Statistical Approach According to Structure of Variables

Autor: Zeki Akkuş, S.Yavuz Sanisoğlu, Mesut Akyol, M. Yusuf Çelik
Jazyk: angličtina
Rok vydání: 2006
Předmět:
Zdroj: Dicle Medical Journal, Vol 33, Iss 2, Pp 101-104 (2006)
ISSN: 1308-9889
1300-2945
Popis: Variable structures are the essential factors to be considered in the first stage of the statistical analysis. If they do not take the stepwise analysis into consideration, researchers make publications full of many errors. Variable structures determine the statistical methods to be chosen. Researchers with limited knowledge about variable structures are apt to have troubles in making decisions about variables.In this study, we aimed to investigate the strong relation between variable structures and statistical methods, those assumed to be the strongest assumptions in statistics. Although researchers deal with either univariate or multivariate structures in their studies, today, it is the reality that the multivariate effects are observed more frequently. All of the univariate or multivariate biostatistical methods propose assumptions related to variables. The basic assumption of the parametric tests is the normal distribution of the variables. The most important characteristic of the normal distribution is being a distribution of the continuous variables. Univariate or multivariate parametric methods need continuous variables. Since they are not normally distributed, non-continuous variables should be treated with “distribution-free methods.”The most important characteristic of the non-parametric methods is that they have no limitation for the researchers. This characteristic provides easiness for the researchers. As these advantages are used correctly, it is obvious that the results of the analysis will be more accurate and stronger.Before the analysis, explicitly definition of the variable structures provides us getting the correct results in the method selection.
Databáze: OpenAIRE
Externí odkaz: