Answer:
Unfortunately, there is no single procedure to check for causality between two variables. But there are three known procedures:
1. Test of association/relationship between the two variables
2. Time ordering of the variables
3. Non-spurious validation
In software or in econometric analysis, we usually use the Granger causality test! But is not really sufficient for an empirical study.
Step-by-step explanation:
1. We must first establish association or relationship between the two variables. If the variables are numeric, we can conduct a correlation analysis and the variables are categorical scale, we can conduct a cross tabulation and do Chi-square statistics.
2. If we are able to establish association or relationship, we go on with time ordering of the variables usually involves; setting up of an hypothesis to carefully determine the direction of relationship between the two variables of interest the and testing of significance, etc., associated with hypothesis testing.
3. We must ensure that relationship or association observed are not spurious or misleading or false relationship.
For instance, there may be relationship or association between age and academic grades among pupils, but does age really tells what academic grade will be?
In this type of scenario, or any other under study, we must come up with careful study design, through a well designed experimental procedure and well structured data collection approach with careful statistical controls. It may go further to conduct data 'crosstabbing' from many other verified sources to establish that the relationship or association observed are non-spurious. It's a painstaking process!
