Contribution of body dissatisfaction and demographic and academic variables in the eating behavior of Brazilian and Portuguese university students

Author(s)

Moema de Souza Santana

• Document type: Master's Dissertation
• Press: Araraquara. 2016-10-05.
• Institution: Universidade Estadual Paulista (Unesp). Faculdade de Ciências Farmacêuticas. Araraquara
• Defense date: 2016-09-23
• Advisor: Juliana Alvares Duarte Bonini Campos

Abstract

Aims: To translate The Three Factor Eating Questionnaire (TFEQ) to Portuguese language and to adapt it according to The Portuguese Language Orthographic Agreement. To evaluate the psychometric properties of complete and 18-item version of TFEQ in Brazilian and Portuguese students. To identify the contribution of the dissatisfaction with body size and of variables demographic and academic to the eating behavior. Methods: TFEQ was translated into Portuguese language by three bilingual native speakers from Brazil. These three versions were combined into a single version and submitted to adaptation by Portuguese and Mozambican researchers to obtain a culturally appropriate instrument for the three countries. The final version was back-translated to English by another two translators and compared with the original version to examine their equivalence. Further, a pilot study was conducted to verify the Incomprehension Index (II) and Feasibility of the instrument. To assess the validity of content, 8 experts in psychology and in eating behavior evaluated each item of TFEQ according to its essentiality. The Content Validity Ratio (CVR) was calculated and considered adequate if ≥ 0.75. The adapted Portuguese version of TFEQ was completed by 1,335 Brazilian and 1,216 Portuguese students. Additionally, it was conducted a Confirmatory Factor Analysis (CFA) using the indexes Chi-Square by Degrees of Freedom Ratio ( 2 /df), Comparative Fit Index (CFI), Tucker-Lewis Index (TLI) and Root Mean Square Error of Approximation (RMSEA). A convergent validity (AVE) and discriminant validity (r 2 ) were also estimated. The reliability was evaluated by composite reliability (CR) and internal consistency (α) analyses. The invariance of independent and cross-cultural (between countries) samples was evaluated ( 2 ). The original complete (TFEQ with 51 items oblique - with factors correlated), short (TFEQ-18 - with factors correlated), unifactorial (TFEQ with 51 items – with factors assessed separately) and orthogonal versions (TFEQ with 51 items – with uncorrelated factors) were tested. After the best model was defined, it was made a structural model for each country separately, and the TFEQ-18 factors (Cognitive Restraint - CR, Emotional Eating - EE and Uncontrolled Eating - UE) were considered as dependent variables, while the dissatisfaction with body size and the variables demographics and academics were considered as independent variables. Finally, the Z-test was used to estimate the significance of trajectories (β) (α=5%). Results: As none of TFEQ item showed II ≥ 20%, the adapted Portuguese version was considered adequate. The Feasibility of the instrument was adequate in both Brazil (96.1%) and Portugal (91.1 to 95.6%), and either 30 items of the TFEQ and 11 of the TFEQ-18 were considered essential to assess eating behavior (CVR≥0.75). The TFEQ version obtained with oblique rotation presented a poor fit to the study samples while the TFEQ-18 and unifactorial models presented good fits. The instrument presented adequate convergent validity (AVE=0.50-0.76) and adequate reliability (CR=0.72-0.86 and α=0.72-0.89) for the fitted models in all samples. Among Portugueses and in the total sample, the discriminant validity in the TFEQ-51 oblique model was compromised by the factors Disinhibition and Hunger. The cross-national invariance was not verified. It was found a strong invariance in independent samples of the TFEQ models (Δχ2>0.05). Brazilian students had higher scores of EE and UE (β=0.159-0.181; p<0.001). The 12 variables gender, desire to reduce body size and use of medication to change body shape had a significant contribution to the CR and EE factors in Brazil and Portugal as well as the variables age, year of the course and the thought of giving up course for the UE factor. Significance of contribution of variables such as Social Economic Stratum, year of the course, performance in course and labour activity together with studies and Body Mass Index (BMI) for the factors of TFEQ- 18 differed between the studied countries. Conclusion: The cross-culturally adapted Portuguese version of TFEQ showed adequate face validity and content. The TFEQ models evaluated (oblique 51-item TFEQ, TFEQ-18 and unifactorial) showed adequate validity and reliability when applied to Brazilian and Portuguese students, and, it was not observed invariance among the countries. Furthermore, the Brazilian students had higher scores of EE and EU, and it was found a different contribution of the dissatisfaction with body size and of variables demographic, academic to the eating behavior of Brazilian and Portuguese students. (AU)

• FAPESP's process: 14/17249-4 - Contribution of body dissatisfaction and demographic variables and learning in the eating behavior of college students
• Grantee: Moema de Souza Santana
• Support Opportunities: Scholarships in Brazil - Master