Hello Razia! By definition, SEM is a system of relations among latent and measured variables and gender and age are none. However, you can do an SEM analysis when you are interested in different groups. Joreskog has an excellent article that might be of help to you. In his study, Joreskog had data for two groups that had taken a scholastic tests and conducted the SEM analysis. In your case, you apply the SEM two-group techique if your two groups are, say, boys and girls, or groups set by age ranges. the article's name is "validating Psychometric Assumptions within and between Several Populations". Education and Psychological Measurement, 1977, 37. Dr. Joreskog has a profile here in ResearchGate and you might be able to download the mentioned article from his profile. If not, please let me know and I will see how I can share a copy with you. I hope this helps. Cheers!

Thanks Ma'am Patricia, you indeed solved my problem,  I will surely read the article you referred, and in case I don't get the access, I will revert back to you. Thanks again

You don't have to go as far as running separate groups (which will test for interaction effects or moderation). All you need to do to include a demographic variable in SEM is to specify a reliability (error term) for it. Things like age and gender are usually treated as measured without error.

Thanks Sir David, Could you please explain how age and gender will be treated in SEM as controls , for example we have 5-point likert scale for all the variables included in SEM, Gender is a nominal variable, male coded 1, female coded 2, whereas age is a continuous variable , say we have five age brackets, how can we use age and gender as controls

How you enter these variables depends on the program you are using, and I am not familiar with AMOS. Note also that there is no problem with having age as a continuous variable.

I did a quick Google search and here is one link that discusses using a two-category variable such as gender in AMOS.

http://www-01.ibm.com/support/docview.wss?uid=swg21478651

On the general topic of control variables, in SEM terms these are typically treated as "exogenous" so you might try search on that term.

Thanks Sir David for your valuable input,  in fact the link is really helpful, thanks again

Dear Razia.

Yes you can use demographic characteristics of the population as a control variable if you have the reason from literature to do so. Otherwise you can run the model without control variable. Normally researchers use demographic characteristics as a moderating variable in the model.

Handling control variable in Amos is easy but quite messy. You have to include the control variable at the very begining of analysis.then, you have treat control variable as exogenous construct and covary it with other exogenous construct to impose on endogenous construct. If estimate of endogenous construct is affected when include the control variable,then, it might be control variable is really exist in these relationships.

Thanks Sir Awang and Sir Wan Mohamad for your valuable feedback on the subject, it is very helpful, thanks again

Hi, if you want enter age and sex as a control variable in the model, you must use a combination of SEM and regression such as MIMIC model or SEM regression model. you can do this model with STATA.

Thanks Sir Ahmad for your insightful guidance

NUMERICAL SOLUTION OF PARTIAL DIFFERENTIAL EQUATIONS (1- AND 2-D)

Hi I am struggling with trying to control demographic variables in full SEM model in Lisrel..can anyone help!

Ashita, if the correlations among study constructs and demographic variableso are not significant, you don't need to control them

I have develop SEM model which examines the the effect of mediator (Z) on the relationship between X and Y. I want to include gender as a control variable in this model. Please give your valuable suggestion in this matter.

Dear Sonal,

Please observe the correlations between IV and demographics, in case the value of correlation coefficient is not high, you don't need to control them

Thank you for your suggestion. My model is build on data which collected from different three state. In this situation, can I go for multi group analysis? please suggest for this. If the comparative net P value is significant, I can run the model from the dataset (which includes data from three states)

Hi, good feedback from all of you.

In my case I used gender as moderator and two-group analysis with AMOS. And I followed main steps to test my hypotheses.

Hi everyone,

Can I please ask how to add control variables in LISREL? What would be the syntax for this? I couldn't find a way to add any CVs...this is very frustrating! Please give me some advice on it. Thank you soooooo much!

Hi,

Now the paper is accepted in the International Journal of Gender and Entrepreneurship indexed in Emerging Sources Citation Index and ranked by AIDEA (Italy), American Sociological Association, Australian Business Deans Council (ABDC) Quality Journal List, Australian Research Council (ERA Journal List), Chartered Association for Business Schools (CABS, UK) Academic Journal Guide 2018, CNRS (France), Scopus.

Article Influences of Role Models and Gender on Saudi Arabian Freshm...

Hi,

When do you need to include demographic variables in SEM analysis, where the interest is to examine relationship between measured predictors and dependent variable/s?

Thanks!