+ Reply to Thread
Results 1 to 3 of 3

Thread: Is multinomial logistical regression appropriate here?

  1. #1
    Points: 5,395, Level: 47
    Level completed: 23%, Points required for next Level: 155

    Location
    Seattle
    Posts
    99
    Thanks
    48
    Thanked 2 Times in 2 Posts

    Is multinomial logistical regression appropriate here?




    I'm wondering about the appropriate statistical test in the following type of situation:

    Let's say you think that mouse coat color (black, white or orange) might be influenced by the size of the mouse (and imagine that size is binary, i.e. either big or small). You set up 6 crosses between a big mouse and a small mouse and 10 crosses between two big mice, and you count how many black, white and orange pups are generated in each cross.

    So now you have essentially 6 repetitions of 1 cross and 10 repetitions of the other cross and you want to know whether the ratio of the three different coat colors is different in the two types of crosses. What type of statistical test is appropriate here? I was thinking that it might be multinomial logistical regression, "multinomial" because there are three possible outcomes and "logistical" because the outcomes are categorical, but I'm grasping for straws here. Any help would be appreciated, including recommendations of R packages that could be useful.

    Thanks.

    Eric
    Last edited by efoss; 08-09-2016 at 11:23 AM.

  2. #2
    Points: 4,664, Level: 43
    Level completed: 57%, Points required for next Level: 86
    kiton's Avatar
    Location
    Corn field
    Posts
    234
    Thanks
    47
    Thanked 51 Times in 46 Posts

    Re: Is multinomial logistical regression appropriate here?

    Eric, if your dependent variable (DV) has three categories, then indeed multinominal logistic regression would be appropriate (Note, your results would be interpreted in relationship to the control group that you choose). However, if your DV is a ratio that would be a different story. Assuming, the values fall between 0 and 1, you should consider beta regression or glm fractional logit (depending on the inclusiveness of 0 an 1).

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

    efoss (08-08-2016)

  4. #3
    Points: 5,395, Level: 47
    Level completed: 23%, Points required for next Level: 155

    Location
    Seattle
    Posts
    99
    Thanks
    48
    Thanked 2 Times in 2 Posts

    Re: Is multinomial logistical regression appropriate here?


    Hi Kiton,

    I think that my ratios in this case would be inappropriate, in part because that would weight all litters of pups equally, regardless of litter size. I'll look further into multinomial logistic regression. Thanks very much for the help. I really appreciate it.

    Eric

+ 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