Hello there!
My question concerns three cases: scientific papers, doctoral thesis and simple presentations. Which data from design of experiment are necessary in each case, which are more than welcome and which are completly redundant?
For example in full 2 level factorial design should I present half-normal probability plot, Pareto plot and ANOVA? Or perhaps ANOVA is better for papers and Shapiro Wilk normality plot better for presentations while it LOOKS better?
Another question, should I present all diagnostics plots (Box Cox, Residuals vs Factors and so on) or just mentioning that all data is correct is enough?
All suggestions are much appreciated!
It really depends on what you are doing. Suppose this is a response surface analysis with the CCD as screening design and the factorial model is what you wish to optimize. in such a case I would put more emphasis on the factorial model and quickly discuss the CCD in a lecture. In a dissertation I would cover everything in detail. It all depends on what your goals are and your audience.. Best wishes David Booth
1/ In the case of a factorial design of experiments, the main of your study is to determine the effects of involved experimental factors on your analysed response. So, the better and essential thing is to present the ANOVA table, rather than other results.
2/ In the case of a Central Composite Design (CCD), which is a Response Surface Methodology (RSM), the target of your study is to find the optimum of your experience. So, in this case, you need to present the following results:
- ANOVA table of factors effects;
- Response surface plots and
- The optimum setting of your involved factors, accordingly to the desired response you are looking for.
I am agree to what mentioned by Tarek Moussaoui. You may consider the following as well:
1- In simple presentations use graphs to show the effects of main factors and also the interactions.
2- In scientific papers and doctoral thesis, In addition to finding the optimum conditions of your experiments by using final model of RSM, you should perform a verifying experiment at that condition, and compare the experimental against the model predicted response.
- Badges
- Science method