Current research in the field of business and management is done using a lot of explanatory approach. According Sugiyono (2012: 21) is an explanatory research study intends to explain the position of the variables that were analyzed and the relationship between one variable with another. While the characteristics of this study is replicating, so the hypothesis test results must be supported by previous studies, which are repeated with other conditions that are more or less equal. Currently regression method is the method most often used to conduct quantitative research. With the development of research methods in the field of business and management, then the method of regression analysis found to be able to answer the issues raised by the study researchers. Structural Equation Modelling (SEM) is one method that is currently used to close the existing weaknesses of the method of regression. Experts mengelompokka SEM research methods into two approaches. The first approach is referred to as Covariance Based SEM (CBSEM) and other approaches is Variance Based SEM or better known as Partial Least Squares (PLS). To perform analysis using CBSEM then the software is often used by AMOS and LISREL while for PLS software that is often used is smartPLS, warpPLS and XLStat.
This module discusses the guides do an analysis in business research and management. In particular, this module will discuss the reason for using PLS, to evaluate the outer model (outer model of evaluation), the evaluation of inner model (inner evaluation models), path analysis, testing mediation / indirect influence and multi comparison group
Why Use SEM
Before operating the PLS as part of the SEM, it would be much better if it is understood terlabih first reason and the purpose of performing an analysis using PLS. As indicated previously that the regression method are weaknesses that reduce komprehensifitas of analysis. The following will be discussed on those weaknesses.
Normality Assumption Data
In the method using regression analysis, there are assumptions that should be examined by the researchers to ensure that the regression equation formed BLUE (Best Linear Unbiased Estimate). One assumption that is often a stumbling researchers is the assumption of normality.
In the study of business and management, especially in marketing and human resources that make the measurement of perception would be hard to to get a normal distributed data. Therefore, by using regression method, the researchers will find it hard to get the regression equation BLUE.
PLS using bootstrapping method or random multiplication. Therefore, the normality assumption would not be a problem for PLS. In addition to the data associated with normality, with bootstrapping the PLS does not require a minimum number of samples. Research has small samples can still use the PLS.
@Rusiadi: sorry for contradicting you, but you don't need the normality assupmtion for OLS being the BLUE. You rather need this assumption for making inference.Furthermore, see https://www.researchgate.net/publication/277816598_PLS_CB_SEM_a_weary_and_a_fresh_look_at_presumed_antagonists_keynote_address
@Gabrielle: I think the following references might be helpful: Sarstedt et al. (2011); Henseler (2012).
Sarstedt M, Henseler J, Ringle CM (2011) Multigroup analysis in partial least
squares (PLS) path modeling: Alternative methods and empirical results. Advances in International Marketing 22(1):195{218
Henseler J (2012) PLS-MGA: A non-parametric approach to partial least squaresbased multi-group analysis. In: Challenges at the Interface of Data Analysis,
Computer Science, and Optimization, Springer, pp 495{501
Conference Paper PLS & CB SEM, a weary and a fresh look at presumed antagonis...
Cite
28th Feb, 2020
Shinaj Valangattil Shamsudheen
International Centre for Education in Islamic Finance
Dear Gabrielle Daniels-Gombert ,
You may refer to chapter 4 of the book 'Advanced Issues in Partial Least Squares Structural Equation Modeling'