Parametric or non-parametric test ?

A

Ahmad 2013

New Member
#1
#1
Dear the members of group.
I have 2 questions regarding my research which has 8 variables in which 6 of them are normally distributed and the rest are not normally distributed.
1. Do I need to use parametric test for those 6 variables and non parametric test for the rest of two variables to obtain the similar study objective ?
2. The 8th variable has several data which are grouped into two subgroups in which the first subgroup is normally distributed but the other subgroup is not . If I would like to compare mean difference between those groups, which test is more suitable, either the parametric or non parametric test ?
I seek your response. Thanks in advance
 
noetsi

noetsi

No cake for spunky
#2
#2
It depends on what you want to do. It is not normally whether a variable is normal or not that matters, but whether the error term is. Individual variables can be non-normal and the error term not be affected. Also most methods are robust to the normality assumption so it has to be signficant to matter. Commonly, as with regression, non-normality does not influence the parameter or SE calculations, but the confidence intervals and statistical tests.

Whether you have to use a non-parametric test depends on what you are doing, if you can transform the data etc. We need more details before answering your question.
 
You must log in or register to reply here.