# Thread: How can I find out, if subscales are independent?

1. ## Re: How can I find out, if subscales are independent?

Originally Posted by LtStarbuck
Hey I have another question. I just read in the book of Andy Field about collinearity. Isnt that something, I can use? Its there to check for the independence of my variables aswell, right?
collineartiy as in in multiple regression? that just means your variables are correlated. if you're doing factor analysis, you need some of that.

2. ## Re: How can I find out, if subscales are independent?

Yes exactly. Remember that the EFA was one suggestions of my prof to me, to check for independence of my subscales. But the independence is mostly important to decide, how I implement them in my mediation (one by one separately or all together. The last method would require the IVs to be independent / uncorrelated).
So instead of going of the way to different methods that dont seem to useful anyway, as you pointed out for the EFA, instead I could just use the check for (multi)collinearity of my IVs.

3. ## Re: How can I find out, if subscales are independent?

I actually had some free time to so something with AMOS (yes we have it on the campus, I did not know), and can make a CFA :-)
Since I have four subscales, that are supposed to be independent, is this here the model, I need for my analysis, including the covariances?

4. ## Re: How can I find out, if subscales are independent?

oh god, i had forgotten how horrible AMOS is.

anyhoo, the key here would be to see what the p-value of the correlations between those latent factors are. if significant = subscales NOT independent.

you could also test the model where you constrain the factors to be orthogonal VS a model where they're freely estimated and see which model fits better. i bet my brownies the model with freely-estimated correlations fits better

5. ## Re: How can I find out, if subscales are independent?

oh god, i had forgotten how horrible AMOS is.
Although I never worked with it before (and never will again), I can say, it is really hideous and a total nightmare concerning ergonomics.

Anyway I have been looking for the correlations and found them unter "Estimates/scalars/correlations". Are those the one I am looking for? I have all my estimates there, but no p-values. Am I looking at the wrong place?

How can make the models with orthogonal and freely estimates? (betting brownies is a good idea, I think :-D )

6. ## Re: How can I find out, if subscales are independent?

Since I have four subscales, that are supposed to be independent, is this here the model, I need for my analysis, including the covariances?
Didn't you say that you want to find out whether 4 already existing
scales are independent of each other? Seemingly, you analyse
assoctiations between latent dimensions, not between scales.
Simply using the total scores from each of the 4 existing scales
and inspecting the 6 pairwise correlations would already answer
your question (but I probably am missing something here).

With kind reagrds

K.

7. ## Re: How can I find out, if subscales are independent?

To be honest, Im not sure anymore, whats the right method *cough and looking down to the floor*.
So I use an inventory with four subscales, that has already been established. For my analysis I need to know, if those subscales are independent or not, because if they are, I can put then all together in my analysis (for example as IVs my mediation or multiple regressions). If they are not independent/uncorrelated, I need to analyse every subscale separatly.
At the start I thought, a correlation-table is enough but since my prof told some stuff about EFA, I thought, that there is more to it. Im not sure.

8. ## Re: How can I find out, if subscales are independent?

So, AFAICS you are using well-established scales, which you won't change as a result of your preleminary analysis, and you just have to look at the degree of interelatedness of the scales within your current sample. Did you describe anywhere in the thread *why* you cannot use all 4 scales simultanously, if they are correlated? Moreover, interrelatedness is not a simple yes/no categorization, ususally the actual degree of interelatedness (the size of the correlation coefficients for example) is what matters.

As stated above, you could have a look at the 6 pairwise correlations. A general measure for the degree of interelatedness of the 4 scales would be Cronbach's alpha (if you treat the 4 scales oft the test like 4 items of a scale).

Regarding EFA, I guess that you have to talk to your supervisor about why this should be necessary here, if you aren't interested in the dimenaionality (latent structure) of the test, just in the correlation between the already existing scales.

With kind regards

K.

9. ## Re: How can I find out, if subscales are independent?

Did you describe anywhere in the thread *why* you cannot use all 4 scales simultanously, if they are correlated?
When I do my mediation-analysis with 4 IVs and they are correlated, it is possible, that my effects are way smaller. So Hayes (the guy, who made the PROCESS-Makro I use for my Mediation in SPSS) give different options on how to use the IVs:
If the are uncorrelated, I use one if the DV as DV and the other three as covariates, and this for times, once for every IV.
If they are correlated, I can not use the other three as covariates, instead I calcuate for every IV completely separately.

Also I have to do regressions, where die subscales are IVs and other variables the DVs. When my subscales are not correlated, I can use multiple regression. But when they are correlated, I again would have a much smaller effect and probably no effect at all from my DV, since they can cancel each other out. Instead of doing three multiple regressions (one for each DV) with my four subscales as IVs, I would have to do 12 normal regressions.

My results for the correlations are, that 5 of the 6 are signifikant, 4 of them with .3 or higher.
Your idea with Cronbachs Alpha seems good. I have to do a reliability-analysis, right? When I do this, my Alpha is .56 (standardised its 0.58). So besindes the not so small correlations, the Alpha tells me, that the for subscale-scores are not very similar. Multicollinearity say the same thing, with VIF around 1.2 for each subscale.

10. ## Re: How can I find out, if subscales are independent?

When I do my mediation-analysis with 4 IVs and they are correlated, it is possible, that my effects are way smaller. So Hayes (the guy, who made the PROCESS-Makro I use for my Mediation in SPSS) give different options on how to use the IVs:
If the are uncorrelated, I use one if the DV as DV and the other three as covariates, and this for times, once for every IV.
If they are correlated, I can not use the other three as covariates, instead I calcuate for every IV completely separately.
I must admit that I don't quite understand what you are undertaking,
especially the last statement sounds wrong.
Also I have to do regressions, where die subscales are IVs and other variables the DVs. When my subscales are not correlated, I can use multiple regression. But when they are correlated, I again would have a much smaller effect and probably no effect at all from my DV, since they can cancel each other out.
No, that is wrong. Adding variables to a regression equation, whether they
are correlated or uncorrelated with the other independent variables, CANNOT
cancel effects out and CANNOT reduce explained variance. Every additional
variable MUST increase explained variance by >= 0% . Or did you mean
something else than explained variance when saying "probably no effect at
all from my DV" ?
4 of them with .3 or higher.
Well, that still looks quite common and generally acceptable.

With kind regards

K.

11. ## Re: How can I find out, if subscales are independent?

Hey K!

Adding variables to a regression equation, whether they
are correlated or uncorrelated with the other independent variables, CANNOT
cancel effects out and CANNOT reduce explained variance. Every additional
variable MUST increase explained variance by >= 0% . Or did you mean
something else than explained variance when saying "probably no effect at
all from my DV" ?
No, thats what I meant. Well sure, the overall effect will get bigger, but the specific effects can get smaller, because if the IVs are highly correlated, the effect on the DV must be shared between the IVs. Anyway, the values of my VIF, Alpha and correlations seem to be fair enough to draw the conclusion, that my IVs are not correlated in a relevant magnitude. Thats the important thing :-)
Thanks a lot for the help!

12. ## Re: How can I find out, if subscales are independent?

if the IVs are highly correlated, the effect on the DV must be shared between the IVs.
Not necessarily, since suppressor effects can increase the regression weight
of a variable if a correlated variable is added. But seemigly yuo are not
interested in the joint performance of all 4 variables together, but in the
univariate relationsips between the respective predictors and the independent
variable - in that case, the whole discussion seems quite pointless from the
beginning. If you are not interested in variance explained by the combination
of the 4 variables, then you just could have perfomed the 4 univariate analyses
and that's it (a multiple regression could be added out of curiosity maybe).

With kind regards

K.

+ Reply to Thread
Page 2 of 2 First 1 2

 Tweet

#### Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts