Help needed in SPSS

#1
Hello !

With a friend, we have a paper to write about the fail of measures to counter the illegal downloading of music in our country, for that we wrote a questionnaire and had about 300 respondents for it. But the thing is that neither of us is really good at statistical analysis and we decided to turn to the internet for answers. Any help will be greatly appreciated!
Here are the steps we did with our data base:
- A factor analysis to reduce the 15 elements (all likert scale) into components
- A Scale reliability analysis to make sure our components were reliable
- We then computed the new variables through the arithmetic mean of their items. Two principals components were created.
- We did an independent sample t-test between the two components
- Then a linear regression on those same components.
- After that we wanted to control a potential mediation of an element on this correlation through the Baron and Kenny method.
- We then did a regression analysis to test the impact of this element on one of our component
- We then did a independent sample t-test to test the impact of another element (Yes/No question) on one the same component
- And finally a one way anova to test the impact of another element (3 choices question) on the same component

Is there any step that seems wrong according to you ? If yes why cause we did try our best? :)
 

Karabiner

TS Contributor
#2
- A factor analysis to reduce the 15 elements (all likert scale) into components
- A Scale reliability analysis to make sure our components were reliable
- We then computed the new variables through the arithmetic mean of their items. Two principals components were created.
I guess it means that you performed a principal component analysis (not a factor analysis)
with 15 Likert-type items, and extracted 2 principal components?
-
We did an independent sample t-test between the two components
Not sure what you mean. What were the independent samples?
Or, did you perform a dependent samples t-test?

- Then a linear regression on those same components.
Since you did not yet explain for what reason you performed it, and how,
and with which variables, I am afraid I cannot comment on this.

With kind regards

Karabiner
 
#3
I guess it means that you performed a proncipal component analysis (not a factor analysis)
with 15 Likert-type items, and extracted 2 principal components?
Not sure what you mean. What were the independent samples?
Or, did you perform a dependent samples t-test?
Since you did not yet explain for what reason you performed it, and how,
and with which variables, I am afraid I cannot comment on this.

With kind regards

Karabiner
Yes Exactly, some of the items were not included in any of the components but two principal one have been created. Basically the questionnaires were in two groups, people would see an awareness campaign aiming to reduce illegal downloading, and the idea was to measure to impact of the campaign on the viewer. So the two components made were the impact measured of the campaign and the propencity of the viewer to engage themselves in illegal downloading in the future.

So a linear regression was made between those two components to measure a potential correlation between the impact the campaign had on the viewer and his or her propensity to engage in illegal downloading in the future.

Most of all, thank you for your time and answer !

Anabelle
 

Karabiner

TS Contributor
#4
I guess you performed your PCA with oblique rotation? Regarding the other steps,
the description is too vague to tell whether you were right or wrong.

With kind regards

K.
 
#5
Hello,

Yes, the PCA was made with oblique rotation. I understand that my explaination is not clear as I don't understand the subject really well myself. But could you answer a question regarding statistiques ?

If we make a PCA and get two components with it, can each component be used as both a dependant variable and/or independant variable ? Or will those components be either an IV or a DV ?
Right now we would like to do it by ourselves but I will explain the other steps clearly if we don't manage.
Thank you for your time and answer in any case.

Anabelle
 

Karabiner

TS Contributor
#6
I can see no reason here which would prevent the use of the principal components as variables (dependent or independent) in further analyses.


With kind regards

Karabiner