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

Chi-Squared Test

  1. Sep 19, 2004 #1
    What should I do when the expected value for a chi squared test is zero, so when I try to calculate the test statistic, i get a divide by zero?
     
  2. jcsd
  3. Sep 21, 2004 #2

    Moonbear

    User Avatar
    Staff Emeritus
    Science Advisor
    Gold Member

    The chi-squared test isn't very accurate when expected frequencies are less than 5 (and definitely not if they are 0). In some cases, you can instead use a Fisher's Exact Probability Test (if you have a 2x2 contingency table and small expected frequencies...I think less than 10). I don't have the formula for that handy though. If that doesn't work for your data, I'm not sure what other alternatives there are.
     
Know someone interested in this topic? Share this thread via Reddit, Google+, Twitter, or Facebook

Have something to add?