Partial Least Square (PLS)

Partial Least Square (PLS) PLS path modeling is commonly used in business and social sciences with the purpose of predicting endogenous latent variables as well as estimating and examining the relationships between these latent variables (causal analysis).

Compared to the covariance-based structural equation modeling techniques, PLS is a vigorous technique with usually less restraining distributional requirements and assumptions. Additionally, PLS-PM allows constructs to be formatively and more willingly measured.

There are no specific prerequisites for this course though some knowledge of statistics is presumed. Furthermore, having some knowledge of linear regression techniques will also be beneficial for the participants.

The main objectives of this course are:

 (1) to provide detailed methodological introduction into the PLS path modeling approach (particularly, the nature of causal modeling, analytical objectives, some statistics)

 (2) the assessment of measurement results, and

 (3) complementary analytical techniques. An essential part of this course is the practical applications and the use of a PLS software application.

This course is perfect for you if you are doing your research based on Structural Equation Modelling (SEM).

However, if you do not have time to attend this course, you can have some private sessions with me or alternatively, you can order your project and we will do that for you.

Structural Equation Modelling (SEM)

Authors (Gefen, et al., 2000) believe this approach has many advantages over other methods, for instance Multiple Regression. SEM is also good in terms of path and factor analysis; especially when we are looking for reliability and validity of a research outcome from different angles, which is available through this approach.

 In SEM approach, Partial Least Squares (PLS) method is one of the best. This method has good advantages compared to others, for example LISREL. Whereas sample size is important in SEM, PLS is good for a small sample size research (Gefen et al., 2000) such as our sample, of 300 people. According to Gefen et al. (2000) and Chin (1998), in PLS the minimum sample size need to be 10 times the number of items related to the most complex variable or constructs. “PLS combines a factor analysis with multiple linear regressions to estimate the parameters of the measurement model (item loadings on constructs) together with those of the structural model (regression paths among the constructs) by minimizing residual variance.” (Gefen and Straub, 2004).

With the help of PLS we are able to test

-          validity of discriminant

-          convergent scales

We have private courses in PLS. However, if you do not have time to attend this course, you can have some private sessions with me or alternatively, you can order your project and we will do that for you.