Rasch analysis applied to a questionnaire on tobacco use among adolescent students

Abstract

AIMS:  To perform the psychometric analysis of a questionnaire on tobacco exposure in schoolers, through Rasch analysis.

METHODS: A cross-sectional study with students between 12 and 15 years was held in the city of Uruguaiana, State of Rio Grande do Sul. Data collection was performed by random sample and the information was obtained by a questionnaire composed of 14 questions adapted from a version in Portuguese of the Global Youth Tobacco Survey (GYTS) questionnaire. Data analysis was performed using psychometric analysis by the Rasch method, a mathematical model that seeks to analyze the probability of a particular item adjustment, depending only on the subject's ability and the item difficulty.

RESULTS: The study consisted of 956 students, which had a mean tobacco exposure of 5.3 (±2.2) logits, average - 0.66 (±0.99) logits, 0.70 logits error, dimensionality internal 0.99 (±0.39) logits and external dimensionality of 1.03 (±0.81) logits. With regard to the questionnaire items, they had a mean of 0.00 logits (± 1.38), 0.09 logits error, internal dimensionality of 0.99 (± 012) logits, external dimensionality of 1.03 (± 0, 30) logits, separation rate of items of 15.62 logits, and 51% reliability.

CONCLUSIONS: The use of  Rasch method enabled us to verify the exposure of each participant to the different tobacco exposure conditions, and showed that the questionnaire that measures tobacco exposure should deal each question with different weights.