+ Reply to Thread
Results 1 to 4 of 4

Thread: New to statistics and Binary Logistics Regression

  1. #1
    Points: 20, Level: 1
    Level completed: 39%, Points required for next Level: 30

    Posts
    2
    Thanks
    1
    Thanked 0 Times in 0 Posts

    New to statistics and Binary Logistics Regression




    Dear Fellows!

    In search of help just found this forum and I hope there would be anyone who will help me.

    I am doing the Binary Logistic regression analysis of an assignment.Having no background of statistics ,I am a bit confused with the result.

    I used SPSS for regression analysis.The constant's value of Exp(B) is 9148.758 and it appears to me very high.

    Would any body can help me?

    Further,If anybody can tell what are the assumption of a chi-square test?

    Waiting for response

  2. #2
    Points: 1,118, Level: 18
    Level completed: 18%, Points required for next Level: 82

    Posts
    44
    Thanks
    1
    Thanked 7 Times in 7 Posts

    Re: New to statistics and Binary Logistics Regression

    I apologise for an earlier post in error. I realised I had expB rather than B in the prediction formula.
    You dont normally look at the expB of the constant in isolation of the other variables so without knowing the rest of your model it is not clear.
    However a large ExpB simply means you have a large B and pushes predictions towards 1. In this case constant b0=log(9148.758) = 9.12

    You will know that to predict your outcome

    outcome = 1/(1 + (exp(-(b0 + b1 * x1 +b2 * x2 ....))))

    Lets ignore all the other variables and just consider the constant

    For large b0, exp(-b0) -> 0 so outcome -> 1
    For b0=0 exp(b0)=1 so outcome =0.5
    For large negatve b0, exp(-b0) -> infinity so outcome -> 0

    In your case without any other variables exp(-9.12) = 0.00019 so 1/1+exp(-b0) is almost 1
    Therefore your large expB for the constant means that in the absence of your other variables (which we dont know) your predictions are pushed towards 1.

    Regarding Chi-squared all you need are two variables with two or more groups each and counts or frequencies for each combination of groups
    Last edited by statsanon; 12-12-2015 at 05:08 AM.

  3. #3
    Points: 20, Level: 1
    Level completed: 39%, Points required for next Level: 30

    Posts
    2
    Thanks
    1
    Thanked 0 Times in 0 Posts

    Re: New to statistics and Binary Logistics Regression

    Mate thanks for a reply.

    But I am sorry ...being new to statistics.....I could not get .....

    But I am really grateful for your time and attention....

    Kind Regards

  4. #4
    Points: 1,118, Level: 18
    Level completed: 18%, Points required for next Level: 82

    Posts
    44
    Thanks
    1
    Thanked 7 Times in 7 Posts

    Re: New to statistics and Binary Logistics Regression


    Quote Originally Posted by Usman Abdullah View Post
    Mate thanks for a reply.

    But I am sorry ...being new to statistics.....I could not get .....


    Kind Regards
    Hi

    Sorry you didnt get it but if you are new to statistics it is hard to explain a concept like logistic regression - that is a more intermediate to advanced modeling method.

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

    Usman Abdullah (12-13-2015)

+ 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