   Last update: 03-Dec-2009 Arch Hellen Med, 26(5), September-October 2009, 699-711 APPLIED MEDICAL RESEARCH Statistical methods of data analysis P. GALANIS Center for Health Services, Management and Evaluation, Department of Nursing, University of Athens, Athens, Greece

Measures of association are the object of epidemiologic studies and are used to quantify the relation between the determinant under study and the outcome, while statistical significance tests are used to identify statistically significant relationships between these two variates. Statistical analysis of data is achieved using a variety of methods. Choice of the appropriate method depends on the scientific hypothesis and the kind of data. Data commonly are distinguished into qualitative (nominal and ordinal data) and quantitative (interval scale and ratio scale data). Con-cerning quantitative variates which follow a normal distribution, parametric methods should be applied. The inde-pendent samples t-test is used to determine whether two population means are significantly different, but for this to apply, the distribution of both populations must be normal and the two samples must not be related to each oth-er. One-way analysis of variance is employed to determine whether more than two population means are different. Paired samples t-test and repeated measures of one-way analysis of variance are the appropriate tests when the sam-ples (two or more than two respectively) are related to each other. Multivariate mathematical models are used to evaluate the effect of one or more characteristics while simultaneously controlling for possible confounding effects of other characteristics. In a multivariate model, the inclusion of several variates results in each term being uncon-founded by the other terms. In epidemiology, the most frequently used multivariate models are the multiple linear and the logistic regression models. The outcome (or dependent variate) in linear regression is a quantitative variate, while in logistic regression it is a dichotomous variate.

Key words: Data analysis, Linear regression, Logistic regression, Parametric methods, Survival analysis.