Discussion: View Thread

Endogeneity: An inconvenient truth

  • 1.  Endogeneity: An inconvenient truth

    Posted 10-10-2011 15:04
    Hi:

    As a service to the community, my university has sponsored a podcast in
    which I talk about the problems of endogeneity.

    These podcasts, which are freely available on Youtube, are primarily
    methodological; however, they have very important implications for the
    development of theoretical models in OB, leadership, and related areas.
    I show with simple and salient examples how endogeneity is engendered by
    omitted varianbles, common-method variance, and the like. I discuss how
    to deal with endogeneity by using "instrumental variables," and also
    show that endogeneity can also be present when testing mediation (which
    means that experiments can also be threatened by endogeneity too).

    There are three versions of the podcast:

    1. Endogeneity: An inconvenient truth (full version)
    (about 32 minutes in length)

    http://www.youtube.com/watch?v=dLuTjoYmfXs

    Description:
    A key assumption of regression analysis (or structural equation
    modeling) is that the modeled independent variables are not endogenous.
    Yet, the problems of endogeneity are not well known to researchers
    working in many social sciences disciplines (e.g., management, applied
    psychology, sociology, etc.). When the independent variable has not been
    exogenously manipulated, there is a strong possibility that its
    relationship to a dependent variable will not be correctly estimated,
    leading to spurious findings. This podcast gives a brief and vivid
    overview to endogeneity and why it is engendered. Prof. John Antonakis
    discusses the problems of endogeneity using non-technical language and
    intuitive explanations; he shows that when the independent variable is
    endogenous--which is also possible in experimental designs (when the
    mediator is endogenous)--the observed relationship that is estimated can
    be very misleading. Prof. Antonakis demonstrates how the problem of
    endogeneity can be solved using procedures borrowed from econometrics
    (i.e., two-stage least square regression estimator).

    2. Endogeneity: An inconvenient truth (a gentle introduction)
    (this is really a gentle introduction on omitted variable bias; does not
    discuss two-stage least squares and mediation; about 19 minutes in length)

    http://www.youtube.com/watch?v=CCiIfjm8qjw

    Description:
    A key assumption of regression analysis (or structural equation
    modeling) is that the modeled independent variables are not endogenous.
    Yet, the problems of endogeneity are not well known to researchers
    working in many social sciences disciplines (e.g., management, applied
    psychology, sociology, etc.). When the independent variable has not been
    exogenously manipulated, there is a strong possibility that its
    relationship to a dependent variable will not be correctly estimated,
    leading to spurious findings. This podcast gives a brief and vivid
    overview to endogeneity and why it is engendered. Prof. John Antonakis
    discusses the problems of endogeneity using non-technical language and
    intuitive explanations; he shows that the observed relationship that is
    estimated can be very misleading when the independent variable is
    endogenous.

    3. Endogeneity: An inconvenient truth (for researchers)
    (Excludes the "gentle introduction" content and discusses the two-stage
    least squares estimator straight away; about 16 minutes in length)

    http://www.youtube.com/watch?v=yi_5M7oUceE

    Description:
    It is well known that endogeneity leads to inconsistent estimates.
    Unfortunately, many researchers working outside of economics are not
    aware of the problem of endogeneity and how to deal with it. Prof. John
    Antonakis shows how the two-stage least squares (2SLS) estimator
    recovers causal estimates in the presence of endogeneity (which includes
    the problem of common-method variance). He also shows that endogeneity
    can even be prevalent in experimental designs, when researchers estimate
    mediation models; that is, where the causal effect of an exogenous
    variable on a dependent variable is mediated by an endogenous variable
    (or a manipulation check).

    The podcasts are an accompaniment to the following papers (which are
    written in very general terms and useful for many disciplines):

    Antonakis, J., Bendahan, S., Jacquart, P., & Lalive, R. (2010). On
    making causal claims: A review and recommendations. The Leadership
    Quarterly, 21(6). 1086-1120.

    Antonakis, J., Bendahan, S., Jacquart, P., & Lalive, R. (submitted).
    Causality and endogeneity in leadership research: Problems and
    solutions. In D.V. Day (Ed.), The Oxford Handbook of Leadership and
    Organizations.

    If interested, please contact me for copies.

    The I trust that you find the podcasts useful, particularly for teaching
    purposes.

    All the best,
    John.

    --
    __________________________________________

    Prof. John Antonakis
    Faculty of Business and Economics
    Department of Organizational Behavior
    University of Lausanne
    Internef #618
    CH-1015 Lausanne-Dorigny
    Switzerland
    Tel ++41 (0)21 692-3438
    Fax ++41 (0)21 692-3305
    http://www.hec.unil.ch/people/jantonakis

    Associate Editor
    The Leadership Quarterly