Statistical analysis flowchart implementation in R

I have seen lots of flowcharts describing the different questions you should ask yourself when doing statistical analysis and the corresponding tests you should run based on your answer. For instance, you can find some here: Flowcharts to help selecting the proper analysis technique and test

Unfortunately I could not find implementations of such flowcharts. By implementation of the flowchart I mean the program could ask you some questions, and based on your answers would run some test. Do you know if such programs exist? (it would be better in R but implementations in other programming languages are also welcome)

Thanks a lot!

• There is no implementations because the flowcharts describe simple heuristics, while real life cases are much more complicated than this. I would say that the flowcharts are most useful for passing your "applied statistics" course, but not much beyond that.
– Tim
Feb 2, 2016 at 14:16
• I agree with @Tim. I just found one (reference suppressed to protect the guilty as well as the innocent) which said "Normal mean, Non-normal median". I suppose the advice is better than the exact opposite, but you don't have to be a statistician to see that statistical life is too complicated to be reduced to such dichotomies. If I have a Poisson or an exponential, I really want to work with the mean! Feb 2, 2016 at 15:14
• Understood ! I guess the fun of doing stats comes from the fact that it cannot be reduced to a flow chart ;) @Tim you can post an answer and I will validate it ! And also, thank you ! Feb 2, 2016 at 16:41

There is no implementations because the flowcharts describe simple heuristics, while real life cases are much more complicated than this. The flowcharts described by you are connected to, often criticized, "cookbook" approach to statistics. Flowcharts are helpful for passing your applied statistics exams, but in real life may not be that helpful given the complicated nature or real-life statistical problems and datasets. Below you can find few quotes dealing with this approach.

Rasmus Bååth:

In Bayesian statistics you can quickly escape the cookbook solutions, once you get a hang of the basics you are free to tinker with distributional assumptions and functional relationships. In some way, classical statistics (as it is usually presented) is a bit like Playmobile, everything is ready out of the box but if the box contained a house there is no way you’re going to turn that into a pirate ship. Bayesian statistics is more like Lego, once you learn how to stick the blocks together the sky is the limit. (http://www.sumsar.net/blog/2014/01/bayesian-first-aid/)

Herbert Spirer and Louise Spirer:

Most textbooks offer "cookbook" approach of analysis of data without advising the reader what will happen to the recipe if one of the ingredients is left out... (Misused Statistics, p. 6)

Michael R. Hulsizer and Linda M. Woolf:

A cookbook approach to teaching of statistics may make the material more accessible to students. (...) Students need to decide what to bake (what to study), collect the ingredients (data collection), prepare the dish (data analysis), finish preparing the dish (data analysis), finish preparing the dish (interpretation of results), and present the food to one's guests (publication/preparation). (...) However, this strategy believes that fact in statistics, data are not always neat or precise and researchers can use more than one procedure to explore a problem. (A Guide to Teaching Statistics., p. 71)

The quotes are quite random, but should give you some overview of the critique.

The same with flowcharts, they may be helpful with simple, "textbook" problems, but may easily fail in more complicated situations. Furthermore, real-life usage of such flowcharts needs its user to make number of subjective decision, e.g. on critical value of $p$-value, or assessing if sample is "normal" enough etc. so they cannot be applied automatically.