Remark on Section on Statistical Thinking

As author of section on Statistical Thinking in book Voices in Statistics, I am glad to see a lot of discussion and collection of material on this topic. I am especially thankful to rktyagi which help me in writing Teaching Tips. Still section of simulation is empty and hope it will be soon filled. We are in search of a tools which may help in increasing feelings of variation.

Some user raised relevance of term `Statistical Thinking’. They asked whether old statistician were not using Statistical Thinking. Certainly they were using `Statistical Thinking’. Statistician like  W.A.Shewhart (link), R.A.Fisher (link) are itself example of integration between context are and statistical method (which is one of foundational base of Statistical Thinking). Need of use of terminology Statistical Thinking emerged in course of time because statistical works are now divided in three groups:

1. Experts of context area: In this group there is less appreciation for variation in data and result.
2. Hard Core Statistician: They are basically tool developers and not concerned with context of data. They discuss their problem only through symbols and judge their results on mathematical profoundness.
3. Applied Statistician: They are hardcore statistician by training  working in domain of different context area. These people are most ignored people. After availability of statistical package, there is no need to know formula and shortcut rule to use complex statistical method. Hence subject expert donot need to consult them. Their emphasis on `variation’ and `sampling plan’ gives impression of unnecessary hovering. Since statistician do not feel confident (due to lack of training) in context area, they find themselves capable for finding problems in applying statistical tools but they fail to provide best possible answers.

Raising voices for statistical activity is an effort to bring these three on same platform. Vast computational capability is main strength of this voice. Now without knowing complicated mathematical proofs, one can get confidence through simulation. Not only this, though simulation one can study the impact of violation of assumptions through simulation. Many assumptions can be relaxed because they are taken for getting analytical solution. Simulation is very strong tool and much useful for tool developers and trainers both. This is the reason, apart from data, variation and integration of statistics with context area, I have included simulation as part of Statistical Thinking.