Use of partial least squares as a theory testing tool-an analysis of information systems papers

Mikko Rönkkö, Kaisa Parkkila, Jukka Ylitalo

    Research output: Chapter in Book/Report/Conference proceedingConference contributionScientificpeer-review

    4 Citations (Scopus)


    Motivated by recent critique toward partial least squares path modeling (PLS), we present a research question if the PLS method, as used currently, is at all an appropriate tool for theory testing. We briefly summarize some of the recent critique of the use of PLS in IS as a theory testing tool. Then we analyze the results of 12 PLS analyzes published in leading IS journals testing if these models would have been rejected in the case that the data used for model testing had very little correspondence with the theorized models. Our Monte Carlo simulation shows that PLS will often provide results that support the tested hypotheses even if the model was not appropriate for the data. We conclude that the current practices of PLS studies have likely resulted in publishing research where the results are likely false and suggest that more attention should be paid on the assumptions of the PLS model or that alternative approached like summed scales and regression or structural equation modeling with estimators that have known statistical properties should be used instead.

    Original languageEnglish
    Title of host publicationECIS 2012 - Proceedings of the 20th European Conference on Information Systems
    PublisherAssociation for Information Systems
    ISBN (Print)9788488971548
    Publication statusPublished - 2012
    MoE publication typeA4 Article in a conference publication
    EventEuropean Conference on Information Systems - Barcelona, Spain
    Duration: 10 Jun 201213 Jun 2012
    Conference number: 20


    ConferenceEuropean Conference on Information Systems
    Abbreviated titleECIS


    • Monte Carlo simulation
    • Partial least squares
    • Theory testing

    Fingerprint Dive into the research topics of 'Use of partial least squares as a theory testing tool-an analysis of information systems papers'. Together they form a unique fingerprint.

    Cite this