Statistical methods analyze how the outcome on the response variable depends on or is explained by the value of the explanatory variable. In statistics, multivariate analyzes were the characteristic of interest in the joint distribution of several variables. The bivariate analyzes, such as ANOVA and Student's t test, are special cases of two variables.
Multivariate analyzes are very different according to the desired objective, the nature of the variables and the formal implementation. We can identify two main families: the descriptive methods (to structure and summarize the information) and the methods for explanatory variables explain a "dependent" (variables to explain) a set of variables called "independent "(explanatory variables).
Involves the analysis of a single variable For example, we can consider the education variable in research that seeks to determine what the average number of years of education that people have completed a business. For this, we first group individuals according to the number of cycles and years of schooling completed (elementary, intermediate, middle, university), then these data can be analyzed in terms of percentages and finally in terms of measures of central tendency (e.g. arithmetic mean). In conclusion, we will say only that the average number of years of education completed by individuals is N (Service 1972).
Comparative analysis involves two variables, one of which modifies the other. Returning to the previous example, we now know not only the average years of education completed, which is the distribution of these results according to the gender of the respondents, In other words look for a relationship destined to let us know what level of education according to the sex of the respondent. If this cause-effect relationship was hypothesized before the interviews, the results obtained will allow verifying the hypothesis, i.e. confirming or rejecting by the data collected (Johnson 1982).
Almost all confirmatory investigations are analyzed by means of multivariate analysis type. In this type of analysis three or more variables are examined simultaneously. Returning to the previous example, we now not only know the level of education is also seeing what happens to the age variable.
Using multivariate methods (also: multivariate analysis) in the multivariate statistics several statistical variables or random variables examined simultaneously. For example, vehicles for the variable number of seats, weight, length, etc. are levied. In the univariate analysis, however, each variable is analyzed individually. Relationship or dependencies between the variables, such as greater number of seats due to a greater weight, with a multivariate, are not recognized by a univariate analysis (Edwards 1958).
Multivariate techniques substantially reduce want contained in a record number of variables and / or observations, without reducing the information contained therein substantially. Also the (context) is structure of the data analyzed. Either there is a structure and verifies that match the data with the given structure (structure Examiners process: part of inductive statistics), or trying to extract the structure of the data (process: part of Explorative statistics).
The classical methods are all linear models, which have special requirements for ...