Should the Null Hypothesis be rejected or accepted?

[arnab]

Just a quick question, should the Null Hypothesis be accepted or rejected?

 

Null Hypothesis : Same concentrations of substance X & Y will have the same effect on substance Z

 

A experiment was carried out to determine whether the same concentration of X and Y will have the same effect on substance Z. Hence 3different concentration of X & Y was made and mixed with substance Z in a different test tubes.

 

Data was collected for each different concentration of X and Y and its mixing with substance Z.

 

It was found, from the Mann-Whitney U test, that the two weakest concentration of X and Y did not have a significantly different effect on Z but the strongest concentration of X and Y had a significant effect on Z.

 

Now my question is, should the Null hypothesis be rejected based on the fact that the 2weakest concentration doesnt follow the null hypothesis?

 

PS : this is for my A2 unit 6 coursework whose deadline is in 2days time. Any help or advice could be much appreciated!

 

Also just as a side question, can there ever be a situation where a Null hypothesis is neither accepted or rejected?

[Suxamethonium]

Statistics was my worst subject- but I believe you need to perform a statistical test (For example a t-test) in order to prove or disprove the null hypothesis mathematically i.e. you can't just assume there is no correlation because it makes sense, you need to prove it using a statistical test with a P value. Apart from that though I'm not much help.

[Bioc]

Hi, the Mann-Whitney is the non-parametric alternative to the t-test. Since they told that there is a significant difference between the effects of X & Y on Z, when the concentrations are high enough, I think there is enough evidence to reject the null hypothesis;at low concentrations there is not enough evidence to reject the null hypothesis (hypotheses are never accepted, only rejected or non rejected).

[arnab]

Hi, the Mann-Whitney is the non-parametric alternative to the t-test. Since they told that there is a significant difference between the effects of X & Y on Z, when the concentrations are high enough, I think there is enough evidence to reject the null hypothesis;at low concentrations there is not enough evidence to reject the null hypothesis (hypotheses are never accepted, only rejected or non rejected).

 

ahh thanks for clearing that up for me. i was thinking that the hypothesis could be rejected as the majority of the data tell us it is?

[CharonY]

The null is not clearly formulated. I.e. what is meant by "same"?

From the way you described the experiment and the respective statistical tests it appears that you formulated three nulls (i.e. one for each concentration). In each case the null would be that at the tested concentrations (let us call it c1, c2 and c3) the compounds x&y have no effect on z.

In other words, each test requires its own null.

Alternatively, one could apply statistical tests that account for more than one group (e.g. ANOVA).

 

Thus, the null with c1 and c2 cannot be rejected (i.e. we do not know whether it has or not). Whereas the null with c3 can be rejected. I.e. at c3 there is a significant effect on z.

 

As already mentioned, a null is never accepted, we either reject, or fail to reject.

Edited May 8, 2012 by CharonY

[arnab]

The null is not clearly formulated. I.e. what is meant by "same"?

From the way you described the experiment and the respective statistical tests it appears that you formulated three nulls (i.e. one for each concentration). In each case the null would be that at the tested concentrations (let us call it c1, c2 and c3) the compounds x&y have no effect on z.

In other words, each test requires its own null.

Alternatively, one could apply statistical tests that account for more than one group (e.g. ANOVA).

 

Thus, the null with c1 and c2 cannot be rejected (i.e. we do not know whether it has or not). Whereas the null with c3 can be rejected. I.e. at c3 there is a significant effect on z.

 

As already mentioned, a null is never accepted, we either reject, or fail to reject.

 

Thank you! i actually realised i made a mistake in my experiment yesterday, so what i said above is not correct. Thanks again for answering my question though.