Questionnaire analysis... constructs or variables???

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

einno
Posts: 2
Joined: Thu Jun 06, 2013 3:33 pm

Questionnaire analysis... constructs or variables???

Postby einno » Thu Jun 06, 2013 4:11 pm

Hi everybody!!!

My question is very stupid (i think) but the matter is that i really have never practiced neither a questionnaire analysis nor SPSS.

Well... I have sent a questionnaire composed by 10 constructs. Each of those constructs represent a feature that I want to analyze, let me say for example...
1°= company financial performance
2°= use of IT
3°= Integration with suppliers
4°= integration with customers

Each construct is composed by at least 4 items (questions) and for each item the respondent has to indicated his agree/not agree choice based on a likert scale from 1 to 7.

In SPSS each item is represented by a different variable... so I have 40 variables when i open the program.

I want to investigate the relation that exists between the different constructs to test some assumptions that i have done before sending the survey.

Let me say i want to test:

Hp1: the higher use of IT, the higher company performance
HP2: the higher supplier integration, The higher use of IT



THE PROBLEM IS: I want to test the constructs relation, not the single items relation!!!

I can view the correlation between all the items, but i can't understand how to establish the correlation between the constructs.

Is necessary to select the most representative items for each construct and use only them?
Or is there the possibility to "clash" all the items for each construct into a unique indicator and use only it to establish the correlation between different constructs?

Thanks a lot!!!

Please help me and don't exitate to ask for more information.


Enrico
statman
Administrator
Posts: 2699
Joined: Tue Jun 12, 2007 12:08 pm
Location: Florida, USA

Re: Questionnaire analysis... constructs or variables???

Postby statman » Fri Jun 07, 2013 1:25 pm

Am not sure how you plan on testing before fielding as noted below
I want to investigate the relation that exists between the different constructs to test some assumptions that i have done before sending the survey.
Will you generated simulated data?

Also see the Note below
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!

Statman
Statistical Services
einno
Posts: 2
Joined: Thu Jun 06, 2013 3:33 pm

Re: Questionnaire analysis... constructs or variables???

Postby einno » Tue Jun 11, 2013 2:03 pm

Windowa Vista, SPSS 20
Study: engineering survey to test the correlation between the use of modular product and the integration with suppliers in a supply chain.



Hi,

Well, I have all the data in my dataset, because i have already sent the survey.

I want just to know how to calculate (from the data that I have collected) which is the correlation between the constructs. Unfortunately if i do a bivariate correlation I only calculate the correlation between the items, not the entire constructs.

Sorry if I wasn't clear enough before.

Enrico


statman wrote:Am not sure how you plan on testing before fielding as noted below
I want to investigate the relation that exists between the different constructs to test some assumptions that i have done before sending the survey.
Will you generated simulated data?

Also see the Note below
natdyas
Posts: 2
Joined: Thu Jul 31, 2014 4:51 pm

Re: Questionnaire analysis... constructs or variables???

Postby natdyas » Thu Jul 31, 2014 6:10 pm

I need the answer to this too!! Does anyone know how to do this?? How do put the variables under constructs and how do i test these constructs??
GerineL
Moderator
Posts: 1477
Joined: Tue Jun 10, 2008 4:50 pm

Re: Questionnaire analysis... constructs or variables???

Postby GerineL » Mon Aug 11, 2014 2:51 pm

what do you mean by constructs? Mean of separate variables/ then first compute these means and then test it.

Please provide more information, otherwise it is impossible to answer your question.

Maybe canonical correlation is the answer, but I think it may be easier than that.

Who is online

Users browsing this forum: No registered users and 4 guests

cron