The interview was given to a convenience sample made up of 1,173 people with ID who were resident in 16 of the 17 autonomous communities in Spain. Women made up 41.6% of the sample. The age of those assessed ranged between 35 and 80 (M=52.27; SD=7.5). In terms of habitual residence, 48.70% of the participants lived in their own homes, while the remaining 51.30% lived in residential accommodation. Considering the degree of recognized disability (based on R.D. 1856/2009, December 4) 88.4% of the participants showed a degree larger than 65% while the remaining 11.6% showed a disability between 33 and 65%. In 28.7% of cases the person with ID themselves provided the information and data about their needs and requirements (self-application), because their verbal ability (i.e. understanding and verbal expression) were adequate to respond to interview items (the participants mostly showed basic skills in reading and writing). The remaining 71.3% of responses to interviews were given by families (18.7%) or professionals who knew the person with ID well (52.6%). In these cases, the requirements for responding the interview were: (a) knowing the participant for at least six months and (b) having sufficient opportunity to observe the participant for long periods of time along several situations.

The structured interview was created with a combination of 128 items allowing the evaluation of the following areas: perceived needs (split into six categories: personal health, economic resources, health resources, social services resources, physical and social obstacles, support measures and networks), personal problems, future outlook, institutional measures and support, and solutions needed. These items were developed by independent experts in ageing and disability according to recent recommendations in the psychometric literature (Moreno, Martínez, & Muñiz, 2015).

A study was carried out in order to refine the initial items and obtain evidence of content validity with the participation of a panel of 20 experts with significant experience in disability in both practical and academic fields. The study was performed in two phases: firstly, to evaluate suitability, all of the items were presented for the experts to assign each one to the area they thought relevant. In the second phase, the same experts judged the relevance of each item in terms of evaluating the area it had been assigned to.

Once the instrument met the psychometric requirements, contact was made with organizations that provide services to people with disability in the various communities in Spain, explaining the aims of the research and requesting participation. The research team communicated with those organizations who showed an interest in participating. The Ethics Committee of the hosting University granted ethical approval. The principles for conducting research contained in the Declaration of Helsinki were respected. Permissions and informed consents were collected by participating organizations or provided by family or professionals. Removing names and using identifying codes that were not known by the researchers guaranteed confidentiality and anonymity.

Validation of the instrument was accomplished by analysing the results of the expert panel, removing those items which failed to achieve agreement from two thirds of the judges (i.e., 14 similar responses) on the appropriateness of their assignment to an area of evaluation. Then, the Content Validity Coefficient was calculated (CVC), which considers the number of participating experts, reducing possible bias in the evaluation (Pedrosa, Súarez & García-Cueto, 2013). As an index of double validation, Aiken's V coefficient was calculated (Aiken, 1980), which allows the estimation of confidence intervals independently of the shape of the underlying distribution. In addition, owing to the low variability in the experts’ responses, the cosine between the scores the judges awarded for items belonging to different areas was calculated in order to analyse consensus among their opinions.

Once the final instrument had been defined (https://goo.gl/Cb0BJI), multidimensional scaling was used to graphically analyse the proximity and relationships as perceived by the participants between the various evaluated areas (Carroll, 1980). The graphical solution was limited to two dimensions for simplicity. The procedure locates each category in each of two dimensions. Large distances between categories in this two-dimensional space indicate dissimilarity between the categories (Borg & Groenen, 2005). This procedure was applied to the data belonging to the perceived needs area, using the Proxscal method to create the data proximities. The goodness of fit to the model was tested using the Stress index (Kruskal, 1964).

Finally, the fit of the data to the normal distribution was tested using the Shapiro-Wilk test. A study was performed of the differences of means for each of the areas making up the interview in terms of predictive variables: sex, age, habitual residence, and assessment method. The total score of each participant was obtained based on the number of perceived needs in each area. Owing to the violation of the assumption of normality, this study of differences of means was performed using the Mann-Whitney U statistic, the effect size was estimated using Cliff's delta (Cliff, 1993).

Items were removed from the original instrument if there was a lack of agreement on their relevance to an area. In addition, the area institutional measures and support was removed as the items in this area were mostly classified by the experts as relevant to the solutions needed area. Following that, items which had a CVC lower than .80 were eliminated, leaving a total of 93 items grouped into the four previously mentioned areas with a total CVC of .89. The areas with lower values of Aiken's V coefficient were solutions needed and social support networks. Both gave a coefficient of .85 [.75-.91; CI=95%], which indicates high representativeness with respect to the content evaluated. In all other cases the coefficients were higher. Furthermore, the values of the cosine matrix between the experts’ scores of item relevance were above .93 in all cases.

Once the final version of the instrument was determined, the multidimensional scaling process was applied. The aim was to analyse the relationship between the different categories making up the area of needs as perceived by the participants and to see which are closest for them. The model gave a Stress value of<.001, indicating an excellent fit according to the criteria of Kruskal (1964).

As can be seen in Figure 1, dimension 1 defines a continuum going from those needs related to personal aspects and the immediate environment to those needs more associated with contextual, more external factors. Needs related to personal health, individual and family economic resources are on the left of the chart, more social aspects in the middle, and health resources on the right. Dimension 2 goes from needs related to service provision, such as economic and health resources, to those which are more relational, such as social services, social barriers, and social support measures and networks.

Figure 1.

Graphical solution of multidimensional scaling of the area of perceived needs.

Note. ER= Economic Resources; PH= Personal Health; HR= Health Resources; SSR=Social Services Resources; PSO= Physical and Social Obstacles; SMN= Support Measures and Networks.

(0.04MB).

The differences in means were statistically significant in practically all areas in terms of the predictive variables (Table 1). In general, men and those participants who use residential services tended to indicate more needs, problems and solutions. In terms of the way the interview was conducted, where participants themselves answered the interview questions, they related more needs, whereas in cases where the answers were given by third parties this tendency was reversed, with the proxies giving more solutions, future perspectives, and personal problems.

With respect to age, bearing in mind previous studies which indicate that as people age, their needs increase considerably (Navas et al., 2014; Ng et al., 2015), the participants were divided into two groups: those 54 and under, which grouped together adults in the early stages of ageing (67.70% of participants), and those over 54 (32.30%), in the later stages of the ageing process. The results show that the older group tended to indicate more needs, except in personal health and health services, where the pattern is reversed. Despite this, statistically significant differences were only observed in personal health, social support measures and networks, and solutions required. While bearing in mind the differences, it is worth underlining that the effect size is small in almost all cases.