14
Journal of Intellectual Disability Research doi: 10.1111/j.1365-2788.2005.00781.x pp © Blackwell Publishing Ltd 335 Blackwell Science, LtdOxford, UKJIRJournal of Intellectual Disability Research0964-2633Blackwell Publishing Ltd, 2005 5335348Original ArticleA multivariable cross-sectional analysisE. A. Nowicki Correspondence: Elizabeth A. Nowicki, Faculty of Education, The University of Western Ontario, Western Road, London, Ontario, N G G , Canada (e-mail: enowick @uwo.ca). A cross-sectional multivariate analysis of children’s attitudes towards disabilities E. A. Nowicki The University of Western Ontario, London, ON, Canada Abstract Background Past research has shown that children can be biased against peers with disabilities, but the association of attitudes with gender, age and disabil- ity preferences, as well as interactions between these variables, are unclear. The objectives of this study were to examine these issues in a cross-sectional, split-plot study to clarify: ( ) if elementary school children’s attitudes towards peers with disabilities are related to age, gender and type of disability; ( ) if interactions between these variables exist; and ( ) if convergent validity could be achieved across three theoretically linked dependent variables. Methods One hundred elementary school children between and years old were assessed for attitudes towards target children with no disability, a physical or an intellectual disability, and a combined intellec- tual/physical disability. Measures were completed in an interview format. Results Attitudes towards a target child with physi- cal disabilities and a target child without disabilities did not differ. There was a significant interaction for age and disability. Attitudes towards target children with intellectual and intellectual/physical disabilities were negatively biased, and were negatively associ- ated with age. Results were consistent across mea- sures except for a main effect of gender in one measure and a gender by age interaction in another. Conclusions Children’s attitudes appear to be asso- ciated with several factors, including age and the presence or absence of disability. Gender differences in attitudes may be because of gender-based response biases rather than disability biases. Because of the multifaceted nature of childhood attitudes, cross- sectional designs with several dependent and inde- pendent variables provide an opportunity to examine consistency of results across measures and potential interactions between factors that may not be uncov- ered when variables are examined in isolation. Keywords attitudes, elementary school children, intellectual disability, physical disability Introduction Research has shown that children can have negative attitudes towards peers with disabilities (e.g. Nabors & Keyes ; Harper ; Nabuzoka & Ronning ; Nowicki ; Norwich & Kelly ; Ring & Travers ), but findings focusing on the relation of attitudes with specific variables such as gender, age and disability type are inconsistent. Discrepancies exist regarding gender differences in children’s atti- tudes towards peers with disabilities (Woodard ; Colwell ; Tamm & Prellwitz ). Nor is it clear if age and attitudes are related (Gash & Coffey ;

A cross-sectional multivariate analysis of children's attitudes towards disabilities

Embed Size (px)

Citation preview

Journal of Intellectual Disability Research doi: 10.1111/j.1365-2788.2005.00781.x

pp

©

Blackwell Publishing Ltd

335

Blackwell Science, LtdOxford, UKJIRJournal of Intellectual Disability Research0964-2633Blackwell Publishing Ltd, 2005

5335348

Original Article

A multivariable cross-sectional analysisE. A. Nowicki

Correspondence: Elizabeth A. Nowicki, Faculty of Education, The University of Western Ontario,

Western Road, London, Ontario, N

G

G

, Canada (e-mail: enowick

@uwo.ca).

A cross-sectional multivariate analysis of children’s attitudes towards disabilities

E. A. Nowicki

The University of Western Ontario, London, ON, Canada

Abstract

Background

Past research has shown that children can be biased against peers with disabilities, but the association of attitudes with gender, age and disabil-ity preferences, as well as interactions between these variables, are unclear. The objectives of this study were to examine these issues in a cross-sectional, split-plot study to clarify: (

) if elementary school children’s attitudes towards peers with disabilities are related to age, gender and type of disability; (

) if interactions between these variables exist; and (

) if convergent validity could be achieved across three theoretically linked dependent variables.

Methods

One hundred elementary school children between

and

years old were assessed for attitudes towards target children with no disability, a physical or an intellectual disability, and a combined intellec-tual/physical disability. Measures were completed in an interview format.

Results

Attitudes towards a target child with physi-cal disabilities and a target child without disabilities did not differ. There was a significant interaction for age and disability. Attitudes towards target children with intellectual and intellectual/physical disabilities were negatively biased, and were negatively associ-ated with age. Results were consistent across mea-

sures except for a main effect of gender in one measure and a gender by age interaction in another.

Conclusions

Children’s attitudes appear to be asso-ciated with several factors, including age and the presence or absence of disability. Gender differences in attitudes may be because of gender-based response biases rather than disability biases. Because of the multifaceted nature of childhood attitudes, cross-sectional designs with several dependent and inde-pendent variables provide an opportunity to examine consistency of results across measures and potential interactions between factors that may not be uncov-ered when variables are examined in isolation.

Keywords

attitudes, elementary school children, intellectual disability, physical disability

Introduction

Research has shown that children can have negative attitudes towards peers with disabilities (e.g. Nabors & Keyes

; Harper

; Nabuzoka & Ronning

; Nowicki

; Norwich & Kelly

; Ring & Travers

), but findings focusing on the relation of attitudes with specific variables such as gender, age and disability type are inconsistent. Discrepancies exist regarding gender differences in children’s atti-tudes towards peers with disabilities (Woodard

; Colwell

; Tamm & Prellwitz

). Nor is it clear if age and attitudes are related (Gash & Coffey

;

Journal of Intellectual Disability Research

E. A. Nowicki •

A multivariable cross-sectional analysis336

©

Blackwell Publishing Ltd,

Journal of Intellectual Disability Research

,

Diamond

et al.

; Tamm & Prellwitz

). Fur-ther, the literature does not present a conclusive pic-ture about children’s attitudes towards different kinds of disabilities (Nowicki & Sandieson

).There may be several reasons for these inconsisten-

cies. First, children prefer to associate with members of their own gender (Sippola

et al.

), thus atti-tudes towards a child with a disability could be influ-enced by an interaction between the gender of the target child and participant. Second, age differences may not emerge if there is a restriction of range in participants’ ages. Third, preferences for some dis-abilities over others could confound results if only one disability condition is presented (Nowicki & Sandieson

). Fourth, few studies have examined these variables simultaneously and it is unclear how they interact with one another. Finally, measures between studies differ and it is difficult to ascertain if discrepancies in outcomes are the result of the variety of representations of disability or of the atti-tude measure.

Attitudes have been described as good predictors of behaviour or behavioural intentions, and can be based on direct or indirect experience, a little or a lot of knowledge, as well as being influenced by the accuracy of knowledge (Eagly & Chaiken

; Kraus

). Researchers have found that children who have limited knowledge about or lack of exposure to individuals with disabilities can and do form attitudes about disabilities, but their attitudes may differ in valence and magnitude to those of children who have regular contact (Townsend

et al.

; Gash & Coffey

; Magiati

et al.

). The acceptance of individ-uals with disabilities is unlikely as long as negative attitudes persist (Antonak & Linveh

) and not surprisingly, inclusive education continues to be vig-orously contested (Fitch

). Thus, it is important to investigate children’s attitudes towards peers with disabilities in a systematic manner. If the literature base presents uncertainty regarding which factors, or for that matter, which levels within a factor, are asso-ciated with children’s attitudes, then designing strat-egies to facilitate inclusion may be all the more difficult.

The purpose of this study was to examine several key variables that have been addressed in the past yet have yielded inconsistent conclusions about chil-dren’s attitudes towards individuals with disabilities. It was hoped that examining these variables in con-

junction with one another and across several attitude measures would provide a better understanding of potential associations and interactions. Children between the ages of

and

years were administered three attitude measures in a one-on-one interview format. This age range was selected to ensure repre-sentation from the beginning to the end of elemen-tary school. Four target children depicting no disability, a physical or an intellectual disability, and a combined intellectual/physical disability were pre-sented so that participants would be able to make comparative judgements. Further, gender of partici-pant and target child were matched to ensure that attitudes towards disabilities were not influenced by gender biases.

Differences in gender, age and disability preferences

Several studies have reported that girls have more positive attitudes towards individuals with disabilities (Townsend

et al.

; Nabors & Keyes

; Tripp

et al.

), some found that boys are more positive (Woodard

; Nabuzoka & Ronning

), while others indicated that gender differences do not exist (Kratzer & Nelson-Le Gall

; Cohen & Lopatto

; Colwell

; Tamm & Prellwitz

). It is well known that children prefer to associate with their own gender (Sippola

et al.

). Thus, the presence or absence of matching gender of the target child and the participant may be an influencing factor in the lack of consistent results across studies. Studies matching gender have found that boys and girls have similar attitudes (Kratzer & Nelson-Le Gall

) but if gender is not matched, children show more biases against opposite-sex target children (Woodard

). Gender differences have not been found in studies with several members of each gender, such as a choir or a group of children (Cohen & Lopatto

; Colwell

).Moreover, the direction or even the existence of an

association between attitudes and age is unclear. Neg-ative associations have been reported by some (Gash & Coffey

; Weirserbs & Gottlieb

), positive associations by others (Kratzer & Nelson-Le Gall

; Townsend

et al.

), and several studies sug-gested that age and attitudes are not related (Tripp

et al.

; Diamond

et al.

; Tamm & Prellwitz

). Methodological factors contributing to the lack of consistent results may include a wide range of

Journal of Intellectual Disability Research

E. A. Nowicki •

A multivariable cross-sectional analysis337

©

Blackwell Publishing Ltd,

Journal of Intellectual Disability Research

,

age groups between studies and a narrow range of ages within studies (Nowicki & Sandieson

). A related concern is that children at the pre-reading/writing stage are often excluded from cross-sectional studies as data must be collected with time consum-ing and expensive one-on-one interviews. Excluding younger children and focusing on participants who can read and write well enough to complete paper-and-pencil measures does not provide an adequate overview of attitudes throughout elementary school.

Of further interest is the association of attitudes with type of disability. Children with intellectual dis-abilities are at risk for social isolation, neglect or rejection by their classmates (Nowicki

; Norwich & Kelly

; Ring & Travers

). Discrimination against individuals with intellectual disabilities is a consistent finding, which is not surprising given that attitudes are frequently negative (Abrams

et al.

; Nabors & Keyes

; Cohen & Lopatto

; Harper

; Okagaki

et al.

). Research address-ing attitudes towards peers with physical disabilities has been less conclusive. Some studies have reported an absence of biases against individuals with physical disabilities (Colwell

; Kratzer & Nelson-Le Gall

; Obrusníková

et al.

), although Magiati

et al.

(

) found that children had positive atti-tudes towards peers with physical disabilities but were less positive about activities that required personal involvement. Negative perceptions of peers with physical disabilities have been reported in other stud-ies (Cohen

et al.

; Bracegirdle

) including negative biases against visible disabilities such as amputation and paralysis (Woodard

). Degree of bias may vary with extent of functional restriction such that more negative attitudes are associated with greater physical limitations (Harper

). Interest-ingly, though, younger children may be more inclined than older children to believe that physical disabilities are malleable over time (Lockhart

et al.

).Few studies with child participants have used a

repeated measures design to compare attitudes towards physical and intellectual disabilities. Some children prefer peers with physical disabilities (Nabors & Keyes

), but others may not be adept at distinguishing between physical and intellectual disabilities (Abrams

et al.

; Tripp

et al.

). Indeed, the concept of disability appears to be undif-ferentiated in children who have participated in struc-tured interventions with children who have different

kinds of disabilities (Maras & Brown ; Magiati et al. ; ). One hypothesis put forth by Abrams et al. was that a global categorization of normal vs. abnormal was used to distinguish between peers with or without disabilities.

Psychometric considerations in the measurement of children’s attitudes

Studies assessing children’s attitudes towards peers with disabilities have used a variety of measures, many of which are unique to a given study. Measures have included interviews and drawings (Tamm & Prellwitz ), forced choice selection of trait descriptors (Bracegirdle ), and the semantic dif-ferential (Abrams et al. ; Townsend et al. ). Other options have included play preferences and degree of liking (Nabors & Keyes ; Harper ). Thus, one of the concerns in drawing collec-tive conclusions from past research is that compari-sons between studies are difficult because of the variety of measures.

To further complicate the situation, measuring atti-tudes in children can be challenging. One of the difficulties is that the task must be fully understood by the respondent. Children may not be aware of or understand formal terminology denoting intellectual or physical disability (Maras & Brown ; Magiati et al. ), although they may be aware of differ-ences in behaviours or more overt physical character-istics (Diamond & Kensinger ). Therefore, attitude measures need to present the presence or absence of disabilities in simple, everyday language. Asking a -year-old about their attitudes towards someone with, e.g. learning, developmental, or intel-lectual disabilities will be inappropriate if the child does not understand the meaning of these terms. Describing a hypothetical person as someone who ‘finds learning is easy’ or ‘finds learning is difficult’ may be more valid in terms of conveying the notion of disability, particularly if that description also con-sists of concrete, behavioural examples. Further, most young children do not have the reading skills, vocabulary or conceptual understanding needed to complete paper-and-pencil measures consisting of checklists, hypothetical scenarios, or numerical rating scales. In order to make comparisons between age groups, all participants need to be administered the same dependent measure in the same fashion. Con-

Journal of Intellectual Disability Research

E. A. Nowicki • A multivariable cross-sectional analysis338

© Blackwell Publishing Ltd, Journal of Intellectual Disability Research , –

sequently, paper-and-pencil measures require adaptations to accommodate oral administration. Although this is expensive, requiring a one-on-one interview format, it is necessary in order to achieve a representative sample of participants.

Given that research is inconclusive regarding the association of attitudes towards disabilities with age, gender and disability preferences, it is beneficial to include several dependent measures of the same atti-tude construct within a study. If each of the measures provides similar results then good outcome validity has been demonstrated. Validation is the process of gathering evidence to provide a sound scientific basis for the interpretation of test scores (American Edu-cational Research Association, American Psycholog-ical Association, & National Research Council on Measurement in Education ). In the case of atti-tude measures, convergent validity is achieved when one scale is correlated with others addressing the same attitude construct. During adulthood, attitudes can consist of one, two or three response classes: affective (feelings), behavioural (behaviours or behav-ioural intent) and cognitive (beliefs and knowledge). Eagly & Chaiken () suggested that using the three classes in attitudes research can foster a better understanding of attitude structure, adding that the three classes are likely to correlated in adults when an attitude consists of more than one class. Little is known about how the three classes are correlated during childhood (Ruble & Dweck ). Thus, the current study also sought to determine if the affec-tive, behavioural and cognitive classes of attitudes are associated in children.

Research questions

Over the past several decades, research into children’s attitudes towards peers with disabilities has provided inconsistent results regarding age and gender differ-ences and disability preferences. If factors associated with these attitudes are not adequately identified, then it is difficult to design effective intervention strategies that facilitate the acceptance of children with disabilities. A systematic analysis focusing on children’s attitudes towards peers with disabilities, from early to later elementary school years, is needed to provide a clearer understanding of associated fac-tors. Few studies have used repeated measures and cross-sectional approaches to examine the association

of attitudes with gender, age and disability type. Between subjects designs do not allow children to make comparisons across different disabilities. Thus, it is difficult to ascertain if evaluations for one con-dition, in isolation, may be the same or different if made in reference to an anchor representing a no disability condition. And a within subjects factorial design controls for between subject error variance and also provides information regarding potential interac-tions between variables. Examining one independent variable at a time will provide information regarding main effects but will not uncover differences between a level of one factor and a level of another. Further, using several attitude measures should provide greater outcome validity if similar results are obtained across dependent variables. Thus, the following research questions were investigated: Are there gender differences in attitudes if partici-pant and target child are matched on gender? Are there age differences in attitudes from early through to late elementary school? What are children’s disability preferences across conditions representing no disability, physical or intellectual disabilities, and a combined intellectual/physical disability? How do differences in age, gender and disability preferences interact? Can consistent results be achieved across three dependent measures of attitudes, each addressing the cognitive, affective or behavioural aspect of attitudes?

Directional hypotheses for age and gender differ-ences were not made because of inconsistencies in the research literature. It was predicted that children would have more favourable attitudes towards chil-dren without disabilities than those with disabilities. Specific predictions regarding differences between physical and intellectual disabilities were not made, but it was anticipated that attitudes would be the least favourable towards an individual with a combined intellectual/physical disability. If attitudes are based upon severity of disability (Harper ), then a com-bined physical/intellectual disability condition should elicit more negative evaluations than a single disabil-ity condition. Predictions addressing potential inter-actions between variables were not made because the literature base focusing on multivariate outcomes is not substantive. Further, it was hoped that results would be consistent across the three attitude measures.

Journal of Intellectual Disability Research

E. A. Nowicki • A multivariable cross-sectional analysis339

© Blackwell Publishing Ltd, Journal of Intellectual Disability Research , –

Participants and method

Participants

One hundred children representing a wide socio-economic spectrum and attending schools in a medium-sized Canadian city participated. The sample consisted of girls and boys. There were children from each of Junior Kindergar-ten (JK, mean age of . ± . months), Grade (mean age of . ± . months), Grade (mean age of . ± . months), and Grade (mean age of . ± . months). All partici-pants attended schools that promoted a full inclusion model. Participants in Grades and had classmates with Individualized Education Plans (IEPs) for intellectual disabilities. Formal identification for IEPs did not occur within the school system until Grade , but teachers of JK and Grade classes indicated that they had likely candidates for future identification. How-ever, none of the participants had classmates with moderate to severe intellectual disabilities (i.e. mental retardation or multiple learning dis-abilities), hearing loss, visual impairments or orthopaedic disabilities.

Further, parents of participants completed a paper-and-pencil survey to provide information concerning the amount of personal, regular, out of school contact their child had with individuals with an intellectual disability (i.e. learning disability, mental retardation or slower learner), a physical (orthopaedic) disability, and combined intellectual/physical disabilities. Definitions of these terms were provided on the survey form and corresponded to the terminology used by the school board. Ninety of the surveys were returned. Of these, % of the participants had no regular contact with an individual with an intellectual disability, % had less than one hour a week of regular contact, and % had more than one hour of contact per week. Sixty-seven per cent of the participants had no reg-ular contact with an individual with a physical dis-ability, % had less than one hour per week, and % had more than one hour per week of regular contact. In regards to regular weekly contact with an individual who had combined intellectual/physi-cal disabilities, % had no contact, % had less than one hour per week, and % had more than one hour.

Target conditions

Eight fictional target children (four girls and four boys) were depicted in coloured drawings. Male tar-get children were drawn from the same template such that facial features, body size, posture, hair and cloth-ing style (i.e. shorts, t-shirt and running shoes) were consistent. Colour of clothing and hair differed for each target child. Female target children were drawn from the same template used for the male target children but they were shown with a pony tail and more feminine facial features. Clothing and hair colour also varied across conditions. Given that of the participants were of White racial background, the target children were depicted with light skin colour. Further, two target children of each gender were shown seated in regular school-type chairs and two children of each gender were shown seated in wheelchairs.

There were four target conditions within gender. Gender of the participant and target child were matched because children have a preference to play in same gender groups and tend to avoid contact with the opposite gender (Sippola et al. ). It was anticipated that matching would ensure that any potential biases directed at a target child would be based on the disability rather than on negative eval-uations of the opposite gender.

Although physical difficulties can be described in a drawing, intellectual disabilities require verbal elab-oration. Many children with intellectual disabilities do not differ in general physical appearance when compared with their same-aged peers. Consequently, it was considered necessary to provide a short, simple verbal description. Descriptions for the target chil-dren were based on Nabors & Keyes (). For the no disability condition, the target child was described as: ‘This girl/boy learns new things easily. S/he knows how to do the things that someone of your age can do such as: ————.’ The blank was filled in with three grade-specific exemplars of learning objectives as described in the provincial elementary school curric-ulum. Exemplars for JK were: can (or is learning to) count to , can understand a story that was read to them and tell it to someone else, knows some letters of the alphabet. Grade exemplars were: is learning how to read, is learning how to write some words, is learning how to add. For Grade participants, exem-plars were: can read a chapter book, can write a story,

Journal of Intellectual Disability Research

E. A. Nowicki • A multivariable cross-sectional analysis340

© Blackwell Publishing Ltd, Journal of Intellectual Disability Research , –

can multiply and divide. Exemplars for Grade were: can read a novel, can write a story using paragraphs, can multiply decimal numbers.

The second target child represented the physical disability condition. This child was seated in a wheel-chair and was described as: ‘This boy/girl gets around in a wheelchair. S/he learns new things easily. S/he knows how to do the things that someone of your age can do such as: ————.’ The blank was filled in with the grade-specific exemplars described above. The third target child represented the intellectual disability condition and was shown seated in a regu-lar chair. The verbal description for this child was: ‘This girl/boy finds learning new things difficult. S/he cannot do some of the things that someone of your age can do such as: ————’. Again, the blank was filled in with the same grade-specific exemplars as above. The fourth target child, representing the intel-lectual/physical disability condition, was shown seated in a wheelchair. The verbal description began with: ‘This boy/girl gets around in a wheelchair’. It was followed with the same description as for the third target child.

It should be emphasized that the terms ‘intellectual disability’ and ‘physical disability’ were not used in these descriptions. Young children, in particular, may not be familiar with formal terminology used to denote disability, thus, descriptions used simple, everyday language. Participants were provided with grade-appropriate exemplars and the phrases ‘finds learning new things easy’, ‘finds learning new things difficult’, and ‘gets around in a wheelchair’ to denote the presence or absence of intellectual or physical disability.

Measures

Participants completed three attitude scales corre-sponding to the cognitive, behavioural, and affective components of attitudes described by Eagly & Chaiken ().

The Multi-Response Attitude Scale

The cognitive component of attitudes was assessed with the Multi-Response Attitude Scale (Doyle et al. ), but the original target children depicting indi-viduals of different ethnic groups were replaced with pictures of the target children described above. It consists of positive and negative adjectives. The

positive adjectives are clean, wonderful, healthy, good, nice, happy, friendly, kind, helpful and smart. The negative adjectives are bad, unfriendly, mean, dirty, cruel, stupid, selfish, sick, naughty and sad. For each adjective, the experimenter read aloud a behav-ioural example, showed the participant four identical cards labelled with the appropriate adjective, and read the descriptor to the participant. The participant was requested to place the card(s) in one or more of four boxes, each one labelled with a picture of a target child depicting one of the four target conditions. The scale provides four separate scores, one for each tar-get condition. Scores are calculated by subtracting the number of negative descriptors from the number of positive descriptors, and can range from − to . A negative score indicates a negative attitude, and a positive score represents a positive attitude. Internal consistency coefficients for the positive and negative items were calculated for each grade level. For par-ticipants in JK, alphas were . for the positive items and . for the negative ones; for Grade , alphas were . and . respectively; for Grade , they were . and .; and for Grade , both were .. The scale was not significantly correlated with the Comprehensive Social Desirability Scale for Children (Walsh et al. ).

Before the Multi-Response Attitude Scale items were administered, participants’ understanding of the task was assessed with the following practice items: () Which child is wearing a hat? () Which child is wearing a red T-shirt? () Which child is wearing shoes? For each practice item, the participant was given four cards, each with a drawing of the relevant item, and was asked to place the card in the appropriate box. During administration of the Multi-Response Attitude Scale items, four catch trial items were included: () Which child finds learning easy? () Which child finds learning difficult? () Which child uses a wheelchair? and () Which child does not use a wheelchair? The practice and catch trial items were not included in sub-scale scores, but all partic-ipants responded appropriately to these items.

Behavioural Intent Scale

The behavioural component of attitudes was assessed with the Behavioural Intent Scale (Roberts & Lindsell ). It consists of questions describing increas-ingly more intimate aspects of childhood friendships

Journal of Intellectual Disability Research

E. A. Nowicki • A multivariable cross-sectional analysis341

© Blackwell Publishing Ltd, Journal of Intellectual Disability Research , –

ranging from, ‘I would go up to (target child) and say hello’ to ‘I would share a secret with (target child)’. All items are keyed in a positive direction. Originally a paper-and-pencil measure, it was adapted for the current study to accommodate children at the pre-reading stage. Four response options were printed on cards. A card with upper case letters followed by an exclamation point (i.e. YES!) represented a response of definitely yes; a card with ‘yes’ written in lower case letters represented a response of maybe yes; a card with a lower case ‘no’ represented a maybe not response; and an upper case ‘no’ followed by an exclamation point (i.e. NO!) represented a definitely not response. Each question was read aloud to the participant who was then asked to select a card (i.e. YES!, yes, no or NO!) to indicate their behavioural intent towards the target child, and to place the card in the box assigned to that target child, as described above. For participants who could not read, the four possible responses were read aloud following each item. The score for each target condition was deter-mined by assigning weights to the responses (i.e. YES! = points, yes = points, no = points, NO! = point) and summing. Scores can range from to . The internal consistency coefficient for the scale was . for JK, . for Grade , . for Grade , and . for Grade . The scale was not significantly correlated with social desirability.

To determine if the participants understood how to select the appropriate responses, they were given the following practice questions prior to the admin-istration of the Behavioural Intent Scale: () Would you wear your pyjamas at night? At breakfast? At school? () Would you go swimming on a very hot day? A warm day? A very snowy day? Participants were asked to select one of the four cards, described above, to indicate their responses. Responses indi-cated that all participants understood the task. Responses that were not expected were queried. For example, one child mentioned that he would wear his pyjamas to school, but only on the school’s annual pyjama day. Another child mentioned that she would go swimming on a snowy day but only in an indoor pool.

Pictographic Scale

The affective component of attitudes was assessed with the Pictographic Scale designed for this study.

Based on a one-item scale by Graffi & Minnes (), it consists of five questions with responses selected from a set of five simple line drawings of faces ranging from happy to sad. For each target condition the participant was instructed to select a face to indicate their responses to each of the following items: () How do you feel about this boy/girl? () How do you feel about helping this boy/girl? () How do you feel about playing with this boy/girl? () How do you feel about this boy/girl asking you to help them? () How do you feel about this boy/girl asking you to play with them? A score for each target condition was deter-mined by assigning weights to each face selected (ranging from for the happiest face to for the saddest) and summing across items. Possible mini-mum and maximum scores for each sub-scale are– respectively. In the current study, the internal consistency coefficient of the overall scale for JK was .; for Grade it was .; it was . for Grade , and for Grade , it was .. The scale did not correlate with social desirability.

Prior to the administration of the Pictographic Scale, participants were given several practice items. They were asked to point to the face that showed how they felt about pizza, oatmeal, and broccoli, going to a birthday party, watching TV, and spiders. Without exception, participants’ responses reflected under-standing of the task. Unexpected responses were que-ried and reflected atypical preferences rather than lack of understanding.

Procedures

The initial visit took place in the participants’ class-rooms. A graduate student described the study and children had an opportunity to ask questions. Letters of information and consent were provided to inter-ested students who were asked to return the signed parental consent form prior to participation. A paren-tal survey, described above, addressing each partici-pant’s amount of regular out of school contact time with individuals with disabilities, was attached to the forms.

Measures were individually administered to each participant over three sessions, each lasting from to min. Data were collected by a graduate student who was an experienced elementary school teacher and school psychometrist. Sessions were conducted in a quiet room at the participant’s own school, and

Journal of Intellectual Disability Research

E. A. Nowicki • A multivariable cross-sectional analysis342

© Blackwell Publishing Ltd, Journal of Intellectual Disability Research , –

were separated by – days depending on scheduling availability. The initial session began with the presen-tation of the four target children: no disability, phys-ical disability, intellectual disability, intellectual and physical disability. Presentation of target condition was partially counterbalanced such that target con-ditions were presented to participants in the afore-mentioned order or the reverse order. Order was consistent for each participant across sessions.

Before the attitude measures were administered, and after the target conditions were described, par-ticipants’ understanding of the conditions was assessed. Participants were asked to point to () the target children who found learning new things easy and to describe some things the target children could do; () the target children who found learning new things difficult and to describe some things the target children could not do; () the target children who found walking easy and to describe some things the target children could do; and () the target children who found walking difficult and to describe some things the target children could not do. If a partici-pant did not respond accurately, the experimenter provided the necessary information and asked the participant to point to each target condition, as above. This procedure was repeated at the beginning of each session, and all participants gave accurate responses prior to the administration of the depen-dent measures.

Results

Differences in gender, age and disability preferences across attitude measures

Descriptive data

Means and standard deviations were calculated for each of the attitude measures across the four grade (age) levels, gender and target conditions (see Table ).

Analysis of variance

Scores by gender, grade and target condition were compared for each dependent measure (i.e. Multi-Response Attitude Scale, Behavioural Intent Scale, and Pictographic Scale) with a split-plot analysis of variance (). The between subject factors were gender and grade. The within subjects factor con-

sisted of the four target conditions. For each of these univariate tests, within subjects degrees of freedom were adjusted with the Greenhouse-Geisser correc-tion factor. The rationale for this adjustment is that F-ratios of within subjects effects tend to have a pos-itive bias, thus a correction reduces the probability of a Type I error (Kirk ).

Multi-Response Attitude Scale (MRA). A significant main effect was found for gender, F(,) = ., P < ., with girls having significantly higher scores than boys (M = ., SD = ., and M = ., SD = . respectively). However, a significant F-ratio shows that the null hypothesis can be rejected but does not provide information about the strength of the relationship between the dependent and inde-pendent variables. Kirk () wrote that statistical and practical significance are not the same. He defined practical significance as ‘whether an observed effect is large enough to be useful in the real world’ (p. ). According to Cohen (), an effect size of around . is small, around . is large, and a large effect size is around .. Thus, for the main effect of gender, the effect size (f) of . is large indicating that gender differences are likely to be strong and consistent when this measure is used.

There were no significant interactions for gender by target condition, gender by grade, or gender by grade by target condition. However, significant main effects were obtained for target condition with a very large effect size, F(., .) = ., P < ., f = ., and grade also with a large effect size, F(, ) = ., P < ., f = .. The target condition by grade interaction also reached significance F(., .) = ., P < ., f = .. Post hoc Tukey HSD tests for the target condition by grade interaction were conducted on cell means (differences in means [Md] significant at the . level are reported in brackets). Significant differ-ences were found at each grade level for the no disability and the intellectual disability conditions (for JK, Md = .; Grade , Md = .; Grade , Md = .; and Grade , Md = .) and the no disabil-ity and the intellectual/physical disability conditions (JK, Md = .; Grade , Md = .; Grade , Md = .; and Grade , Md = .). All children pre-ferred the child without any disabilities. Similar dif-ferences were also found between the physical disability condition and both the intellectual disabil-

Journal of Intellectual Disability Research

E. A. Nowicki • A multivariable cross-sectional analysis343

© Blackwell Publishing Ltd, Journal of Intellectual Disability Research , –

ity and intellectual/physical disability conditions for all participants but for those in Grade (respectively, JK, Md = . and Md = .; Grade , Md = . and Md = .; Grade , Md = . and Md = .). For the Grade children, the physical disability condition was more positive compared with the intellectual disability condition (Md = .). Comparing across grades but within target condi-tions, JK children had lower scores than children attending Grades (Md = .) and (Md = .) for the intellectual disability condition. For the intel-lectual/physical disability condition, JK children’s attitudes were more negative than those of children in Grades and (Md = . and Md = . respectively).

Behavioural Intent Scale. Gender differences were not significant nor were any of the interactions involv-ing gender. The main effect of grade was not signifi-cant but significant differences were found for target condition, F(., .) = ., P < ., f = ., and the target condition by grade interaction, F(., .) = ., P < ., f = .. Both were accom-panied with substantial effect sizes. Tukey post hoc tests conducted on this interaction (differences in means [Md] significant at the . level are reported in brackets) showed that JK and Grade children were more biased against the intellectual and intel-lectual/physical disability conditions compared with the no disability condition (JK, Md = . and Md = .; Grade , Md = . and Md = .

Table 1 Means (M) and standard deviations (SD) for attitude measures by target condition and grade

None Physical IntellectualIntellectual/physical

M SD M SD M SD M SD

Multi-Response AttitudeJK Girls 6.9 1.8 4.8 3.1 −3.4 3.1 −3.7 3.2

Boys 7.6 1.6 4.6 4.6 −2.6 4.9 −4.2 4.5Grade 1 Girls 8.2 2.1 6.9 2.0 1.6 6.1 0.8 5.7

Boys 6.8 2.9 5.0 3.7 −1.2 6.2 −2.6 5.5Grade 3 Girls 6.6 4.7 5.1 3.2 2.9 5.5 2.3 5.3

Boys 4.1 5.7 4.5 4.1 −1.3 6.1 0.5 6.1Grade 5 Girls 5.4 4.6 6.7 2.3 0.8 4.7 2.9 3.7

Boys 8.0 2.6 6.7 1.6 −1.0 6.7 0.4 5.9

Behavioural IntentJK Girls 34.0 7.7 32.3 7.4 26.8 10.1 28.9 8.7

Boys 35.2 4.3 33.8 4.9 31.4 5.3 30.4 6.4Grade 1 Girls 36.3 4.0 35.3 3.9 29.9 8.9 29.8 8.9

Boys 37.3 2.9 35.0 4.0 25.6 11.5 24.3 10.4Grade 3 Girls 34.6 5.2 35.4 3.4 33.2 6.4 35.1 4.8

Boys 33.8 6.0 35.2 4.5 32.5 6.9 32.0 7.3Grade 5 Girls 33.4 5.2 34.1 3.8 32.6 5.2 33.7 3.6

Boys 32.8 4.7 32.3 4.7 30.1 6.5 31.8 5.0

PictographicJK Girls 21.8 3.7 19.4 3.7 12.8 5.5 12.2 5.5

Boys 22.4 3.4 20.1 3.6 17.4 6.6 16.8 6.7Grade 1 Girls 22.4 2.5 21.2 2.7 18.8 4.6 18.1 5.4

Boys 22.8 2.0 21.1 4.6 15.3 7.6 15.8 6.5Grade 3 Girls 22.7 2.1 22.4 1.8 22.3 1.7 21.2 2.6

Boys 21.3 3.7 20.5 3.0 19.2 5.4 17.5 5.4Grade 5 Girls 21.5 2.9 21.6 2.7 21.8 1.9 21.6 2.7

Boys 22.6 2.3 20.8 2.2 20.1 5.1 19.8 1.9

JK, Junior Kindergarten.

Journal of Intellectual Disability Research

E. A. Nowicki • A multivariable cross-sectional analysis344

© Blackwell Publishing Ltd, Journal of Intellectual Disability Research , –

respectively). Grade children were also more biased against the intellectual and intellectual disability conditions than the physical disability condition (Md = . and Md = .). Further, children in Grade had lower scores than children in Grade for the intellectual disability condition (Md = .). They also had lower scores for intellectual/physical disability condition than children in Grade (Md = .) and Grade (Md = .).

Pictographic Scale. Although the main effect of gen-der was not significant, or were the gender by target and gender by grade by target interactions, the gen-der by grade interaction did produce significant results, F(, ) = ., P < ., f = .. Post hoc analysis (differences in means [Md] significant at the . level are reported in brackets) indicated that girls in JK had more negative affect towards all target children than did girls in Grade (Md = .), Grade (Md = .), and Grade (Md = .). Also, Grade girls showed more positive affect than Grade boys (Md = .).

The main effect of target was significant as was the target by grade interaction. Tukey post hoc analysis on the interaction indicated that children in JK and Grade showed a preference for the target child without a disability compared with the intellec-tual disability condition (Md = . and Md = . respectively) and the intellectual/physical disability condition (Md = . and Md = .). They also pre-ferred the child with the physical disability to the intellectual (Md = . and Md = .) and intellectual/physical disability (Md = . and Md = .) condi-tions. JK and Grade participants had more negative attitudes to the target child with the intellectual dis-ability than did participants in Grade (Md = . and Md = .) and Grade (Md = . and Md = .), and towards the target child with the intellectual/physical disability than did participants in Grade (Md = .) and Grade (Md = .).

Correlations between attitude measures

A set of correlations were performed to assess the convergent validity of the three attitude measures. This was accomplished by examining the correlations between measures but within each target condition. There were significant inter-correlations between all three measures for the no disability condition [Multi-

Response Attitude and Behavioural Intent, r() = ., P < .; Multi-Response Attitude and Pictographic, r() = ., P < .; Behavioural Intent and Pictographic, r() = ., P < .]; the intellectual disability condition [Multi-Response Attitude and Behavioural Intent, r() = ., P < .; Multi-Response Attitude and Picto-graphic, r() = ., P < .; Behavioural Intent and Pictographic, r() = ., P < .]; and the intellectual/physical disability condition [Multi-Response Attitude and Behavioural Intent, r() = ., P < .; Multi-Response Attitude and Pictographic, r() = ., P < .; Behavioural Intent and Pictographic, r() = ., P < .]. For the physical disability condition, significant cor-relations were obtained for the Multi-Response Atti-tude and the Behavioural Intent scales, r() = ., P < .; and the Behavioural Intent and Picto-graphic scales, r() = ., P < ..

Discussion

Previous research shows that children may have neg-ative attitudes towards individuals with disabilities but it has been inconclusive regarding the association of these attitudes with gender, age and disability pref-erences. It is also difficult to compare results across studies because of the variety of measures and little information about their convergent validity. This study attempted to clarify these issues by examining age, gender and disability preferences in terms of main effects and interactions across three theoreti-cally linked attitude measures. Assessing several dependent and independent variables in one study reduces error variance and enhances the conceptual understanding of relationships between variables that cannot be ascertained when variables are isolated across studies. It was hoped that gathering these vari-ables into one study would provide a sound basis for resolving past inconsistencies.

Descriptions of the target children in the current study were based on simple, age-appropriate lan-guage (i.e. finds learning is easy or difficult, gets around in a wheelchair). These descriptions were accompanied with concrete, behavioural exemplars taken from grade-appropriate curricula in order to provide easy to understand information about what the target children could or could not accomplish.

Journal of Intellectual Disability Research

E. A. Nowicki • A multivariable cross-sectional analysis345

© Blackwell Publishing Ltd, Journal of Intellectual Disability Research , –

Formal terminology, such as learning, developmen-tal, intellectual or physical disability was not used because it was anticipated that some children would not understand the meaning of these terms.

Study objectives were presented through a set of five related questions. The first question asked if gen-der differences in attitudes exist when gender of par-ticipant and target child are matched. The answer appears to depend on the measure. A clear gender difference was found for the Multi-Response Attitude Scale with girls selecting more positive patterns of descriptors than boys. It was consistent across all age groups and all target conditions suggesting gender-specific response biases rather than gender differ-ences in attitudes towards peers with disabilities. This is a critical distinction and one that needs to be kept in mind when interpreting results of other studies. Indeed, gender differences may be influenced by measures that elicit response biases favoured by one gender. Including gender and several target condi-tions and age groups within a study allows patterns to emerge that might otherwise remain hidden.

Although girls and boys did not differ in the affec-tive response class of attitudes, the Pictographic Scale revealed that the youngest girls had the most negative affect towards the target children, in general, than the other girls. However, when cell means are compared across gender, disability and grade, similar trends are evident for boys and girls, but the JK girls do show a marked negativity towards the two target children with intellectual disabilities. This result requires fur-ther study to determine why such differences exist.It also remains to be seen if younger girls’ negative affect extends to boys with intellectual disabilities. This question could be answered by purposefully mismatching gender to determine if girls or boys within specific age groups are more or less biased towards children of the opposite sex with disabilities.

Thus, matching gender of participant and target child may have controlled for same-sex positive biases and opposite-sex negative biases that are common during childhood, thereby ensuring that responses were directed at the disability conditions. It needs to be noted, however, that girls and boys may differ in how they respond to attitude measures that tap into certain classes of responses and that these differences may be more indicative of gender-based response biases rather than gender differences in disability preferences. Also, age differences in attitudes within

a gender may be greater than differences in attitudes between genders.

In response to the second question, age differences in attitudes from early through to late elementary school were evident. Younger children’s attitudes towards children with intellectual disabilities were more negative than those of older children. However, grade and disability conditions interacted, providing a clear answer to the third query: What are children’s disability preferences across conditions? All children were most biased against the target children repre-senting the intellectual and intellectual/physical dis-ability conditions compared with the no disability and physical disability conditions. Nabors & Keyes () also reported preferences for physical vs. intel-lectual disabilities. However, participants did not view the target child with the combined intellectual/physical disability any differently to the target child with the intellectual disability. Data were collected in a school setting where learning is encouraged and rewarded. All participants, regardless of age, were probably aware of the expectations to learn, and that failure to do so is undesirable. From a contextual perspective, the intellectual disability may have been of greater salience than the physical disability. Had the study taken place in another context, such as a playground or sports venue, attitudes may have been different. That is, physical disabilities may be per-ceived as more limiting in such situations and might be more negatively evaluated than intellectual dis-abilities. Previous research has suggested that atti-tudes are dependent upon severity of disability or degree of functional loss, but these assertions are based on experimental conditions that have focused on physical or sensory disabilities rather than on intellectual disabilities (Diamond et al. ; Harper ). Perhaps, though, context, severity and type of disability are important. It is not known if mild intel-lectual disabilities are more or less negatively evalu-ated than severe physical disabilities in general, or in specific contexts. Thus, research is needed to deter-mine the relationship of attitudes towards varying degrees of severity of physical and intellectual disabilities.

The fourth research question was: How do differ-ences in age, gender and disability preferences inter-act? As previously mentioned, there was a main effect of gender for the Multi-Response Attitude Scale indi-cating that girls have a tendency to use a greater

Journal of Intellectual Disability Research

E. A. Nowicki • A multivariable cross-sectional analysis346

© Blackwell Publishing Ltd, Journal of Intellectual Disability Research , –

proportion of positive descriptors than boys, regard-less of the presence or absence of disability. Further, -year-old girls displayed more negative affect towards the intellectual disability conditions than the older girls. Otherwise, gender differences in attitudes towards disabilities did not emerge as a significant result. Age had a negative association with attitudes, but only towards the two target children with intel-lectual disabilities. Younger children were more biased in these attitudes, regardless of measure, than were the older children. It has been suggested that younger children’s understanding of people is built around a global categorization process (Abrams et al. ; Heyman et al. ). Their perceptions are formed by using the anchors of an evaluative dimen-sion, such as good vs. bad, while ignoring the middle ground. Younger children are more inclined to latch onto one outstanding feature when making an evalu-ative judgement whereas older children may be less likely to over-generalize. This observation is under-scored by younger children’s evaluation of the target child with both the physical and intellectual disability. If failure to learn is a highly salient characteristic, then peers who do not learn new things easily may be viewed as globally bad by younger children, regardless of the presence or absence of other disabil-ities or characteristics. Older children may have been using multiple referents in making judgements (Magiati et al. ), thereby tempering negative evaluations.

The last research question focused on the consis-tency of outcomes between dependent measures. Results were generally consistent across the three dependent measures, which is of particular signifi-cance given the inconsistent outcomes described in the literature. Thus, including more than one depen-dent variable addressing the same construct enhances both internal and external validity. Good attitude measures need to be both reliable and valid (Eagly & Chaiken ). The measures used in this study had very good internal consistency. They also appear to have face and content validity with each tapping into one of the three response classes of attitudes. That is, the Behavioural Intent Scale consists of items that reflect the intent to behave is a certain way with someone with or without a disability, the Picto-graphic Scale focuses on feelings towards individuals with or without disabilities, and the Multi-Response Attitude Scale requires children to indicate if they

consider a target child to be, e.g. kind, mean, happy or sad, thereby reflecting cognitions or beliefs. Fur-ther, with one exception (i.e. no significant correla-tion in the physical disability condition between the Multi-Response Attitude and Pictographic scales), the three measures were significantly inter-correlated with one another across target conditions thereby demonstrating convergent validity and some degree of construct validity.

Good attitude measures also need to be age-appropriate. Given that young children cannot read or write, large scale collection of data using paper-and-pencil measures is not feasible. Although measuring attitudes of young children is resource expensive, this study demonstrated that children as young as years old can be reliable research partici-pants as long as task demands are developmentally appropriate, but at the same time are not overly sim-plistic to discourage older participants. Had the younger participants not understood what they were expected to do, responses between and within mea-sures would likely not have been consistent. Language describing attitude objects needs to be age-appropriate, but meaning can be lost with substitu-tions. It is not known if children conceptualized learn-ing difficulties as mild, moderate or severe. However, participants were provided with grade-appropriate exemplars of tasks the target children could or could not accomplish, and hopefully, this information helped children to understand general differences in intellectual ability across target conditions.

Implications and future directions

The most important implication of this study is that children’s attitudes towards their peers are complex. Past inconsistencies in study outcomes are common in this domain and may be averted with future research that focuses on relationships between mul-tiple variables rather than focusing on few at a time. Using multiple target conditions, a wider range of ages, gender matching, and several theoretically linked measures provides insight into what variables may or may not be important in forming attitudes. In this study, nature of the disability and its interac-tion with age was a consistent finding with younger children showing more bias against intellectual dis-abilities than older children. An important finding

Journal of Intellectual Disability Research

E. A. Nowicki • A multivariable cross-sectional analysis347

© Blackwell Publishing Ltd, Journal of Intellectual Disability Research , –

concerning gender response bias was also found that transcended age and the presence or absence of dis-ability. Thus, attitude measures may produce unan-ticipated gender differences that are best understood when studies include multiple variables and several levels within variables. Further, results supported the convergent validity of the three attitude measures. Each measure tapped into one of the three classes of evaluative responses. Undoubtedly there are other variables that also contribute to a big picture under-standing of children’s attitudes towards peers with or without disabilities such as amount of contact (Mara & Brown ), exposure to the media (Diamond et al. ), vigor of inclusion policies (Townsend et al. ), and whether a child has a disability of his or her own. As more studies focus on the intercon-nections between multiple variables, a reliable and valid configuration of attitudes and associated factors will develop.

Further, this study has shown that children’s biases towards intellectual disabilities may be quite pro-nounced at the beginning of elementary school. If attitudes are good predictors of behaviour (Kraus ), then interventions designed to facilitate inclu-sion need to be implemented from the first day of school. Children who are coping with academic dis-abilities do not need the additional burden of endur-ing negative social evaluations by their classmates. Negative biases need to be addressed and changed early on so that children can view their peers with intellectual disabilities as participating members of the classroom community.

Several directions for further investigation have been mentioned, such as mismatching gender of tar-get child and participant, the role of context, and exploring the relationship of attitudes towards vary-ing degrees of severity of physical and intellectual disabilities. Further, attitude measures provide infor-mation about valence and magnitude, but do not necessarily provide information about amount or accuracy of knowledge. Children’s understanding about the aetiology of disabilities is beyond the scope of the attitude measures described in this study and should be explored. Another extension is to deter-mine if age-related differences in attitudes towards peers with disabilities are associated with quantity and quality of social behaviours. Also, the majority of children in this study had little contact with peers with severe disabilities and it is not known if these

results would apply to children with more extensive contact.

Understanding the underlying dimensions of neg-ative attitudes is an important step towards the devel-opment of intervention strategies promoting positive attitude change (Antonak & Linveh ). Much of the confusion in the literature regarding the roles of age, gender and disability type in children’s attitudes has been the result of the tendency to examine these issues in isolation from one another. The variety of dependent measures used between studies has also been problematic because it can be difficult to ascer-tain if outcomes are valid representations of the atti-tude construct. This study addressed these concerns by using a wide age range, gender matching, allow-ing for evaluative comparisons across physical and intellectual disability conditions, and through the administration of three theoretically related attitude measures. Given the multifaceted nature of attitudes, multivariable studies provide a viable means to better understand the nature of children’s attitudes.

References

Abrams D., Jackson D. & St. Claire L. () Social iden-tity and the handicapping function of stereotypes: chil-dren’s understanding of mental and physical handicap. Human Relations , –.

American Educational Research Association, American Psy-chological Association, & National Council on Measure-ment in Education () Standards for Educational and Psychological Testing. American Educational Research Association, Washington, DC.

Antonak R. F. & Linveh H. () Measurement of atti-tudes toward persons with disabilities. Disability and Rehabilitation , –.

Bracegirdle H. () Children’s stereotypes of visually impaired targets: an empirical study. British Journal of Occupational Therapy , –.

Cohen J. () Statistical Power Analysis for the Behavioral Sciences, nd edn. Erlbaum, Hillsdale, NJ.

Cohen T. & Lopatto D. () Preschool children’s com-prehension of the word ‘handicapped’. Perceptual and Motor Skills , –.

Cohen R., Nabors L. & Pierce K. () Preschoolers’ eval-uations of physical disabilities: a consideration of attitudes and behavior. Journal of Pediatric Psychology , –.

Colwell C. () Effects of information on elementary band students’ attitudes toward individuals with special needs. Journal of Music Therapy , –.

Diamond K., Hestenes L., Carpenter E. & Innes F. () Relationships between enrolment in inclusive classes and

Journal of Intellectual Disability Research

E. A. Nowicki • A multivariable cross-sectional analysis348

© Blackwell Publishing Ltd, Journal of Intellectual Disability Research , –

preschoolers children’s ideas about people with disabili-ties. Topics in Early Childhood Special Education , –.

Diamond K. & Kensinger K. R. () Vignettes from Sesame Street: preschoolers ideas about children with Down syndrome. Early Education and Development , –.

Doyle A., Beaudet J. & Aboud F. () Developmental patterns in the flexibility of children’s ethnic attitudes. Journal of Cross-Cultural Psychology , –.

Eagly A. H. & Chaiken S. () The Psychology of Attitudes. Harcourt Brace Jovanovich, Inc., Fort Worth, TX.

Fitch F. () Inclusion, exclusion, and ideology: special education students changing sense of self. Urban Review , –.

Gash H. & Coffey D. () Influences on attitudes toward children with mental handicap. European Journal of Spe-cial Needs , –.

Graffi S. & Minnes P. M. () Attitudes of primary school children toward the physical appearance and labels associated with Down syndrome. American Journal on Mental Retardation , –.

Harper D. () Children’s attitudes toward physical dis-ability in Nepal. Journal of Cross Cultural Psychology , –.

Heyman G. D., Gee C. & Giles J. W. () Preschool children’s reasoning about ability. Child Development , –.

Kirk R. E. () Experimental Design: Procedures for the Behavioral Sciences. Brooks Cole Publishing Co., Toronto, ON.

Kratzer L. & Nelson-Le Gall S. () Understanding com-petencies and limitations of wheelchair-bound helpers: developmental changes during early childhood. Journal of Applied Developmental Psychology , –.

Kraus S. J. () Attitudes and the prediction of behavior: a meta-analysis of the empirical literature. Journal of Personality and Social Psychology , –.

Lockhart K. L., Chang B. & Story T. () Young chil-dren’s beliefs about the stability of traits: protective opti-mism. Child Development , –.

Magiati I., Dockrell J. E. & Logotheti A. () Young children’s understanding of disabilities: the influence of development, context, and cognition. Applied Develop-mental Psychology , –.

Maras P. & Brown R. () Effects of contact on children’s attitudes toward disability: a longitudinal study. Journal of Applied Social Psychology , –.

Nabors L. & Keyes L. () Preschoolers’ reasons for accepting peers with and without disabilities. Journal of Developmental and Physical Disabilities , –.

Nabuzoka D. & Ronning J. () Social acceptance of children with intellectual disabilities in an integrated school setting in Zambia: a pilot study. International Jour-nal of Disability, Development and Education , –.

Norwich B. & Kelly N. () Pupils’ views on inclusion: moderate learning difficulties and bullying in mainstream and special schools. British Educational Research , –.

Nowicki E. A. () A meta-analysis of the social compe-tence of children with learning disabilities in inclusive classrooms: comparisons with average to high and low achieving classmates. Learning Disability Quarterly , –.

Nowicki E. A. & Sandieson R. () A meta-analysis of children’s attitudes toward individuals with intellectual and physical disabilities. International Journal of Disability, Development and Education , –.

Obrusníková I., Válková H. & Block M. E. () Impact of inclusion in general physical education on students without disabilities. Adapted Physical Activity Quarterly , –.

Okagaki L., Diamond K., Kontos S. & Hestenes L. () Correlates of young children’s interactions of classmates with disabilities. Early Childhood Research Quarterly , –.

Ring E. & Travers J. () Barriers to inclusion: a case study of a pupil with severe learning difficulties in Ireland. European Journal of Special Needs Education , –.

Roberts C. M. & Lindsell J. S. () Children’s attitudes and behavioral intentions toward peers with disabilities. International Journal of Disability, Development, and Edu-cation , –.

Ruble D. N. & Dweck C. S. () Self-conceptions, person conceptions, and their development. In N. Eisenberg (Ed.) Social Development: Review of Personality & Social Psychology. London, UK: Sage Publications.

Sippola L. K., Bukowski W. M. & Noll R. B. () Dimen-sions of liking and disliking underlying same-sex prefer-ences in childhood. Merrill-Palmer Quarterly , –.

Tamm M. & Prellwitz M. () ‘If I had a friend in a wheelchair’: children’s thoughts on disabilities. Child: Care, Health and Development , –.

Townsend M. A. R., Wilton K. M. & Vakilirad T. () Children’s attitudes toward peers with intellectual disabil-ity. Journal of Intellectual Disability Research , –.

Tripp A., French R. & Sherrill C. () Contact theory and attitudes of children in physical education programs toward peers with disabilities. Applied Physical Activity Quarterly , –.

Walsh J. A., Tomlinson-Keasey C. & Kleiger D. M. () Acquisition of the social desirability response. Genetic Psy-chology Monographs , –.

Weirserbs B. & Gottlieb J. () Perceived risk as factor influencing attitudes toward physically disabled children. Journal of Developmental and Physical Disabilities , –.

Woodard R. () The effects of gender and type of dis-ability on attitudes of children toward peers with physical disabilities. Therapeutic Recreation Journal , –.

Accepted August