+ Reply to Thread
Results 1 to 4 of 4

Thread: Question about sampledistribution

  1. #1
    Points: 237, Level: 4
    Level completed: 74%, Points required for next Level: 13

    Posts
    5
    Thanks
    1
    Thanked 0 Times in 0 Posts

    Question about sampledistribution




    Hi, I have a question I was hoping someone could answer me.

    It goes as follows:
    The correlation r between two variables x_1 and x_2 equals 0.33. For testing the significance of the correlation, H_0 : p = 0, the sampledistribution of the correlationcoefficient has to be determined. Wat goes for the sampledistribution of the correlation whilst testing the 0-hypothesis? The sample distribution whilst testing this hypothesis is:
    a, skewed to the left
    b, skewed to the right
    c, symmetric around 0.33
    d, symmetric around 0

    (I hope I translated everything in a way that makes sense..)

    I don't really understand the question so I'm hoping someone could answer it and explain why that is the answer.

  2. #2
    TS Contributor
    Points: 6,786, Level: 54
    Level completed: 18%, Points required for next Level: 164

    Location
    Sweden
    Posts
    524
    Thanks
    44
    Thanked 112 Times in 100 Posts

    Re: Question about sampledistribution

    I'd say that the distribution of the sample correlation coefficient is independent of the null hypothesis. Can you specify exactly what confuses you?

    Here's a couple of simulations that might help you.


    jpg images

    click image upload
    Code: 
    samplingdist <- function(M,n,r) {
      corr <- numeric(M);
      for (i in 1:M) {
        x <- mvrnorm(n, rep(1, 2), matrix(c(1,r,r,1),2,2))
        corr[i] <- cor(x[,1],x[,2])
      }
      hist(corr,freq=F,breaks=80); list(median(corr),mean(corr))
    }
    samplingdist(20000,100,0.33)
    Code: 
    samplingdist <- function(M,k,n) {
      corr <- numeric(M); meandiff <- numeric(k); r <- numeric(k)
      for (j in 0:99) {
        for (i in 1:M) {
          x <- mvrnorm(n, rep(1, 2), matrix(c(1,j/100,j/100,1),2,2))
          corr[i] <- cor(x[,1],x[,2])
        }
        r[j] <- j/100
        meandiff[j] <- median(corr)-mean(corr)
      }
    plot(r,meandiff,ylim=c(-0.0035,0.0035)); abline(mean(meandiff),0)
    }
    samplingdist(1000,100,500)

  3. #3
    Devorador de queso
    Points: 95,540, Level: 100
    Level completed: 0%, Points required for next Level: 0
    Awards:
    Posting AwardCommunity AwardDiscussion EnderFrequent Poster
    Dason's Avatar
    Location
    Tampa, FL
    Posts
    12,930
    Thanks
    307
    Thanked 2,629 Times in 2,245 Posts

    Re: Question about sampledistribution

    I'm thinking the question is asking about the sampling distribution under the null hypothesis (so assuming the null is true what would we expect to see).
    I don't have emotions and sometimes that makes me very sad.

  4. #4
    TS Contributor
    Points: 6,786, Level: 54
    Level completed: 18%, Points required for next Level: 164

    Location
    Sweden
    Posts
    524
    Thanks
    44
    Thanked 112 Times in 100 Posts

    Re: Question about sampledistribution


    Quote Originally Posted by Dason View Post
    I'm thinking the question is asking about the sampling distribution under the null hypothesis (so assuming the null is true what would we expect to see).
    Ah, of course.

    The distribution of the sample correlation coefficient under the null is always symmetrical around zero when the errors are normally distributed. When the errors are not normally distributed the sampling distribution is normal and symmetric in large samples. The sampling distribution in small samples when the errors aren't normal depends on the distribution.

    Have a look on a proof of why the beta coefficient is normal and you'll find your answer of why this is the case.

+ 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