For studies in which you will apply statistical analysis is important to characterize the aggregation to be studied. Aggregation is the group to which they make generalized conclusions until the sample is this group that isdirectly tested. In some cases, the two groups almost coincide. The challenge is that to make valid general conclusions of aggregation on the basis of the data samples. Before making tests to define the aggregation and a sample that reflects this complexity. Also, there are tests that can determine the sample size, which can achieve a certain effect of the demonstration.
Choosing a statistical method but the processing of the data depends on the type of data being analyzed – there are categorical or continuous values. The type of data (variables) is determined by the accuracy of the values, characteristics and importance that is put in the values.
In studies independent variable (sometimes called the resultant) is one that has been studied as a function of the other. In the experiments, the dependent variable is the one which is expected to change as a result of the experiment. Not necessarily dependent variable is a direct result of an experiment - there may be a case of variation of one variable (dependent) on other variables (nezavisima). These terms have meaning in summary statistics. In some cases, this anchoring is directly determined. Statistics solves the problem of finding relationships between variables when they are not directly Defines. In other cases, studies aimed at characterizing the relationship between variables and not causality or predictions. For them, no matter the designation of dependent and independent variables.
Methods in statistics
processing is performed
types of analysis - variation and alternative. In the first
analysis using numerical data, while the second variable is categorical.
Relationships between parameters and variables are tested by the following approaches:
- Correlation analysis – displays the strength and direction of relationships
- Coefficient of Pearson - rank coefficient for categorical indicators
- Coefficient of Spearman - rank coefficient for categorical indicators
- Cluster analyses – this is a method to group - assigning an object to a group. Cluster analysis is a collective concept of multiple classification algorithms. The object of the analysis is to be natural groups formed on the basis of several parameters (variables, signs, etc.). One of the most common procedures is a pre-set number of clusters (groups). Setting of the number of groups in this case is usually expert judgment. This method is known as the method of k-mean. Objects identified by the indicators are appropriate points in space.
- Discriminant analysis - used for classification of cases into groups of predetermined number. Using discriminative analysis is based on the concept that if the discriminant function is effective for a given set of data, it is effective for a high percentage of correct classification of new cases. Discriminant analysis is used to solve the following tasks - classification of cases into groups on the basis of predictive equations discriminant, testing the correctness of classification, exploring differences between certain groups, determination of the most parsimonious way to distinguish between the groups, finding of the variance of the dependent variable by the independent variables.