For the analysis of construct validity, every source of information was submitted into a confirmatory factor analysis (CFA), which considered the existence of a measurement model with eight first-order factors and three second-order factors. Given the ordinal characteristics of the items on the scale, we estimated polychroric correlation coefficients and used a Maximum Likelihood estimation (MLE), which assumed the multinormality of observed distributions and the factor scores were distributed with a normal model of μ=0 and σ2=1. The model that had a better adjustment was the one that was related to the reports by professionals, although the other two models did not have remarkably worse adjustments. These are normal results for confirmatory models, which supports the acceptance of this proposed model as being the structure of the construct. The adjustment results for the three measurement models can be seen in Table 2. Additionally, this table shows standardized factor loading values that were associated with every factor and source of information, with all of them being statistically significant (p<.001). The values that were obtained guarantee significant factor loadings of every item on the theoretically assigned factor.

Also, we obtained the final summations for every factor and source of information. The distributions that were observed for every factor are shown in Table 3 for first- and second-order factors.

In order to estimate the reliability in terms of internal consistency, Cronbach's α was used for every factor and source of information. Naturally, in this case, the results of the values should be considered as merely descriptive, since the POS is hetero-administered for all the informants and it would therefore not be strictly an internal consistency estimation. Results from this analysis are shown in Table 4.

The values that resulted from this analysis correspond to appropriate values for most of the factors, particularly for the second-order factors.

Finally, Pearson's correlation values were estimated from the values of every factor (first- and second-order), and these values were considered to be continuous functions. Results shown in Tables 5 and 6 point to there being a degree of dependence among the factors, which is similar to what the model proposed.

In order to estimate the convergent validity between the POS and Gencat, Pearson's correlations were obtained between the direct values of the different factors that were defined in both scales in order to verify their concurrence according to the general criteria (Izquierdo, Olea, & Abad, 2014). Global index values were ignored because they were not the object of this analysis. Table 7 shows the correlation values and distinguishes between the different POS sources by providing the Bonferroni correction for their significance in order to reduce the family wise error rate.

In this study, the psychometric properties for the three informers of the POS were examined. This measurement instrument assesses QoL in adults with ID in terms of personal outcomes. The reliability study found a proper internal consistency for the items for every factor and respondent. The values for the second-order factors are slightly higher than those in the original scale, particularly for the Independence factor, both in the self-report and in the direct observation (Claes, Vandevelde et al., 2012).

In regard to the validity study, construct validity and convergent validity were examined. In regard to the CFA, we can assume that the model that adjusts better to the construct is that of the professionals’ reports, although the models for people with ID and family members are also considered to be appropriate. The results were consistent with previous studies in regard to the multi-dimensionality of the QoL construct (Jenaro et al., 2005; Schalock et al., 2005) and the presence of three second-order factors (Wang et al., 2010).

In regard to convergent validity, as previously explained, the Gencat Scale was used because it was also used in the validation of the first original POS version. However, although the Gencat assesses QoL from the professional's objective perspective, first-order factors were analyzed for the three kinds of respondents. Results showed an acceptable correlation with the measurement criterion that this adapted scale intends to evaluate.

Regarding this psychometric study, the authors believe that two points need to be highlighted. Firstly, as seen in the reliability and validity results for the self-report, we have a good measurement instrument that allows for the direct participation of the individual with ID in the assessment of their QoL. These results were obtained with appropriate representation from people with ID due to the instructions for the proper application of the scale were strictly followed and the guidelines for the assessment of QoL were considered (Claes et al., 2010). Therefore, we agree with Claes, Vandevelde et al. (2012) that if the people with ID can speak for themselves, the self-report should be directly answered by them. If the individual with ID has communication problems, this part being answered by a proxy (professional or family member) cannot be justified because it was not created for this purpose. Moreover, we have to consider that, in relation to the degree of agreement among the different POS respondents, the correlations between self-reports and the information from the other respondents may be low (Claes, Vandevelde et al., 2012). Therefore, in those situations when self-reports cannot be assessed, applying the report of others is proposed, which has satisfactory psychometric properties, both for professional and family versions.

Secondly, we assert that we have an instrument with a sufficient amount of psychometric qualities, for the self-report and for the report of professional and family member. It allows us to assess QoL from subjective and objective perspectives that are proposed in relevant literature in this field (Schalock & Verdugo, 2002; Schalock et al., 2007). Thus, the organizations in Spain that rely on this QoL model and work for the development of ISP have a valid and reliable instrument at their disposal, which will allow them to assess the impact of these plans on the expected personal outcomes.

This study has a few limitations. Firstly, the organizations that participated from the different autonomous communities are not at the same levels in terms of their knowledge and application of Schalock & Verdugo's (2002) QoL model and their guidelines that are defined in relation to the support paradigm (Schalock et al., 2008; Thompson et al., 2009). Thus, although some training was carried out in regard to the theoretical background and administration of the POS, the degree of comprehension of the items may have been influenced the study, depending on the organization's degree of involvement in using the QoL model and their commitment to promoting social inclusion. A second limitation lies in the fact that most of the participants with ID live with their families, and there is a low presence of those who live in supervised homes or independent homes. This situation is due to the fact that for this study, we needed the participation of the family, and indeed, the organizations have had more access to those parents who live with the individual with ID. In some cases, when the individual with ID lived in a supervised home, the legal tutor was considered to be a family member because their role was similar to that of a family member. Nevertheless, this was not the case for all of the cases because these cases were specific to participants who did not have a family or had parents who were rather old. In regard to the last limitation, we highlight the low participation of people with severe and/or profound ID due to their limitations in comprehension and communication. It is true that the administration of this scale through an interview has increased the participation of people with high comprehension difficulties; however, the representation of such people was low in our study. Otherwise, the use of the Classical Theory of Test can be a methodological and technical limitation in order to obtain more adjusted results in terms of generalizability so, for this type of instrument, it may be an interesting approach according to Rupp (2013) propositions.

Likewise, the results of this study provide new ways and opportunities for future research. Firstly, the Spanish version of the POS can be used to examine the degree of agreement between the self-report and the information from professional and family and, in this sense, to advance the knowledge of the significance of QoL assessment (Balboni et al., 2013; Claes, Vandevelde et al., 2012; Gómez, Arias, Verdugo, & Navas, 2012). Secondly, some studies show that personal outcomes are influenced by specific support strategies, environmental factors and personal characteristics (Claes, Van Hove, Vandevelde, van Loon, & Schalock, 2012). Thus, the Spanish POS can be used to assess the impact of support strategies that are provided by services such as was done with the original version (Claes, Van Hove et al., 2012) and to adopt improvement measures (van Loon et al., 2013). Finally, it is becoming more necessary to know the relationship between QoL and certain personal conditions of disability. We have to consider whether the dimensions, subdimensions, and indicators that are proposed in an evaluation system are equally relevant for all of the people with ID or whether there should be different or more detailed specifications (Petry, Maes, & Vlaskamp, 2009; Verdugo, Gómez, Arias, Navas, & Schalock, 2014). It is important to note that a more thorough analysis is needed of the convergence among sources of information and, even more, the divergence among them. That divergence may be very informative in evaluation and systematization processes when using the Spanish version of the POS. Some proposals and indexes have been put forward for this type of situation (Carbó-Carreté, Guàrdia-Olmos, & Giné, 2015) which we should also apply to this scale as future complementary analysis research.

This research has been funded by the Ministerio de Educación y Ciencia de España (DEP2005-00202-C04-03/ACTI).