+ Reply to Thread
Results 1 to 13 of 13

Thread: Continuous variable and its effect on dependent categorical variable, Which test?

  1. #1
    Points: 18, Level: 1
    Level completed: 35%, Points required for next Level: 32

    Posts
    4
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Red face Continuous variable and its effect on dependent categorical variable, Which test?




    Hi, I question as a new SPSS user (I use version 20)
    I want to find a relation/correlation/association between an independent continous parametric variable and dependent categorical variable
    Example:
    Effect of systolic BP readings ( continuous independent parametric variable ) On Stroke event (Dependent categorical variable, either stroke happened =1 or not 0 ). i.e. The question is there any association between higher readings of systolic BP and stroke events, in other words the higher systolic BP reading the more stroke happened.

    Thanks

  2. #2
    TS Contributor
    Points: 18,889, Level: 87
    Level completed: 8%, Points required for next Level: 461
    CowboyBear's Avatar
    Location
    New Zealand
    Posts
    2,062
    Thanks
    121
    Thanked 427 Times in 328 Posts

    Re: Continuous variable and its effect on dependent categorical variable, Which test?

    Independent samples t-test would be a simple conventional choice. You'd be testing whether there is a significant difference in mean systolic BP between the two groups (stroke patient, non -stroke patient).

    If you get to the point of wanting to control for any covariates, you might want to switch to logistic regression.

  3. #3
    Human
    Points: 12,676, Level: 73
    Level completed: 57%, Points required for next Level: 174
    Awards:
    Master Tagger
    GretaGarbo's Avatar
    Posts
    1,362
    Thanks
    455
    Thanked 462 Times in 402 Posts

    Re: Continuous variable and its effect on dependent categorical variable, Which test?

    As I understood OP the interest is in investigating if the blood pressure possibly influences the probability of of an stroke event. Then a logistics regression model = logit regression, seems reasonable, with stroke as the dependent variable and BP as explanatory variable.

    Not that the stroke event would influence the blood pressure, where a t-test would be natural.

    The most interesting question at the moment is: What is your sample size?

  4. #4
    TS Contributor
    Points: 17,779, Level: 84
    Level completed: 86%, Points required for next Level: 71
    Karabiner's Avatar
    Location
    FC Schalke 04, Germany
    Posts
    2,542
    Thanks
    56
    Thanked 640 Times in 602 Posts

    Re: Continuous variable and its effect on dependent categorical variable, Which test?

    The question is there any association between higher readings of systolic BP and stroke events, in other words the higher systolic BP reading the more stroke happened.
    From this I would rather conclude that a t-test (or U-test, if necessary)
    would answer the question. Or you calculate a point-biserial correlation
    coefficient (if the stroke/non-stroke ratio isn't too extreme).

    With kind regards

    K.

  5. #5
    Human
    Points: 12,676, Level: 73
    Level completed: 57%, Points required for next Level: 174
    Awards:
    Master Tagger
    GretaGarbo's Avatar
    Posts
    1,362
    Thanks
    455
    Thanked 462 Times in 402 Posts

    Re: Continuous variable and its effect on dependent categorical variable, Which test?

    Here we can see how different one can read a post.

    Since OP had talked about stroke as the dependent variable I wanted to clarify for the OP that logit seemed reasonable, not because I wanted to be confrontational towards CoyboyBear.

    (I had done a bad reading of Gianmarcos post yesterday, reading in something that wasn't there, so it is very possible that I do the same mistake again. I believed that high blood pressure was a known risk factor to stroke (short googling seems to confirm that) so there is a possibility that I was reading in something.) The title seems to imply a hypothesis of a direction of causality from BP to stroke.

    So the issue is what is the dependent variable, Stroke or blood pressure? And what are we conditioning on: are we looking at the probability of stroke given the patients blood pressure? Or are we looking at how the blood pressure varies given the event of a stoke or not?

  6. #6
    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: Continuous variable and its effect on dependent categorical variable, Which test?

    I would think you could use logistic regression and look at the odds ratio and associated p values.
    "Very few theories have been abandoned because they were found to be invalid on the basis of empirical evidence...." Spanos, 1995

  7. #7
    TS Contributor
    Points: 18,889, Level: 87
    Level completed: 8%, Points required for next Level: 461
    CowboyBear's Avatar
    Location
    New Zealand
    Posts
    2,062
    Thanks
    121
    Thanked 427 Times in 328 Posts

    Re: Continuous variable and its effect on dependent categorical variable, Which test?

    Quote Originally Posted by GretaGarbo View Post
    Since OP had talked about stroke as the dependent variable I wanted to clarify for the OP that logit seemed reasonable, not because I wanted to be confrontational towards CoyboyBear.
    You didn't seem confrontational, no worries

    I guess my thinking was that the t-test looks at whether there's a mean difference, regardless of the direction of causality.

    But now that I think about that more I'm not so convinced. E.g., if BP causes stroke, we might not have reason to think that BP would be normally distributed within the two categories (stroke/no stroke). And anyway the logistic regression would be useful because it allows for the adding of control variables without changing method, so I think we're all agreed on that!

  8. #8
    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: Continuous variable and its effect on dependent categorical variable, Which test?

    Wouldn't logistic regression (but not t test) also show how much impact a variable had (not just if), the direction, goodness of fit and other useful information? I have never used t tests to do this so I was wondering.

    I always thought you had to have interval data for a t test, is that not true?
    "Very few theories have been abandoned because they were found to be invalid on the basis of empirical evidence...." Spanos, 1995

  9. #9
    TS Contributor
    Points: 18,889, Level: 87
    Level completed: 8%, Points required for next Level: 461
    CowboyBear's Avatar
    Location
    New Zealand
    Posts
    2,062
    Thanks
    121
    Thanked 427 Times in 328 Posts

    Re: Continuous variable and its effect on dependent categorical variable, Which test?

    Quote Originally Posted by noetsi View Post
    Wouldn't logistic regression (but not t test) also show how much impact a variable had (not just if), the direction, goodness of fit and other useful information?
    Yeah the logistic regression tells you more. You would see the direction from the t-test. But the magnitude of the effect of the continuous variable would only be shown in a very indirect way. So yes hopefully the OP chooses logistic, sorry for the confusion!

  10. #10
    Points: 18, Level: 1
    Level completed: 35%, Points required for next Level: 32

    Posts
    4
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: Continuous variable and its effect on dependent categorical variable, Which test?

    145 sample size, third had stroke ( group 1 )

  11. #11
    Points: 18, Level: 1
    Level completed: 35%, Points required for next Level: 32

    Posts
    4
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: Continuous variable and its effect on dependent categorical variable, Which test?

    Quote Originally Posted by GretaGarbo View Post
    Here we can see how different one can read a post.

    Since OP had talked about stroke as the dependent variable I wanted to clarify for the OP that logit seemed reasonable, not because I wanted to be confrontational towards CoyboyBear.

    (I had done a bad reading of Gianmarcos post yesterday, reading in something that wasn't there, so it is very possible that I do the same mistake again. I believed that high blood pressure was a known risk factor to stroke (short googling seems to confirm that) so there is a possibility that I was reading in something.) The title seems to imply a hypothesis of a direction of causality from BP to stroke.

    So the issue is what is the dependent variable, Stroke or blood pressure? And what are we conditioning on: are we looking at the probability of stroke given the patients blood pressure? Or are we looking at how the blood pressure varies given the event of a stoke or not?
    Sample size is 143, 1/3 of them had stroke. what i am looking for is a higher systolic bp would that be more significant in those who had the stroke. i.e. is higher bp associated with more strokes ??
    thanx

  12. #12
    Points: 18, Level: 1
    Level completed: 35%, Points required for next Level: 32

    Posts
    4
    Thanks
    0
    Thanked 0 Times in 0 Posts

    Re: Continuous variable and its effect on dependent categorical variable, Which test?

    So,
    Shall i use logistic regression or categorical regression ??
    Thx
    Sorry a bit beginner in SPSS

  13. #13
    TS Contributor
    Points: 18,889, Level: 87
    Level completed: 8%, Points required for next Level: 461
    CowboyBear's Avatar
    Location
    New Zealand
    Posts
    2,062
    Thanks
    121
    Thanked 427 Times in 328 Posts

    Re: Continuous variable and its effect on dependent categorical variable, Which test?


    What is categorical regression? I think our consensus here is to use logistic regression.

+ 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