RESUMO

BACKGROUND: Given the relevance of social support on the mental health of older adults, having an instrument to evaluate this variable is essential for research in the area. However, mainly, having instruments with suitable evidence of their psychometric properties is critical. For this reason, this study sought to evaluate the factorial and reliability structure of the Multidimensional Scale of Perceived Social Support within autonomous older adults from the Province of Concepción, Chile. METHODS: We surveyed 399 older adults using quote sampling. They answered a Spanish version of the Multidimensional Scale of Perceived Social Support, plus a sociodemographic questionnaire. We performed confirmatory factor analysis using Weighted Least Squares Means and Variances adjusted estimation (WLSMV) to compare the factor models proposes by previous studies. To evaluate reliability, we calculated Cronbach's alpha and McDonald's omega. RESULTS: The Confirmatory factor analysis found that the 3-factors models showed the best fist index between the models with CFI = 0.991, TLI = 0.989, and SRMR = 0.035, even though RMSEA were over the cutoff point. The factors presented reliabilities from α = 0.858 to α = 0.941, and from ω = 0.937 to ω = 0.972. CONCLUSIONS: The results support the existence of three factors for the Multidimensional Scale of Perceived Social Support (MSPSS), differentiating the support perceived from Family, Friends, and significant others. All factors present good or excellent reliability. This solution is theoretically consistent and coherent with the literature, and it presents evidence in favor of the use of MSPSS as a measurement to distinguish the support perceived source.

Assuntos

RESUMO

A two-week-long intervention study was performed in two classrooms in an elementary school in Costa Rica. Split-cooling air-conditioning (AC) units were installed in both classrooms. During the first week, the air temperature was reduced in one classroom while in the other (placebo) classroom the fans were operated but no cooling was provided. During the second week, the conditions were exchanged to create a 2 × 2 crossover design in which each pupil was their own control. A total of 37 children performed tasks similar to school work and completed questionnaires reporting their thermal sensation and perceptions. Operating the AC units reduced classroom temperature by about 5 K, from about 30 to 25°C. Thermal sensations changed from hot to neutral and slightly cold, and the percentage of children rating the thermal conditions as acceptable increased significantly. Neutral temperature was estimated to be about 27°C. The 11-year-old children performed the language and logical-thinking tasks significantly better in terms of speed at the lower temperature, while the less able pupils performed better on all tasks at the lower temperature. There were no significant effects on accuracy. These results confirm published findings from moderate climates and extend their validity to the tropics. They indicate that acclimatization can increase the optimal temperature for learning.

Assuntos

RESUMO

[RESUMEN]. Objetivos. Exponer las ventajas de la utilización del alfa ordinal para situaciones en que no se cumplan los supuestos del alfa de Cronbach y mostrar la utilidad del alfa ordinal con la versión chilena del AUDIT, así como proporcionar los comandos en el lenguaje de programación R para realizar los cálculos respectivos. Métodos. Se compararon el alfa de Cronbach y el alfa ordinal en el caso del cuestionario AUDIT aplicado a 419 estudiantes de una universidad chilena. Resultados. En todos los dominios, el alfa ordinal resultó ser mayor que el alfa de Cronbach, lo que es congruente con lo encontrado en la bibliografía. Al considerar el alfa ordinal se acepta la confiabilidad de cada uno de los dominios y de la encuesta completa, lo que no ocurre completamente al trabajar con el alfa de Cronbach. Conclusiones. Para tomar decisiones sobre la confiabilidad de una dimensión se debe considerar un índice que cumpla los supuestos necesarios para su utilización, ya que no hacerlo puede desembocar en conclusiones erróneas y afectar de manera notable el desarrollo de una investigación. Los valores del alfa de Cronbach subestiman la confiabilidad de un instrumento al trabajar con escalas ordinales.

[ABSTRACT]. Objectives. Demonstrate the advantages of using ordinal alpha when the assumptions for Cronbach’s alpha are not met; show the usefulness of ordinal alpha with the Chilean version of the WHO Alcohol Use Disorders Identification Test (AUDIT); and provide the commands in R programming language for performing the respective calculations. Methods. Cronbach’s alpha and ordinal alpha were compared in an AUDIT questionnaire administered to 419 students at a Chilean university. Results. In all domains, the result using ordinal alpha was greater than with Cronbach’s alpha. This finding was consistent with the literature on the subject. When ordinal alpha was chosen, the reliability of each domain and the survey as a whole was accepted, which does not always happen with Cronbach’s alpha. Conclusions. In making decisions about the reliability of a dimension, consideration should be given to an index that meets the required assumptions for using it in order to avoid erroneous conclusions that can have an appreciable effect on research. The values of Cronbach’s alpha underestimate the reliability of an instrument when working with ordinal scales.

[RESUMO]. Objetivos. Apresentar as vantagens do uso do alfa ordinal nos casos em que não são cumpridos os pressupostos do alfa de Cronbach e demonstrar a utilidade do alfa ordinal com a versão chilena do instrumento AUDIT e criar comandos na linguagem de programação R para realizar os respectivos cálculos. Métodos. Foi realizada a comparação entre o alfa de Cronbach e o alfa ordinal com o uso do instrumento AUDIT aplicado a 419 alunos de uma universidade chilena. Resultados. Em todos os domínios, o alfa ordinal foi maior que o alfa de Cronbach, o que é condizente com o descrito na literatura. Ao considerar o alfa ordinal, aceita-se a confiabilidade de cada um dos domínios e da pesquisa completa, o que não ocorre completamente quando se usa o alfa de Cronbach. Conclusões. Para tomar decisões sobre a confiabilidade de uma dimensão, deve-se considerar um índice que cumpra os pressupostos necessários para seu uso, porque do contrário pode-se chegar a conclusões errôneas e afetar consideravelmente o desenvolvimento de uma pesquisa. Os valores do alfa de Cronbach subestimam a confiabilidade do instrumento por se basear em escalas ordinais.

Assuntos