+ Reply to Thread
Results 1 to 5 of 5

Thread: How to treat missing IV cases in logistic regression?

  1. #1
    Points: 12, Level: 1
    Level completed: 23%, Points required for next Level: 38

    Posts
    2
    Thanks
    1
    Thanked 0 Times in 0 Posts

    How to treat missing IV cases in logistic regression?




    Hey folks,

    quick question: I am analyzing consumers advertising click behavior (click/no-click =DV) and two IVs in this model are the location of the ad that was clicked on and what ad was clicked on. However, whenever no click occured of course there are missing values in these cases. Problem is, the logistic regression can't handle missing values. Therefore the question is: Anybody has an idea how to solve this problem?

    Best regards,
    Oniontoast

  2. #2
    Omega Contributor
    Points: 38,396, Level: 100
    Level completed: 0%, Points required for next Level: 0
    hlsmith's Avatar
    Location
    Not Ames, IA
    Posts
    7,001
    Thanks
    398
    Thanked 1,186 Times in 1,147 Posts

    Re: How to treat missing IV cases in logistic regression?

    How big is your dataset?
    And you don't know the available ads and their locations for non-clickers?


    If so, your conclusions get very limited whatever you opt to do and some questions wont be answerable. Since you don't have missing data per se than completely missing variables for a major category.

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

    oniontoast (09-30-2015)

  4. #3
    Points: 12, Level: 1
    Level completed: 23%, Points required for next Level: 38

    Posts
    2
    Thanks
    1
    Thanked 0 Times in 0 Posts

    Re: How to treat missing IV cases in logistic regression?

    Thanks hlsmith for the quick response!

    Quote Originally Posted by hlsmith View Post
    How big is your dataset?
    Around 35.000 cases. (1.200 clicks to rest no-clicks, so I'm kind of dealing with a logistic regression for rare events.)

    Quote Originally Posted by hlsmith View Post
    And you don't know the available ads and their locations for non-clickers?
    The variables indicate which banner ad and location the first click occured during a visitor session. For non-clickers these variables become redundant, since they did not click on any ad within their visiting session.

    Quote Originally Posted by hlsmith View Post
    If so, your conclusions get very limited whatever you opt to do and some questions wont be answerable. Since you don't have missing data per se than completely missing variables for a major category.
    The idea was to control with these two variables if the location and type of banner ad had an influence on consumers click response. I just realize that might not be able to analyse on a session level.. I'm kinda lost and desperate here.. any ideas how that would still be possible?

    edit: Well I could give the missing values an additional variable that could stand for "didn't matter what banner/location the consumer didn't click either way". Surely that would do the magic to run the logistic regression.. but would it make sense though?
    Last edited by oniontoast; 09-30-2015 at 12:11 PM. Reason: see edit

  5. #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: How to treat missing IV cases in logistic regression?

    Logistic regression can handle missing values many different ways. The most common, but not ideal, way is to simply delete the values. Multiple imputations using FCS (chained equations) is a much better way. If they are not showing up because they shouldn't show up like skip behavior, it might be MAR which makes life a lot easier. But even if they are missing MI handles that.
    "Very few theories have been abandoned because they were found to be invalid on the basis of empirical evidence...." Spanos, 1995

  6. #5
    Omega Contributor
    Points: 38,396, Level: 100
    Level completed: 0%, Points required for next Level: 0
    hlsmith's Avatar
    Location
    Not Ames, IA
    Posts
    7,001
    Thanks
    398
    Thanked 1,186 Times in 1,147 Posts

    Re: How to treat missing IV cases in logistic regression?


    But they don't have missing data - that is a miss interpretation. It sounds like these data just don't exist.


    I think you just need to write out your question and that will solve your issue. Also, if you have multiple clicks for a person, than you need to use a multi-level logistic approach that controls for person's correlated data. Or only use the first page and not subsequent page clicks. Meaning each person would only contribute one observation.


    Another question may be, of those that clicked (as a count or Poisson distribution), what predicted it. Still may need to be multi-level - unsure.

+ 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