What statistical technique should be adopted in which scenario?
This is the crucial question that usually research students feel inappropriate to answer and feel difficulty in adopting the correct technique. This article will put lights on the methods and it will explain the scenario where they can be best applied.
For the simplicity, divide the techniques in two groups. One will comprise those techniques which are used for grouping and the second will be comprises of those techniques which are used for finding the relationship between variables.
· 1st group (for grouping the data or reducing the size of the data):
o Factor analysis: This technique is used to make factors of the related or similar variables. It makes the grouping of those variables which are closely linked and then treat them as on variable.
o Cluster analysis: This technique is use to make the groups of entity by studying the proximity between them and try to reduce the spread of the data.
o Multi-dimension scaling MDS: This technique is use to group the data on the basis of both the similarity and the proximity by explaining the scale.
· 2nd group (for finding the relationships):
o Chi-square analysis: This technique is used to find out the relationship between the single dependent and single independent variable. The dependent variable can be continuous or categorical as well as the independent variable too.
Where it can be applied:
It is best applicable when there are only two variables under study.
o Regression analysis: This technique is used to find the relationship between the variables. The major assumption for this test is that: the dependent variable must be continuous variable and dependent variable must be one. Independent variables can be more than one and can be continuous or categorical (i.e. scaled variables).
Where it can be applied:
It can be applied in a research where one want to find the relationship between the continuous dependent variable like time and one or more than one independent variable.
o Logistic analysis: This technique is used to find the relationship between the one dependent which have two or more levels and more than one independent variable. The dependent variable like brand switching resulting into two possible answers (Yes or No) will be the case of binary logistics and the dependent variable for more than two possible answers will come under the case of Multiple logistics. The independent variable can be continuous or can be categorical.
Where it can be applied:
It can be applied when the dependent variable has categorical levels like the dependent variable is consumer preference resulting in categories as local brand or international brand will be considered under the case of logistic regression.
o Multiple discriminant analysis: This technique is must similar to that of logistic regression. It follows the same assumptions as of logistic regression. Here the dependent variable has more than two categories.
Where it can be applied:
It is best suited when the dependent variable has more than two categories like the dependent variable is channel choice which have categories like entertainment, sports, news or cooking is the case of multiple discriminant analysis.
o Structural equation modeling: This technique is used to find the relation between variables where there is more than one dependent variable. The dependent variable itself acts as the independent variable and is affecting the other variables. This techniques is use to find out the hidden variables that are affecting the dependent variable.
Where it can be applied:
This technique is best suited when the researcher wants to find out the relationship between variables with more than one dependent variable and the same time the researcher wants to do the path analysis.
