Dear Editor:
We have intently read the article “Reliability and validity of the EASYCare-2010 Standard to assess elderly people in Portuguese Primary Health Care”.1 It is a very important study for the health field. However, it presents some methodological weaknesses in the instrument validation process which could affect the results of future research. Although there is no rigid guide for the process of construction and validation of an instrument, there are basic criteria that come from the psychological and educational sphere2 that should be considered in the instrument validation process to guarantee their results.
First of all, it is widely accepted that large samples are needed for quantitative studies in order to reduce the probability of error of the parameter and that the number of questions should be considered when validating an instrument, with a common recommendation that there should be 5–20 participants for each question or simply no less than 400 participants.3 Apparently in the study these criteria were not considered, as there are a lot of questions and only 245 participants. We believe that due to the lack of evidence for instrument reliability in other countries, a larger number of participants is necessary.
Second, factorial analysis (FA) is presented in the article. However, the results of the analysis of sample adequacy measurements are not reported; of Kaiser–Meyer–Olkin (KMO), whose value should be higher than 0.80 and of the Barlett's test of sphericity, whose significance should be less than 0.05. These are necessary procedures for the extraction of factors. In addition, it is mentioned that the instrument has two components that explain only 40.9% of the variance, which is inadequate because it is recommended that the explained variance be greater than 50%.4
Third, we believe that the use of correlations to evaluate the construct validity of an instrument is insufficient and that the exploratory factor analysis (EFA) does not definitively determine the number of factors because when analyzing the metric properties of the instrument in a new sample the questions may present a different behavior. A confirmatory factor analysis is required, which is a more rigorous statistical analysis procedure, where absolute and incremental indices are considered, such as; the standardized root mean residual (SRMR), normed fit index (NFI), non-normed fit index (NNFI), comparative fit index (CFI) and root mean square error of approximation (RMSEA). These serve to analyze instrument construct validity.5
Finally, it is reported that Cronbach's alpha was used to evaluate the reliability of the instrument, but its use is not specified for the total or partial instrument. For an instrument with polytomic and dictomic questions, the techniques of Cronbach's Alpha and the Kuder–Richardson Formula 20 (KR-20), respectively, should be used.6
We emphasize the importance of EASYCare-2010 to detect the needs of care and welfare for adults. In addition, we believe that it is an instrument whose metric characteristics should continue to be evaluated in different populations, under rigorous and appropriate statistical procedures to demonstrate its validity and reliability and guarantee measurements in future research projects.
References
- 1.Piedade M., Martins L., Philp L., Fonseca M. Reliability and validity of the EASYCare-2010 Standard to assess elderly people in Portuguese Primary Health Care. Ate Primaria. 2017;40:576–585. doi: 10.1016/j.aprim.2016.11.016. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Hogan T.P. 2nd ed. El Manual Moderno; México: 2015. Pruebas psicológicas: Una introducción práctica. [Google Scholar]
- 3.Campo-Arias A., Oviedo H.C. Propiedades psicométricas de una escala: la consistencia interna. Rev Salud Pública. 2008;10:831–839. doi: 10.1590/s0124-00642008000500015. [DOI] [PubMed] [Google Scholar]
- 4.Ferrando P.J., Anguiano-Carrasco C. El análisis factorial como técnica de investigación en psicología. Pap Psicólogo. 2010;31:18–33. [Google Scholar]
- 5.Kline R. 4th ed. Guilford Publications; New York: 2016. Principles and practice of structural equation modelling. [Google Scholar]
- 6.da Silva F.C., Gonçalves E., Arancibia B.A., Bento G., Castro T.L., Hernandez S.S. Estimadores de consistencia interna en las investigaciones en salud: el uso del coeficiente alfa. Rev Peru Med Exp Salud Publica. 2015;32:129–138. [PubMed] [Google Scholar]