact. Categories (SLD and mainstream) were explicit, clearly defined, and maintained throughout the program.
In addition, a number of the precursors proposed by Allport (1 979) and more
recently by Hewstone and Brown as essential for successful contact such as
institutional support, were also evident in the scheme.
Thus, the research reported had two main aims: (a) The measurement over
time of changes in childrens attitudes towards learning disability and intergroup perceptions as a function of different amounts and types of contact, (b) a
quasi-evaluation of an exchange program.
Data are presented that explore the criteria children use to categorize their
judgments about and their sociometric preference for unknown peers with and
without disabilities. Our principal hypothesis was that the children participating in the program would show more positive attitudes toward their disabled
peers than those not participating. In addition, in line with Hewstone and
Browns (1 986) model, the experimental children should show more evidence
of generalization of this positive attitude change.
The Integration Program
The integration program between the special school and the mainstream
primary school4 had been in operation for a number of years. The two schools
are within the same catchment area of a London borough. Children in the
special school have all been identified as having SLD, although this classification encompasses children with a wide range of physical and cognitive
41n the U.K., primary schools are schools for children aged 5 to 1 1 years. Schools for
pupils over 1 1 years are generally referred to as secondary schools.

EFFECTS OF CONTACT

21 17

impairments. Integrating children are selected from a number of classes, and

selection is based on recommendations of and discussions between the head,
deputy, and class teachers in the special school. Twenty children with SLD
were selected to integrate in the period reported in this paper.5
Integrated sessions took place on the same afternoon each week, the form
of involvement being determined by random allocation, and the allocation
being under the supervision of the researchers.6 The integrated program involved structured sessions with mainstream children being paired and grouped
with SLD children. All work was collaborative, and ND children were given
information about how to approach each task-for example, by being given
checklists of tasks to be completed, and both mainstream and special school
teachers were involved in the sessions.
Method

Design
The study utilized a 2 x 3 design (Program Participation [Experimental vs.
Control] x Time [Time 1 vs. Time 2 vs. Time 3]), the second factor being
within-subjects.

Participants
Fifty children from two national curriculum Year 4 classes participated in
the study-28 girls and 22 boys. The mean age of the children was 8.8 years
(range 8 to 10 years, mode 9 years).7 Twenty-six children from one of the
classes participated in the integrated program (experimental group). Twentyfour children from the second class served as the control group. The decision
to be in either the control or experimental group was random as children were
randomly assigned to the two classes earlier in the year. However, it is important to note that the first phase of this study (Time 1) was conducted in the first
week of the new school year, and thus the children had only been together for
a minimal amount of time.
'Sadly, one of the children with SLD who was involved in the program died during the
term the program was studied.
6The integrated sessions took place in three forms: (a) 10 mainstream children went to
the special school; (b) five mainstream children spent time with SLD children in the mainstream
staff room; and (c) the remaining children stayed in the mainstream classroom. However, the
numbers involved were too small to permit any subgroup analysis.
'Parental permission to participate in the study was obtained prior to random selection.
Ten children moved or were absent for at least one of the three sessions.

21 18 MARAS AND BROWN

Procedure
Stimuli. Various stimuli were pretested. The data described here used
stimulus photographs of unknown children with and without disabilities to
elicit childrens reactions (standard stimuli, SS). The photographs showed
children with physical disabilities (PD) sitting in wheelchairs, children with
hearing impairment (HI) wearing body-worn hearing aids, and children with
learning disabilities (LD) who had Downs syndrome. The ND children were
matched for age, gender, and other features such as posture; piloting determined that the different disabilities were obvious to children of this age. In
addition, photographs were taken of all the children with SLD involved in the
integrated sessions, and these were utilized to determine changes in sociometric choice over time.
Measures. Three measures were used for this research. The first, a new
measure consisted of five different sized balloons. This was designed for
measuring amount of certain physical and psychological attributes. The second, a five happylsad faces Likert-type scale, was modified for measuring
affect or liking. The third measure consisted of five postboxes into which the
children were asked to post the photographs of the known and unknown
individual children depending on how much they wanted to play with them
(always, a lot, sometimes, not much, or never), thus supplying a measure of
sociometric preference.
Categorization: Sorting tasks. The children sorted the SS into piles in an
unstructured sorting task with the instructions Take a good look at the
photographs, then put into piles any photographs that are alike, that go
together . . .. The purpose of this task was to explore how the children
organized their perceptions of the children represented in the SS. Children were
told to sort the photographs any way they wanted . . . . into as many piles as
they liked.
Measures of affect and evaluation: Smileyface and balloon tasks. The
children used the specially designed scales (affect = Likert-type 1-5, and how
much of 1-5) to respond to items on a questionnaire. The first question measured liking (affect); the remainder measured psychological and physical attributes of unknown children (DIS and ND). To elicit these stereotypical
responses, the unknown stimuli photographs (SS) were used and modified to
show the children in pairs.
Sociometric preference: Play with task. The children indicated, using
the postbox measure of sociometric preference, how much they would or
would not like to play with each of the SS and all of the SLD children by
posting the photographs in the five boxes provided. This was then converted to
a 5-point scale.

EFFECTS OF CONTACT

21 19

Table 1
Stress Valuesfor n-Dim<ensionalMultidimensional Scaling Solutions by
Condition and Time
Dimension 2

Dimension 1
Condition

Time

Stress

RSQ

Stress

RSQ

Experimental

1
2
1
2

.46 1
.356
.263
.320

.483
.613
320
.767

.015a
.196a
.155a
.216a

.875
.766
.884
.750

Control

Note. Stress values for one- and two-dimensional solutions are shown in Columns 1 and 3.
RSQ values: proportion of variance of scaled data (disparities) accounted for by their

corresponding distances.
aselected solutions.
Results
Categorization
Multidimensional scaling techniques (MDS-alscal) were applied to the sort
data. The input to this program took the form of similarity matrices based on
the frequency of co-occurrence of each pair of stimuli. Each similarity matrix
represented the aggregarte frequency of one group of children (experimental,
control) at one point in time. The MDS procedure involves fitting of points,
in our case individual photographs of unknown children with and without
disabilities (SS), to metric space that indicates similarity (when SS are close
together) and dissimilarity (when S S are far apart).8 Thus, the MDS locates in
an n-dimensional space the underlying structure of similarity judgment.9 The
distance between plots indicates the degree to which the children tended to
categorize the stimuli as similar. For economy of presentation, data from just
Time 1 and Time 3 are presented below for each of the experimental and control
groups (Figures 1 and 2). Kruskals (Kruskal & Wish, 1978) stress formula and
R2 values are given for each solution at Time 1 and Time 3 in Table 1.
8The Alscal procedure on SPSS was used to perform these analyses.
9For a fuller explanation, see Kruskal and Wish (1978).

2120

MARAS AND BROWN

CE:DER

2.1

1.0

.#HI

-*

WPD)

MU)
DISABILITY
0.0

.F(PD)
-1.0

-2.1

OF

*F(HI)

F -F

F(U)

-.

2.1

1.0

I
GENDER

OH

on
On

@ H(PD)

0.0
F(H1)

F 8 F
-1.0 -

- 2 . 1 -

O F

YIS*BILITY
F(PD)

0 ULD)

Figure 1. Multidimensional scaling plots of sorting standard stimuli at Time 1 in

experimental and control groups.

EFFECTS OF CONTACT

2121

As can be seen in Figure 1, at Time 1 the children in the Experimental

group are sorting the photographs on two main dimensions: gender and
disability. Notice that the girls and boys are clustered tightly in DIS and ND
groups. At Time 3 the picture changes somewhat. There seems to be little
evidence of sorting on the same two general dimensions. Rather, the disabled
PD and HI children seem to be subtyped, while the LD boy is placed among the
other disabled stimuli and the LD girl is placed nearer to the ND children. One
interpretation could be ihat disability remained salient but became more differentiated; and the children seemed to become more aware of different kinds of
disability. Gender, by contrast, seemed to become much less salient at Time 3 .
As can be seen in Fiigure 1, at Time 1 the control children's categorization
of the SS are consistent with the pattern already seen in the experimental group.
The children are sorting on the two main dimensions gender and disability with
the DIS children clustered more tightly than the ND children. At Time 3 the
picture is almost identical to that at Time 1; once again, the children are
employing just the two main dimensions of gender and disability and are
exhibiting little evidence of any differentiation within the latter category. This
is in complete contrast to the experimental group. Correlational analysis of
loadings for the two dirnensions in the two groups lends further support to this
finding. Correlations between loadings at Time 1 and Time 3 for both dimensions are highly significant in the control condition Dimension 1 (disability),
r = .97 p < .01; Dimension 2 (gender), r = .82 p < .01. In contrast, in the
experimental condition.,although correlations are high on Dimension 1 ( r = 3 6 ,
p < .01) on dimension two there is no significant correlation ( r = .26, ns).
Thus, the contact experience afforded by the integration program seems to
have markedly affected the way the children order their social environment.
The Evaluative Measures
ANOVA using the MANOVA procedure on SPSS was employed to analyze
the eight dependent measures tapping group evaluations. In addition to the two
main independent variables in the design there were one further betweensubjects factor (gender of participating child) and two further within-subjects
factors (stimuli [HI, LD, PD, ND]; and gender of stimuli).lO

'OAs is often the case in applied research, some preexisting differences between the groups
were found at Time 1 despite random assignment to treatments. In order to ascertain that these
were not artifactually inflating subsequent differences between the groups at Time 2 and Time 3,
analyses of covariance were conducted on all measures reported in this paper using evaluation
at Time 1 as the covariate. Findings on all measures remained statistically significant in all
of these analyses.

2122

MARAS AND BROWN

2.1

1.0

n@
0 F(U))

-1.0

-2.1

UPD)

nm

0.0

DISABILITY

OH(U))

@F

.M(PD)

OF

N(H1)

mF(H1)

2.1

CENI

1.o

0.0

-1

.o

-2.1
-2.5

-1.5

-0.5

0.5

1.5

2.5

(2b) Control group

Figure 2. Multidimensional scaling plots of sorting standard stimuli at Time 3 in

experimental and control groups.

3.22*
7.55**
4.20*
9.58***
4.88***
3.50**
3.97**

2.64*

2.47*

F(6,258)

SxT

6.56**

F(2, 86)

CxT

Physical Education. Friends

3.24*
2.76*

F(3, 129)

cx s

Interactions

Principal significant effects

Note. Like = Respondents liking of target children. Schwk = Schoolwork. PE

Wkhard = Working. Run = Running. Hear = Hearing. Think = Thinking.
*p < .05. **p < .01. ***p < ,001.

4.77*

19.30***
210.75***
77.16***
82.44***

7.32***

93.48***
21.82***

7.00**

4.2 1*
6.03**

16.50***
69.94***

5.18*

Like
Schwk

PE
Friends
Wkhard
Run
Hear
Think

F(2,86)

F(3, 129)

F(1,43)

Measure

Time (T)

Stimuli (S)

Within subjects

Contact (C)

Between
subjects

Summary of Principal Signi9cant Effects on Eight Judgement Measures

Table 2

Making Friends.

3.80** (2.35)
2.89* (2.58)

F(6,222) (MSe)

CxSxT

h)
G)

!2

-4

20

8
z

0
-4
v,

rn

n
n

rn

2124

MARAS AND BROWN

The principal significant effects identified in these analyses are summarized

in Table 2 . Of most interest to our hypothesis are interactions involving the
contact variable-particularly Contact x Time x Stimulus-since these indicate
differential changes in attitude toward particular stimulus groups over time as
a function of the integration program. Before discussing these interactions a
few general points are worth noting in Table 2 . First, on every measure there
was a highly significant main effect due to stimulus photographs being evaluated. The consistent trend across all measures was for ND children in the
photographs to be rated more favorably than the DIS children with mostly
nonsignificant differences among the latter. There were also consistent main
effects for time, indicating a general increase in rating between Time 1 and
Time 2 and then a leveling off. The three measures revealing reliable interactions involving the contact variable were: run, hear, and think. These
are discussed in turn.
Running

On this measure, there were both two-way (Contact x Stimulus) and threeway (Contact x Stimulus x Time) interactions involving the contact variable.
We concentrate here on the theoretically more relevant three-way effect. There
was virtually no difference at Time 1 in the ratings between the experimental
and control groups; as usual, the ND stimuli were evaluated higher than the DIS
stimuli in both groups, although the difference is only marginally significant in
control. However, ratings in the experimental group show a marked rise at
Time 2 and are significantly higher for LD, HI, and PD at Time 3, while in the
control group, there is little difference as an effect of time. This finding is
particularly relevant, as it is LD children with whom the experimental children
have contact and it is within the LD and HI stimuli that the most positive and
significant changes occur. The fact that the positive effects in the LD group are
reflected in the HI group also bode well for generalization across subcategories
of disability, that is from the known group LD to the unknown group HI. It is
thus indicative of a reduction in stereotypes specifically in relation to the HI
and LD children who at Time 3 are not rated any less able to run than the ND
children by the experimental group. In the control children, however, the clear
differentiation between ND and DIS children remains visible at Time 3 even if
it does appear to diminish slightly at Time 2 (Table 3).
Hearing

The same interactions were also identified in ANOVA of the hearing

measure. In this instance, as can be seen in Table 4, a similar pattern to the

EFFECTS OF CONTACT

2125

Table 3
Mean Evaluations of Running
How well can these children run?
Stimuli (SS)
Integrated
contact

Time

HI

LD

PD

ND

Marginals

Yes

1
2
3

3.4b
4.4
4.0ab

3.43
4.3
4.0ab

3.3b
3.8
3.71,

4.1,
4.6
4.4,

3.4
3.8
3.7

No

1
2
3

3.2
3.8
3.3b

3.3
3.5
3.2b

3.2
3.4
3.1b

3.8
3.3
4.4,

3.4
3.7
3.3

3.4

3.2

1.6

4.4

Marginals

Note. HI = hearing impaired. LD = learning disability. PD = physical disability. ND =

nondisabled. Means in row not sharing the same subscript are significantly different by
Tukeys HSD test, p < .05.

running measure can be seen but with one key difference. As before, there is
little difference between the experimental or control groups at Time 1. However, ratings in the experimental group become greater over time for only LD
and PD children, and there is little change in their ratings of HI children. This
is particularly interesting given that LD children are integrated in this school
and a number of them are in wheelchairs (i.e., they appear similar to PD stimuli
children). In the no-contact control group, there is little difference in the
childrens ratings of HI, LD, or PD as an effect of time, with ND remaining
consistently higher than the DIS stimuli. In this instance, therefore, contact
seems to be changing stereotypes about LD and PD childrens ability to hear,
while retaining a realistic picture of the hearing ability of HI children.
Thinking
The three-way interaction, Contact x Stimuli x Time was not found in the
more abstract concept of thinking. The Stimuli x Time interaction shown in all

2126

MARAS AND BROWN

Table 4
Mean Evaluations of Hearing
How well can these children hear?
Stimuli (SS)
Integrated
contact

Time

HI

LD

PD

ND

Marginals

Yes

1
2
3

2.0b
2.5c
2.2b

3.0,
4.0ab
3.6,

2.11,
3.2bc
4.1,

4.6,
4.5,
4.3,

3.5
3.9
3.8

No

1
2
3

2.21,
2.2b
2.2b

2.9b
3.4,
3.2,

2.5b
2.4b
2.6b

4.4,
4.5,
4.6,

3.5
3.6
3.7

2.4

3.3

2.8

4.5

Marginals

Note. HI = hearing impaired. LD = learning disability. PD = physical disability. ND =

nondisabled. Means in row not sharing the same subscript are significantly different by
Tukeys HSD test, p < .05.

of the measures above was found for this measure, and examination of the mean
scores reveals that in both experimental and control groups, ratings of the DIS
stimuli change over time, although not of the ND children. A Contact x Time
interaction was also identified. Children in the experimental group rate the HI,
LD, and PD stimuli significantly lower than do those in the control group at
Time 1, but higher at Time 2. At Time 3 there is little difference between
experimental and control group ratings of HI and LD, though children in the
experimental group rate PD higher than in the control group (Table 5).
Sociometric Preference
The final measures (sociometric preference) utilized the postbox measure.
A number of significant interactions were identified when the play with data
were analyzed. In line with past research (e.g., Maccoby & Jacklin, 1976), there
was a significant interaction between gender of participating child and gender
of stimuli, F(2, 35) = 7 2 . 0 3 , ~< .001, which was revealed in a consistent and

EFFECTS OF CONTACT

2127

Table 5
Mean Evaluations of Thinking
How well can these children think?
Stimuli (SS)
Integrated
contact
Yes

No

Time

HI

LD

PD

ND

Marginals

1
2
3

2.0,
3.81,
3.5b

1.8bC
3.5b
3.21,

1.3b
3.3b
3.4b

4.2,
4.8,
4.5,

2.9
4.2
3.9

3.2,
3.01,
3.7,

2.51,
2.7b
3.2,b

2.4b
2.5b
2.8b

4.4,
4.1,
4.4,

3.5
3.4
3.8

3.2

2.8

2.7

4.4

2
3
Marginals

Note. HI = hearing impaired. LD = learning disability. PD = physical disability. ND =

nondisabled. Means in row not sharing the same subscript are significantly different by
Tukeys HSD test, p < .05.

strong same gender preference throughout (Table 6 ) . More relevant for the
hypothesis are the interactions involving contact, stimuli and time, F(6,222) =
9.1 1 , p < .001. First, contact as a whole had a main effect (3.2 vs. 3.9), but this
was qualified by time and type of stimulus as can be seen in the increased
ratings of both DIS and ND stimuli in the experimental group between Time 1
and Time 2 (see marginal columns in Table 7 which show time effects just for
the experimental group, collapsing across type of disability). In the control
group, by contrast, the ratings showed little change over time.
Generalization
In order to examine ,generalizationfrom the individual to the group, a comparison was made between ratings of how much children wanted to play with known
and unknown individual children with disabilities. Mean scores were computed
for the childrens sociometric choice for the male and female known SLD
children, and these were correlated with mean scores for sociometric preference

2128 MARAS AND BROWN

Table 6
Participating Children 's Mean Sociometric Preference
How well can these children think?
Stimuli (SS)
Disabled
Integrated
contact
Yes

Time

Participants

Girls

Boys

Girls

Boys

Marginals

Girls
Boys
Girls
Boys
Girls
Boys

3.81,
2.31,
4.21,
2.9b
4.2b
3.21,

3.0,
2.6b
4.01,
3.3b
3.8,
3.4b

4.41,
2.5b
4.6b
3.2b
4.5b
3.2b

1.9,
3.9,
4.0,
4.4,
3.7,
4.2,

3.2
2.8
4.2
3.5
4.1
3.5

Girls
Boys
Girls
Boys
Girls
Boys

3.71,
2.4b
3.9b
2.3,
2.91,
2.4ab

2.9,
2.6b
3.11,
2.6b
2.81,
2.71,

3.5&
2.61,
3.8,
2.81,
3.5,
2.7b

3.2ab
4.2,
3.9,
4.0,
2.9ab
3.9,

3.3
3.0
3.7
2.9
3.0
2.9

3.1

3.0

3.3

3.7

2
3

No

I
2
3

Marginals

Nondisabled

Note. Means in row not sharing the same subscript are significantly different by
Tukey's HSD test, p < .05.

of unknown girls and boys at Time 2 and Time 3, separately for experimental and
control children.ll At Time 2 there were significant correlations in both the
control and experimental groups, but the latter relationship was significantly
higher than the former (2 = 2.57, p < .05). At Time 3 the correlations in each
group were also high, but there was no difference between them (Table 8).
"As noted earlier, one child from the special school died during the period the study was
conducted. Data for sociometric preference at Time 1 are not included in these analyses as
they were collected prior to this child's death.

EFFECTS OF CONTACT

2129

Table 7

Participating Children ',sMean Sociometric Preference for Disabled and

Nondisabled Standard Stimuli in Experimental Condition Only
Stimuli ( S S )
Time

Disabled

Nondisabled

Marginals

1
2
3
Marginals

2.8
3.5
3.6
3.2

3.2
4.0
3.8
3.6

3.0
3.7
3.7

Discussion
The results from this natural experiment demonstrate some consistent and
powerful effects of contact in the applied context of integrated education for
children with disabilities. As already noted, little recent research has focused
on real situations, tending instead to be laboratory based. In our study, at the
beginning of the prograim both experimental and control groups categorized on
the same two main dimensions: gender and disability. However, by Time 3
children in the experimental group had ceased to use gender but continued to
use disability, if in a more differentiated way. The control children persisted in
using the broad categories of gender and disability. Thus, after only 3 months
of structured and planned contact with the SLD children, children in the
experimental group were using the category of disability in a more differentiated way and subtyping the different disabled groups. For example, in the study
the category with which the children are having contact (LD) became more
salient and generalized attitudes became more positive.
The evaluative data also provided some interesting changes over time. As
with the categorization task, there were highly significant changes in the
experimental (contact) group's attitudes over time. These changes were in a
positive direction and were particularly evident on the measures of running,
hearing, and thinking. On the first two dimensions, ratings of LD and PD
children were significamtly higher at Time 2 and Time 3. Findings on the
thinking measure were islightly less clear than on running and hearing, perhaps
reflecting the abstract nature of the concept being rated. In addition, the
children in the experimental group were aware that they would be involved in

2130

MARAS AND BROWN

Table 8

Correlations Between Sociometric Preference for Known and Unknown

Children With Disabilities

Experimental
Control
Difference between groups ( Z score)

Time 2

Time 3

0.90**
0.53**
2.57*

0.77**
0.77**
0.00

Note. Z scores * = significantp < .05, one-tailed.

* p < .05. **p < .01.

the program and a number expressed uncertainty about participation in terms

of their expectations and knowledge of the SLD children. This uncertainty may
well be reflected in the low ratings of thinking given to the stimuli-particularly the PD children-at Time 1 of the study, and the increase in ratings of the
stimuli over time. This is a particularly hopeful finding as the integrating
children in the scheme all had SLD. In addition, a number of them were in
wheelchairs and had PD as well as LD, so this might go some way toward
explaining the similar rise in evaluations of the PD group.
The increased ratings over time on some of the evaluative measures were
reflected in how much the children in the experimental group wanted to play
with the SS. Over time, this increased significantly and bodes well for integrated contact of this kind. After all, the evaluative measures were more
abstract constructs, while playing is probably one of the most important and
salient activities for children. The data on sociometric preference also allowed
us to look at the issue central to two models of contact: generalization. Again
(though in this instance only partially) in support of the intergroup model,
preference for known and unknown disabled children was more highly correlated in the experimental group than in the control group at Time 2 but by
Time 3 there was no difference between the correlations. There are two
possible reasons for these slightly less clear findings; first, the sample size in
these analyses was rather low, and second, although the control class was not
involved in the integration scheme, they did see the SLD children at playtimes
and may well have played with them during these times.
The quasi-experiment lends support for integration when contact is organized in a planned and structured way, and its findings have important implications for policy and practice in terms of integration. In this study numbers of

EFFECTS OF CONTACT

21 31

children involved in the program were small, the collaborative work they
engaged in was carefully planned and implemented, and there is a strong ethos
in the schools of the importance of the integration of children with special
educational needs. These are all factors identified by Allport (1954/1979) and
others since as being important for successful contact. For example, recent
work in support of Hewstone and Browns (1986) model has found that
cooperation is an important component in improving attitudes (Vivian &
Brown, 1993; Vivian et al., in press). In addition, after only 3 months of
structured and planned contact with the SLD children, children in the experimental group were using the category of disability in a more differentiated way
and subtyping the different disabled groups, a finding that lends some support
to both models: Hewstone and Browns because the category with which the
children are having contact (LD) has become more salient, a feature they would
predict to be conducive to positive generalized attitude change; and Brewer and
Miller (1984) because it might be theorized that this subtyping was a move
toward personalization, and thus be in line with their view that categories have
to be broken down to promote positive attitude change.
Our findings thus have a number of important implications for integrated
schemes and programs. Ideally classes should be small and physical access
should be as easy as possible. Nonteaching staff should be used effectively to
ensure all children are adequately supported. Children with disabilities should
be integrated into their local schools. Sessions involving integrated contact
should be structured and involve cooperative tasks in which all participants
should have clearly defined roles. Mainstream children and teachers should be
well prepared for integration and be given basic information along with the
opportunity to ask questions, and vice versa for children with disabilities and
their teachers.
Although the children involved in the integrated program were prepared for
the program, this preparation was not optimal and there was little preparation
for the mainstream teachers or other children in the mainstream primary. This
perhaps goes some way toward explaining that, while attitudes did significantly
improve over time, DIS children were still consistently rated unfavorably in
comparison to ND peers. In conclusion, anecdotal evidence from this study
revealed that mainstream teachers and children had very mixed feelings about
participating in future integrated projects, often feeling unprepared and underresourced. It is likely that these attitudes of mainstream children and teachers
will have an effect on the children with disabilities with whom they are having
contact in the future. ]Findings from our work suggest that future research
should focus both on pireparing mainstream children and teachers for integration and on the effects of attitudes on children with disabilities integrated into
mainstream schools.

2132

MARAS AND BROWN

References
Acton, H. M., & Zarbatany, L. (1988). Interaction and performance within
cooperative groups: Effects on nonhandicapped students attitudes toward
their mildly mentally retarded peers. American Journal of Mental Retardation, 95, 16-23.
Allport, G . W. (1979). The nature of prejudice. Cambridge, MA: AddisonWesley. (Original work published 1954)
Amir, Y. (1969). Contact hypothesis in ethnic relations. Psychological Bulletin, 71,319-342.
Armstrong, B., Johnson, D. W., & Balow, B. (1981). Effects of cooperative vs.
individualistic learning experiences on interpersonal attraction between
learning-disabled and normal-progress elementary school students. Contemporary Educational Psychology, 6, 102-109.
Bettencourt, B. A., Brewer, M. B., Rogers-Croak, M., & Miller, N. (1992).
Cooperation and the reduction of intergroup bias: The role of reward
structure and social orientation. Journal of Experimental Social Psychol00,
28, 301-319.
Brewer, M. B., & Miller, N. (1984). Beyond the contact hypothesis. In N.
Miller & M. B. Brewer (Eds.), Groups in contact: The psychology of
desegregation (pp. 28 1-300). New York, NY: Academic.
Brinker, R. P., & Thorpe, M. E. (1984). Integration of severely handicapped
students and the proportion of IEP objectives achieved. Exceptional Children, 51, 168-175.
Bryan, T., Donahue, M., & Pearl, R. (1981). Learning disabled childrens peer
interactions during a small-group problem solving task. Learning Disability Quarterly, 4, 13-22.
Cavallaro, S. A., & Porter, R. H. (1980). Peer preferences of at risk and
normally developing children in a pre-school mainstream classroom.
American Journal of Mental DeJciency, 84,352-357.
Cheder, M. A. (1965). Ethnocentrism and attitudes toward the physically
disabled. Journal of Personality and Social Psychology, 2, 877-882.
Cook, S. W. (1978). Interpersonal and attitudinal outcomes in cooperating
interracial groups. Journal of Research and Development in Education, 12,
97-1 13.
Cowie, H., & Rudduck, J. (1990). Learning from one another: The challenge.
In H. C. Foot, M. J. Morgan, & R. H. Shute (Eds.), Children helping
children (pp. 235-255). London, England: John Wiley & Sons.
Department of Education and Science (DES). (1981). The Education Act.
London, England: Her Majesties Stationary Office.

EFFECTS OF CONTACT

2133

Department of Education and Science (DES). (1987). Third report of the House
of Commons Educcztion, Science & Arts Committee. London, England:
Her Majesties Statijonary Office.
Donaldson, J. (1980). Changing attitudes toward handicapped persons: A review and analysis of research. Exceptional Children, 46, 504-5 14.
Esposito, B. G., & Reed, T. M. (1986). The effects of contact with handicapped persons on young childrens attitudes. Exceptional Children, 53,
224-229.
Foot, H. C., Morgan, M. J., & Shute, R. H. (Eds.). (1990). Children helping
children. Chichester, England: John Wiley & Sons.
Fortini, M. E. (1987). Attitudes and behavior toward students with handicaps
by their nonhandicapped peers. American Journal of Mental Deficiency,
91, 78-84.
Furnham, A., & Gibbs, M. (1984). Schoolchildrens attitudes towards the
handicapped. Journal of Adolescence, 7, 99-1 17.
Hazzard, A. (1983). Childrens experience with, knowledge of and attitudes
towards disabled peers. Journal of Special Education, 17, 131- 139.
Hewstone, M., & Brown, R. (1986). Contact is not enough: An intergroup
perspective on the contact hypothesis. In M. Hewstone & R. Brown (Eds.),
Contact and conflct ,inintergroup encounters (pp. 1 4 ) . Oxford, England: Basil
Blackwell.
Hornby, G. (1992). Integration of children with special educational needs: Is
it time for a policy review? Support for Learning, 7, 130-134.
Johnson, D. W., & Johnson, R. T. (1981). The integration of the handicapped
into the regular classroom: Effects of cooperative and individualistic instruction. Contemporary Educational Psychology, 6,344-353.
Johnson, R., Rynders, J., Johnson, D. W., Schmidt, B., & Haider, S. (1979).
Interaction between handicapped and nonhandicapped teenagers as a function of situational goal structuring: Implications for mainstreaming.
American Educational Research Journal, 16, 161- 167.
Katz, I., Hass, R., & Biailey, J. (1988). Attitudinal ambivalence and behavior
toward people with disabilities. In H. E. Yuker (Ed.), Attitudes toward
persons with disabilities (pp. 47-57). New York, NY: Springer.
Kruskal, J. B., & Wish, M. (1978). Multi-dimensional scaling. Beverly Hills,
CA: Sage.
Langer, E., & Chanowitz, B. (1988). Mindfulness/mindlessness:A new perspective for the study of disability. In H. E. Yuker (Ed.), Attitudes toward
persons with disabilities (pp. 68-8 1). New York, NY: Springer.
Lewis, A., & Lewis, V. (1987). The attitudes of young children towards peers
with severe learning difficulties. British Journal of Developmental Psychology, 5,287-292.

2134

MARAS AND BROWN

Lewis, A., & Lewis, V. (1988). Young childrens attitudes after a period of
integration, towards peers with severe learning difficulties. European
Journal of Special Needs Education, 3, 161-172.
Lindsay, G. (1989). Assessing children with learning difficulties in the new
ERA. Support f o r Learning, 4,209-215.
Maccoby, E. E., & Jacklin, C. N. (1976). Gender segregation in childhood.
Advances in Child Development and Behaviour, 20,239-287.
McConkey, R., McCormack, B., & Naughton, M. (1983). A national survey of
young peoples perceptions of mental handicap. Journal of Mental DeJiciency Research, 27, 17 1 - 183.
Miller, N., Brewer, M. B., & Edwards, K. (1985). Cooperative interaction in
desegregated settings: A laboratory analogue. Journal of Social Issues,
41, 63-79.
Pettigrew, T. F. (1971). Racially separate or together? New York, NY:
McGraw Hill.
Strauch, J. D. (1970). Social contact as a variable in the expressed attitudes of
normal adolescents towards EMR pupils. Exceptional Children, 36,
485-494.
Strohmer, D. C., Grand, S. A., & Purcell, M. J. (1984). Attitudes towards
persons with a disability: An examination of demographic factors, social
context, and specific disability. Rehabilitation Psychology, 29, 13 1 - 145.
Tajfel, H. (1 978). Intergroup behavior. Individualistic perspectives. In H.
Tajfel & C. Fraser (Eds.), Introducing social psychology (pp. 401-422).
London, England: Penguin.
Vivian, J., & Brown, R. (1993, September). Salience of membership and
intergroup contact: The effects of perceived typicality and outgroup
homogeneity. Paper presented to the general meeting of the European
Association of Experimental Social Psychology, Lisbon, Portugal.
Vivian, J., Hewstone, M., & Brown, R. (in press). Intergroup contact: Theoretical
and empirical developments. In R. Ben-Ari & Y. Rich (Eds.), Understanding and enhancing education f o r diverse students. An international
perspective (pp. 1-47). Jerusalem, Israel: Bar-Ilam University Press.
Voeltz, L. M. (1980). Childrens attitudes towards handicapped peers. American
Journal of Mental Deficiency, 84,455-464.
Voeltz, L. M. (1984). Effects of structured interactions with severely handicapped peers on childrens attitudes. American Journal of Mental Deficiency, 88, 380-390.
Wright, B. A. (1988). Attitudes and the fundamental negative bias: Conditions
and corrections. In H. E. Yuker (Ed.), Attitudes toward persons with
disabilities (pp. 3-22). New York, NY: Springer.