The Student Room Group

last minute Mann-Whitney U confusion

My alternative hypothesis states that processed anti-bacterial substances are more effective than natural anti-bacterial substances, making it a one-tailed test, and my null hypothesis states that there will be no significant difference.

When conducting Mann-Whitney U I used processed substances as 1st set of data, and natural as 2nd. Therefore I assumed that my value of U1 would be the one to be used, as I am comparing processed vs natural, and to use U2 as well would be testing to see if natural is better than processed, which is not my alternative hypothesis.

Howver all the sources about MannWhitney say I should always use the lowest value, but this doesn't make sense...for example in one case natural was much more effective than processed, therefore my alternative hypothesis cannot be accepted, but I have values 256 and 0...and if I use the lowest, 0, it means I can reject my null hypothesis??

I hope someone can shed some light on my confusion, thanks
Reply 1
AnnaBelle142
My alternative hypothesis states that processed anti-bacterial substances are more effective than natural anti-bacterial substances, making it a one-tailed test, and my null hypothesis states that there will be no significant difference.

When conducting Mann-Whitney U I used processed substances as 1st set of data, and natural as 2nd. Therefore I assumed that my value of U1 would be the one to be used, as I am comparing processed vs natural, and to use U2 as well would be testing to see if natural is better than processed, which is not my alternative hypothesis.

Howver all the sources about MannWhitney say I should always use the lowest value, but this doesn't make sense...for example in one case natural was much more effective than processed, therefore my alternative hypothesis cannot be accepted, but I have values 256 and 0...and if I use the lowest, 0, it means I can reject my null hypothesis??

I hope someone can shed some light on my confusion, thanks


Mann Whitney is for comparison; surely there should only be one figure?
Reply 2
do you mean one U value or one comparison?
I compared various sets of data to each other and then overall, and Ive come across this problem for a particular set.
For the U value you usually get 2 U values.

Latest

Trending

Trending