Organization and Management Theory OMT

Back to discussions
Expand all | Collapse all

PDW Announcement: The Case Against Null Hypothesis Significance Testing: Flaws, Alternatives, and Action Plans

  • 1.  PDW Announcement: The Case Against Null Hypothesis Significance Testing: Flaws, Alternatives, and Action Plans

    Posted 08-02-2006 18:27

    Professional Development Workshop (Sponsors: RM, BPS)

    "The Case Against Null Hypothesis Significance Testing: Flaws, Alternatives, and Action Plans"

    Academy of Management Meeting in Atlanta, GA
    Saturday, Aug 12, 2006 - 4:00PM-6:00PM
    Atlanta Marriott Marquis in International B

    Organizers:
    William H. Starbuck; University of Oregon
    Andreas Schwab; Louisiana State University
    Ingo Holzinger; York University

    Presenters:
    William H. Starbuck; University of Oregon
    Raymond Hubbard; Drake University
    Eric Abrahamson; Columbia University
    Lisa Schurer Lambert; Georgia State University
    Jose M. Cortina; George Mason University

    The purpose of this PDW is to increase the awareness among management researchers of the severe limitations of Null Hypothesis Significance Testing (NHST) and to introduce alternative approaches based on effect size measures and confidence intervals. NHST has been criticized by methodologist on various grounds (Craver, 1978; Cohen, 1994; Schmidt, 1996; Thompson, 1996; Cortina & Folger, 1998). Their scepticism suggests that the extensive use of NHST in quantitative management research has led to the accumulation of deceptive findings. Essentially, NHST creates the misleading impression that meaningless findings are important while meaningful findings are often portrayed as "not significant." As a result, management journals are full of "statistically significant" findings that are too small to be practically relevant and/or to be replicated by other studies. At the same time, studies with important findings do not get published because of their lack of "statistically significant" results. In a field that aspires to provide useful advice to managers, we need to focus on practically important effects that are robust across a wide variety of settings. To identify practically meaningful findings, methodologists have recommended the reporting of effect sizes and confidence intervals as an alternative to NHST. Recently, approaches for estimating confidence intervals for effect size measures have also been introduced (Kline, 2004; Cummings & Finch, 2001). The application of NHST alternatives by management researchers, however, has been limited by both a lack of sensitivity regarding the inherent problems of NHST and a limited familiarity with alternative approaches. Consequently, our proposed two-hour workshop addresses these issues in the following three modules: Critique of NHST (2 speakers); alternatives to NHST (1-2 speakers); round-table discussion/panel for hands-on advice.

    Members of all divisions are welcome.  There is a $5 fee for non-members of RM.

    Pre-registration recommended. Please contact Andreas Schwab at aschwa3@lsu.edu.