mann-whitney u test

i have been told to do a mann whitney u test as part of my coursework. no idea what it is. looked it up and it looks very complicated. anyone kno how to do a very basic version??

Reply 1

DENNIS_MENACE

Erm.. this is general discussion. Move this to the psychology/statistics section

Reply 2

Knight-of-Ni

2

eeeeew mann-whitney u is bad. everyone i know who did it found their results were not significant! i did t-test though, much easier! i suggest you look it up on the net... theres loads about it! :biggrin:

xxx

Reply 3

xxxjase

OP

Knight-of-Ni

eeeeew mann-whitney u is bad. everyone i know who did it found their results were not significant! i did t-test though, much easier! i suggest you look it up on the net... theres loads about it! :biggrin:

xxx

i have a major problem tho. i only have 2 sets of results. male and female. and a mean for each.

Reply 4

mipmapped

13

eeeeew mann-whitney u is bad. everyone i know who did it found their results were not significant! i did t-test though, much easier! i suggest you look it up on the net... theres loads about it! xxx

You can't just use whatever test you feel like.

Reply 5

Fire Point

12

It's relatively easy to do but I think you need someone to show you how it's done first.

Reply 6

mipmapped

13

i have a major problem tho. i only have 2 sets of results. male and female. and a mean for each.

Have you got scores or frequencies?

Reply 7

xxxjase

OP

mipmapped

Have you got scores or frequencies?

my results are recorded in this table. left the results blank in the very unlikely event someone doing the same cwk as me. i have a mean at the bottom as well

Right, they're scores, so you can't use a T-Test.

It's an experiment, rather than a correlational design, so you can't use Spearman's Rho.

There's 1 IV.

Two groups, and the data are [I presume] non-parametric.

You have to use Mann-Whitney.

Gimme 5 mins to find a tutorial.