The elephant in the room: Predictive performance of PLS models

doi:10.1016/j.jbusres.2016.03.049

Journal of Business Research

Volume 69, Issue 10, October 2016, Pages 4552-4564

https://doi.org/10.1016/j.jbusres.2016.03.049 Get rights and content

Abstract

Attempts to introduce predictive performance metrics into partial least squares (PLS) path modeling have been slow and fall short of demonstrating impact on either practice or scientific development in PLS. This study contributes to PLS development by offering a comprehensive framework that identifies different dimensions of prediction and their effect on predictive performance evaluation with PLS. This framework contextualizes prior efforts in PLS and prediction and highlights potential opportunities and challenges. A second contribution to PLS development lies in proposed procedures to generate and evaluate different types of predictions from PLS models. These procedures account for the best practices that the new framework identifies. An outline of the many powerful ways in which predictive PLS methodologies can strengthen theory-building research constitutes a third contribution to PLS development. The framework, procedures, and research guidelines hopefully form the basis for a more informed and unified development of the rigorous theoretical and practical applications of PLS.

Section snippets

Introduction and motivation

Business research has adopted PLS path modeling, known simply as PLSPM or just PLS, as a methodological tool that bridges exploration with explanation. The popularity of PLS stems partly from its ability to produce parameter estimates of complex models without many of the distributional and other constraints of traditional parametric methods. PLS achieves these estimates by using an iterative estimation procedure considered “predictive” in nature (Hair, Ringle, & Sarstedt, 2011). Yet, the PLS

Two key features of predictive performance in predictive analytics

As early as the 1950s, thinkers of the philosophy of science recognized the fundamental difference between explanation and prediction and the necessity of both (Dubin, 1969). Management research typically focuses on explanatory models. But with the recent popularity of Big Data and predictive analytics, some predictive modeling now appears in the management literature.

Before undertaking an examination of predictive performance in the context of PLS models, please note that in contemporary

Predictive performance of PLS models: A conceptual framework

Unlocking the full potential of PLS predictions requires understanding the full picture of what is possible with predictions in PLS, where current proposals fit into this picture, and which of these proposals supports the goal of predictive performance evaluation of a PLS model.

A procedure for evaluating predictive performance of a PLS model

The lack of established techniques and tools to generate and evaluate predictions from PLS has hindered progress on creating predictive methodology for PLS. This study proposes a set of procedures for prediction and evaluation that together create a framework for evaluating the predictive performance of PLS models. The framework is tailored to the desirable properties of predictive performance evaluation highlighted earlier: out-of-sample, item-level, and operative. Specifically, the procedure

Enhancing PLS models using theory-informed prediction

As demonstrated, PLS models can be used not only to quantify and test a causal theory but also to generate point predictions and prediction intervals for manifest items both in-sample and out-of-sample. These predictions are based on a causal model, unlike predictions typically generated by machine learning algorithms. The ability of PLS to generate predictions and evaluate their performance gives PLS a self-diagnostic capability that permits use of predicted outcomes and metrics to evaluate

Conclusions and future directions

Being able to generate empirical case-wise out-of-sample predictions from a model and to evaluate the predictive power of explanatory models is vital to theory building and evaluation. Prediction using PLS models differs from machine learning algorithms and even from regression models in that PLS allows a researcher to incorporate a theoretically justified causal structure into the prediction process. Although regression models are useful for prediction, “[they are] most effective when dealing

Acknowledgments

The authors thank Kwok-Kee Wei and Loretta Fung for their feedback and suggestions. Thanks also to the organizers and participants of the 2nd International Symposium on Partial Least Squares Path Modeling (Seville, 2015) for discussions and feedback on prediction and PLS. Special thanks (in alphabetical order) to Jean-Michel Becker, Gabriel Cepeda, Wynne Chin, Nicholas Danks, Theo Dijkstra, Jorg Henseler, Christian Ringle, Ed Rigdon, José L. Roldán, Mikko Rönkkö, and Marko Sarstedt for sharing

References (39)

J.S. Armstrong
Illusions in regression analysis
International Journal of Forecasting
(2012)
T. Dijkstra
Some comments on maximum likelihood and partial least squares methods
Journal of Econometrics
(1983)
J.-P. Lam et al.
Bootstrap prediction intervals for single period regression forecasts
International Journal of Forecasting
(2002)
M. Tenenhaus et al.
PLS path modeling
Computational Statistics and Data Analysis
(2005)
A.G. Woodside
Responding to the severe limitations of cross-sectional surveys: commenting on Rong and Wilkinson's perspectives
Australasian Marketing Journal
(2011)
A.G. Woodside
Moving beyond multiple regression analysis to algorithms: calling for adoption of a paradigm shift from symmetric to asymmetric thinking in data analysis and crafting theory
Journal of Business Research
(2013)
R.J. Ball
The significance of simultaneous methods of parameter estimation in econometric models
Journal of the Royal Statistical Society, Series C
(1963)
D. Barclay et al.
The partial least squares (PLS) approach to causal modeling, personal computer adoption and use as an illustration
Technology Studies
(1995)
R.M. Baron et al.
The moderator–mediator variable distinction in social psychological research – conceptual, strategic, and statistical considerations
Journal of Personality and Social Psychology
(1986)
J.-M. Becker et al.
Predictive validity and formative measurement in structural equation modeling: embracing practical relevance, thirty fourth international conference on information systems
(2013)

W.W. Chin

The partial least squares approach to structural equation modelling

W.W. Chin

Bootstrap cross-validation indices for PLS path model assessment

W.W. Chin

On partial least squares variance-based SEM (VBSEM) vs. covariance-based SEM (CBSEM) for confirmatory analysis: It's all about the components and variance explained

G.A. Churchill

A paradigm for developing better measures for marketing constructs

Journal of Marketing Research

(1979)

J. Dana et al.

The superiority of simple alternatives to regression for social science predictions

Journal of Educational and Behavioral Statistics

(2004)

A. Diamantopoulos et al.

Guidelines for choosing between multi-item and single-item scales in construct measurement: a predictive validity perspective

Journal of the Academy of Marketing Science

(2012)

T. Dijkstra

Latent variables and indices: Herman Wold’s basic design and Partial Least Squares

T.K. Dijkstra et al.

Consistent partial least squares path modeling

MIS Quarterly

(2015)

R. Dubin

Theory building: A practical guide to the construction and testing of theoretical models

(1969)

Cited by (796)

Harnessing green innovation via green transformational leadership in Italian luxury hotels: Key strategic takeaways
2024, International Journal of Hospitality Management
The increasing focus on environmental issues has resulted in the widespread acceptance and prominence of green innovation. However, existing literature on hotel management has overlooked exploring the key factors that drive green innovation specifically through the mediating mechanisms of green dynamic capability (GDC) and green environmental orientation (GEO). To fill the gap in the existing hospitality literature this study utilized the natural resource-based view (NRBV) to investigate the correlation between green transformational leadership (GTL) and green innovation (GI). Two research studies were carried out in Italian luxury hotels to assess the efficacy of the theoretical framework. The research findings emphasize that the enhancement of a hotel’s GDC and GEO can be facilitated by providing GTL. This in turn may lead to the enhancement of GI which improves the CA and GP of a hotel.
What drives higher active customer engagement in luxury brands’ social media? Measurement and contingencies
2024, Journal of Retailing and Consumer Services
This study investigates customer engagement behaviors related to luxury brands on social media, focusing on the motives driving active and passive engagement. Applying a sequential mixed methods approach, drawing on netnographic and survey data analyses, the study identifies 10 emergent themes across intrinsic and extrinsic motives that shape overall motivation for social media engagement with luxury brands. Findings from the netnographic analysis provide empirical insights into customer motivations, which inform the development of hypotheses tested through quantitative analysis. The results highlight the roles of intrinsic motivations, such as entertainment, authenticity, gratitude, experience sharing, nostalgia, aesthetics, and lifestyle inspiration. Further, extrinsic motivations such as social support, utilitarian rewards, and social status seeking are revealed to influence customers' attitudes and behaviors toward luxury brands on social media. The findings regarding the necessary condition and importance-performance matrix analyses provide additional nuanced insights into the necessity and importance of intrinsic and extrinsic motivations for active and passive engagement on social media by customers. The study also clarifies the moderating effects of customers’ value for money orientation and brand identification on the relationship between overall motivation and engagement behaviors. The findings provide a granular understanding of luxury customer behavior on social media, offering valuable insights for luxury brand managers aiming to foster stronger customer–brand relationships.
Unveiling the link between digital entrepreneurship education and intention among university students in an emerging economy
2024, Technological Forecasting and Social Change
Digital Entrepreneurship (DE) is becoming increasingly relevant due to advancements in technology and infrastructure, which offer many opportunities for entrepreneurs. This research seeks to elucidate the influence of Digital Entrepreneurship Education (DEE) on the interrelations among Attitude (ATT), Subjective Norms (SNs), Perceived Behavioral Control (PBC), Personal Innovativeness (PI), and Digital Entrepreneurship Intention (DEI) among university students in emerging nations. The Theory of Planned Behavior (TPB) serves as the theoretical foundation for this study. The study adopts a quantitative research design, utilizing a survey questionnaire to garner information from a subset of business college students in Vietnam. A total of 334 valid responses were analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM), facilitating the testing of the research hypotheses. The study revealed a positive correlation between TPB constructs, PI, and DEI among university students in developing economies. However, the relationship between PBC and DEI was inversely affected by DEE. Furthermore, the study found no evidence to support the moderating effects of DEE on the relationship between ATT, SNs, PI, and DEI. This research contributes significantly to entrepreneurship studies by investigating the impact of DEE on students' inclination to embark on DE in developing countries. It extends the theoretical framework of TPB by integrating PI and underscores the importance of DEE in fostering DE. Moreover, the findings of this study hold several implications for academic administrators, policymakers, and educators. These insights can be leveraged to devise strategies and initiatives to enhance support and advancement for DEE.
Unleashing the university potential: Exploring the impact of staffing autonomy on organizational performance through strategy implementation
2024, International Journal of Educational Research Open
Drawing on the resource-based view, the current study examines the direct effect of staffing autonomy (SA) on strategy implementation (SI) and organisational performance (OP) in public and private universities. The study also investigates the direct effect of SI on OP. It further explores whether SI mediates the link between SA and OP. Moreover, the study examines whether the type of universities moderates the relationship among the constructs. The hypotheses were tested based on survey data from 254 leaders of public and private universities in India. PLS-SEM-based results reveal that SA has a significant positive impact on both SI and OP. Further, the results show that SI has a significant positive impact on OP. Besides, the results find that SI mediates the relationship between SA and OP. Furthermore, the multi-group analyses show that the university type has no moderating effect on the effect of SA on SI and OP, as well as on the impact of SI on OP. The results of the current study support the anticipations derived from a resource-based view. The theoretical and practical contributions are discussed, and limitations and new lines of future research are presented.
Investigating pre-service teachers’ artificial intelligence perception from the perspective of planned behavior theory
2024, Computers and Education: Artificial Intelligence
There is a need for teachers who are prepared to teach Artificial Intelligence (AI) across the K-12 learning contexts. Owing to the dearth of teacher education programmes on AI, it is helpful to explore factors to be considered in designing an effective AI programme for future teachers. We posit that understanding how to encourage pre-service teachers to learn AI is thus critical for practitioners and policymakers while designing effective instructional AI teacher education programmes. This exploratory study examined the perceptions of pre-service teachers and their behavioral intention to learn AI, by identifying factors that might affect learning and promoting AI in teacher preparation programmes. This study proposed a research model supported by the theory of planned behavior and expanded with other constructs. The factors that were examined include basic knowledge of AI, subjective norm, AI for social good, perceived self-efficacy, self-transcendent goals, personal relevance, AI anxiety, behavioral intention to learn AI, and actual learning of AI. Using a duly validated questionnaire, we surveyed 796 pre-service teachers in Nigerian Universities. Through structural equation modeling approach analyses, our proposed model explains about 79% of the variance in pre-service teachers' intention to learn AI. Basic knowledge and subjective norm were found to be the most important determinant in pre-service teachers’ intention to learn AI. All our hypotheses were supported except for self-efficacy and personal relevance, personal relevance and social good, and behavioral intention and actual learning behavior. The findings provide practitioners, researchers, and policymakers with valuable information to consider in designing effective AI teacher education programmes.
Connecting the dots between brand equity and brand loyalty for retailers: The mediating roles of brand attitudes and word-of-mouth communication
2024, Journal of Business Research
Managing and developing brand equity is of crucial interest to retailers. To provide a better understanding of retailer brand equity, we introduce a specific nomological network that brings together operational retailer brand equity dimensions and marketing performance outcomes. More specifically, we hypothesize that consumer attitudes toward the retailer and word-of-mouth communication individually and sequentially mediate the relationship between retailer brand equity and consumer loyalty toward the retailer. Partial least squares path modeling (PLS) was used to test the proposed model with 335 US consumers from the home improvement retail sector. A systematic validation procedure with explanatory (i.e., in-sample) and predictive (i.e., out-of-sample) power assessments and robustness tests such as the confirmatory tetrad analysis and model comparison confirms these specific relationships. The results show that the refined model allows retail managers to link brand equity dimensions to marketing performance more accurately and to predict customer loyalty more reliably.

View all citing articles on Scopus

View full text

The elephant in the room: Predictive performance of PLS models

Abstract

Section snippets

Introduction and motivation

Two key features of predictive performance in predictive analytics

Predictive performance of PLS models: A conceptual framework

A procedure for evaluating predictive performance of a PLS model

Enhancing PLS models using theory-informed prediction

Conclusions and future directions

Acknowledgments

International Journal of Forecasting

Journal of Econometrics

International Journal of Forecasting

Computational Statistics and Data Analysis

Australasian Marketing Journal

Journal of Business Research

The significance of simultaneous methods of parameter estimation in econometric models

Journal of the Royal Statistical Society, Series C

The partial least squares (PLS) approach to causal modeling, personal computer adoption and use as an illustration

Technology Studies

The moderator–mediator variable distinction in social psychological research – conceptual, strategic, and statistical considerations

Journal of Personality and Social Psychology

Predictive validity and formative measurement in structural equation modeling: embracing practical relevance, thirty fourth international conference on information systems

The partial least squares approach to structural equation modelling

Bootstrap cross-validation indices for PLS path model assessment

On partial least squares variance-based SEM (VBSEM) vs. covariance-based SEM (CBSEM) for confirmatory analysis: It's all about the components and variance explained

A paradigm for developing better measures for marketing constructs

Journal of Marketing Research

The superiority of simple alternatives to regression for social science predictions

Journal of Educational and Behavioral Statistics

Guidelines for choosing between multi-item and single-item scales in construct measurement: a predictive validity perspective

Journal of the Academy of Marketing Science

Latent variables and indices: Herman Wold’s basic design and Partial Least Squares

Consistent partial least squares path modeling

MIS Quarterly

Theory building: A practical guide to the construction and testing of theoretical models