Adaptation and Psychometric Analysis of the Scale of Attitudes Towards People with Disabilities (EAPCD) in Work Contexts in Chile

Abstract

Workers’ attitudes towards people with disabilities can influence labor inclusion. Despite the relevance of learning about these attitudes, in Chile, there are no reliable and validated instruments to measure them in workers. An instrumental study was carried out to adapt and psychometrically analyze the Spanish version of the Scale of Attitudes towards People with Disabilities (EAPCD) in the Chilean population. By means of a non-probabilistic convenience sampling, 310 people between 19 and 76 years of age (M = 39.4; SD = 11.2) from different regions of Chile participated in the study. The linguistic adaptation included the change in Spanish idioms and expressions to the Chilean context, whose relevance was validated with cognitive interviews. Descriptive analysis, internal consistency analysis, and CFA (confirmatory factor analysis) were performed. The CFA indicated an adequate fit to the three-factor model proposed in the original scale (χ²(431) = 808.070, p < 0.01; RMSEA = 0.053; CFI = 0.93; TLI = 0.93). The Social Relations and Normalization dimensions were reliable; however, it is recommended that the items of the Intervention Programs dimension be revised or eliminated, since they could be confusing in the local context. It is concluded that the scale is reliable and valid to evaluate attitudes towards the inclusion of people with disabilities in Chile in the work context.

1. Introduction

Attitude is a multidimensional construct defined as an evaluation formed from the interaction of cognitive and affective processes, reflecting judgments about oneself, other people, subjects, and objects. This evaluation extends along a dimension ranging from positive to negative [1,2]. Attitudes are not fixed dispositions but rather constructed and adjusted according to contextual conditions, highlighting their potentially adaptive nature [3].

People with disabilities often face several barriers to accessing work [4,5]. One of the main barriers is society’s negative attitudes towards their participation in work settings [6,7,8,9]. These negative attitudes may originate from multiple factors, such as lack of knowledge regarding disability [10], type of disability [11], previous experiences with people with disabilities [12,13], and the organizational culture of the workplace [6,14].

For those people with disabilities who do manage to access employment, the tendency to receive lower average wages than workers without disabilities has been reported [15,16]). In addition, they are more likely to be in part-time, temporary, or non-standard jobs, which offer lower wages and few benefits [17]. These inequalities in work experiences are reflected in job satisfaction levels, which are generally lower compared to workers without disabilities.

Stone and Colella [18] propose a multidimensional model to analyze attitudes toward people with disabilities in organizational contexts. The model suggests that personal characteristics (attributes of the individual with a disability and attributes of the observer), environmental factors (such as regulations and legislation), and organizational characteristics (such as norms, values, policies, nature of work, and reward system) interact to influence the treatment of people with disabilities at work.

Personal characteristics include attributes, such as the nature of the disability, previous performance, gender, and interpersonal style of people with disabilities. These factors affect the observer’s perception, which influences his or her expectations and how people with disabilities are treated at work. Environmental factors, such as legislation, can influence organizational characteristics, modifying policies, technologies, and culture to facilitate the inclusion of people with disabilities. Lastly, organizational characteristics, such as norms, values, policies, and work practices, determine the type of people who are hired, as well as job design and reward systems.

The observer’s psychological processes include the automatic categorization of people with disabilities and the use of stereotypes to make inferences about their attributes and capabilities. The interaction and feedback suggest that people with disabilities' responses may modify an observer’s expectations and organizational characteristics, impacting future interactions and perceptions.

One of the most widely used instruments in Spanish to measure attitudes towards disability is the Scale of Attitudes towards People with Disabilities (Escala de Actitudes hacia las Personas con Discapacidad or EAPCD in Spanish) [19]. This instrument, designed for the Spanish population, has a multifactorial structure composed of five dimensions: assessment of abilities and limitations, recognition/denial of rights, personal involvement, generic qualification, and role assumption, distributed in 37 items.

Despite its widespread use in Spanish-speaking countries, such as Spain, Peru, Mexico, and Chile—mainly in educational contexts, with undergraduate students [20,21,22,23], graduate students [24], mothers and fathers of kindergartens [25], and in the military population—an insufficient adaptation of language and context to the specific realities of each country has been observed.

For example, Atoche-Silva et al. [26] used the original version of the EAPCD instrument in Peruvian university students, while Quiroz et al. [27] used it to assess attitudes towards disability in Mexican university students, without major modifications. In other studies, no psychometric properties are reported [21,26,28,29,30,31,32].

In Chile, the EAPCD instrument has been used to evaluate attitudes of medical students towards people with disabilities; although, the reported analyses have been limited to ANOVA, Student’s t-test, and Z tests of proportions [33]. Araya-Cortés et al. [34] conducted a reliability analysis in university students mainly linked to social sciences, without further adaptation of the original items. To guarantee the validity of the interpretations derived from an instrument, it is essential that its psychometric properties are adapted for the population being evaluated, in this case, the Chilean population.

In this context, Arias et al. [35] attempted to address some of these limitations by updating and validating the EAPCD, particularly in a Spanish sample of 976 health and education professionals. Their results showed a three-factor model, with one main dimension and two secondary dimensions. The recent study by Simón et al. [24] reported the psychometric analysis of this scale in a sample of Spanish master’s degree students. Their results confirm the three-factor structure proposed by Arias et al. [35] and, in turn, evaluated a second-order structure, concluding that the scale can be interpreted both dimensionally and as a whole.

This evidence underscores the need for an instrument adapted to the population of interest and the importance of evaluating and adapting the scale to other cultural contexts. Specifically in Chile, there is a lack of research on attitudes towards the labor inclusion of people with disabilities using validated and updated instruments. Therefore, the present study aims to evaluate the psychometric properties of the Scale of Attitudes Towards People with Disabilities (hereafter, EAPCD) [35] in the Chilean general population and to identify eventual adaptations that may be required. For this purpose, the internal consistency and factorial structure of the scale was evaluated.

2. Materials and Methods

An instrumental design [36] was used to examine the psychometric properties of the EAPCD instrument and to adapt it [35] in the Chilean working population. In its implementation, the guidelines for test use were followed, as well as the recommendations established by the International Test Commission [37].

2.1. Sample

Participants were invited through a non-probabilistic convenience sampling, selected from the general population residing in different regions of Chile. Subjects were contacted via e-mail and through dissemination in social networks. Inclusion criteria included being over 18 years of age, having a trade or profession, and being employed. Exclusion criteria included having any type of disability, being a student, or being unemployed.

All participants were informed of the general objective of the study, as well as their rights to anonymity and data confidentiality. They agreed to participate by signing a digital informed consent form before answering the questionnaire. The procedures and methods used in the present study conformed to the ethical guidelines defined by the Declaration of Helsinki (World Medical Association, 2013) [38]. The research project was approved by the Ethics, Bioethics, and Biosafety Committee of the Universidad de Concepción, Chile (CEBB code 1349-2023).

2.2. Description of the Sample

To collect sociodemographic information, a brief questionnaire was prepared asking about age, gender, educational level, and work activity. In addition, questions about the frequency of contact with people with disabilities in the work environment, hiring preferences, and the perception of the level of knowledge according to the type of disability were asked.

Initially, the sample consisted of 319 people, but 9 cases were excluded because they did not meet some inclusion criteria, leaving a final sample of 310 working participants, aged between 19 and 76 years (M = 39.4; SD = 11.2). Of these participants, 107 were men (34.5%), and 198 were women (63.8%), while 5 participants (1.7%) chose not to indicate their gender.

In terms of educational level, the sample consisted of 12 participants who reported having only secondary education (3.8%), and 7% of participants reported having trade-level training. Additionally, 25 participants (8%) reported having incomplete university studies, while 126 (40.6%) reported having completed university studies. Finally, 125 participants (40.3%) indicated having completed postgraduate studies, reflecting a sample predominantly made up of individuals with high educational levels. The most representative professions were teachers with 82 (26.5%), followed by psychologists with 48 (15.5%), and finally, engineers with 41 (13.2%). In addition, 59 participants indicated that they worked in a trade, representing 19% of the sample.

Most of the participants worked in large companies (47.4%), with an employment contract (64.1%, n = 199), in a full-time, dependent job. In relation to work contact with people with disabilities, 51.6% reported having contact in their work environment. Out of participants who reported having contact with people with disabilities at work, 23.2% (n = 72) reported daily contact with people with disabilities, followed by several times a week (13.2%), several times a month (8.3%), and at least once a month (7%). Finally, when asked how they feel about this contact, 51.9% (n = 161) reported feeling “very comfortable” in the presence of people with disabilities.

In relation to hiring, 28% (n = 87) of all subjects in the sample reported holding a managerial position, of which 8.3% (n = 26) had hired a person with a physical disability, 6.4% (n = 20) with a visual disability, 6.1% (n = 19) with a hearing disability, 4.8% (n = 15) with an intellectual disability, 4.1% (n = 13) with a psychiatric disability, and 2.5% (n = 8) with multiple disabilities.

Regarding perceived knowledge about the different types of disability, the highest perception of knowledge was recorded in intellectual disability with 121 participants reporting a high level of knowledge. In contrast, the lowest perception of knowledge was observed in multiple challenges, where 70 people reported having a knowledge level of “none”, followed by 53 participants with the same perception regarding psychiatric disability.

2.3. Instruments

For the present study, the Spanish Scale of Attitudes Towards People with Disabilities (Escala de Actitudes hacia las Personas con Discapacidad or EAPCD) developed by Arias et al. [35] was used. The instrument was designed to assess attitudes towards disability and has been widely used in higher education to assess the attitudes of undergraduate students towards disability, as well as in work contexts, particularly health and education. However, the scale can be applied to professionals and workers in different fields.

Although the scale presents 34 items in the original article [35], an error in the correlative numbering was verified, finding that the scale is composed of 31 items distributed in 3 dimensions. The original instrument is answered using a 4-point Likert scale, with the options (a) strongly agree, (b) agree, (c) strongly disagree, and (d) totally disagree. However, for the adaptation of the scale and to facilitate understanding in the Chilean context, the order of the options was reversed, as detailed in Appendix A (see Spanish version in the Supplementary Materials).

The first dimension, which explains 42.44% of the common variance (α = 0.858; ω = 0.949), is entitled “Social and personal relationships with people with disabilities” or “Social Relationships”. This factor comprises 13 items focused on feelings, behavioral intentions, and thoughts when interacting with people with disabilities (example item: “I think I would find it easy to relate to a person with a disability”).

The second dimension, which explains 11.96% of the common variance (α = 0.822; ω = 0.921), is titled “Normalization”. This dimension includes 13 items that address the rights to lead a full and equal life and the ability of people with disabilities to function in various aspects of life (example item: “People with disabilities can enjoy sports as much as anyone else”).

Finally, the third dimension, which represents 7.71% of the common variance (α = 0.603; ω = 0.802), is titled “Intervention Programs”. This factor is related to specific actions to favor the integration and full inclusion of people with disabilities, as well as evaluations of the economic profitability of such actions (example item: “Something should be done to achieve greater integration of people with disabilities, for example, by facilitating their access to public places”).

2.4. Procedure

First, permission was requested via e-mail from the original author for the adaptation and validation of the instrument, and their consent was obtained. Subsequently, in order to ensure the understanding of the instructions, the content of the items, their format, and the response scale, 10 cognitive interviews were carried out with working people between the ages of 20 and 55, after obtaining their informed consent, following the recommendations of Smith-Castro and Molina [39]. Convenience sampling was employed, selecting accessible participants who expressed interest in being part of the study [40].

The interviews used three cognitive tests: (1) instructions, (2) clarity of questions or items, and (3) sensitivity. The first consists of asking the respondent to identify problems in the instructions from the respondent’s perspective (“before moving on to the questions, could you repeat in your own words the instruction you just heard or read?”). The clarity of questions or items test seeks to identify problems related to the meaning of the items (“what does the word (term) in this question mean to you?”). Finally, the sensitivity test asks the respondent to review questions or words of a sensitive nature (“is it okay to talk about these topics in a survey or do you feel very uncomfortable?”, “how do you feel about these types of questions?”). These techniques aim to detect problems in the four moments of the question–answer process: comprehension, information retrieval, estimation, and execution [39].

Each interview lasted approximately 40 min. Some were conducted virtually and others in person. Each participant was given the questionnaire, which was read aloud while the interviewer (main researcher) recorded the session, asked questions, and took notes. The answers provided were then analyzed, making necessary adjustments where difficulties in comprehension were identified. The objective was to seek clearer alternatives for the formulation of the items. Specifically, linguistic adaptations were made to the original items 5, 6, 10, 21, 22, 25, and 33 (see

Table 1. Modifications after cognitive interviews.

Item	Original	Modification
5	Something should be done to achieve greater integration of people with disabilities, such as facilitating their access to public places.	* 4. There should be concrete actions to achieve greater inclusion of people with disabilities, such as facilitating their access to public spaces.
6	People with severe disabilities should be confined to specialized centers.	* 5. People with severe disabilities should be in specialized residential centers.
10	If I had a meeting with a person with a disability, I would worry about what people might think.	* 9. If I were to meet with a person with a disability, I would worry about what people might think.
21	I would like my child to interact with children with disabilities.	* 18. I would like my daughter/son to interact with children with disabilities.
22	Workers with disabilities should receive the same salary as those without disabilities.	* 19. Workers with disabilities should receive the same salary as those without any type of disability.
25	The behavior of people with disabilities is generally irritating.	* 22. The behavior of people with disabilities is generally bothersome.
33	More money should be spent on removing the physical barriers that continue to make life difficult for people with disabilities.	* 30. More money should be invested to eliminate the physical barriers that continue to make life difficult for people with disabilities.

Note: * The number of items in the scale differs due to changes in the correlative order between the original scale that was in error and the new version that corrects the error in the correlative numbering.

).

After modifications were made, the scale was digitized in the Survey Monkey platform and administered via a link shared by the research team through social networks and emails using the snowball method. Data collection took place between May and July 2023, and the reception of questionnaires was automatically managed through the platform.

2.5. Data Analysis

A qualitative analysis of the cognitive interviews was performed. Following this, descriptive data were obtained for the total sample. The internal consistency of the factors was evaluated by means of Cronbach’s alpha and McDonald’s omega coefficients using the statistical program JASP version 18.3 [41].

A confirmatory factor analysis (CFA) was performed using Mplus version 7.0 [42], using the WLSMV (Weighted Least Squares Mean and Variance adjusted) estimator for its robustness and capacity to model categorical or ordinal data without requiring the assumption of normality in the variables [43]. To evaluate the fit of the models in the confirmatory factor analysis, the Chi-square index (p > 0.05), root mean square error of approximation RMSEA (good fit < 0.05–0.08), the comparative fit index (CFI), and Tucker–Lewis index (TLI) (expected > 0.95) were used [43,44].

The convergent and discriminant validity of the factors was evaluated according to the criteria laid out by Fornell and Larcker [45] by calculating the shared variance between the latent variables of the model. For convergent validity, two indicators were used: average extracted variance (AVE) and composite reliability (CR). The AVE, which measures the level of variance captured by a construct compared to the variance attributable to the measurement error, is considered good with values greater than 0.7 and acceptable with values greater than 0.5. The CR, less biased than Cronbach’s alpha coefficient, is acceptable with a value of 0.7 or higher. Meanwhile, discriminant validity [45] compares the square root of the AVE with the correlations between the latent constructs. Each construct must explain more variance by itself than that shared with other constructs, so the square root of each AVE must exceed the correlations between constructs.

3. Results

3.1. Linguistic Translation of the Instrument and Cognitive Interviewing

The results of the cognitive interview showed that most of the items were well understood; although, room for improvement was identified in seven of them. Terms were adjusted to be more specific and less negative, e.g., “confined” was changed to “being”, “spent” to “invested”, “special centers” to “specialized residential centers”, and “integration” to “inclusion”. In addition, both genders were included in the items, using “son/daughter” and “girls/boys”.

3.2. Descriptive Analysis

The sample did not present any missing data. Before conducting statistical procedures, the scores of the 13 negatively worded items were reversed, most of which were concentrated in the first dimension. Subsequently, descriptive statistics were calculated for each item and each dimension. The means ranged from 2.29 to 3.94, indicating a tendency toward positive responses on a scale of 1 to 4. However, for the reversed items, the interpretation of this result is the opposite, meaning it reflects the negation of the item. The item total correlations varied, and some items showed low correlations, particularly in the Intervention Programs dimension, suggesting possible issues with internal consistency in this dimension (see

Table 2. Descriptive statistics by item and by dimension.

Dimension	Item	M (SD)	Min–Max	Correlation Item-Total	Alpha If the Item Is Deleted
Social and personal relationships with people with disabilities (SR)	6 (inv) *	3.84 (0.41)	1–4	0.488	0.763
	8 (inv) *	3.77 (0.57)	1–4	0.289	0.779
	9 (inv) *	3.84 (0.41)	1–4	0.398	0.769
	11(inv) *	3.80 (0.48)	1–4	0.292	0.777
	14	3.55 (0.66)	1–4	0.443	0.764
	15 (inv) *	3.31 (0.69)	1–4	0.355	0.775
	16	3.47 (0.64)	1–4	0.489	0.759
	18	3.73 (0.48)	1–4	0.475	0.762
	21 (inv) *	3.87 (0.42)	1–4	0.333	0.774
	22 (inv) *	3.67 (0.55)	1–4	0.437	0.764
	25	3.69 (0.69)	1–4	0.347	0.776
	28 (inv) *	3.71 (0.51)	1–4	0.544	0.755
	29 (inv) *	3.64 (0.53)	1–4	0.527	0.756
	(n = 13)
Normalization (N)	1	3.50 (0.71)	1–4	0.378	0.769
	2	3.73 (0.58)	1–4	0.311	0.773
	3 (inv) *	3.47 (0.67)	1–4	0.342	0.772
	7	3.94 (0.29)	1–4	0.247	0.776
	10	3.43 (0.59)	1–4	0.465	0.757
	12	3.87 (0.38)	1–4	0.501	0.759
	13	3.71(0.52)	1–4	0.413	0.763
	17	3.45 (0.60)	1–4	0.512	0.752
	19	3.81 (0.45)	1–4	0.388	0.766
	20	3.85 (0.46)	1–4	0.359	0.768
	24	3.48 (0.64)	1–4	0.403	0.765
	26	3.64 (0.56)	1–4	0.532	0.751
	27	3.74 (0.48)	1–4	0.497	0.756
	(n = 13)
Intervention Programs (IP)	4	3.88 (0.39)	1–4	0.125	0.255
	5 (inv) *	3.01 (0.74)	1–4	0.160	0.211
	23 (inv) *	2.29 (0.92)	1–4	0.090	0.330
	30	3.84 (0.44)	1–4	0.125	0.251
	31 (inv) *	3.75 (0.51)	1–4	0.237	0.164
	(n = 5)

Note: own elaboration. (inv) *: inverted item. For interpretation, it represents the negation of the original item.

).

3.3. Reliability Analysis

Internal consistency reliability using Cronbach’s alpha and McDonald’s omega is presented in

Table 3. Reliability estimates for each dimension.

Dimension	Alpha (IC 95%)	Omega (IC 95%)	Items
SR	0.78 (0.74–0.81)	0.78 (0.75–0.82)	6,8,9,11,14,15,16,18,21,22,25,28,29
N	0.77 (0.74–0.81)	0.78 (0.74–0.81)	1,2,3,7,10,12,13,17,19,20,24,26,27
IP	0.28 (0.15–0.39)	0.24 (0.13–0.34)	4,5,23,30,31

Note: SR: Social Relationships; N: Normalization; IP: Intervention Programs.

for each dimension of the scale. The Social Relationships and Normalization dimensions show good reliability estimates (alpha and omega > 0.70), indicating acceptable internal consistency. However, the Intervention Programs dimension shows very low reliability estimates (alpha = 0.28, omega = 0.24), suggesting a lack of consistency among the items of this dimension.

3.4. Confirmatory Factor Analysis

The confirmatory factor analysis for the original model proposed by Arias et al. [35] of three factors and 31 items showed an appropriate fit to the data (χ²(431) = 808.070, p < 0.01; RMSEA = 0.053 [0.047–0.059]; CFI = 0.93; TLI = 0.93).

The factor loadings for the model are presented in

Table 4. Factor loadings for the 3-factor Professionals’ Attitudes Toward Persons with Disabilities Scale.

Item	Social Relationships	Normalization	Intervention Programs
1		0.513 (0.051)
2		0.554 (0.053)
3		0.527 (0.042)
4			0.721 (0.087)
5			0.265 (0.072)
6	0.754 (0.046)
7		0.910 (0.089)
8	0.650 (0.042)
9	0.740 (0.045)
10		0.634 (0.038)
11	0.669 (0.043)
12		0.892 (0.042)
13		0.644 (0.050)
14	0.669 (0.039)
15	0.537 (0.043)
16	0.679 (0.038)
17		0.595 (0.040)
18	0.785 (0.033)
19		0.690 (0.045)
20		0.722 (0.052)
21	0.715 (0.045)
22	0.690 (0.037)
23			0.051 (0.066) *
24		0.540 (0.045)
25	0.674 (0.041)
26		0.711 (0.038)
27		0.780 (0.037)
28	0.774 (0.032)
29	0.713 (0.033)
30			0.691 (0.056)
31			0.731 (0.048)

Note: Values in parentheses represent the standard error. Values indicated with * correspond to non-significant p-values.

. The loadings for factor 1 (SR) are between 0.537 and 0.785, for factor 2 (N) 0.513 and 0.910, and for factor 3 (IP) 0.051 and 0.731, all being statistically significant (except for item 23 of factor 3).

3.5. Discriminant Validity and Convergence Indicators

The convergent and discriminant validity indicators, according to the criteria laid out by Fornell and Larcker [45], are presented in

Table 5. Reliability and discriminant validity of the 3-factor model.

Model	CF	AVE	Social Relationships	Normalization	Intervention Programs
3 Factors
SR	0.924	0.488	0.698
N	0.913	0.464	0.803	0.681
IP	0.640	0.320	0.817	0.827	0.565

Note: CF: composite reliability. AVE: average variance extracted RS: Social Relationships; N: Normalization; IP: Intervention Programs; square root of the SPV is in bold on the diagonal.

. The composite reliability (CF) values for the Social Relations (0.924) and Normalization (0.913) dimensions are high and exceed the acceptable threshold of 0.70, indicating good internal consistency. This suggests that the items in these dimensions are well correlated and consistently measure the underlying construct. However, the average variance extracted (AVE) values for Social Relationships (0.488) and Normalization (0.464) are slightly below the recommended threshold of 0.50. This indicates that these dimensions are not capturing enough variance explained by the items relative to the error; although, the values are close to the recommended threshold. In contrast, the Intervention Programs dimension presents less satisfactory results. The CF for this dimension (0.640) is below the acceptable threshold, suggesting weak internal consistency among items. In addition, the AVE for Intervention Programs (0.320) is considerably low, indicating significant convergence problems. This suggests that the items in this dimension are not adequately measuring the construct they are intended to represent.

3.6. Normative Data for the Validated Scale: Normative Scores by Category

Normative data allow for the interpretation of individual scores in relation to a normative sample. The categories for the EAPCD instrument are presented in

Table 6. Normative scores for the Scale of Attitudes Towards People with Disabilities (EAPCD) in a Chilean sample.

Direct Score	Qualitative Description (Levels)	Percentile	Z Score	T Score
79	Very negative attitude	20	−3.216	17.84
82	Very negative attitude	20	−2.808	21.92
83	Very negative attitude	20	−2.672	23.28
84	Very negative attitude	20	−2.536	24.64
85	Very negative attitude	20	−2.400	26.00
86	Very negative attitude	20	−2.264	27.36
87	Very negative attitude	20	−2.128	28.72
89	Very negative attitude	20	−1.855	31.44
90	Very negative attitude	20	−1.719	32.81
91	Very negative attitude	20	−1.583	34.17
92	Very negative attitude	20	−1.447	35.53
93	Very negative attitude	20	−1.311	36.89
94	Very negative attitude	20	−1.175	38.25
95	Very negative attitude	20	−1.039	39.61
96	Very negative attitude	20	−0.903	40.97
97	Very negative attitude	20	–0.767	42.33
98	Negative attitude	40	−0.631	43.69
99	Negative attitude	40	−0.495	45.05
100	Negative attitude	40	−0.359	46.41
101	Negative attitude	40	−0.222	47.78
102	Negative attitude	40	–0.086	49.14
103	Neutral attitude	60	0.050	50.50
104	Neutral attitude	60	0.186	51.86
105	Neutral attitude	60	0.322	53.22
106	Neutral attitude	60	0.458	54.58
107	Positive attitude	80	0.594	55.94
108	Positive attitude	80	0.730	57.30
109	Positive attitude	80	0.866	58.66
110	Very positive attitude	100	1.002	60.02
111	Very positive attitude	100	1.138	61.38
112	Very positive attitude	100	1.274	62.74
113	Very positive attitude	100	1.410	64.10
114	Very positive attitude	100	1.546	65.46

Note: Direct score provides the score obtained on the scale. Levels provide a useful qualitative interpretation for a quick understanding of the meaning of the direct score. Percentile contextualizes the raw score within the assessed population. Z-score shows the standardization of the raw score, useful for statistical comparisons. T-score provides a standardized scale by transforming the Z-scores to a scale with M = 50 and SD = 10.

. In Table 6, they include direct scores, percentiles, qualitative description, Z-scores, and T-scores. The percentiles indicate the distribution of scores within the population studied. For example, a direct score of 103 corresponds to the 60th percentile, indicating a neutral attitude.

Raw scores were transformed into Z scores to standardize the results using the formula $Z={\displaystyle \frac{(X-\overline{X})}{S}}$, where X represents the raw score, $\overline{X}$ is the mean of the raw scores, and S is the standard deviation of the raw scores. Subsequently, the Z scores were transformed into T scores using the formula T = 50 + 10Z. This transformation facilitates the interpretation of the scores, placing the M = 50 with a SD = 10. This process was performed using JASP statistical software, which allowed for the accurate and efficient standardization of the scores.

The qualitative description of the levels is as follows: (a) Very negative attitude: This level indicates a highly negative and resistant attitude towards labor inclusion of people with disabilities. Individuals at this level may actively oppose inclusive policies and practices and hold strong prejudices or negative stereotypes. (b) Negative attitude: At this level, professionals and workers have a negative attitude towards the labor inclusion of people with disabilities. They may show resistance to inclusive policies and have prejudices or stereotypes that hinder inclusion. (c) Neutral attitude: People with a neutral attitude do not have a dominant position for or against labor inclusion of people with disabilities. Their attitude is indifferent, and their support or resistance may depend on external factors or the specific situation. (d) Positive attitude: At this level, professionals and workers show a positive attitude towards labor inclusion. They accept and support the inclusion of people with disabilities; although, they may not be as proactive as those at the very positive level. (e) Very positive attitude: This level indicates a highly positive and proactive attitude towards labor inclusion of people with disabilities. Individuals at this level not only support inclusion but also actively promote inclusive policies and practices in their work environment.

4. Discussion

The present study evaluated the psychometric properties of the Scale of Attitudes Towards People with Disabilities (“Escala de Actitudes hacia las Personas con Discapacidad” or EAPCD) [35] in the working population, using the digital self-report format. The results obtained show an adequate general fit, confirming its applicability in the Chilean population. The main findings are discussed below.

Linguistic adaptation is crucial to ensure that the questionnaire items are culturally relevant and understandable. The modifications made reflect an effort to make the language more accessible and less stigmatizing. These changes are important to ensure that participants understand and respond accurately, which in turn, improves the validity of the data collected.

The three-factor model was adequate, with the Social Relationships (SR) and Normalization (N) dimensions showing Cronbach’s alpha and McDonald’s omega coefficients above 0.770. These values suggest that the items within these dimensions are cohesive and measure homogeneous constructs, which is indicative of good internal reliability [46].

In contrast, the Intervention Programs (IP) dimension presented very low reliability estimates, with values indicating a significant lack of internal consistency among the items of this dimension, possibly due to item heterogeneity or problems related to their interpretation. This suggests the need to review and possibly reformulate these items to improve their reliability. These findings are consistent with a similar recent study that, while confirming the validity of the three-factor model proposed by Arias et al. [35], emphasizes that the IP factor has limitations and recommends further revision of its items for accuracy and consistency [24].

The results of the confirmatory factor analysis (CFA) provide evidence of the factor structure of the instrument and its ability to measure attitudes toward people with disabilities in three dimensions. The overall model fit indices (χ², RMSEA, CFI, and TLI) suggest that the three-factor model is adequate. The χ² value is significant, which is common in large samples, but the RMSEA of 0.053 is below the threshold of 0.06, indicating a good fit [47]. The CFI and TLI indices, both at 0.93, fall below the 0.95 threshold, which, according to the literature, can be interpreted as a moderate or acceptable fit. This indicates that the model has a reasonable capacity to fit the observed data; although, it is not an ideal fit [43,44]. In such cases, it is advisable to consider other fit indices, such as RMSEA and Chi-square, which, by suggesting a good fit, support the interpretation that a value of 0.93 is acceptable, as in the case of this study.

However, the variability observed in the factor loadings of the IP factor suggests the need to reformulate some items within this dimension to improve its reliability and validity. In addition, the non-significant loading of item 23 indicates that this item could be eliminated. Item 23 (“Rehabilitation programs for people with disabilities are excessively expensive for what little is achieved”) addresses a specific and subjective issue, as is the perception of the cost-benefit ratio of rehabilitation programs for people with disabilities. This approach may lead to varying interpretations among respondents, some of whom may focus on the economic aspect, while others may adopt a more emotional perspective on rehabilitation, beyond cost. In addition, by introducing a value judgment on the effectiveness of rehabilitation programs, this item departs from others that explore more general attitudes or social behavior, suggesting a possible misalignment with the factor to which it belongs.

Composite reliability for the Social Relationships and Normalization dimensions was high, indicating that the items within these dimensions are highly correlated and measure consistent constructs. The average extracted variance for these dimensions, although adequate, is slightly below the desired threshold (>0.50), with values of 0.488 and 0.464, respectively. This suggests that less than half of the variance is explained by the factor. However, these values are close to the threshold and are considered acceptable.

In contrast, the Intervention Programs dimension shows both a somewhat low composite reliability (0.640) and an average extracted variance (0.320) well below the desired threshold (>0.50), suggesting an unreliability of the factor and a low proportion of variance explained. This result underscores the need for a review and possible restructuring of the items that compose this dimension.

Regarding discriminant validity, the results indicate that the Social Relationships (SR) and Normalization (N) dimensions do not discriminate well from each other, with square roots of their AVEs (0.698 and 0.681, respectively) that are lower than the interfactor correlations. The same is true for the IP dimension, with interfactor correlations of 0.817 and 0.827, higher than the square root of its AVE (0.565). The above suggests that, in general, the three dimensions are not clearly differentiated from each other and may be measuring similar or overlapping constructs.

When comparing the findings from the survey applied to the Chilean population with the results of the original scale applied to a Spanish sample, notable differences emerge. The study by Arias et al. [35] focused on a sample of 976 professionals from various fields in Spain, with a three-dimensional factoriazl model that demonstrated good fit and internal reliability across most dimensions. In contrast, the adapted version used in the present study targeted a broader occupational context, including a variety of professions and trades, and was based on a smaller sample of 310 participants. This contextual change revealed some differences in the reliability of the dimensions, particularly in the dimension of “Intervention Programs”, suggesting the need to revise certain items. This highlights the importance of tailoring scales to the cultural and social particularities of each population.

It is worth noting that, in the study by Arias et al. [35], the instrument was primarily administered to professionals in the health and education sectors. However, the scale is flexible enough to be used with professionals and workers from diverse fields. Our findings are consistent with those of [35] as a significant proportion of our participants also come from the health and education sectors. Nevertheless, our study includes a notable representation of professionals in the engineering field, which is an advantage. This suggests that, in these occupational sectors, there is experience, or at least interest, in the labor inclusion of people with disabilities.

Moreover, the labor inclusion of people with disabilities is often regarded as a morally correct and socially beneficial issue, which may introduce a social desirability bias [5]. This bias can influence the outcomes of studies using self-report surveys, as participants may underestimate their biases or, conversely, exaggerate their support for the labor inclusion of people with disabilities [7]. To mitigate this bias, indirect approaches are recommended, such as qualitative interviews or methodologies that capture implicit or unspoken attitudes. Additionally, it is suggested to include statistical controls that help identify inconsistent response patterns [15]. In the context of this study, the cultural adaptation of the instrument contributes to reducing biases and improving the accuracy of responses.

It is important to highlight that these results may contribute to promoting the labor inclusion of people with disabilities in Chile, considering that, despite legislative progress and existing public policies, they continue to face barriers. According to the 2022 National Disability Study conducted by the National Disability Service (SENADIS), only 46.9% of people with disabilities of working age are employed, compared to 65.3% of the non-disabled population, reflecting the labor exclusion of this group. The 2014 Study on Labor Inclusion of People with Disabilities in the State revealed that only 2.2% of public sector employees have some form of disability, underscoring their low representation in state employment. In line with this, despite Law No. 21.015, enacted in 2018, which requires companies with more than 100 employees to hire at least 1% of people with disabilities, its implementation has been insufficient [48]. The ConTrabajo Foundation, a disability specialist organization, reports that as of May 2023, only 45.8% of contracts established under this law remain active, indicating a lack of job stability. Furthermore, people with disabilities often occupy informal positions, earn lower wages, and face challenges related to accessibility, lack of reasonable accommodations, and continuous training opportunities.

Consequently, the validation and adaptation of this scale are expected to guide employers in creating more inclusive environments by assessing attitudes toward labor inclusion in their workplace contexts. This would provide valuable information for designing equitable job opportunities for people with disabilities and fostering their inclusion in the labor market. Additionally, future research is encouraged to evaluate the psychometric properties of this adaptation by reformulating the items in the Intervention Programs dimension.

However, this study has certain limitations. First, the use of non-probability convenience sampling may limit the generalizability of the results, since participants were selected based on their availability and accessibility. A second limitation is the gender distribution in the sample, as it does not reflect the proportion of men and women in the general population. This could introduce a bias in the results, especially if attitudes or perceptions related to the workplace inclusion of people with disabilities vary by gender.

In addition, the diversity of the sample, both in its educational level and the occupations and professions of the participants, must also be considered. It is known that a higher educational level correlates with more positive attitudes towards the work performance of people with disabilities [49]. People with previous experience working with people with disabilities also show more positive attitudes, so the trend towards favorable scores for inclusion could be explained by the professionals linked to the area of education and psychology being overrepresented in the sample. Future studies should consider a more balanced representation of different work settings.

Lastly, the use of digital media, such as email and social media, to contact participants may have limited access to certain groups, particularly older working populations who might have less accessibility or familiarity with these online platforms. Future research should consider incorporating more diverse contact methods to achieve a more representative sample of the entire population, including those with less access to digital technologies.

5. Conclusions

In conclusion, although the three-factor model proposed by Arias et al. [35] shows an adequate overall fit and most of the items have significant loadings, it is necessary to revise some items of the Intervention Programs (IP) factor to improve its consistency and validity and to eliminate item 23. This study contributes to the literature on attitudes towards disability, providing a valid tool to assess and improve the labor inclusion of people with disabilities in the Chilean context.