Dismiss Notice
Join Physics Forums Today!
The friendliest, high quality science and math community on the planet! Everyone who loves science is here!

Statistical Tests

  1. Dec 11, 2007 #1
    Hi everyone,

    I can't remember which statistical tests are needed for the following two tests. I thought maybe the null hypothesis test but wanted to check on here first... Maybe something Bayesian?

    Consider you have a plot of y against x with several data points on it, each with y error bars. I would like to calculate the probability that:

    a) These points are randomly distributed

    b) These points have the same y-value = constant, i.e. y is not a function of x.

    Thanks for any advise.

  2. jcsd
  3. Dec 11, 2007 #2


    User Avatar
    Science Advisor
    Homework Helper

    Describe "error bar."

    For part b, my first approach would be to estimate the linear relationship y = a + b*x using least squares and show that b is not statistically significantly different from zero. But "y is not a function of x" is a very general statement; you may have to iterate with polynomial functions (of x), until you run out of data points.
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook