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

Normal Distribution v. Student's T Distribution

  1. May 30, 2007 #1
    The "Empirical Rule" states that if your data is normally distributed, 95.45% of that data should fall within "2" standard deviations of your Mean. There doesn't appear to be any reference to sample size in the literature regarding the Empirical Rule and a Normal Distribution.

    By contrast, however, the Student's T Distribution table, for a two-tailed test, has multipliers that differ from the Empirical Rule. Although where N=10000, at 9999 degrees of freedom, the .0455 level is "2" sd like the Empirical Rule, where N=20, at 19 degrees of freedom, the .0455 level is "2.14" sd.

    In sum, then, I don't understand the difference between the "normal distribution" and the "Student's T-Distribution". Is the difference that the Empirical Rule assumes that your data is both normal and "stationary" whereas the Student's T Distribution (i.e., degrees of freedom) assumes that your data is not stationary and that your Mean and Standard Deviations for any period of N will shift with the addition of new data? It's the only thing I can think of since the formulas for confidence intervals for Means and prediction intervals for individual outcomes use the numbers from the Student's T-Distribution.

    Thanks in advance.

  2. jcsd
  3. Jun 1, 2007 #2


    User Avatar
    Science Advisor
    Homework Helper

  4. Mar 28, 2011 #3
    I believe that the Student distribution does not assume that the sample mean is the true (underlying) mean. So it is not just the variance or SD that is taken from the data, and I would say that the fact that the sample mean is used is more important than that the sample standard deviation is estimated from the data.
Share this great discussion with others via Reddit, Google+, Twitter, or Facebook