It is marginal effect. That does not mean that you can accept the null hypothesis. You need to dig deeper into the data.

Thank you very much sir

Dear Ataul Vaia,

I think we can not say it significant literally.

Regards

Bezon Kumar

Please do not equate a lack of significance as proof that the null hypothesis is true. There actually is no proof in any statistical test no matter the p-value.

In statistical testing we are estimating the probability of a result as large or larger given that the null hypothesis is true. Typically we have decided that if that probability is less than 0.5 we will reject the null-hypothesis. If we use this approach to conclude that the null hypothesis is true, then we will do so more frequently than we should. I can also scam the system by proving the null hypothesis by using an inappropriate model or by using a small sample size. This just ends very badly for everyone.

Be very careful about relaxing the significance level from 0.05 to 0.10. Consider that your manuscript has 10 statistical tests, and ask what is the alpha for your manuscript if you use 0.05 versus 0.10? The answer is 0.40 versus 0.65 respectively. Or think about a person that has a manuscript with 100 independent variables and will now run 100 univariate tests for each variable against a treatment variable with 6 pairwise comparisons for each test. Even if the null hypothesis is true in every case, it is likely that this person will find at least one statistically significant outcome. That said, it is also possible that having found one significant outcome that this one outcome highlights the only real relationship in the entire data set. Based on the data from one experiment it would be impossible to know which outcome was correct. Is the significant outcome real or an artifact of testing a large number of variables?

Thank you very much sir Dr. Timothy A Ebert ,from the valuable comment you add here, I have come to know that a large set of variables can do great good rather rejecting the null hypothesis at 0.10 level.

Maybe I don't quite understand the comment. Relaxing alpha from 0.05 to 0.10 causes a big problem (not good).

If the null hypothesis is actually true (as in the data from treatment 1 comes from a random number generator set as Normal (mean=12,sd=4) and treatment 2 comes from a random number generator set as Normal (mean=12, sd=4)) and if significance is declared if the p-value of the ANOVA F-test

Thank you Dr. Timothy A Ebert