In minitab i apply normality anderson darling test on my data it show normal data value P value is greater then 0.05 but when i apply ANOVA on same data its results shows non significant then what should i do ?
please suggest me i will be very thank full to you
Large p-value do NOT demonstrate that the tested hypothesis is true. They just indicate that the data is not conclusive regarding the tested hypothesis. So p>0.05 of the AD test does NOT mean, as you wrongly stated, that the data is a sample from a normal distributed population. It only means that your data is not conclusive to see differences. Formal (hypothesis-)tests of assumptions are nonsense.
The failure to reject the hypothesis tested in the ANOVA has the same interpretation: the data is not conclusive regarding the tested hypothesis. To answer your question ("what should I do?") with the words of RA Fisher: "get more data". Or state that your data is not conclusive.
I don't understand the question. Are you referring to the F statistic in ANOVA? Or something else?
Thank you so much Jochen Wilhelm, Sir actually i want to check which treatment is effective as compare to control in this situation which statistical test is used to check, is hypothesis is true or not ?
Thank you so much Daniel Wright , Sir actually i applied 8 treatments and 1 control .Now, I have to check which treatment is more effective with comparison to control. At this stage i can't understand which statistical test would give me information regarding that , which treatment is more effective as compare to control ??
Thank you so much Mugumaarhahama Yannick , No sir i did not test variance equality
Noshia Arshad,
" which statistical test is used to check, is hypothesis is true or not "
This reveals some misunderstanding about what tests do. They do not tell you if a hypothesis is true or not. The hypothesis is a mathematical assumption within a matehmatical model. This is not identical to something that is physically true or not. Tests measure how compatible your observed data is with such a (given!) hypothesis in some (given!) model (model and hypothesis are mental constructs!). If the model makes sense and is flexible enough to give a sufficiently general description of the data, finding your observations being statistically incompatible under the tested hypothesis tells you that this particular hypothesis "significantly" lowered the ability of the model to describe (or "explain") these data. You would then reject this hypothesis.
Note that this is a statistical issue. A model that is restricted to a given hypothesis is neccesarily less well performing in "exlaining" data. Thus, the entire meanig of a test is to see if the data (that is, the information in the data that is relevant to the hypothesis in the given model) is sufficient to identify this "loss of performance". The question answerd by a test is: "do I have enough data to interpret it relative to the hypothesis?".
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA