# What is the most appropriate statistical procedure?

#### ornat turin

##### New Member
I have several independent variables from different scales: gender, income, educational institution, age,
And I want to estimate which of them affect the dependent variable which is,
Attitudes towards a political issue.

And I want to learn about their differential power in relation to the dependent variable
Do I need logistic regression? Is there another suitable procedure that provide an answer?

Thank you,

Ornat

#### Karabiner

##### TS Contributor
And I want to estimate which of them affect the dependent variable
Do you mean, you want to know the association with the dependent variable for each
of the variables separately?
which is, Attitudes towards a political issue.

And, second most important question, how large is your sample size?

And I want to learn about their differential power in relation to the dependent variable
I am not sure whether I understand this.

With kind regards

Karabiner

#### ornat turin

##### New Member
Thank you for your willingness to help

About five hundred students completed a political attitudes questionnaire, I want to know what is the connection between their position and: the academic institution where they study, their age, gender and income
I can do this with various statistical tools based on the type of independent variable... say Pearson with the income and chi2 l in relation to gender.
But how do I compare or standardize, how do I evaluate what is the most powerful background variable among the independents I have measured?

#### hlsmith

##### Less is more. Stay pure. Stay poor.
HOW IS THE DEPENDENT VARIABLE FORMATTED??? This dictates the approaches we recommend. You need to likely use a regression, but which one depends on the dependent variable.

Also, if you have students clustered in school/institutions, you also likely need to control for this via multilevel modelling. So us an example of what these data look like - if they are confidential, you can make up the values.

#### gw1

##### Member
Agreed. Asked another way, What options did Attitudes towards an issue have?

#### ornat turin

##### New Member
The dependent variable is a series of averages.
The average level of liberalism, the average level of equality towards women, the average of the willingness to promote minorities, and other averages of political positions in clusters of topics.

In addition, I already used factor analysis for types of political positions, which also constitute a dependent variable.

So what kind of regression should I run, in your opinion?

#### gw1

##### Member
How all of the parts you just described fit together, to make a clear research question, is still unclear, to me. To make it really simple, with one dependent variable, are you asking Which variable(s) from gender, income, educational institution and age explain more of the variance of the average level of liberalism?
I ask you, how are you calculating the average level of liberalism?

#### ornat turin

##### New Member
This is the average of a series of statements to which consent was expressed across the Likert scale

#### gw1

##### Member
As hlsmith suggests, there may be a higher likelihood of model considerations / suggestions if you post an example of the data