How do we get this chest thumping 600 pound gorilla to a more thoughtful head scratching one?
Let me see whether I can bring out the latent dimensions of prediction admonitions and how statisticians can protect from lame blames of managers who are the real culprits in prediction problems, other than some really really uncool statisticians. Of course, the statisticians can not be purely tactical executors and that is where the problem is. What are the tools that statisticians/mathematicians/economists/ data scientists have to consider to get the management scratching its head? How to culturize the management to use right metrics and monitor them. How to get the management drive their decisions on the basis of well collected quality data. Obserations do not lie, the problems in faulty data collection design and the interpretations do.
This is in response to blogs like:
Statisticians and economists are not the final call makers: However, they can do certain things to get the 600 pound gorilla scratching its head:
- Whether we like it or not, we can not live with out prediction
- There are statistical errors (which means structure is already imposed) and there are managerial errors (all those who do not want to be put in the bucket of “analysts” including VP, SVP, CXOs, and policy makers)
- The prediction industry blames the whole thing for sub-prime mortgage to climate problems. What a set of lame brilliants. These are the people who mislead as managers. Managerial errors are the 600 pound gorillas
- Statisticians have the responsibility to challenge the 600 pound gorilla and always gorilla has the way out; else statistician is out
- Statisticians can not be in the business of writing research articles or questioning assumptions in business context but just do the analytics for immediate business problem and create reports for the 600 pound gorilla; else statistician is out
- However, what statisticians can do is the following: make available and get attention to macro trends; this is innocent looking and that is where most of the intelligence is and nobody will challenge that as time consuming. Note that firms have to watch out and have to spend money and time on few important macro level metrics that will affect their own existence.
- Also a company should look at strategic statistical analysis; it is not enough to ceate reports based on normal and logistic regression models for propensities, and campaign analysis. This is not difficult if the company has a long culture of encouraging statistical thinking and promotion to senior position in such companies can be based on strategic metrics, studies, and influences. Get the 600 pound gorilla scratching its head – monitor strategic metrics and create strategic reports
- For any industry, some of the common strategic metrics are: the industry key measure and how these metrics are differing across segments of consumers with in one’s own enterprise vs. the competition. For example, in mortgage industry, we need to monitor the sub-prime mortgage lending volume (frequency and severity) by region, key mortgage bankers, generic consumer segments.. Get the 600 gorilla scratching its head; challenge the gorilla politely with well formatted reports and metrics
- We also lack a broad collection of macro metrics that all CEOs should be reading and get their junior gorillas scratching their head
- It is disparaging to note that even the Fed Chairman did not scratch his head when the national debt to the world was skyrocketing to $4.5 trillion in the last 30 years, from 250Billion… mm… on top of it 9/11 happened and well before quantitative easing, credit easing happened and we sold another unknown number of trillions of mixed risk to the whole world in a short period of 6 years to help out every one’s economy. The vibrations are still felt around the world
- Do we need a statistician to tell? Was it difficult to predict a catastrophe?
- Of course, the ruling party was adamant saying that there is no hair on head and no need to scratch…