QUOTE(Blofeld @ May 9 2014, 12:09 PM)
Go to Transform, then click Compute Variable
Key in (Q1+Q3+Q5+Q10)/4 and you will get a summated score for Construct Belief.
Anyway, have you run factor analysis and reliability analysis for your variables in the first place?
That's not right, you're just measuring the relationship between the mean score of all those questions. I don't think that is what the TS is askingKey in (Q1+Q3+Q5+Q10)/4 and you will get a summated score for Construct Belief.
Anyway, have you run factor analysis and reliability analysis for your variables in the first place?
The correct answer is that you cannot test a multivariate factor with a t test unless there is valid reason that all the 5 questions are simply different samples of the same variable. Them you take the sample mean and test it.
If it's 5 questions testing slightly different things, you need to use multiple factor analysis to find the underlying variable that is affecting the dependent variable.
This post has been edited by mumeichan: Oct 23 2014, 10:47 PM
Oct 23 2014, 10:38 PM

Quote
0.0166sec
0.48
6 queries
GZIP Disabled