Dear Manel Miloudi

I would suggest you to use the structural equation modeling (SEM) which is a combination of factor analysis and path analysis.

Using (SEM), you can also test if there is a significant moderation effect from any of the control variables e.g. Age, Gender,... And more.

I would like to briefly hight light that there are two approaches of SEM, which are called covariance-based structrual equation modeling (CB-SEM) or variance based that so often called partial least square structural equation modeling (PLS-SEM).

Here is a link where you can find a question about using (CB-SEM) or (PLS-SEM) to give you a glance about the two approaches. https://www.researchgate.net/post/Whether_analysis_with_CB-SEM_by_AMOS_will_have_better_results_comparing_with_PLS-SEM

You can run the SEM using either

(CB-SEM) via AMOS, R (laavan) or

(PLS-SEM) via SmartPLS or R (plspm)

I would recommend you begin with some simpler approaches. Assuming both of your variables (self efficacy and leadership style) are measured on discrete scales, you might consider a contingency table analyzed with the Chi Square statistic. This would tell you if there is a significant relationship and follow ups could tell you the direction. If one of the variables is continuous, a relatively simple approach would be to use regression analysis with ordinary least squares.

Good luck,

Jim McLean

Data on ordinal scale (likert scale) with more dependent variables included in model with 1 independent variable is a very specific combination for the analysis. It would require a form of Multivariate Ordinal Regression (stands for multiple dependent variables and ordinal scale).