Identifying predictors of violent behaviour among students using the conventional logistic and multilevel logistic models

Bidemi Yusuf, Olayinka Omigbodun, Babatunde Adedokun and Odunayo Akinyemi

Journal of Applied Statistics, 2011, vol. 38, issue 5, 1055-1061

Abstract: Analysing individual-, school- and class-level observations is a good and efficient approach in epidemiologic research. Using data on violent behaviour among secondary school students we compared results from the conventional logistic modelling with multilevel logistic modelling approach using the gllamm command in Stata. We illustrated the advantage of multilevel modelling over the conventional logistic modelling through an example of data from violence experience among secondary school students. We constructed a logistic model with a random intercept on the school and class levels to account for unexplained heterogeneity between schools and classes. In the multilevel model, we estimated that, in an average school, the odds of experiencing violence are 3 (OR=2.99, 95% CI: 1.86, 4.81, p >0.0001) times higher for students who use drugs as opposed to the odds of experiencing violence for students who do not use drugs. However, the estimates in the conventional logistic model are slightly lower. We estimated that a normally distributed random intercept for schools and classes that accounts for any unexplained heterogeneity between schools and classes has variances 0.017 and 0.035, respectively. We therefore recommend the multilevel logistic modelling when data are clustered.

Date: 2011
References: View complete reference list from CitEc
Citations:

Downloads: (external link)
http://hdl.handle.net/10.1080/02664761003759008 (text/html)
Access to full text is restricted to subscribers.

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:taf:japsta:v:38:y:2011:i:5:p:1055-1061

Ordering information: This journal article can be ordered from
http://www.tandfonline.com/pricing/journal/CJAS20

DOI: 10.1080/02664761003759008

Access Statistics for this article

Journal of Applied Statistics is currently edited by Robert Aykroyd

More articles in Journal of Applied Statistics from Taylor & Francis Journals
Bibliographic data for series maintained by Chris Longhurst ().