+ Reply to Thread
Results 1 to 5 of 5

Thread: How to report a non-significant regression?

  1. #1
    Points: 82, Level: 1
    Level completed: 64%, Points required for next Level: 18

    Posts
    5
    Thanks
    0
    Thanked 0 Times in 0 Posts

    How to report a non-significant regression?




    Hi guys,

    For my final year research project, I have a few regression models with non-significant results.

    I was just wondering what the standard APA protocol would be for reporting this? For now, I've just provided the F value and the model summary (R square and adjusted R square)in the text. I've then placed a table of the corresponding regression coefficients in the Appendix. Do you think this is enough, or should I explicitly report the beta values in the text?

    I'm just trying to save as much words as possible I suppose, but it's not worth it if I'll end up being marked down for it!

    Thanks

  2. #2
    Points: 3,006, Level: 33
    Level completed: 71%, Points required for next Level: 44

    Posts
    177
    Thanks
    1
    Thanked 29 Times in 29 Posts

    Re: How to report a non-significant regression?

    are you submitting this as a manuscript or is this a class assignment?

    As for reporting non-significant values, you report them in the same way as significant.

    Something akin to-
    Predictor x was found to be significant (B =, SE=, p=). Predictor z was found to not be significant (B =, SE=, p=).

    I caution against using phrases that quantify significance. I.e. almost, nearly, very, strongly. Use qualifiers for effect sizes, not for p values.

  3. #3
    Points: 82, Level: 1
    Level completed: 64%, Points required for next Level: 18

    Posts
    5
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: How to report a non-significant regression?

    Quote Originally Posted by the42up View Post
    are you submitting this as a manuscript or is this a class assignment?

    As for reporting non-significant values, you report them in the same way as significant.

    Something akin to-
    Predictor x was found to be significant (B =, SE=, p=). Predictor z was found to not be significant (B =, SE=, p=).

    I caution against using phrases that quantify significance. I.e. almost, nearly, very, strongly. Use qualifiers for effect sizes, not for p values.
    Thanks for the reply

    It's for my undergraduate dissertation.

    Well my project supervisor mentioned making a table of the Beta values (B, SE, p etc) and putting that in the Appendix (to save word space).
    So I was just wondering whether the beta values need to still be explicitly referred to in the text, or whether I can just say that the model was non-significant (F values and R^2 and adjusted R^2), and then just say that all the predictors were non-significant and refer to the table in the Appendix. But you seem to think I should refer to them in the text?

  4. #4
    Points: 3,006, Level: 33
    Level completed: 71%, Points required for next Level: 44

    Posts
    177
    Thanks
    1
    Thanked 29 Times in 29 Posts

    Re: How to report a non-significant regression?

    It all depends, if the various beta coefficients of your predictors were the focus of your research questions, then they should be discussed.

    Here is a small excerpt from a manuscript I wrote recently-

    "The results for the model containing the transformed variable Total Gifted Funding saw Total White (β=0.002, p<.001) and Total Hispanic (β=0.001, p<.001) students as significant predictors. Other ethnic variables were not significant predictors of Total Gifted Funding. Both Urban (β=-1.475, p=.005) and Rural (β=-1.409, p<.001) locales were found to be statistical significant predictors. In interpretation of both variables, it must be cautioned that locale variables were binary while ethnic variables were continuous. "

    This paragraph was directly after a table detailing all the results.

  5. #5
    Points: 82, Level: 1
    Level completed: 64%, Points required for next Level: 18

    Posts
    5
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: How to report a non-significant regression?


    Quote Originally Posted by the42up View Post
    It all depends, if the various beta coefficients of your predictors were the focus of your research questions, then they should be discussed.

    Here is a small excerpt from a manuscript I wrote recently-

    "The results for the model containing the transformed variable Total Gifted Funding saw Total White (β=0.002, p<.001) and Total Hispanic (β=0.001, p<.001) students as significant predictors. Other ethnic variables were not significant predictors of Total Gifted Funding. Both Urban (β=-1.475, p=.005) and Rural (β=-1.409, p<.001) locales were found to be statistical significant predictors. In interpretation of both variables, it must be cautioned that locale variables were binary while ethnic variables were continuous. "

    This paragraph was directly after a table detailing all the results.
    Thanks for your reply, that was a big help I think I best put the coefficients in then, best to be on the safe side.

+ 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