Conditionally independent data structureAssociations Analysis is as powerful tool for Statistical Way of Making Hypothesis . "We can use as the analysis Asymmetry of Category Data , Analysis Using Category Data and Robust Analysis .
But it is not easy to use output of this tool for the analysis of cause-effect.
We can get the hint to consider Kinds of Cause and Effect (AND, OR) by the output is if-then rules, "If ### then +++." But one if-then rule is not equal to cause-effect.
If-then is often changed to arrow graph. The arrow does not tell us the cause-effect.
The graphs in Seven QC Tools and New Seven QC Tools and SEM and Covariance Structure Analysis use the arrows to express cause-effect. But the graph by association analysis is different.
If the root of arrow is on cause side, the cause is OR condision .
If the root of arrow is on effect side, the cause is AND condision .
Association analysis does not tell us all condition of real phenomena.
In real analysis, we do not have all data to study all conditions. But if we get the information "AND or OR", it is the strong hint to consider the area of no data.
If we have infomation of causes and effects for the variances of data, we can study "AND or OR" condisions.
But in real analysis, we often start without the information. For example, when we do not have Meta Knowledge and when we only have the informaion about the variance of the last effect.
It is good to understand that the if-then rules as the output of association analysis are hints to understand the data structure and phenomena at the first stage of the analysis. In this stage, it is not inportant to defeine "AND or OR" or "Cause or Effect".
NEXT Conditionally independent data structure
