The sign test

The sign test

1. inferential statistics

   • test the probability (p) that a particular set of data could have occurred by chance
   • they establish significance: whether the differences or associations we have observed in our samples are likely to be true in the population as a whole
   • tests account for sampling error but not sample bias which is a problem of the research design

2. can be used to test for difference in paired or related data (repeated measures or matched pairs design

3. 1) state hypothesis

   • 2) for each pair of data, add a sign column (+ or - depending on which pair is greater, leave blank if no difference)
     • 3) find the calculated value of the test statistic, S, which is the frequency of the less frequent sign e.g. if there are less + then S = +
       • 4) find the critical value of S from the table of critical values, where N = total number of scores (ignoring blank) Then choose one or two tailed test & a level of significance
         • 5) compare calculated value with critical value following the instruction below table to see if difference is significant, then decide if result is in expected direction so whether the hypothesis can be accepted
           • 6) conclude
             • example = the calculated value of S (3) is equal to critical value of S (3) where N=13, p <0.05, one tailed , so the results are significant. The null hypothesis can be rejected & the experimental hypothesis can be accepted.