Numerical Stat Tests

Numerical Stat Tests

1. 1 group

   • 1 sample from a frame and 1 variable
     • Parametric
       • 1 sample t-test
         • Compares means. H₀: μ = μ₁
           • example:
     • Non-Parametric
       • Wilcoxon signed ranks test
         • ranks values/differences ignoring signs. H₀; Median difference =0
           • Example:

2. 2 Groups

   • Paired
     • Parametric
       • Paired t-test
         • Compare the mean of the diferences between the two measurements. If =0 then two groups the same. H₀; μ of diffs = 0
           • Example: Recruits with tooth decay at onset and 1 year later. Or Case Control study
     • Non-parametric
       • Wilcoxon signed ranks test
         • Rank the differences, ignoring signs. H₀; differences between -ve and +ve =0
   • Independent
     • Parametric
       • unpaired t-test
         • compare means of the two samples. H₀: means in both groups equal
     • Non-Parametric
       • Wilcoxon rank sum test. Mann Whitney U test
         • Sum the ranks in the two groups and compare

3. >2 Groups

   • Parametric
     • ANOVA
       • between group variance (a) should equal within group variance (b) under the Null. ie HOMOSCDASTIC
         • Therefore H₀: a = b. H₁: a > b
           • Based on F -Ratio
   • Non-parametric
     • Kruskal-Wallis test
       • Same as Wilcoxon rank sum but take into account sample sizes