what if i get non significant test result in the one way ANOVA and there appears to be a significant difference between treatment means when post test(esp. LSD) is run? is it only after ANOVA result one can go to post hoc test?
Trust your ANOVA! If the ANOVA shows no significance, the test ends there. Post hoc tests are only legitimate if the ANOVA shows p < 0.05 :)
You still can look at and interpret post-hoc tests even when the ANOVA wasn't significant, if you are using an appropriate multiple-comparison control (LSD is not appropriate because it does not control for multiple comparisons).
For a more detailed answer see here
https://stats.stackexchange.com/questions/9751/do-we-need-a-global-test-before-post-hoc-tests
Trust your ANOVA! If the ANOVA shows no significance, the test ends there. Post hoc tests are only legitimate if the ANOVA shows p < 0.05 :)
I agree with the previous answers, but I would like to add that significance testing is not everything. How was the power of your test, were all assumptions met, how does the distribution look like, were the outliers, how large was the effect size of the ANOVA and the effect sizes of the post hoc tests, etc...? Describing your data with p-values is really not everything. Would you omit a conclusion if you have a p-value of .07, but the effect size would be moderate and of practical interest? You should also have a look at other aspects of your data to coome to the right conclusions (although I know that in case of a publication it is hard do get something published with p > .05).
Sometimes, we can get the nonsignificant results from ANOVA. First, you can concentrate which statistical tests are suits to your work, then you can be referred latest published articles related your topics (you might be studying the effects of data size and concentrate the limitations of one way ANOVA). You must focus this way, it will definitely help your research.
1. Describing and exploring the data to be analyzed
2. Find out the model the data relates and demonstrate your model using any statistical software like min tab, SPSS and etc.
If the ANOVA shows no significance, no need to run Post hoc tests
No need to run post hoc test if your ANOVA shows no significant defference
I agree with what Rainer Duesing suggests. First of all, P< or = 0.05 is due what Fisher stated it at the first time; what if it was 0.04 or 0.06 therefore, better not to highly rely on statistical significant rather try to consider a number of other things too
The ANOVA is run to show wether or not there is a significant difference between more tnan two variables.Where there are significant differences you than run the post hoc to find out the varables that are responsible for the difference so there is no need to run post hoc test if your ANOVA shows no significant defference
The post hoc tests do take into account multiple comparisons. But that's not the point. If the ANOVA has p > 0.05, then you cannot continue. It's not fair to trust the ANOVA when p is < 0.05, but not when it is > 0.05! You have to be consistent and perform the test properly. Sadly, ANOVA has to be abandoned if p is > 0.05 (if that is the criteria). Also, LSD does account for multiple comparisons, but not as strictly as others, such as Bonferroni
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA