Combine items

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

Posts: 3
Joined: Mon Jun 29, 2015 5:47 pm

Combine items

Postby bram1988 » Mon Jun 29, 2015 6:05 pm


I'm working on a study and i started all enthusiastic, but along the way this enthusiasm was replaced with me just being realistic.

So the problem is: amongst other concepts, 3 concepts and 12 measures are of interest. Blame (3 measures), Injustice(1 measure) and norm violations(8 measures). I did some Anovas and those basicaly indicated, you are not going to find any significant results. My supervisor recommended to find out if i could combine them all into 1 concept (12 measures).

I can't seem to find any literature on this, or find any published work that did this before. Ofcourse i combined the (3 and 8) measures for blame (and norm violations) to one concept(Blame & Norm Violations), but that has already been done in previously published work. So that seems easier and more legit to do.

My question is, what conditions have to be met, before i can take the mean of all 12 measures and combine it in 1 variable.
Inter-item correlations are also high (majority above 0,6) Lowest value 0,3.
After combining them, Kurtosis and skewness are fine.

Can anyone tell me what i should consider before i do this? and does anyone know a paper/article where the author(s) did something similar?

Thanks a lot!
You do not have the required permissions to view the files attached to this post.
Posts: 2699
Joined: Tue Jun 12, 2007 12:08 pm
Location: Florida, USA

Re: Combine items

Postby statman » Tue Jun 30, 2015 12:13 pm

If clear as to what you are saying and/or showing, all this notes is that your respondents are consistent but NOT that there isn't any sig results.

Exactly what is your premise, your study hypotheses
See the note below

NOTE: Please read the Posting Guidelines and always tell us your OS, the SPSS version and information about your study and data!

Statistical Services
Posts: 3
Joined: Mon Jun 29, 2015 5:47 pm

Re: Combine items

Postby bram1988 » Sat Jul 11, 2015 2:34 pm

Oke, hope I can clear some things up.

This is what i measured, among other things.

Blame (3 measures), Injustice(1 measure) and norm violations(8 measures). Theory suggests, that the attribution of blame and norm violations are elements of injustice. So i want to see if i can combine these 3 measures into 1 item (feelings of injustice). So I conducted a Factor analysis. But now i'm stuck as i don't know how to interpreted it.

Pattern Matrixa
1 2 3
n6 ,875 -,072 ,030
n3 ,843 -,193 ,107
n1 ,785 ,124 -,131
n4 ,753 -,262 ,040
n2 ,685 ,144 -,191
b1 ,020 -,913 -,025
b2 ,125 -,904 ,121
b3 ,080 -,769 -,225
u1 -,061 -,590 -,419
n5 ,082 ,011 -,891
n8 ,008 -,077 -,888
n7 ,237 -,107 -,701
Extraction Method: Principal Component Analysis.
Rotation Method: Oblimin with Kaiser Normalization.
a Rotation converged in 7 iterations.

This suggests a 3 factor model.

Factor 1: n6,n2,n1,n4 and n2
Factor 2: b1,b2,b3,u1??? (scores on two factors)
Factor 3: (u1),n5,n8,n7.

Can these 3 factors still be combined as 1 item? I feel like i'm missing one step.

For example,

Kmo and bartlett above 0,5 and significant etc etc, so we conducted a factor etc etc.. we observed the scree in the scree plot at three factors, thus indicating three-factor model of injustice. etc etc patern matrix, etc etc factor loadings above 0,5 etc etc

The three factors score ... on .... ?? so they could be safely combined.

Please bare with me, i'm not a SPSS guru, but I really can't find the answer i need.

Who is online

Users browsing this forum: No registered users and 1 guest