We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Logaritmik Doğrusal Modeller ve Uyum Analizinin Birlikte Kullanımı: Lise Öğrencilerinin Sigara İçme Alışkanlıklarını Etkileyen Faktörlerin Belirlenmesi.
- Authors
YILMAZ, Veysel; KAŞKIR KESİN, Fisun
- Abstract
Objective: Log-linear analysis (LLM), and correspondence analysis (CA), a statistical technique to facilitate interpretation of the relationship between categorical variables. Chi-square analysis in the analysis of the contingency table (CT) can become unusable especially if the number of categories belonging to row and column variables is large, interactions and co-changes can not be analyzed. For this reason, techniques such as CA and LLM are needed to reveal the interaction between variables. Since there is no significance test support in CA, it should be checked with an analysis like LLM. The aim of this study is to apply backward hierarchical LLM to the categorical variables such as gender, age, parental education status, school type, daily smoking amount, school achievement status of 662 high school students with smoking habit summarized by the multivariate CT, to identify significant interactions, to apply CA to the components of the most appropriate model so that the relationship between the subcategories of the relevant variables can be clearly seen and to graphically interpret the interactions of subcategories of components with each other. Material and Methods: In this context, in 2010, 662 high school students in the city center of Eskişehir, collected data on smoking habits. The variables are preceded by a backward stepwise LLM, and the CA is applied to the components of the resulting model. Conclusion: As a result, maximum three interactions backward stepwise LLM was statistically significant. Backward hierarchical LLM as a result of smoking in the next 5 years * success in school * cigarette is harmful to health and the amount of cigarettes smoked daily * whether or not the family is informed about smoking * School type triple interactions were found significant. Results: It has been demonstrated that LLM and CA are complementary methods of analysis.
- Publication
Turkiye Klinikleri Journal of Biostatistics, 2018, Vol 10, Issue 1, p65
- ISSN
1308-7894
- Publication type
Academic Journal
- DOI
10.5336/biostatic.2017-57095