+ Reply to Thread
Results 1 to 4 of 4

Thread: Meta-analysis - can I combine/use these studies?

  1. #1
    Points: 41, Level: 1
    Level completed: 82%, Points required for next Level: 9

    Location
    The Netherlands
    Posts
    5
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Meta-analysis - can I combine/use these studies?




    Hi!

    I'm trying to set up a meta-analysis with my thesis partner and we ran into trouble which our supervisor isn't able to solve. So I figured I'd turn to the internet :) Googling hasn't helped me much, so now I'm on a forum.

    In our research, we are looking for the effect of responsibility on interpersonal behavior in people with high compared to low power.
    We have found 50 studies, roughly dividable in two types. The first type is the studies that manipulate power (high/low) and measure some type of interpersonal behavior (for example, allocation of work load to subordinates). This gives us F, Chi-square or t-statistics, sometimes even an effect size, just for the effect of power. Which is good and which we can code. We code these studies on the type of interpersonal effect found and on whether or not they have (unknowingly) primed a sort of responsibility in participants.

    The second type of studies we have manipulate power (high/low) and some other factor related to responsibility (accountability, moral identity.. stuff like that). Then in the results section we find main effects of both these variables (sometimes) and (most times) interaction effects on behavior too. We are wondering if it's possible to use both these studies in a meta-analysis, and if so, how these should be coded? Do we use the interaction-effect sizes or just the main effects, possibly both? Maybe something else?

    Any ideas and even questions on the matter are greatly appreciated!
    Thanks!

  2. #2
    TS Contributor
    Points: 12,501, Level: 73
    Level completed: 13%, Points required for next Level: 349

    Posts
    951
    Thanks
    0
    Thanked 103 Times in 100 Posts

    Re: Meta-analysis - can I combine/use these studies?

    Statistical questions aside, do you really want to combine those two studies? eg Would it be too unacceptable in your field to combine the studies?

    Back to statistics: you can start by combining the studies, then looking at the heterogeneity tests that you get from meta analysis.
    All things are known because we want to believe in them.

  3. #3
    Points: 41, Level: 1
    Level completed: 82%, Points required for next Level: 9

    Location
    The Netherlands
    Posts
    5
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: Meta-analysis - can I combine/use these studies?

    Thanks for taking the time to reply
    About the statistics, with the risk of sounding quite dense: how do I combine these studies? What effects do I use in my meta-analysis, when one type of studies only has main effects and another type of studies has main and/or interaction effects?

    And I'm not sure about your first question.
    I know my supervisor wants me to use them all in the meta-analysis, if possible. She just doesn't know if this is at all possible, and neither do my thesis partner and I.
    Perhaps it is unacceptable, but luckily it's 'just' a master thesis, so it's mostly practice.

  4. #4
    TS Contributor
    Points: 12,501, Level: 73
    Level completed: 13%, Points required for next Level: 349

    Posts
    951
    Thanks
    0
    Thanked 103 Times in 100 Posts

    Re: Meta-analysis - can I combine/use these studies?


    Quote Originally Posted by Jiske View Post
    About the statistics, with the risk of sounding quite dense: how do I combine these studies? What effects do I use in my meta-analysis, when one type of studies only has main effects and another type of studies has main and/or interaction effects?
    Use the main effects, and ignore the interaction effects. Disclaimer(?): In the times I've done meta analysis, I've only used one effect.

    Usually a meta analysis has data that looks like this:
    http://effectsizefaq.com/2010/05/30/...ust-2-minutes/

    My first question was getting at why you would not want to combine the studies. Optimally, all studies run similar analyses (eg same dependent variables, they all include interaction effects, etc.). If the studies are too dissimilar then you may be discouraged from using them all, even if the statistics say their results do not break homogeneity.
    All things are known because we want to believe in them.

+ Reply to Thread

           




Tags for this 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