OBJECTIVETo analyze the direct and indirect outcomes of influencing factors on smoking behaviors among adolescents.

METHODSSelf-administered questionnaires were used to collect smoking-related information from 2021 respondents. Measurement models were built by confirmatory factor analysis and parameters were estimated by ERLS method. The final structural equation model was determined by comprehensive evaluation and necessary modification.

RESULTSThree latent variables were extracted from 10 manifest variables of environment, while only one latent variable was identified from 9 manifest variables of attitudes. The goodness of fit for the structural equation model was satisfactory that all indices had met corresponding requirements. The final model could explain 38.8% of the variance of smoking behaviors. Four factors (smoking environment, smoking restriction from parents and teachers, determination of cigarette refusal and attitudes toward smoking) were directly affecting the smoking behaviors, while another three factors (grade, health knowledge and school environment) had indirect impacts. According to the percentages of their contribution, the risk factors were ranked as follows: smoking environment (45.76%), attitudes toward smoking (19.88%) and grade at school (0.44%). Similarly, the top protective factor were: determination of cigarette refusal (16.61%), followed by smoking restriction from parents and teachers (10.51%), health knowledge (3.89%) and school environment (2.92%).

CONCLUSIONSHeath knowledge had minor effect on smoking in adolescents but could indirectly affect their smoking behaviors through changing their belief. Grade at school had a doubled influence on smoking, but mainly served as a risk factor. Tobacco control measures for adolescents should not only be limited to health education but environment factors as well.