Correlation

#1
Hi!

I was wondering if someone could give me some help. I am struggeling to understand how to run my correlation analysis using SPSS

This is my question:
To what extent do you agree or disagree on the following statements about advertising being entertaining.

1. strongly agree 2. agree 3. neither 4. disagree 5. strongly disagree

ENTERTAINMENT
Advertising is entertaining
Advertising is enjoyable
Advertising is pleasant
Advertising is fun
Advertising is exciting


1. strongly agree 2. agree 3. neither 4. disagree 5. strongly disagree

Over all attitude towards advertising is positive

What I want to do is to calculate how the entertainment (all five variables) are correlated with overall attitude towards advertising. i would like to get only one correlation value for all of the entertainment variables, not separate values... could anyone help me out? I am getting a bit fustrated...

Thanks in advance:)
 

Lazar

Phineas Packard
#3
Bit more information would be good. Does the overall attitude toward advertising contain the items you posted (or is it made up of the items you posted).

Also what is the question that you are trying to answer.
 
#4
Hi Lazar!

Thank you for your reply, I am really new with statistical tests.. I tried to ask my supervisor but he wasn't a great help:/

I am testing a hypothesis: "Entertainment of advertisement on web is positively associated with the attitude towards advertising"

I have in total 5 variables for enternainment and one variable for attitude.
I understood how to calculate the correlation between different entertainment variables and attitude towards advertising. However, I do not understand how could I calculate only one correlation for all four variables in relation to attitude towards advertising...

I don't know if this helped, but if you need more info to help me just let me know..:)
 

Lazar

Phineas Packard
#5
Ok so if I have this right you want to know whether entertainment (as a single factor represented by your five entertainment items) is associated with general attitude toward advertising (measured by a single item). Correct?

If so, it seems quite natural that you would first see if your five entertainment items form a scale by looking at the reliability. I don't use SPSS so much these days but, I think this should do the trick (replace *text* with the names of your variables):

RELIABILITY
/VARIABLES=*put your the names of your five entertainment items here separate by a space*
/SCALE('AdvertEnt') ALL
/MODEL=ALPHA
/STATISTICS=DESCRIPTIVE SCALE
/SUMMARY=TOTAL.

If this gives you a decent alpha level (over .75), which it most likely will because you have basically asked the same question 5 times, create a scale score:

COMPUTE AdvertEnt=MEAN(*your items here separated by a comma*).
EXECUTE.

then look at the correlation between your newly created item and the single item measuring advertising attitude:

CORRELATIONS
/VARIABLES=AdvertEnt *name of attitude variable*
/PRINT=TWOTAIL NOSIG
/MISSING=PAIRWISE.

Hope this helps.
 
#6
Lazar, you just save my life:))
I figured out how to do it with spss with your help! Thank you big time:)

Isn't cronbach's alpa only employed to test reliability to multi-scale items? I asked this my supervisor and he said I have to ask someone who knows better. WHAT A KNOWLEDGEABLE supervisor...
 

Lazar

Phineas Packard
#7
Lazar, you just save my life:))
I figured out how to do it with spss with your help! Thank you big time:)

Isn't cronbach's alpa only employed to test reliability to multi-scale items? I asked this my supervisor and he said I have to ask someone who knows better. WHAT A KNOWLEDGEABLE supervisor...
I assume this is the case with your study (you have a multi-item scale made up of the five items you first presented predicting the single attitude item which is not included in that scale). Your items are so similar that my feeling is that there is little point in treating them as separate. Creating a scale consisting of those five items called 'Entertainment value of Advertising' (or something) is most likely to be your best option in this regard. Hopefully this should be enough for your purposes. There are concerns in using cronbach's alpha with small scales like yours but I don't think you need to worry about this in your context.