Hi Kanika,

How are you doing ? Well, I hope.

Regarding your question, my understanding is that occupational choice is a variable of the model you want to study (or not ?).

This was not that clear for me.

If this is a variable ot the model (IV or DV), this could not be a control variable.

A control variable is a variable that the researcher purposely neutralizes or cancels in a research in order to prevent it from interfering in the analysis of the relationship between the independent and dependent variables.

For instance, if company size is a control variable this means that you must use just one type of company size in your analysis in such way there will be no variation regarding company size when deal with the statistics. The parameters to define the company size you will use is yours, but important is that you will deal with company sizes quite similar.

In this sense, you don`t have to use the control variable in your model since it was controlled previously to data collection.

On the other hand, if you didn't do such control but you suspect that a certain variable will influence the responses of your model, you may check if this variable is a moderator, influencing the relation IV --> DV depending upon its intensity.

Using the same example of company size, you have large companies and small companies in your sample, you must build a general model, w/o using company size in it and after that check if there is any difference in the relation IV --> DV when you consider a subsample of small size and another one of large size.

In your case, if you want to compare the model for occupational choice of skilled returnees against occupational choice of non-migrant, you build a general model, w/o occupational choice in it, and apply the model to both groups, checking if there is any difference between the two.

This can be done by Testing for Multigroup Invariance and Byrne's book (

Byrne, B. M. (2010). Structural Equation Modeling with AMOS: basic concepts, applications, and programming (2nd ed.). New York: Routledge.), chapter 7 (Testing for the factorial equivalence of scores from a measuring instrument - First-order CFA model) should be quite useful for that.

If you are not using AMOS but SmartPLS, this can also be done using

Vinzi, V. E., Chin, W. W., Henseler, J., & Wang, H. (Eds.). (2010). Handbook of Partial Least Squares. Berlin: Springer. https://doi.org/10.1007/978-3-540-32827-8, section 21.4.2 - Multigroup Analysis.

Hope this helps.

If you need anything else, feel free to call me.

Regards.

The role of control variables in a quantitative analysis pertains to spuriousness. If you have a regression model with just two variables, then you're likely to see higher effects. For example, if I examine the impact of education on income, then I may see a strong relationship--those who have higher education tend to have higher income. If I add control variables such as years of experience, then I will get a reduced effect size for the education variable. That's because years of experience also predicts income. Control variables are there to help account for potential spuriousness in your model since you cannot experimentally vary any of these variables.

To answer your other questions:

1. If you include occupation of choice, then you cannot do traditional regression analyses. You may want to make occupation of choice an outcome variable and try to predict it using a procedure called multinomial logistic regression. It basically checks to see which variables increase or decrease the odds of having a certain nominal outcome when compared to a reference category.

2. Your question about analyzing non-migrant as a control variable simply requires that you dummy code (code as zero for no or one for yes) whether or not the respondent qualifies for your definition of migrant. Include it in the analysis. You typically want the affirmative to be coded as one, but you can also base it on which is most common. For example, women are often coded as one in regression analyses in criminal justice research due to their reduced frequencies.

I hope this helped!

If you questions is whether going abroad affects job choice then you need a group who have gone overseas and a group who have not and you want both groups to be as comparable as possible on other things (your controls if you like but this is not an experiment so I prefer the term comparators) except for going abroad. One way of getting comparators is via matching

have a look at

https://gking.harvard.edu/presentations/matching-methods-causal-inference-3

So the role of comparators/ controls is to achieve equipoise, that is to hold off the effect of other variables related both to the outcome (job choice) and the exposure ( the key variable of causal interest) going abroad.

King makes the argument that matching then regression is more robust rather than regression alone in holding off the effect of other variables.

Thanks Sir for this detailed explanation. Certainly helped in my doubt

Dear Kanika.

In quantitative models, a control variable is the one that allows you to isolate the selection bias in a certain observation group. This aims to your statistical inferences are controlled by certain variables that could absorb the explicability of your model, or in other words, increase your error. The most common control variables are the demographic characteristics of individuals such as gender, age and education level.