Announcement

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

  • Linear regression test in dtable with more than two variables

    Dtable has the tests option, which delivers statistical tests for differences in the specified varlist by group category (see below for a code example).

    Code:
    dtable varlist, by(group, tests)
    By default, dtable uses linear regression when testing for differences in continuous variables and posts a single p-value for the difference between groups. As the documentation indicates, this is equivalent to pooled t tests when the grouping variable has two categories. But what about when it does not?

    I'm using dtable when comparing across the groups, and it still delivers a single p-value. It's unclear what this is in reference to, because a linear regression for the difference in the var across three categories should produce two coefficients and two p-values. What exactly is dtable doing in this instance? How would it be appropriate to describe in a paper?

  • #2
    I just realized the title should read "more than two groups" rather than "more than two variables" - sorry about that. The question still stands though. Thanks for any help anyone can provide.

    Comment

    Working...
    X