+ Reply to Thread
Results 1 to 2 of 2

Thread: Estimate or Contrast Statement in ANOVA

  1. #1
    Omega Contributor
    Points: 38,303, Level: 100
    Level completed: 0%, Points required for next Level: 0
    hlsmith's Avatar
    Location
    Not Ames, IA
    Posts
    6,993
    Thanks
    397
    Thanked 1,185 Times in 1,146 Posts

    Estimate or Contrast Statement in ANOVA




    Hey,


    I was asked to review a student's paper. In it they have 2 exposures groups and multiple continuous outcomes they are examining.


    They used a bunch of ANOVAs, with one independent variable with 3 categories formatted as following:
    -Exposure 1 (E1)
    -Exposure 2 (E2)
    -Exposure 1 and Exposure 2 (E1&2)


    The means of the continuous outcomes are about comparable for E1 and E2 and drastically lower for E1&2 group. They state there is an interaction. However, there was not interaction term or testing, just the obvious difference when both are present, which is confirmed with post-hoc t-tests.


    Are there any estimates or contrast statements that can be constructed that shows E1&2 is additive, multiplicative, synergistic, etc? Could each exposure group be entered as its own term (binary). One limitation is that the E1&2 group has an n-value of 6 with other groups having larger n-values.

  2. #2
    Omega Contributor
    Points: 38,303, Level: 100
    Level completed: 0%, Points required for next Level: 0
    hlsmith's Avatar
    Location
    Not Ames, IA
    Posts
    6,993
    Thanks
    397
    Thanked 1,185 Times in 1,146 Posts

    Re: Estimate or Contrast Statement in ANOVA


    Correction, categorical group is:


    -Neither Exposure 1 (E1) or Exposure 2 (E2)
    -E1 or E2
    -Both E1 & E2


    So I should be able to create the four needed groups: none; E1; E2; both and do the basic approach presented here:


    J Mol Cell Cardiol 30, 723–731 (1998), the statistics of synergism


    So it seems fine to just run a two-way model with an interaction term. Does it seem like there may be power issues given the sample size of the four groups: 50; 20; 20; 5, respectively. I think it should be fine - but the generalizability of the interaction may be limited outside the this sample given the small n-value.




+ Reply to Thread

           




Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts






Advertise on Talk Stats