Dear colleague,
how to interpret when the main effect of factors is not significant but its quadratic effect is significant?
- Biserka Runje there are linear, interaction and quadratic effects. Sometimes cubic effects affect the model. The strength of the effects is based on the sum of squares from ANOVA. See the clear equations in https://doi.org/10.1007/s13399-021-01971-3
Try to experiment with Cubic effects of second order model of RSM.
You have to retry inputting the range values of your independent variables based on your pre-experimental data
No DO NOT mess around with p-values. if you chose an appropriate level it is still appropriate. You are asking the wrong questions. My fav resources on DOE and RSM are given in this link: https://www.amazon.com/Analysis-Experiments-Enhanced-Abridged-Companion/dp/1119593409/ref=sr_1_1?keywords=Douglas+C+Mo and
https://www.amazon.com/Response-Surface-Methodology-Optimization-Experiments/dp/1118916018/ref=sr_1_1?keywords=Response+Surface+methodology&qid=1578227486&s=books&sr=1-1
Please refer to these. Now your results. It appears to me that you have chosen the values for your IVs incorrectly. It looks like the temp spread is too wide. Perhaps a smaller spread but higher temps might be better. concentrate on setting the IV levels and repeating the experiment. Your figure is no help at all Good luck, David Booth PS this link may also be helpful: https://www.google.com/search?q=Box+Behnken+designs&rlz=1C1CHBF_enUS874US874&oq=Box+Behnken+designs&aqs=chrome..69i57.16692j0j4&sourceid=chrome&ie=UTF-8
Hello dear Biserka Runje ,
What I understood from your question (correct me if I am wrong) is that your main effects are not significant,
Well, you will have to try the model reduction with the auto-selection based on p-values, this may bring their p-values to less than 0,05.
If it didn't, then that is okay to have some effects with non-significant p-values as long as they are needed for hierarchy.
Then, to interpret; you will say that each factor is not really significant on its own but their quadratic effect is significant.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS