Life-Styles of Spanish School-Aged Children: Their Evolution over Time. Use of Multiple Correspondence Analysis to Determine Overall Trends over Time in a Sequential, CrossSectional Study

In using usual univariate and bivariate analyses of a population’s life
styles researchers will encounter major limitations because these
phenomena are essentially complex. This article uses multivariate statistical
analysis, specifically Multiple Correspondence Analysis, with a two-fold
purpose. First of all, it aims to analyse the general evolution of life styles in
11,13 and 15-year old Spanish schoolchildren between 1986 and 1994.
Secondly, it evaluates the usefulness of MCA in studying this evolution.
The study uses the data collected within the framework of the Spanish
study “Conductas de los escolares relacionadas con la salud”, which is part
of the study “Health Behaviour in School-Age Children”, a WHO CrossNational Study. These data stem from three surveys of Spanish schoolchildren in 1986, 1990 and 1994. All three surveys used a similar sampling method (random, proportional, stratified and multi-stage) and data collection procedure (a questionnaire filled in anonymously by the children in their classrooms). The analysis was based on 57 variables which
were common to all three surveys. Forty-seven of these variables are
related to various features of life styles, adaptation to school and integration
in the family circle. The remaining ten are socio-demographic. The resulting
matrix included data on these 57 variables collected from 10,603 subjects.
Research findings suggest that 11 and 13 year old Spanish schoolchildren’s life styles have evolved positively over the years studied.
However, there was scarcely any positive evolution in 15 years olds’ life
styles. Indeed, 15 year old girls apparently registered no improvement at all.
In an attempt to explain this, the article explores several hypotheses in
detail.
MCA proved very useful in obtaining a global view of the evolution of a
population’s life styles. Using a procedure similar to the one described in
this paper, MCA allows researchers to simultaneously summarise evolution
over time on the basis of the two most discriminatory dimensions.
Nevertheless, the MCA technique has some fairly significant limitations: it
only provides an overall view of evolution over time, does not go into
details about particular variables and it is affected by the sizes of the
successive samples or sub-samples. These problems are easy to control,
which makes the technique very useful when studying the evolution over
time of multi-variable phenomena such as life styles.