+ Reply to Thread
Results 1 to 5 of 5

Thread: multinomial logistic regression: independent continuous variable,help to interpretate

  1. #1
    Points: 1,885, Level: 25
    Level completed: 85%, Points required for next Level: 15

    Posts
    56
    Thanks
    25
    Thanked 0 Times in 0 Posts

    multinomial logistic regression: independent continuous variable,help to interpretate




    Hello I was running this statistic but I have some problems to interpretate it

    my dependent variable is categorical and it expresses difference between two tests, one at time zero, a second one at time 1. So it can basically have three values, improved, same level, worsened

    independent variable: blood level of a certain substance (blood test) at time zero. My aim is to estimate this variable as a possible predictor of the general trend in my sample. So it is continuous.

    covariates: sex, age, another blood test

    My first thought was to use a multinomial logistic regression, using as base value the "same level" for the dependent variable

    I found out some significant associations but my question is: since the independent variable is a continuous one how to interpretate it?
    if for intance the improved category of my dependent variable is associated to the independent variable (blood levels) and the coeff is negative, does it mean that the lower is the blood test value, the bigger are the chances to improve the result? I am a bit confused about how to read it!

    Thanks in advance!

  2. #2
    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: multinomial logistic regression: independent continuous variable,help to interpre

    I would use ordinal logistic regression because your dependent variable categories appear ordered to me. That aside the best way to interpret logistic regression is normally with odds ratios not the slopes of the predictor variables (which aside from the sign are largely uninterpretable).
    "Very few theories have been abandoned because they were found to be invalid on the basis of empirical evidence...." Spanos, 1995

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

    Rerun (12-16-2013)

  4. #3
    Points: 1,885, Level: 25
    Level completed: 85%, Points required for next Level: 15

    Posts
    56
    Thanks
    25
    Thanked 0 Times in 0 Posts

    Re: multinomial logistic regression: independent continuous variable,help to interpre

    I excluded the OR at the beginning because my independent variables is represented by a blood level. there are some foundings that say that high values of it avoid test to worsen.
    Can I use the OR to somehow interpretate the result of test in terms of probability to have a certain result considering the increasing or decreasing of the blood values?

    let`s make an example:

    among the three categories I found out that improving is associated to the independent variable, the OR is negative. Does it mean that lower levels of this blood value are more likely to predict an improvement in the test or not? Sorry, this is my main confusing point at this level. Probably because I have always ran this sort of statistic with independent categorical variables, so it was easy to interpretate it because basically you had to think of wheter the variable was positive or negative, now I have some problems to interpretate it because this one is a continuous.

    vmt anyway

  5. #4
    Points: 1,885, Level: 25
    Level completed: 85%, Points required for next Level: 15

    Posts
    56
    Thanks
    25
    Thanked 0 Times in 0 Posts

    Re: multinomial logistic regression: independent continuous variable,help to interpre

    Of course I meant OR less than 1 rather than negative

  6. #5
    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: multinomial logistic regression: independent continuous variable,help to interpre


    If the odds ratio is greater than one than as the IV goes up the odds of being in the higher state (in ordered logistic regression) is higher. If not the reverse is true. Of course it depends on how exactly you have ordered the DV what this means specifically. I am not certain that is what you are asking however.

    I don't work much with nominal logistic regression so I am not clear how you interpret odds ratio in that. You can not interpret odds ratios, except in very special cases, as a probability. For that you need relative risk. Unfortunately there is no simple way to generate relative risk that I know of (it is used primarily in medicine which I don't work much with). I spent a lot of time trying to find ways to generate relative risk - and the answer was always very complex and required manual manipulation to do.
    "Very few theories have been abandoned because they were found to be invalid on the basis of empirical evidence...." Spanos, 1995

+ 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