Bob Luttman, Robert Luttman & Associates
Cascade Effect
Operational Issues: Documentation and Reporting
Traditional Pathway / Variance Documentation
Cascade Effect
The cascade effect, one variance triggering a series of variances, is the leading cause of multi-collinearity.
Conversely, as we will discuss in upcoming sessions, the cascade effect is also a powerful process improvement tool. If you can eliminate the cascade trigger you can eliminate a great many variances. This is called leverage. Finding and mitigating these variances is a priority for process improvement efforts.
For variance analysis purposes, though, it greatly complicates your ability to identify causal relationships using traditional statistical tools. To simply correlate each variance with some outcome is invalid, they are not independent. And, without the proceeding diagram or other process knowledge, the correlation matrix will not identify the cascade trigger.
The larger issue here, as with multi-collinearity, is: statistical knowledge without process knowledge is like a Ferrari without wheels. Looks cool but it ain't goin' anywhere!
Again, for the biostatisticans in the crowd, Yes statistical tools and techniques exist that mitigate the problem. But, again, can the average staff person use these tools in the heat of battle?
These more sophisticated tools are also limited by two other issues: sample size and the halo effect.
