+ Reply to Thread
Results 1 to 5 of 5

Thread: What if you are not interested in a variable that has been found to matter in the lit

  1. #1
    Fortran must die
    Points: 58,790, Level: 100
    Level completed: 0%, Points required for next Level: 0
    noetsi's Avatar
    Posts
    6,532
    Thanks
    692
    Thanked 915 Times in 874 Posts

    What if you are not interested in a variable that has been found to matter in the lit




    The article I am working for stresses the importance of a variable that my organization has no interest in at all. It is an important control, should you always include such variables even if you are not interested in it?

    When is it legitimate not to add a variable that the literature suggests is an important contributer to the result? I am sure in theory most DV are influenced by many variables, that is they are statistically significant. Do you (should you) include all signficant variables in your model?
    "Very few theories have been abandoned because they were found to be invalid on the basis of empirical evidence...." Spanos, 1995

  2. #2
    Points: 4,664, Level: 43
    Level completed: 57%, Points required for next Level: 86
    kiton's Avatar
    Location
    Corn field
    Posts
    234
    Thanks
    47
    Thanked 51 Times in 46 Posts

    Re: What if you are not interested in a variable that has been found to matter in the

    Quote Originally Posted by noetsi View Post
    The article I am working for stresses the importance of a variable that my organization has no interest in at all. It is an important control, should you always include such variables even if you are not interested in it?
    In my experience working with the reviewers (MIS field), important controls, especially if they are (a) emphasized in the prior literature (including theoretical aspects), and (b) available for the researcher, have to be included in the model even if you are not interested in them. Otherwise, it "raises a flag" for them. For instance, what if your variable of interest looses significance or changes sign, in case that control is included?

    Quote Originally Posted by noetsi View Post
    When is it legitimate not to add a variable that the literature suggests is an important contributer to the result? I am sure in theory most DV are influenced by many variables, that is they are statistically significant. Do you (should you) include all signficant variables in your model?
    I'd assume that if the estimates for the predictors of interest are robust to inclusion/exclusion of the controls, then you can omit the ones you are not interested in (yet, mentioning that you checked the robustness). Additionally, I'd examine the correlations between the predicted values for full and reduced models.

  3. The Following User Says Thank You to kiton For This Useful Post:

    noetsi (07-31-2015)

  4. #3
    Omega Contributor
    Points: 38,253, Level: 100
    Level completed: 0%, Points required for next Level: 0
    hlsmith's Avatar
    Location
    Not Ames, IA
    Posts
    6,989
    Thanks
    397
    Thanked 1,185 Times in 1,146 Posts

    Re: What if you are not interested in a variable that has been found to matter in the

    Typically you want to always include significant variables, otherwise your estimates may be biased. If your colleagues are obtuse about it, don't report its value and confuse them. Just say results given controlling for standard variables or something broad. You don't want to produce less than clean results just because they don't know what they really should want in the model. It will just be your little secret unless you all make your results available to public then you would want to mention it some place.


    Kiton, good point about robustness!
    Stop cowardice, ban guns!

  5. The Following User Says Thank You to hlsmith For This Useful Post:

    noetsi (07-31-2015)

  6. #4
    Fortran must die
    Points: 58,790, Level: 100
    Level completed: 0%, Points required for next Level: 0
    noetsi's Avatar
    Posts
    6,532
    Thanks
    692
    Thanked 915 Times in 874 Posts

    Re: What if you are not interested in a variable that has been found to matter in the

    I don't have collegues this is for a group of senior leaders. Its not going to or for an academic publication its going to take action on in a business environment. But if I understand the comments above if it is a legitimate variable then you should include it.

    Of course the problem with that, to me, is that you might well have 50 variables and running all would be time consuming and difficult to interpret.
    "Very few theories have been abandoned because they were found to be invalid on the basis of empirical evidence...." Spanos, 1995

  7. #5
    TS Contributor
    Points: 12,227, Level: 72
    Level completed: 45%, Points required for next Level: 223
    rogojel's Avatar
    Location
    I work in Europe, live in Hungary
    Posts
    1,470
    Thanks
    160
    Thanked 332 Times in 312 Posts

    Re: What if you are not interested in a variable that has been found to matter in the


    hi,
    if you have enough data points you could try to do cross -validation and pick the model with the best predictive properties, If the "unwanted" variable increases the predictive performance you habve the arguments why you need it, if it doesn't you can exclude it without further need for justification.

    regards

+ 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