Making decisions and judgments on disability: the disability representation of parents, teachers,
and special needs educators
Department of Human and Educational Sciences, University of Perugia, Italy
CIRID, Interdisciplinary Centre for Integrated Research on Disability, University of Rome ‘La Sapienza’
Rome – Perugia, Italy
CIRID, Interdisciplinary Centre for Integrated Research on Disability, University of Rome ‘La Sapienza’
Correspondence: Stefano Federici, Department of Human and Educational Sciences, University of Perugia, Piazza G. Ermini, 1 - 06123
Perugia, Italy. E-mail: email@example.com
Representations of disability, shaping the opinions, attitudes, and behaviour of parents, teachers and special needs educators were
investigated. Two experiments were carried out: in the first one the disability representations of 90 participants were investigated through
both qualitative analyses – consisting of textual interpretation of the data gathered by 17 focus groups, assisted by the Atlas.ti software –
and by quantitative analyses, consisting of Chi-square statistical testing. In the second experiment, the disability representations were
investigated by administering the Implicit Association Test (IAT) to 36 participants. The results obtained demonstrated that, on an implicit
level, disability is strongly and stereotypically associated with a negative and unpleasant dimension of existence, compared to the richness
of scripts showed in explicit explanatory disability opinions, attitudes and behaviour of the participants in different contexts.
EYWORDS: models of disability, parental attitudes, ICF, biopsychosocial model, Computer Assisted Qualitative Data Analysis Software,
Atlas.ti, special needs educators, IAT – Implicit Association Test.
The disability models are categorical representations
in which the social relations are understood, built, and given;
frames in which everyone finds his own identity, and also, as
scripts, in which the identities of other individuals are
represented in that complex system of attribution that defines us
and lets us to make decisions and judgments.
Therefore, following our perspective, disability
models should not only be considered as useful access tools to
deal theoretically and practically with a too complex reality, but
as the possibility itself to access that reality called disability.
The idea of disability should be brought back to
etiopathological causes, to social structures, to cultural
discrimination, to divine punishment, pertains to the diversity of
the models that are not a simple interpretation of a
malfunctioning, but the means of experiencing it.
In this work we will present the results of two
experiments about a qualitative and a quantitative investigation
on the representations of disability which orient the opinions
and attitudes of people who spend large amounts of their time
with a disabled person on a daily basis. Two reasons led us to
conduct this investigation: first, to study more in depth the
relationship between disability, adaptation, socialization, and
development and, second, to verify the spread of the
biopsychosocial model proposed by the International
Classification of Functioning, Disability, and Health .
Aims of descriptive research and quasi-
The implementation of the first experiment is based
on the following assumption: the disability models that
enlighten the social relations of a disabled student can be
inductively inferred by the representations of disability
orienting the opinions and attitudes of people.
The quasi-experimental results were analysed using
qualitative analyses, text coding and interpretation, and
quantitative software-assisted data analyses. The latter were
carried out, with explorative purposes, starting from two
1. In the quantitative analyses, the disability model attributed
to each role-homogeneous group, depends on the highest
number of expressions coded as consistent with that
2. The disability model, significantly prevailing in each role-
homogeneous group, correlates with the participants’
modalities of expression of their personal experience.
The enrolment of participants was carried out
exclusively in schools and local educational agencies, or in local
associations and organizations dedicated to care, assist, and
educate disabled people.
A total of 90 participants (69 F; 21 M; mean age: 47
yrs) were involved in the study: 30 (23 F, 7 M) parents of
disabled students, 7 (6 F, 1 M) parents of non disabled students,
20 (18 F, 2 M) non-specialized teachers, 14 (11 F, 3 M)
specialized teachers, 6 (3 F, 3 M) professional special needs
educators, 13 (8 F, 5 M) social-health workers.
The focus group was chosen as the main tool to gather
all the data for this study. Several authors [2-7] highlight the
expediency of the focus group as an investigation tool when the
researcher is concerned about exploring complex ideas,
connected with the dynamics of personality, such as opinions,
emotions, and motivations.
The choice of the tool for the analysis was largely
reliant on the aim of the study itself, in addition to the type of
data we gathered: the assessment of the underlying disability
models, starting from the coding of the main representations
emerging from the focus groups, is a consistent operation to the
approach supported by the scientific foundations of the
Grounded Theory . The software Atlas.ti was designed
within the frame of the Computer Assisted Qualitative Data
Analysis Software (CAQDAS) Networking Project [9-11] at the
beginning of the 1990s. The software analyse data in a
qualitative approach, by means of interpretation, compatible
with the theoretical basis of the Grounded Theory .
We carried out 17 focus groups. All the focus groups
conversations were audio recorded for a total of approximately
17 hours. The conversations were then transcribed, analysed,
and interpreted, following methods and procedures described in
the following paragraph.
Qualitative data analysis procedures and results
Manual text analysis procedures
The purpose of the manual analysis of the text was to
verify the perspectives on disability according to the different
roles of the participants.
At first, all the 17 hours of audio recordings from the
focus groups were transcribed. Then, the transcripts were
carefully read and indexed, namely the most significant verbal
expressions were identified and a code/index was assigned to
each one. The code categorised the sentences according to the
three main disability models: medical, social, and
biopsychosocial. Overall, 37 codes emerged from this phase: 17
referring to the social model, 8 to the medical one, and 12 to the
After the manual coding, we calculated the most
frequent indexes for each focus group, and we assigned to each
educational role the prevailing disability model.
Manual text analysis results
The text analysis manually performed highlighted
different perspectives on disability, according to the educational
role held by each group of participants. In every role-
homogeneous group, except for the special needs educators’
one, the social model appeared to prevail. However in more
than one group the evaluation was not so unambiguous.
Among the parents of disabled and the non-
specialized teachers, the social disability model was evaluated
as the foremost, since the codes/indexes relative to that model
were the most frequent ones in almost all the focus groups.
It was not possible to attribute a single model to the
other groups. Among the parents of non disabled children, even
though the social model seemed to prevail, the presence of the
medical model was detected. Among the specialized teachers
and the social-health workers emerged a fluctuation between the
social and the biopsychosocial models. This would confirm
what is reported in the literature about the appearance of a so-
called transitional model that would account for aspects of both
Results of the qualitative analysis assisted by
The text analysis carried out with the software
partially confirmed the results emerged from manual indexing,
showing some additional nuances within and between the
The social model is widespread in all the role-
homogeneous groups, except from the specialized teachers that
fluctuate through all the three disability models. The parents of
non disabled students oscillated between the social and the
medical models, confirming what came out from the manual
analysis. However, only the parents of disabled students and the
non-specialized teachers seemed to strongly claim their
adhesion to the social model. The transitional model seems to
spread among the special needs educators and the social-health
workers, with no preference for the biopsychosocial model,
despite the presence of several assumption relative to it.
The parallel analysis developed on a ‘dimensions of
experience’ level, allowed us to discover the existence of a
relationship between disability models and participants’
modalities of expression. Parents, for example, seem to support
some assumptions of the biopsychosocial model, but only at a
level of ‘opinions’, while when they refer to ‘behaviours’ those
assumptions disappear and the social model reappears. In
almost all the groups, the medical model is often expressed with
‘behaviours’, along with emotions like ‘anguish’ and ‘sadness’.
Finally, the social model is expressed through ‘opinions’,
‘behaviours’, and emotions such as ‘anger’ and ‘disgust’.
Quantitative analysis hypothesis
We performed an additional analysis to test if the
qualitative interpretation aiming at assigning a disability model
to the participants of each focus group according to the majority
of the opinions of their components is true. If so, then we
should find a greater number of those codes (attributed to the
texts of each focus group) united in one ‘family’ relative to the
model appearing in each focus group.
Results of quantitative analysis using χ²
Statistical analysis highlighted significant differences
between the groups regarding the distribution of 2 out of 3
disability models: the social and the medical ones (see
Appendix 1). No significant differences were observed for the
biopsychosocial model, equally spread over the groups divided
by role. In general, significant differences in the distribution of
the 3 models appeared in all the groups, with the exception of
parents of non disabled and social-health workers.
The distribution of the 5 dimensions concerning
personal experience resulted significantly different within all
the groups (see Appendix 2). Parents, teachers, social-health
workers, and special needs educators expressed themselves
mainly showing opinions and behaviours, while emotions like
anger, sadness, and joy appeared only in parents and non-
As indicated in Appendix 2, the analysis on the
emotions was carried out only on 6 role-homogeneous groups.
In the groups in which opinions and behaviours showed a
prevalence of the social model, these dimensions were
emotionally expressed as well; in groups in which the
biopsychosocial model prevailed, the participants expressed
mainly opinions, while in groups with preference for the
medical model, the participants expressed both behaviours and
opinions. Lastly, the distribution of opinions, behaviours, and
emotions, resulted significantly different, while the opposite
occurred for the distribution of attitudes and myths/stereotypes.
Discussion of the first experiment results
Results of experiment 1 confirmed our hypothesis that
there is a different perspective on disability according to the
different educational roles.
The results obtained from the qualitative analysis
showed that the horizon of perspectives is wider compared to
the partition into medical, social, and biopsychosocial models.
According to the results obtained we can conclude that:
- The social-health workers and the special needs
educators, as interpreted on the basis of the
qualitative analysis results, refer more to a
- The parents of disabled students, the non-
specialized teachers, and the majority of the
specialized teachers are the groups that adhere the
most to the premises of the social model;
- The parents of non disabled students, as interpreted
on the basis of the qualitative analysis results,
oscillate between the medical and the social model,
although this outcome is not confirmed by the χ²
In the quantitative analysis carried out with the χ² test,
the statistical hypotheses were confirmed: the attribution of a
specific model to each group depends on the greater scores of
the codes united in the ‘family codes’ referring to the model
occurring in each group. The statistical analysis of the personal
experiences distribution in the groups highlighted an interesting
correlation among the foremost disability model in each group
and the modalities of expression of the personal experiences of
the participants. This would explain even more the presence,
within the groups, of different perspectives on disability.
The presence of different models within the groups
leads to some additional conclusions:
- There is a cultural predisposition to the adhesion to
the biopsychosocial model, since the participants
have the tendency to define disability in terms of
individual functioning, rather than using
- Among the parents of non disabled students, the
presence of a perspective on disability partially
different compared to the one that parents of
disabled students have, let us think over the role
played by the experience of disability on the
Aims and hypotheses
As Focus Group methodology is suitable to bring out
judgments and opinions for which participants have a certain
degree of consciousness, in the following experiment we
thought to widen the research question to the so-called
The aim of experiment 2, was to try to establish the
strength of the association that participants create between the
idea of “disabled” and “non disabled” and a series of
dimensional qualities considered as explanatory of two of the
three disability models: the medical and the social one. The
decision of excluding the biopsychosocial model from this
experiment is due to the fact that, being a complex and non
linear model, it is not possible to “stereotypy” it and to bring it
back to an “aut-aut” logic. On the contrary, both the medical
and the social models can be simplified to their basic
assumptions according to attributes of opposite couples:
healthy/ill for the medical model and outcast/integrated for the
social model. Moreover, we introduced a third couple of
attributes, "good/bad", to better evaluate the strenght of the
association that participants create between the categories
“disabled” and “non disabled” and a more basic dimension of
the human nature. Our hypothesis is that, in a condition of
temporal constrain, participants will rely on the automatic
component of the cognitive elaboration when making
associations, generating less variable and more stereotypical
responses, compared to the ones provided by the focus group.
The aim of this second experiment is to verify if there
is a dominant or a prevailing frame in the automatic processes
to elaborate information concerning disability in a different way
compared to what it is observed for the controlled processes, in
which it appears to be a complex of frames which act as scripts,
(i.e. adapted answers and environmental conditions) both
exogenous and endogenous.
36 participants (31 F; 5 M; mean age: 42 yrs) were
involved in the study: 8 (8 F) parents of disabled students, 6 (4
F, 2 M) parents of non disabled students, 6 (4 F, 2 M) non-
specialized teachers, 5 (5 F) specialized teachers, 5 (5 F)
professional special needs educators, 6 (5 F, 1 M) social-health
The IAT [15,16] assess the association between a
target-concept discrimination and an attribute dimension. The
procedure starts with an introduction of the target-concept
discrimination task. The initial discrimination consist in to
distinguish image those who are recognizable as “normal” from
those recognizable as “disabled”. This discrimination task and
the subsequent ones are performed by assigning one category
responding with the left hand and the other responding with the
right hand to the presented stimuli. The second step is the
introduction of the attribute dimensions, also in the form of a
two-category discrimination (good-bad; healthy-ill; outcast-
integrated). After the target discrimination and the attribute
dimension tasks, attributes and categories are superimposed in
the third step, in which stimuli for target and attribute
discriminations appear on alternative trials. In the fourth step,
the respondent has to perform a reversal of response
assignments for the target discrimination, and the fifth (final)
step combines the attribute discrimination (not changed in
response assignments) with the reversed target discrimination. If
the target categories are differentially associated with the
attribute dimension, the subject should find one of the combined
tasks (of the third or fifth step) to be considerably easier than
the other. The measure of difficulty provides the measure of
implicit attitudinal difference between the target categories.
In this study we created 3 different versions of the
IAT: for all the versions the target concept couple was the
normal vs. disabled one, whether in the first test the couples of
attributes were good/ bad, in the second healthy/ill and in the
The hypothesis underlying the IAT  is that
participants’ responses are faster when making associations in
line with their beliefs, emotion and implicit motivations. The
peculiarity of this method consists in the possibility to evaluate
only attitudes without the involuntary tendency of people to
alter their response in order to protect self representation. The
intensity of the association between two concepts is equivalent
to the difference in response time in a classification task
between two blocks. Blocks are defined as compatible if the
experimental task is congruent with the personal association of
the participant, incompatible if the task is not congruent.
In conjunction with the IAT we administered three
scales measuring the semantic differential with the intent to
evaluate, together with the implicit attitudes, the explicit ones
towards disability . Participants are administered a set of
paired attributes which are regarded as a description and a
quantification of the object in a specific dimension. The
semantic differentials used in the present study were constructed
using the same couple of attributes included in each IAT.
Participants were asked to indicate on a 7 point scale, which one
of two extreme was closer to the status of a disabled person.
The mean score indicating a neutral position (“neither…nor”),
was 4. The mean score of the whole scale was 20. All
participants were administered the first 3 IAT followed by the
three semantic differential.
The three implicit association tests (IAT), were
administered with the intent to measure indirectly the individual
differences on the intensity of the association between the
category “disabled” and the attributes couples “good/bad”,
“healthy/ill”, “outcast/integrated”. The computational algorithm
used to measure this association, called “d-biep”, is based on
the calculation of the difference in the means of response times
between the sessions compatible (e.g. non disabled /healthy Vs
disabled/ill) and the session non compatible (e.g. disabled
/healthy Vs non disabled/ill).
For all the IAT, the one sample T-Test (DoF=35,
p<0,000), calculated using the scoring obtained using the “d-
biep” (see Table 1) showed that response latencies for the
compatible sessions are significantly lower compared to the
ones for the non compatible session (see Table 2).
Table 1 – Measures of the d-biep or IAT general effect.
d-biep IAT good/bad 36
d-biep IAT healthy/ill 36
d-biep IAT outcast /integrated 36
Table 2 – One sample T-Test on the d-Biep scoring.
t DoF p
d-biep IAT good/bad
19,57 35 0,000
d-biep IAT healty/ill
18,54 35 0,000
d-biep IAT integrated/castout
10,75 35 0,000
In order to analyse the correlation between the 3 IAT,
we calculated Pearson’s r coefficient on the d-biep measures for
each IAT. The “good/bad” IAT correlated significantly with the
other two: IAT “healthy/ill” (r = .39, p<.05) and IAT “outcast/
integrated” (r = .41, p<.05). The IAT “healthy/ill” correlates
with IAT “good/bad” ” (r = .39, p<.05) and IAT “outcast/
integrated” (r = .54, p<.01). Participants’ performance on the
“healthy/ill” IAT is weakly associated with performance on the
other two IAT, which are strongly correlated. Regarding the
semantic differentials, the mean scores obtained by participants
for each one of them ( “good/bad”= 21,61; “healthy/ill” =
19,30; “outcast/integrated” = 19,27) does not differ significantly
from the overall mean score. Participants did not show a
significant preference for one item of the couple of attributes,
scoring close to the overall mean values.
A good indicator of the validity of the Implicit
Association Test is the correlation with explicit measures. .
In the present study, we found that the semantic differential
“outcast/ integrated” correlates significantly both with the IAT
“healthy/ill” (r = -.37, p<.05), and the “outcast/ integrated” one
(r = -.39, p<.05).
Moreover, participants were divided into two groups:
relatives of a disabled or not. We did so in order to verify the
presence of significant differences on the performance on
implicit and explicit tests as a function of this variable. We
selected 13 participants as relatives of a disabled on a total of
36. An independent sample T Test highlighted the significant
differences between the two groups on the IAT “healthy/ill”
(t(34) = 2,84, p = 0,01), showing a stronger association for the
compatible category and attribute in the group of relatives of a
disabled (“normal”, “healthy”) compare to non relatives.
Discussion of the second experiment results
Results obtained from the administration of the 3 IAT
and the 3 semantic differentials showed that participants tend to
establish preferential associations between the category
“disable” and the attributes “bad”, “ill” and “outcast”. Such
associations, instead, did not emerge clearly from performance
on the explicit tests. Participants’ response is more stereotypical
and less variable on the implicit tests whereas, on an explicit
level, disability in not seen as distinctive of a group of people
with clear boundaries like those demarcated by attributes such
as healthy/ill, limited/able, outcast/ integrated and so on, but is
laid on a continuum.
On an implicit level, the categorisation of the disabled
as “bad” seemed to be the strongest one. D-biep scores are
higher for the dimension “good/bad” compared to the
dimensions “healthy/ill” and “integrated/ outcast”. The bipolar
dimension “good/bad” seem to better discriminate between the
two categories “disabled” and “non disabled” compared to the
dimension “healthy/ill”. The latter appears to discriminate more
compared to the “outcast/integrated” one. Results of the
correlational analysis, appeared to confirm the better capacity of
the dimension “good/bad” to represent the categorical
discrimination between “disabled” and “non disabled”. Among
the implicit tests, the IAT “healthy/ill” and the IAT “outcast/
integrated” showed a robust correlation, while there was a poor
correlation between those dimension with the IAT “good /bad”.
As for the correlation between implicit and explicit
tests, only a small correlation appeared between the two IAT
“healty/ill” and “outcast/ integrated” with the scale of semantic
differential “outcast/ integrated”. This result might be
interpreted as indicative of the degree of separation between a
categorisation executed on an unconscious level and the mental
models, more complex and variegated, that people
unconsciously express when describing their beliefs and
experiences with disability.
Finally, the comparison between the scoring of
participants in both tests – implicit and explicit - as a function of
the presence of a disable person in the family, highlighted the
fact that relatives of a disable tended to associate more then
others disability as a disease. This might indicate that people
with a disabled relative are reinforced in the association of the
category disabled with the attribute “bad” as a function of the
experience of medicalisation reserved from our society to
people with disability, in particular during the first years of their
The aim of this study was to investigate if and in
which measure the representations of disability, elaborated by
parents, teachers and educators, reflected the different theorical
models of disability. Some significant data emerged from our
results. First, conscious and unconscious visions do not overlap.
In fact, from a conscious perspective people do not appear to
rely only on a specific model in order to describe and think
disability but they prefer to elaborate more complex points of
view, defined by some authors as “in transit” models between
different theoretical models . From an unconscious
perspective instead, the stereotypical and bipolar dimension
appeared to be the most discriminative between disability and
non disability and the “good/bad” one. In other words, on an
implicit categorization level, the strongest association is the one
between disability and the negative and unpleasant dimension of
existence: disability is considered, regardless the role or
familiarity, as “bad”. It is also interesting to point out that the
association disabled/bad seemed “reinforced” in a different way
on a social environmental level: people with a relative with
disability on an implicit tests tend more then others to relate
disability with the disease, whether on an explicit level in the
focus group, they express with more intensity the “tragic” of the
disabled condition. This information might be interpreted in the
light of the strong medicalisation experience that our society
reserves to disabled people, especially during the first years of
The correlational results between implicit and explicit
tests found in the second experiment, confirmed a certain degree
of differentiation between the categorisation made at an
unconscious level and the mental models, more complex and
variegated, that people consciously use in describing their
beliefs and experiences with respect to disability. This result is
not necessary negative. In fact, it might be explained as an
index of a cultural predisposition to comply with the
biopsychosocial model, more evolved and complex, compared
to the medical and social ones.
Only a capillary and universal action of information
and education, inspired by the principles of the universal model
of disability, will allow to make salient, for the individual and
the group, the instances of a new “culture of diversity”.
 WHO, International Classification of Functioning,
Disability and Health: ICF, Geneva: WHO, 2001.
 D. J. Warr, ‘It was fun... but we don’t usually talk
about these things’: Analyzing Sociable Interaction in Focus
Groups, Qualitative Inquiry, 11, 2005, pp. 200-225.
 S. Wilkinson, Focus Groups in Health Research.
Exploring the Meanings of Health and Illness, Journal of
Health Psychology, 3, 1998, pp. 329-348.
 D. M. Linhorst, A Review of the Use and Potential of
Focus Groups in Social Work Research, Qualitative Social
Work, 1, 2002 pp. 208-228.
 L. F. Chiu, Transformational Potential of Focus Group
Practice in Participatory Action Research, Action Research, 1,
2003, pp. 165-183.
 D. L. Morgan, Focus groups as qualitative research,
London: Sage, 1998.
 V. L Zammuner, I focus group, Bologna: Il Mulino,
 A. L. Strauss, & J. Corbin, Basics of qualitative
research: Grounded theory procedures and techniques,
London: Sage, 1990.
 N. G. Fielding, & R. M. Lee, Computer analysis and
qualitative research, London: Sage, 1998.
 N. G. Fielding, & R. M. Lee, Computing for
qualitative research: Options, problems and potential, in: N. G.
Fielding, & R. M. Lee (Eds.) Using computers in qualitative
research, London: Sage, 1991.
 A. Coffey, B. Holbrook, & P. Atkinson, Qualitative
data analysis: Technologies and representations, Sociological
Research Online, 1, 1996.
 T. Muhr, Atlas.ti - A prototype for the support of text
interpretation, Qualitative Sociology, 14, 1991, pp. 349-371.
 J. Brett, The experience of disability from the
perspective of parents of children with profound impairment: Is
it time for an alternative model of disability?, Disability &
Society, 17, 2002, pp. 825-843.
 E. B. Jenks, Explaining Disability: Parents’ Stories of
Raising Children with Visual Impairments in a Sighted World.
Journal of Contemporary Ethnography, Special Issue:
Ethnography and Disability Studies, 34, 2005, pp. 143-169.
 A. G. Greenwald, D. E. McGhee, & J. L. K.
Schwartz, Measuring Individual Differences in Implicit
Cognition: The Implicit Association Test. Journal of
Personality and Social Psychology, 74(6), 1998, pp. 1464-
 A. G. Greenwald, B. A. Nosek, & M. R. Banaji,
Understanding and using the Implicit Association Test: I. An
improved scoring algorithm. Journal of Personality and
Social Psychology, 85, 2003, pp. 197-216.
 C. E. Osgood, G. J. Suci, & P. R. Tannenbaum, The
Measurement of Meaning. Illinois: University of Illinois
S. Federici, & M. Olivetti Belardinelli, Un difficile
accordo tra prevenzione e promozione, Psicologia Clinica dello
Sviluppo, 10, 2006, pp. 330-334.
Appendix 1. Table of distribution of observed and expected frequencies (among and within) of the 3 disability models in the groups and relative significance scores,
applying the χ² test (p < .05).
Medical model Social model Biopsychosocial model
Expected value Expected value Expected value
Focus Group [FG]
value among within
value among within
Parents of non-
Sig. 0.00 0.00
Appendix 2. Table of distribution of observed and expected frequencies (among and within) of the 5 dimensions of personal experience in the groups and relative significance scores,
applying the χ² test (p < .05).
Dimensions of personal experience
Opinions Attitudes Behaviours Myths and stereotypes Emotions
Expected value Expected value Expected value Expected value Expected value
Focus Group [FG]
value among within
value among within
value among within
value among within
Parents of non