CONSTRUÇÃO DE INSTRUMENTO DE AVALIAÇÃO DO CONHECIMENTO SOBRE O CONTEÚDO DE UM GUIA ALIMENTAR PARA CRIANÇAS DE 2 A 3 ANOS DE IDADE

Abstract

The aim of this study was to build an instrument to assess the knowledge of parents responsible for the home feeding of children (2–3 years old) on the content of a dietary guideline for children in this age group. The target population was of parents responsible for children home feeding (rchf) institutionalized in municipal daycare centers of Rio de Janeiro city. This is quali-quantitative cross-sectional study with the construction of four prototypes. Prototype 1 consisted of a database with 70 items applied to rchf of a Basic Health Unit. An inferential analysis of the statements of five focus groups (n=46 rchf), refinement techniques of items in the semantic perspective and from the view of endorsement tax allowed to create the Prototype 2 (20 items). Those items were submitted to six expert judges who were able to sort items by dimensions and refine them to define the Prototype 3 (17 items). A pretest was performed (n=60 rchf) with prototype 3. This allowed the removal of three items according to the subjective analysis of the theoretical postulates of dietary conducted by the researcher bringing the prototype 4 (14 items). This was subjected to an exploratory factor analysis, in which used KMO criterion that confirmed the possibility of factorization of matrix scores observed. The multivariate test indicated distribution of the scores of the items tending to the normal distribution. The main factor estimator showed the Kaiser-Guttman criterion, the retention of two factors (factor 1 and factor 2 = 3.7514 = 1.3963). The Scree plot graph pointed the two-dimensional, but showed the possibility of another factor, since the slope of the curve. Two simulations with 2 and 3 factors, using the estimator main factor and oblique Promax rotation presented a correlation between f1-f2 = 0.5764 for the two-dimensional model and the f1-f2 = - 0.5764; f1-f3 = 0.0040 for the three-dimensional model. The factor loadings were more parsimonious for the two-dimensional model, although with items presenting above error variance of 0.60. The Cronbach alpha coefficient for the two-dimensional model, assumed point estimate of 0.7724 (95% CI 0.7327513 - 0,805745) for dimension 1 and 0.7798 (95% CI 0.7098758 – 0.820341) for dimension 2. Even though this is an exploratory study, the scale with two factors (factor 1 with eight items and factor 2 with six items) may empirically represent the knowledge of those responsible for healthy feeding of institutionalized children of 2-3 years old. This achieves (results) may assist better nutritional practice and encourage further studies for its replication. DOI: http://dx.doi.org/10.12957/demetra.2014.14068