0.00000 significance value

[marson93]

Hi everyone,

New to Talk Stats, so be nice! I have looked for an answer to my problem on previous posts but have not been able to find anything suitable.

Basically I am comparing the data between three different groups who have all performed a particular type of examination. I have carried out a Kruskal-Wallis test as my data was positively skewed to the right and I have more than two groups to compare.

My results for one of my variables was .001 and for the other it was .00 which I have managed to expand to .000000000002.

These values have thrown me off a little bit so I need some help, I'm presuming I need to reject the null hypothesis as p=<0.05?

And I was then planning on using the Mann-Whitney U test to compare what two groups have a significant difference.

Is this along the right lines? Using SPSS btw.

Thanks all in advance for any help

 

[hlsmith]

You would reject the null hypothesis that none of the groups are different. Yes, then you would typically perform pairwise comparisons of the groups to discern which groups in particular significantly differ.

 

[Miner]

Don't forget to adjust for family-wise error rates.

 

[noetsi]

Although with those p values FW error probably won't matter. p can never actually be 0. When you see that you have not asked your software for enough decimal places.

When you have a p value like that you reject the null. Then you go for one of the test that tells you which effect varied [there are many test for this including contrast].

 

[marson93]

Thanks very much for your help!