Online Article

Choosing between PLS-SEM and EFA for theory-driven instrument development

9th March 2024

Related topic: Quantitative research

Author: Ady Hameme N. A.

In the field of research, the development and validation of reliable and valid instruments are crucial endeavours. These instruments serve as tools for measuring complex theoretical constructs, enabling researchers to gather accurate and meaningful data. When developing an instrument, researchers often have an established theoretical or conceptual framework that guides the construction of the instrument and hypothesised factor structure. In such cases, the choice between Partial Least Squares Structural Equation Modelling (PLS-SEM) and Exploratory Factor Analysis (EFA) during the instrument development process can significantly impact the outcome.  

PLS-SEM is a composite-based approach that aligns well with theory-driven instrument development processes. When a strong theoretical foundation exists, PLS-SEM allows researchers to specify and evaluate a measurement model based on this established framework. This approach enables researchers to test the hypothesised factor structure and assess the extent to which the instrument accurately measures the intended constructs.

One of the key strengths of PLS-SEM is its comprehensive assessment of construct validity. It evaluates convergent validity, which measures the extent to which items truly represent the intended construct, through metrics such as average variance extracted (AVE) and factor loadings. Additionally, PLS-SEM assesses discriminant validity, which examines the distinctiveness of constructs from one another, using criteria like the Fornell-Larcker criterion and the Heterotrait-Monotrait (HTMT) ratio. By providing a thorough examination of construct validity, PLS-SEM ensures that the developed instrument accurately captures the theoretical constructs of interest.  

Moreover, PLS-SEM offers flexibility in model specification and modification, which can be advantageous during the iterative refinement of the instrument. As researchers analyse the data and gain insights, the ability to modify the measurement model in PLS-SEM facilitates the incorporation of necessary changes, ultimately leading to a more robust and valid instrument.

During the instrument development process, it is common to refine or adjust items based on theoretical considerations, expert feedback, or pilot study results. In cases where a few items from the original theoretical construct are changed or dropped, but the underlying constructs themselves are retained, it is still appropriate to choose PLS-SEM over EFA for measurement model validation.  

PLS-SEM, being a theory-driven approach, allows for the specification and evaluation of a measurement model based on the theoretical foundation. Even if some items are modified or removed from the initial item pool, the theoretical constructs can still be modelled and tested using PLS-SEM. One of the advantages of PLS-SEM is its flexibility in model specification and modification. As the instrument is refined by changing or dropping items, PLS-SEM allows researchers to modify the measurement model accordingly, while still maintaining the overall theoretical framework.

In contrast, Exploratory Factor Analysis (EFA) is a data-driven approach used to explore the underlying factor structure of observed variables when there is no a priori hypothesised model or theory. While EFA can be useful in the early stages of instrument development when researchers are still exploring the factor structure of items, it may be less suitable when a strong theoretical foundation already exists. When a theoretical construct guides the instrument development process, EFA may yield results that deviate from the established theoretical framework, potentially leading to inconsistencies or misalignments. Additionally, EFA has limitations in assessing construct validity and typically requires larger sample sizes to ensure stable factor solutions and reliable estimates.

It is worth noting that while PLS-SEM is often preferred when a prior theoretical construct is available, there may be instances where a combination of PLS-SEM and EFA can be beneficial. For example, researchers may conduct an initial EFA to explore the factor structure of the items and then follow it up with PLS-SEM to confirm and refine the measurement model based on the theoretical framework. However, if substantial changes are made to the theoretical constructs themselves, or if the changes to the items result in a significant departure from the original theoretical model, then revisiting the factor structure using EFA may be warranted.

In conclusion, when developing an instrument guided by a prior theoretical construct, and when refining or adjusting items while retaining the underlying constructs, the use of PLS-SEM is often a well-justified choice. Its alignment with theory-driven approaches, comprehensive assessment of construct validity, and flexibility in model modification make it a powerful tool for ensuring the developed instrument accurately measures the intended constructs. However, the decision should be made in consideration of the specific research context and goals, and researchers may consider a combination of PLS-SEM and EFA to leverage the strengths of both techniques.

Cite this article: Ady Hameme, N. A. (2024, March 9). Choosing between PLS-SEM and EFA for theory-driven instrument development. Retrieved <insert month> <insert date>, <insert year>, from https://www.myadvrc.com/publications/article-17

< previous next >