Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Regression Model for Moderation Effect

    Hi everyone,

    I am new to using Stata and have a problem with setting up a linear regression model with a moderating variable. The hypothesis is as follows:
    "The female gender of journalists increases the relationship between media attention and sales"

    DV=sales
    IV=number of articles published in the WSJ
    moderating variable: female gender


    I am not sure whether the model should include an interaction term like this:
    DV= b0 + b1 * IV + b2 * Moderating variable + b3 * (IV *Moderating variable) + e

    Or if the following command sufficiently describes the Moderation effect:
    reg DV IV ModeratingVariable


    Thank you so much for your help,

    Lena




  • #2
    The phrasing of the hypothesis strongly suggests it is the former that you'd want. A statement for the latter would be 'female journalists have increased sales'. The difference being that gender affects sales via media attention.
    The proper way to implement this regression in Stata is with factor variable notation. See https://www.stata.com/manuals13/u25.pdf , especially sections 25.2 and onwards.

    For you case, roughly:
    Code:
    reg DV c.IV##Moderanting_variable
    edit: include prefix "c." before the IV presuming it is a continuous variable, see section 25,2,9 of the manual linked to
    Last edited by Jorrit Gosens; 13 Dec 2017, 03:54.

    Comment

    Working...
    X