Within variable differences?

Moderators: statman, Analyst Techy, andris, Fierce, GerineL, Smash

wmurry
Posts: 3
Joined: Tue Aug 28, 2012 12:56 am

Within variable differences?

I have a series of variables that are 4 part Likert type scales [1,2,3,4 response set] and they have been dichotomized as group 1=Very Poorly (1) and Poor(2) and group 2=Good (3) and Very Good (4). I am wanting to examine the "within" variable differences, i.e., differences between those that responded Very Poorly-Poorly (dichotomized as equaling 1) versus those that responded Good-Very Good (dichotomized as equaling 2) for variable X. What is the best way to determine if there is a statistical difference between group 1 and group 2?
MOram
Posts: 5
Joined: Wed Jan 23, 2013 2:07 pm

Re: Within variable differences?

Hi,

As I understand it, you

1) asked participants "How poorly do you feel?" [1-4]
2) Split them into 2 groups (response 1/2 and response 3/4)

The first thing is to make sure you have a variable that lists the grouping for each participant.

To ask if another variable (X?) differs between the groups, we need to know the nature of that variable. If the variable is

"scale" and normally distributed - use a between-subjects t-test
"scale" but not normally distributed - use non-parametric test (Mann-Whiney U test)
"ordinal" - use non-parametric test (Mann-Whitney U)

Be wary of the interpretation of the Mann-Whitney U - use exact probabilty calculations, particularly if you have only a limited number of response options (i.e. variable X is also a Likert scale)

Hope this helps
neigurniscoug
Posts: 3
Joined: Wed Jan 09, 2013 8:17 am
Location: Russia

Re: Within variable differences?

What excellent words
Aaronwag
Posts: 14
Joined: Mon Sep 04, 2017 7:13 pm