Multivariate statistical analysis: a brief introduction.

* Associate Professor Department of Community, Occupational and Family Medicine National University of Singapore Address for Reprints: Dr Chia Kee Seng, Department of Community, Occupational and Family Medicine, National University of Singapore, Lower Kent Ridge Road, Singapore 119260. Introduction In modern medical research, it is rare to limit data analysis to merely two variables: a single exposure (independent) variable and a single outcome (dependent) variable. Most outcomes are multifactorial and a string of exposure variables are needed to explore their relationships with one another and with the outcome variable. A cursory search of articles in the electronic BMJ from January 1996 to December 1998 yielded 2425 hits for the phrase “multivariate analysis”. Multivariate statistical analysis is designed to handle many independent variables (IVs) and several dependent variables (DVs) all interrelated with one another to a certain degree. For example, several tumour markers were monitored among patients undergoing two different chemotherapy regimes. Several other IVs such as age, gender, ethnicity and stage of disease also affect the tumour marker levels. Multivariate analysis can assist in answering questions like: 1. Which tumour marker show the greatest change after taking into account the various IVs? 2. How are the various tumour markers related to one another after taking into account the various IVs? 3. How do the IVs relate to one another?