# Family functioning in students of health sciences in four Latin American countries: a study of the structure and factorial invariance of the FACES III scale. A cross-sectional study

## TL;DR

This study confirms the reliability of the FACES III scale for measuring family functioning among health science students in three Latin American countries, but not in Peru.

## Contribution

The study provides evidence of strict factorial invariance of the FACES III scale's bi-factor model across three Latin American countries.

## Key findings

- The bi-factor model of the FACES III scale fits well in Colombia, Chile, and Mexico but not in Peru.
- The bi-factor model showed strict factorial invariance across Colombia, Chile, and Mexico.
- The model demonstrated adequate reliability in the three countries where it fit.

## Abstract

Psychometric studies of the FACES III scale in Spanish-speaking countries show a lack of agreement on the factorial structure of the scale. In addition, most of the studies have only performed exploratory analyses of its factorial structure.

The objective of the present study was to confirm the structure and factorial invariance of the FACES III scale in nursing and obstetric students from Chile, Colombia, Peru, and Mexico.

A total of 3303 students from the four countries participated in this study (Colombia = 1559, Chile = 1224, Peru = 215, Mexico = 305).

The results of the study showed that the Bi-factor model presents the best-fit indexes to the data from Colombia, Chile, and Mexico, but not from Peru. In addition, it was found that this model showed evidence of being strictly invariant among the three countries in the sequence of the invariance models proposed: metric invariance (ΔRMSEA = .000), scalar (ΔRMSEA = .008), and strict (ΔRMSEA = .008). The bi-factor model also showed adequate reliability indexes in the three countries.

It is concluded that the FACES III scale shows adequate psychometric performance under a bi-factor model in nursing and obstetric students from Colombia, Chile, and Mexico. The lack of fit of the model in Peru could be associated with the small sample size.