Multivariate Data Analysis

Multivariate data analysis methods are complementary methods to sort out ideas or to put a new light on a problem, or to point out aspects which would not come out in a classical approach. Multivariate statistics help the researcher to summarize data and reduce the number of variables necessary to describe it.

Multivariate techniques can be further classified into two broad categories/situations i.e a) when researcher knows specifically about the dependent variable and independent variable and tries to assess relationship among dependent variable and independent variable such as in case of multiple regression, discriminate analysis, logistic regression and MANOVA etc. and b) situation when researcher doesn’t have any idea about the interdependency of variables and have large set of data; he tries to reduce the data by assessing a commonality among variable and tries to group variable/cases according to commonality such as factor analysis, cluster analysis and multidimensional scaling.

Further in situation where researcher have an idea about the interdependency of variable multivariate research statistics can be further classified based on nature of dependent variable i.e whether its metric or non-metric in nature (please refer diagram on next page). In latter case of data reduction technique also categorization depends on nature of data type i.e in case of metric data factor analysis, cluster analysis and metric MDS could be performed whereas in case of non-metric data non-metric MDS and conjoint analysis are preferred.