I need help choosing my tets

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

Neeks
Posts: 1
Joined: Mon Mar 16, 2015 4:51 pm

I need help choosing my tets

Postby Neeks » Mon Mar 16, 2015 5:11 pm

Hi everyone!

I have trouble choosing the correct tests to carry out my analysis.
My survey consists of:

- 34 items, each with a 5-point interval scale; they all measure 1 construct;
- items asking about gender (male or female - nominal), place of birth (3 levels - nominal), education level (4 levels nominal), education type (2 levels - nominal);

I need to find out how the variables interact:
if gender, place of birth, education type, level of education have any effect on the 34-item construct and if so - what is it?

Problem is, I'm not sure which test to choose from. I was thinking about the Univariate analysis (from gen. linear model) with the construct I'm measuring as the dependant variable and the rest as factors. But I'm not sure if that is correct or what other tests to try out.

Also, since the survey is translated into a different language, I was thinking of giving the original survey (in English) to another group of people in the US (via internet) to see if there are any significant differences. How can I compare the results between the 2 tests?

Lastly, I should do a factor analysis (I was thinking Pr. Axis factoring) to make sure they do indeed measure what they are supposed to.

Can you tell me if I'm going the right way, what other tests I can try out (regression, T-test maybe, I dunno) if any and if I'm missing out on something? Thx
GerineL
Moderator
Posts: 1477
Joined: Tue Jun 10, 2008 4:50 pm

Re: I need help choosing my tets

Postby GerineL » Fri Mar 27, 2015 3:38 pm

Neeks wrote:Hi everyone!

I have trouble choosing the correct tests to carry out my analysis.
My survey consists of:

- 34 items, each with a 5-point interval scale; they all measure 1 construct;
first compute the scale using "compute"
- items asking about gender (male or female - nominal), place of birth (3 levels - nominal), education level (4 levels nominal), education type (2 levels - nominal);

I need to find out how the variables interact:
if gender, place of birth, education type, level of education have any effect on the 34-item construct and if so - what is it?

Problem is, I'm not sure which test to choose from. I was thinking about the Univariate analysis (from gen. linear model) with the construct I'm measuring as the dependant variable and the rest as factors. But I'm not sure if that is correct or what other tests to try out.
yes, you can do that via univariate anova.
you can also look at interactions between the predictors with this test. look at some casestudies online.
Also, since the survey is translated into a different language, I was thinking of giving the original survey (in English) to another group of people in the US (via internet) to see if there are any significant differences. How can I compare the results between the 2 tests?
you can also do that via anova.
Lastly, I should do a factor analysis (I was thinking Pr. Axis factoring) to make sure they do indeed measure what they are supposed to.
factor analysis does not tell you if the variables measure what they supposed to, it only tells you if the items measure the same thing.
whether or not you should do that depends on common practice in your field, and whether or not you use an existing scale or not.

Who is online

Users browsing this forum: No registered users and 1 guest

cron