+ Reply to Thread
Results 1 to 4 of 4

Thread: "two variables are too similar and thus there is no linear interaction found"-Effect?

  1. #1
    Points: 131, Level: 2
    Level completed: 62%, Points required for next Level: 19

    Posts
    3
    Thanks
    0
    Thanked 0 Times in 0 Posts

    "two variables are too similar and thus there is no linear interaction found"-Effect?




    Hi there (german verision below)

    I did a multiple Regression.
    There are two variables where no linear interactions was found. This is a little suprising, because those two veriables seem to be very simliar (its self-efficacy and competence).

    Is there a name for this effect? And are there studies that I can quote on this effect: if two veriables are too similar and thus there is no linear interaction found?

    Thx so much
    MrPeer
    PS (a friend wrote me):
    "I am not surprised that the competence did not predict self-efficacy as both concepts might be measuring almost the same thing and therefore when entered in a regression model there is a high chance that it will not work."

    in german:

    Hallo,

    Ich habe in einer multiplen Regressionsanalyse 3 unabhängige Variablen (Autonomie, Kompetenz und Soziale Eingebundenheit) und 2 Abhnängige Variablen (Gruppe und Selbstwirksamkeit).
    Einen linearen zusammenhang konnte ich statistisch nur bzgl. Autonomie und Selbstwirkamkeit herausstellen.

    Das ist recht Verwunderlich, da Kompetenz udn Selbstwirksamkeit fast das selbe sind (in Studien wurde eine Hohe Korrelation von .50 gefunden).

    Ich suche jetzt für meine Diskussion eine Erklärung die belegt (!), dass wenn sich zwei Kontrukte sich zu sehr ähneln, dass die wahrscheinlichkeit höher ist, dass es nicht in einen linearen Zusammenhang gibt.
    Gibt es dafür einen Namen, dieses Effekts und ggf. Studien die genau das beschreiben?

    Besten Dank
    MrPeer

    2 Konstrukte ähneln sich zu sehr und daher kein Zusammenhang

  2. #2
    Omega Contributor
    Points: 38,289, Level: 100
    Level completed: 0%, Points required for next Level: 0
    hlsmith's Avatar
    Location
    Not Ames, IA
    Posts
    6,992
    Thanks
    397
    Thanked 1,185 Times in 1,146 Posts

    Re: "two variables are too similar and thus there is no linear interaction found"-Eff

    Tell us more about the model. Did you test for multi-collinearity between the terms?
    Stop cowardice, ban guns!

  3. #3
    Points: 131, Level: 2
    Level completed: 62%, Points required for next Level: 19

    Posts
    3
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: "two variables are too similar and thus there is no linear interaction found"-Eff

    uff its hard to translate all those terms into english (i am struggling enough with the terms for itself)

    But I did a muliple Regression.
    The dependent variable: Self-efficacy and the group (intervention and control) (Self efficacy by Bandura)
    The indepentend variable: Autonomy, Kompetence and Relatedness (Basic Needs by Deci and Ryan)

    There was a linear interaction between: Self-efficacy and Autonomy.
    BUT the theoretical research showed that Self-efficacy and Competence is so close it is nearly the same in some models. There was also found a korrelation for .50 in one study.

    So now I am looking for a statistical explaination

  4. #4
    Points: 2,719, Level: 31
    Level completed: 80%, Points required for next Level: 31

    Posts
    4
    Thanks
    0
    Thanked 1 Time in 1 Post

    Re: "two variables are too similar and thus there is no linear interaction found"-Eff


    I don't know who or what gave your friend the idea that it could be possible that 2 constructs are too similar, so that they cannnot have a linear relationship. This is a very awkward idea, in my opinion.

    If you do 3 single regressions instead of 1 multiple regression with 3 predictors, then you will probably find that there exists a linear relation between competence and self-efficacy.

    As hlsmith already told you, you should search for "collinearity" in regression models. I suppose that the other predictors autonomy and/or relatedness overlap with (are correlated with) competence? So, probably the variance in self-efficacy which could be explained by the predictor competence was already explained by the predictors autonomy and/or relatedness. There was't any variance left in self efficacy to be explained by autonomy, when simultaneously controlling for autonomy and relatedness.

+ 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