MULTIVARIATE DATA ANALYSIS AS A TOOL FOR CHARACTERIZE CHILDREN AND FAMILY DIET HABITS TYPOLOGY IN AN INDUSTRIAL AND RURAL AREA: PRELIMINARY RESULTS

The installation of the Estarreja Chemical Complex (ECC) in the middle of the last century provided the settlement of people and the development of relevant residential areas in the municipality. This is a county that has been changing over the past decades, from a mainly rural county to one with more industrial characteristics. However, despite industrial growth, in some localities agricultural practices and cattle raising still remain, particularly for self-consumption.

Although ECC was extremely important for county growth in terms of creating employment, changes were also felt in the environment through pollution of soil, water and air. This last point is of utmost importance in terms of public health, mainly because we are dealing with a city that lives with two realities: urban and rural living side by side.

In order to evaluate lifestyle and consumption habits of the population, questionnaires were applied to residents through children with ages varying from 6 to 15.

The questionnaire contains 3 blocks of questions: one related to student characterization (Block A, referring to sex, age, development, health status, etc.), other directed household characterization (Block B, referring to socio-economic characteristics of the families), and the third related with alimentary habits (Block C, referring to type of meat, fish, vegetables, etc., and their origin). After the conventional treatment of the questionnaire using histograms and cross-tabulations, these preliminary results were combined with a multivariate statistics methodology, in order to establish quantified relationships between the two sets of qualitative variables. Such a methodology relies on the ‘supplementary projection’ of Block B and C variables onto the axes provided by the eigenvalue decomposition of Block A matrix, using a Correspondence Analysis type algorithm. The ‘supplementary projection’ produces a set of quantified relationships between Blocks B and C modalities and the factorial axes given by Block A variables (which resume their optimum combination, according to a qui-square distance).

As a result of the joint interpretation of the two procedures, some preliminary typologies have emerged. These results will be combined to urine analysis performed simultaneously, both to children and their mothers.