Welcome to PlagueFest.com! Log in or Sign up to interact with the Plague Fest community.
  1. Welcome Guest! to interact with the community and gain access to all the site's features.

Statistics? Anybody want to help with a question?

Discussion in Everything & Anything started by Seraphine, Sep 19, 2011

  1. Jul 23, 2010
    Posts
    ORIGINAL QUESTION.


    I am finding the correlation coefficient and the linear regression line (equation) using the STAT feature on my calculator. I noticed that when MathXL walks you through this problem, you have the option to choose if the line is a good fit, is not a good fit, or is horizontal. According to the book, it states a correlation coefficient value close to 1 or -1 is a good fit and a
    correlation coefficient value close to 0 is a bad fit. What correlation coefficient would make this line horizontal?


    I asked my teacher, but she responded with..verbatim...

    "I don't know...I'll ask around."

    Anybody have an idea? I find it weird considering I've never seen the horizontal option before, granted..I'm not mapping the line either. The problem itself is cake, I just don't understand this horizontal option.
  2. Jul 23, 2010
    Posts
    I hate you. :verymad:
  3. Jan 21, 2011
    Posts
    The answer of course is onions.

    From what I just googled the range of the coefficient is -1 to +1. 0 it would seem is horizontal and values close to 0 such as .1 are bad.
  4. Jul 23, 2010
    Posts
    Can you link me? This seems to make sense. I couldn't figure it out on Google.
  5. Jan 21, 2011
    Posts
    http://en.wikipedia.org/wiki/Correlation_and_dependence

    The picture of the plots shows a horizontal line with no correlation coefficient under it. The picture explanation says that's because the correlation coefficient is undefined being that there is no change in y.
  6. Jan 5, 2007
    Posts
    Nothing new there. :whistling:
  7. Jul 23, 2010
    Posts
    So..In order for the line to be horizontal, the correlation coefficient would have to be an exact zero. This line would be complete statistical independence, otherwise a perfect non correlation. If the correlation was not zero, such as .2 or -.2 this would cause the line to either angle upwards or downwards, depending on if the correlation value was a positive or negative.

    This makes sense. I guess it's just a bit confusing considering you can have a scatterplot with an absolute array of points with no correlation.
  8. Jul 23, 2010
    Posts
    If you relate this to slope, a line with zero slope has a value of 0. Unlike a vertical line which has an undefined slope.